Kernel panic during boot in VirtualBox with kernel 2.6.26.*-generic

[Jon Packard]

Kernel 2.6.26.* / 2.6.27.* (intrepid) panics during boot in Virtualbox.

The problem occours with VT-x/AMD-V disabled. Enabling the VT-x/AMD-V (on a host that supports this) and the problem doesn't show up.

Expected result: System boots correctly.

Actual result: System fails to boot with kernel panic errors.

Errors on screen (see attached for screen shot):
Code: Bad EIP value.
Kernel panic - not syncing: Attempted to kill the idle task!

[Michael Lok (munhoe-lok)]

I'm having the same problem as well. Hardy as host (Kubuntu, 2.6.24-19-generic) and Intrepid Alpha1 on VirtualBox 1.62. In addition to getting the "Attempted to kill the idle task!" message, I'm also getting: "Kernel panic - not syncing: Fatal exception in interrupt". Kernel panic occurs when booting with 2.6.26-3. 2.6.26-2 works once in a while.

[hatchetman82 (hatchetman82)]

same here.
attempting to install kubuntu 8.10 alpha 1 from *.iso image in virtualbox 1.6.2. host is windows xp sp2

[Cleber Santz (clebersantz)]

Same here.

I posted the screenshot`s here:
https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/146684/comments/10
https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/146684/comments/11
https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/146684/comments/12
https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/146684/comments/13

[hatchetman82 (hatchetman82)]

here's what i think is the virtualbox bug : http://www.virtualbox.org/ticket/918
at least it matches the error i got from virtualBox

[Kjell Braden (afflux)]

This issue is still showing up on todays (20080709 ubuntu-alternate-i386 images (and maybe others).

[Trip McKay (trip-mckay-deactivatedaccount)]

Hi,

I had this issue, too. But after trying some options in the virtual machine do the following to get a much higher successful boot rate (at least it works for me):

Go into the Settings of the VM, choose the Advanced Tab in General and check "Enable VT-X/AMD-V". I am virtualizing with Windows Vista Home Premium SP1 as Host and an AMD Athlon 64 X2 Dual Core Processor 4800+ 2.50 GHz Brisbane. After unchecking the option I will get the "bad EIP" error codes again.

Hope this helps nailing down the problem.

[Trip McKay (trip-mckay-deactivatedaccount)]

Update: I tryed to boot up the machine today as of this bug report after disabling and reenabling the VT-X/AMD-V option. Now the VM does not find the Hard disk and is stuck at

Begin: waiting for root file system...

and drops to a BusyBox shell. It says first "ALERT! /dev/disk/by-uuid/<placeholder> does not exist. Dropping to a shell!". Fill <placeholder> with the uuid of the hd. Updates are of as yesterday, and it did work yesterday before applying updates just fine.

[MikeBee (michaelbeeson)]

This also happens with Intrepid Ibex Alpha 2 on mac host virtualbox 1.6.2. Tried messing about with the options, but to no avail...

[Jacopo Moronato (jmoronat)]

Similar problem here: VirtualBox on Windows Vista Home Premium host.

With kernel 2.6.26-2 i have "BUG: Unable to handle kernel <1>"
With kernel 2.6.26-3 i have "kernel panic - not syncing: fatal exception in interrupt"

[ANDREA (andrea54)]

i've experienced the very same problem. in fact virtualbox often closes with the virtual machine "booting" and leaves log files in my home folder. the most recent one contains this information (if it's of any help) :

Log created: 2008-07-14T15:46:37.801203000Z
Executable: /usr/lib/virtualbox/VirtualBox
Arg[0]: /usr/lib/virtualbox/VirtualBox
Arg[1]: -comment
Arg[2]: Development
Arg[3]: -startvm
Arg[4]: 8584ee88-3448-4011-bb9a-616db97c989f
internal REM fatal error: rc=VERR_INTERNAL_ERROR cpu_outb

!!Assertion Failed!!
Expression: <NULL>
Location : /build/buildd/virtualbox-ose-1.5.6-dfsg/src/recompiler/VBoxRecompiler.c(4679) remAbort
internal REM fatal error: rc=VERR_INTERNAL_ERROR cpu_outb

whenever the virtual machine does not get closed, this error appears:
Code 3c 62 20 00 6f 64 65 3a 20 00 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 0a 00 31 61 2f 30 78 32 30 0a 00 30 20 74 61 73 <6B> 3d 63 30 34 33 63 33 34 30 20 74 61 73 6b 2e 74 69 3d 63 30
EIP: {<c04ee7eb>] printk_buf.24875+0x2b/0x400 SS:ESP 0068:c046f780
Kernel panic - not syncing: Fatal exception in interrupt

host: hardy 8.04 , kubuntu, using virtualbox. if i can be of any further assistance please let me know.

ANDREA

btw: i get 0 success rate with the options previously mentioned, i have not even once been able to correctly boot using 2.6.26 ...

[Kjell Braden (afflux)]

The virtualbox bug is not about 2.6.26 but the 2.6.23 kernel (the few comments at the end are from hatchetman82, he seems to have missed this).

[Kjell Braden (afflux)]

closing the upstream task for now, as there is no upstream bug (yet).

[Chris (chris-pm)]

I can confirm the above.

Trying to boot the Kubuntu Alpha2 iso (intrepid-alternate-i386.iso) gives the same EIP errors.
Trying to run the 'Check CD for defects' also results in EIP errors.

Host system is Hardy, 2.6.24-19-generic

[Thomas Novin (thomasn80)]

I can also boot 2.6.26-2 but not -3. However, booting -2 there is some problem with X so it doesn't start.

VirtualBox 1.6.2 on Hardy host.

[Cleber Santz (clebersantz)]

If help, i believe the same problem is occurring too with Fedora 10 (Kernel 2.6.26-0)

[Alessio Treglia (quadrispro)]

I can confirm the bug, I tried to boot an intrepid alpha2 (alternate i386) iso with VirtualBox and I received the same result.

[IceWil (icewil)]

Same problem here (host : Ubuntu 8.04 64-bit).

[Adam Niedling (krychek)]

I don't think that any more confirmation is needed.

[ANDREA (andrea54)]

... in case anybody cares even with .4 it's still failing here ...

[Chris (chris-pm)]

Thanks Andrea. I'm keeping an eye on this as I want to try out/help with Intrepid. But unfortunately I don't have a spare partition to use so vm is my only option at the moment.

[Brian Curtis (bcurtiswx)]

Just adding to the long list of confirmations. Virtualbox VM Ubuntu 8.10 07-17-08 updates. I use the old kernel to get it to work (2.6.24), for the time being.

[peter76 (pf-moll)]

This is weird, as I don't have problems ( and never had ) at all running Intrepid in VB under Hardy; both have all the latest updates as of today. What info should I point out here to solve why it's working for me and not others?

[TAC one (tacone)]

In response to Peter76: today's nightly build doesn't work for me as much as the previous ones.
On a side note the error message changed to: "Bug: Unable to handle kernel"

[Jon Packard (jonpackard)]

peter76: Thanks very much for your input on this. I would be interested to know if you have either VT-x or AMD-V enabled (Virtual machine settings > General > Advanced). If so, try disabling it and see if you are still able to boot the latest Intrepid kernel. Also, which version of Virtualbox or Virtualbox-OSE (Open Source Edition) are you using? Thanks!

[Scott Wegner (swegner)]

I'm also able to boot without any problem. I'm using VirtualBox 1.6.2
(not OSE), with x86 Hardy alpha 2, kernel version 2.6.26-4. I do have
"VT-x/AMD-V" enabled, and I believe I also was able to boot alpha 1
with this option as well.

I'm not home this weekend, so I won't be able to test with VT-x/AMD-V
off. If nobody else has responded by late Sunday, I can give it a
try.

On Sat, Jul 19, 2008 at 3:02 PM, Jon Packard <email address hidden> wrote:
> peter76: Thanks very much for your input on this. I would be interested
> to know if you have either VT-x or AMD-V enabled (Virtual machine
> settings > General > Advanced). If so, try disabling it and see if you
> are still able to boot the latest Intrepid kernel. Also, which version
> of Virtualbox or Virtualbox-OSE (Open Source Edition) are you using?
> Thanks!
>
> --
> Kernel panic during boot after upgrading to kernel 2.6.26.3-generic
> https://bugs.launchpad.net/bugs/246067
> You received this bug notification because you are a direct subscriber
> of the bug.
>
> Status in The Linux Kernel: Invalid
> Status in "linux" source package in Ubuntu: Confirmed
>
> Bug description:
> Running Intrepid Alpha1 in a Virtualbox 1.6.2 guest. Host is Hardy.
>
> linux-image-2.6.26-3-generic:
> Installed: 2.6.26-3.9
> Candidate: 2.6.26-3.9
> Version table:
> *** 2.6.26-3.9 0
> 500 http://us.archive.ubuntu.com intrepid/main Packages
> 100 /var/lib/dpkg/status
>
> Expected result: System boots correctly.
>
> Actual result: System fails to boot with kernel panic errors.
>
> Errors on screen (see attached for screen shot):
> Code: Bad EIP value.
> Kernel panic - not syncing: Attempted to kill the idle task!
>
> I can boot to linux-image-2.6.26-2-generic without any problems after the upgrade.
>

[TAC one (tacone)]

I am not able to boot. Playing with "VT-x/AMD-V" doesn't change it.

[peter76 (pf-moll)]

@ Jon Packard: yes, I have Vt-x enabled; and when I disable it I get a kernel panic as well, which ends the same as ANDREA's. I'm running VB 1.6.2 under Hardy. My processor is a C2D T5600. Attached is a dmesg from Intrepid. What looks weird to me is:

weird, boot CPU (#0) not listedby the BIOS.

If I can do anything else, let me know,

Peter

[Bobby (lostinquestions)]

This problem also exists, for me, in Alpha 3.

[Saivann Carignan (oxmosys)]

In my case, under VirtualBox, intrepid always boot with VT-x enabled, and it always get a kernel panic whitout VT-x enabled.

Careful : When the VT-x box have a grey check, it is still not enabled. You must make sure that VirtualBox is really configured to use VT-x.

[Saivann Carignan (oxmosys)]

Bug reported on VirtualBox bug tracking system :
http://www.virtualbox.org/ticket/1875

[Christoph Korn (c-korn)]

Ubuntu also crashes when VT-x is enabled.

CPU: Intel Core2Duo T7300

[lylepratt (lylepratt)]

At first I was experiencing the same error. Then i decided to try it with amd-v enabled. Turns out my bios didn't support it, so I flashed it with a new version, turned on amd-v and have successfully installed ubuntu 8.10 in virtual box.

[Ross Peoples (deejross)]

I had VT-x disabled and got [0.491424] BUG: unable to handle kernel <1> which repeats itself all the way down the screen. I turned VT-x on and was able to boot. KEEP IN MIND, that when you check the Enable VT-x/AMD-V checkbox once, that does NOT enable it!!! That just puts it into a grayed state. Clicking it a second time enables it.

[TAC one (tacone)]

Thank you Ross, but (really) enabling VT seems not to work for me and guess a number of people. It just changes the error message.

[Christoph Korn (c-korn)]

I forgot to enable VT-x in the BIOS.

Now intrepid boots.

[Phil (philheaton)]

Works fine for me.

Host system Hardy 8.04.1 amd64 with Virtualbox 1.6.2.

Host Specs:

Samsung X22 Notebook
Intel Core 2 Duo T7500 2.2Ghz
2GiB RAM

Virtualising Intrepid Alpha 3 32bit guest. Have enabled VT-x/AMD-V and PAE/NX and given the VM 1Gib RAM, 128MB Video RAM, SATA Controller, ALSA Audio Driver and USB/USB 2.0 Controller via Virtualbox.

Boots into the Live CD Environment and runs without problems. Installer works as it should, NAT connectivity works fine through Intel 4695 WiFi on host machine as does sound. Complete install of VM took approx 20 mins.

I found that after installation, the VM needed to be completely turned off not restarted as this would cause problems with X.

Once booted into the new VM, VirtualBox Guest Additions refused to install the Xorg modules (see VBAdditionsXorg screenshot).

Hope this is of some use.

Cheers, Phil.

[Alexandre Kraft (alexandre-kraft)]

As everyone said before, this problem is relative to the VT-x / AMD-V setting.

This setting is relative to hardware virtualization support either by your CPU and, as lylepratt wrote, your motherboard. You can find more information about Intel Virtualization Technology (Intel VT) and AMD Virtualization (AMD-V) supported CPUs on the Hardware support section of this Wikipedia x86 virtualization page (http://en.wikipedia.org/wiki/AMD-V).

Activating this option on an hardware which don't support it (CPU and/or motherboard) will not change anything and you will not be able to boot any of these three first alpha.

As found on the Ubuntu KVM Wiki page (https://help.ubuntu.com/community/KVM): To see if your processor supports one of these, you can run the following command. If nothing is printed, it means that your CPU doesn't support hardware virtualisation. Otherwise, it does - but you still need to make sure that virtualisation is enabled in the BIOS.

egrep '(vmx|svm)' /proc/cpuinfo

For information, I'm running Hardy 8.04.1 AMD64 and VirtualBox 1.6.2 on an AMD Athlon 64 X2 3800+ (no AMD-V support).

[Thomas Novin (thomasn80)]

I have enabled virtualization in my BIOS (Toshiba T61) and also enabled VT-x in VirtualBox. I still cannot boot (using daily alternate 070729).

model name : Intel(R) Core(TM)2 Duo CPU T7250 @ 2.00GHz
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc arch_perfmon pebs bts pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr lahf_lm ida

[Craig73 (funrun73)]

@Alexandre Kraft - My CPU doesn't support VT-x/AMD-V and I have turned the VT-x/AMD-V option off and I still can't boot the newer kernel. How does this fit into your theory that this is indeed the problem?

[Saivann Carignan (oxmosys)]

Craig : Intrepid will boot only with VT-x enabled. If it is disabled or if your CPU doesn't support it, intrepid won't boot.

[Alexandre Kraft (alexandre-kraft)]

@Craig: No no that's not what I wanted to say.... sorry, my english isn't very good. As Saïvann has clarified, if the VT-x/AMD-V option is disabled or if your CPU(motherboard doesn't support it, the first three alpha won't boot.

@Saïvann: Thanks for the clarification :-)

Is the support of virtualization will be required to start Intrepid in its later versions?

[Thomas Novin (thomasn80)]

Since we are a couple of people having this problem even though we have virtualization enabled in both BIOS and VirtualBox, maybe we have come across another bug?

[Christoph Korn (c-korn)]

Seems that in VirtualBox 1.6.4 the bug is still present.

Is it a hardy, kernel or VirtualBox bug?

[Christoph Korn (c-korn)]

I mean intrepid of course.

[Didier Roche-Tolomelli (didrocks)]

Same issue for me even if I enabled VT-x in my BIOS and checked the VT-x/AMD-V.
The VT-x works on my Inspiron 9400 with kvm + qmenu.

Trying to boot intrepid alpha 3 guest with a Hardy up to date.

[Jacopo Moronato (jmoronat)]

I've enabled VT-x, but the problem is still present.
I tried to boot the latest kernel: now no more kernel panic, but it freezes after logging in.

I'm using VB 1.6.2

[Alexandre Kraft (alexandre-kraft)]

I can confirm the bug is already present using VirtualBox 1.6.4.

[Carroarmato0 (carroarmato0)]

Has anyone analysed the changes that were made in the 2.6.26-4.11 kernel and the last kernel known to work?

[Jon Packard (jonpackard)]

It seems that this problem has been fixed in 2.6.26-5-generic. I was able to boot Intrepid in Vitrualbox 1.6.2 without using AMD-V or VT-x. Can anybody else confirm this?

[Christoph Korn (c-korn)]

Still unable to boot with VirtualBox 1.6.4 and kernel 2.6.26-5.11.

[Saivann Carignan (oxmosys)]

Same here

[Jon Packard (jonpackard)]

Please see my full kernel version below. Is anybody running 2.6.26-5.14 and still having the problem?

Full version info on my kernel:
 linux-image-2.6.26-5-generic:
  Installed: 2.6.26-5.14
  Candidate: 2.6.26-5.14
  Version table:
 *** 2.6.26-5.14 0
        500 http://us.archive.ubuntu.com intrepid/main Packages
        100 /var/lib/dpkg/status

[Jon Packard (jonpackard)]

After a few reboots, I noticed that I am still getting an occasional kernel panic. If you are getting a kernel panic with 2.6.26-5.14, try rebooting your client a few times. This did not work with 2.6.26-3 and 2.6.26-4 for me.

[Christoph Korn (c-korn)]

Even after trying 10 times having VT-x disabled booting always failed with the kernel panic (see the screenshot above).

Kernel: 2.6.26-5.14
VirtualBox 1.6.4

Shouldn't developers know when they fixed it?
Or don't they read here?

[Jon Packard (jonpackard)]

@Cristoph Korn
I wonder why mine can boot and yours cannot. My installation was from Alpha1 alternate installer and I have upgraded it through today. Which ISO did you use to install?

[Christoph Korn (c-korn)]

I used alpha3 alternate i386 iso.

I also only disabled VT-x in VirtualBox and not in my BIOS.
If that is of any concern.

Does intrepid also boot with VT-x disabled in VirtualBox 1.6.4.

[Cleber Santz (clebersantz)]

Jon, i make update to kernel 2.6.26-5.14-generic and the boot still crash with kernel panic.

[Cleber Santz (clebersantz)]

I`m testing various boot. Appears when open VirtualBox for the first time it boot fine. When Reboot/PowerOff VirtualMachine won`t boot. I make this test 3 times.

[Cleber Santz (clebersantz)]

Jon, forget my last message. This is not a rule, even when start VirtualBox sometimes crash. Sometimes system boot and most of time won`t boot.

[Carlos Nucette (cjnucette-gmail)]

I have the same problem, i've enabled VT-x on my Optiplex 755 BIOS and in VirtualBox with no luck, but i've noticed that even though i have VT-x enabled on BIOS the "egrep '(vmx|svm)' /proc/cpuinfo" doesn't return a thing.

[tamalet (tamalet)]

I experienced both error messages (BUG: Unable to handle kernel and Kernel panic - not syncing) with VT-x enabled on Virtual Box until I realized that the option that has to be set is on the virtual machine itself!
To be clear, the option that has to be enabled is not the one on File->Preference>General. You have to get the settings of the particular virtual machine, and in the General->Advance click twice on the VT-x/AMD-V option (so that it is not grayed).

It may also help adding some video memory.
I am now running the installer on the VM. I hope this helps others too!

[EmyrB (emyr)]

I am also getting the Kernel Panic screen with Intrepid Alpha 3 on virtual box 1.6.2 and also 1.6.4. Hosts is Vista Ultimate with SP1. The VT-x/AMD-V option has never been enabled on these VM's.

Hardware spec of PC is AMD64 X2 4400+ with 2GB of RAM.

[Christoph Korn (c-korn)]

New alpha, same bug.

So is this a kernel, intrepid or virtualbox bug?

[Rick Gatewood (george-gatewood)]

I tried alpha4 desktop iso and I was able to boot and run from the iso after enabling the VT-x/AMD-V option. It ran and installed fine, but I get kernel panic trying to boot the new install.

I always trial new releases in a virtual machine so this is a little disappointing. I may have to burn a cd and try it on a real pc. I want to see how well the new X works.

[Jonathan Blackhall (johnny-one-eye)]

@Christoph Korn - It appears to be either a kernel or vbox bug because the Fedora 10 alpha gives the same kernel panic. I haven't messed around with it with VT-x to know whether that fixes things.

[Luiz Capitulino (lcapitulino)]

I'm a Mandriva kernel team member and I would like to let you know that we have the same bug in Mandriva.

Our bug report is at:

https://qa.mandriva.com/show_bug.cgi?id=42776

I could reproduce this problem with two (patched) Mandriva kernels 2.6.26.1 and 2.6.26.2. And also with an unpatched 2.6.27-rc2 kernel, which is upstream.

The important information is that each kernel crashes in a different site, which makes me think this is a virtualbox bug.

Now, some people have reported that it works with older kernels. I will try to find a working kernel version and then try to locate the patch (if any) which is triggering the bug.

Would be great if someone could try to downgrade his (or her) virtualbox version to check whether it works with older versions.

I'm running virtualbox OSE 1.6.4 on Mandriva Cooker. My host kernel is a (patched) 2.6.26.2.

[Philip Wyett (philwyett)]

I am running the below version under Ubuntu 8.04.1 below and suffer the issue:

https://launchpad.net/ubuntu/hardy/+source/virtualbox-ose/1.5.6-dfsg-6ubuntu1

Complete changelogs:

https://launchpad.net/ubuntu/intrepid/+source/virtualbox-ose/+changelog

[Saivann Carignan (oxmosys)]

Luiz Capitulino : Thank you for attaching your bug!

[Ricky Campbell (cyberdork33)]

Trying to boot the Alpha4 alt iso in VirtualBox on WinXP host.

kernel panics normal as mentioned above, but if I edit the kernel options with vga=769 (and other values) it does not panic although the display is not good. I could see enough to install though, after which the system would boot OK.

Note: I did not enable virtualization extensions (and I do not have that feature on the computer I tried this with - Pentium M).

[Andreas Wenning (andreas-wenning)]

I have an intrepid install in virtualbox that has been upgraded from hardy. The 2.6.24-16 kernel works while the 2.6.26-2 fails to boot; that's the closest i can narrow it down.

[Andreas Wenning (andreas-wenning)]

Just found the alternate 2.6.25 kernel avaible in the linux-ports source: https://launchpad.net/ubuntu/+source/linux-ports
And this one works too.

[Luiz Capitulino (lcapitulino)]

I have found that kernel 2.6.25 works, while kernel 2.6.26-rc1 does not.

Please, refer to the Mandriva ticket for more information.

On Fri, Aug 15, 2008 at 8:10 PM, Andreas Wenning <email address hidden> wrote:
> Just found the alternate 2.6.25 kernel avaible in the linux-ports source: https://launchpad.net/ubuntu/+source/linux-ports
> And this one works too.
>
> --
> Kernel panic during boot in VirtualBox with kernel 2.6.26.*-generic
> https://bugs.launchpad.net/bugs/246067
> You received this bug notification because you are a direct subscriber
> of the bug.
>
> Status in The Linux Kernel: Invalid
> Status in Virtualbox: New
> Status in "linux" source package in Ubuntu: Confirmed
> Status in linux in Ubuntu Intrepid: Confirmed
> Status in Mandriva Linux: Unknown
>
> Bug description:
> Running Intrepid Alpha1 in a Virtualbox 1.6.2 guest. Host is Hardy.
>
> linux-image-2.6.26-3-generic:
> Installed: 2.6.26-3.9
> Candidate: 2.6.26-3.9
> Version table:
> *** 2.6.26-3.9 0
> 500 http://us.archive.ubuntu.com intrepid/main Packages
> 100 /var/lib/dpkg/status
>
> Expected result: System boots correctly.
>
> Actual result: System fails to boot with kernel panic errors.
>
> Errors on screen (see attached for screen shot):
> Code: Bad EIP value.
> Kernel panic - not syncing: Attempted to kill the idle task!
>
> I can boot to linux-image-2.6.26-2-generic without any problems after the upgrade.
>

--
Luiz Fernando N. Capitulino

[sanassar (sanassar)]

Disable the ACPI Feature in the Advanced tab.This should solve the issue for now.

[Christoph Korn (c-korn)]

I disabled ACPI but the bug remains.
(VT-x also disabled)

[Kevin Valentine (kejava)]

Just tested with Alternate Alpha 4 iso (I stopped burning CDs). Turned off all Extended features and it booted without the kernel panic ... once. It's very strange. It would work at other times if I made what seemed like large config changes in vbox, like SATA to IDE drive type. I've tried using boot options like noapic and nolapic but nothing sticks. It all seems very random. Is there a good way to step through the boot process? I've tried boot options like BOOT_DEBUG=3 and DEBCONF_DEBUG=5 but they don't seem to do anything.

[Fotis Tsamis (ftsamis)]

Same problem here with intrepid alpha 4 desktop 32-bit ...

Kernel Panic

[Samuel Gyger (gyger)]

Confirmed for Intrepid Alpha 4 alternate 32bit.

Worked with VT-x/AMD-V enabled.

[Samuel Gyger (gyger)]

I have to withdraw my info. Install worked out after activating VT-.... but no boot afterwards.

[Wojciech Miłkowski (wojtekm)]

VirtualBox 1.6.4 and Interpid Alpha 4 _works_ for me w/o VT-x/AMD-V support. My processor (Athlon 64 3800+) has no virtualization support. It occasionally crashes (I noticed it once), but generally it seems stable after boot sequence.
I have enabled SATA controler, and support for PAE/NX.

[Christoph Korn (c-korn)]

I can _not_ confirm that.

I also enabled SATA Controller and PAE/NX but intrepid keeps crashing with the same bug.

My processor:
Intel Core2Duo T7300

Dell Latitide D830 Notebook

[The Jay Mann (themann)]

I think I may know the cause to this (though I can't say this with any certainty).

I think this may be due to a feature I've termed "SMP on a single chip" (i.e. multicore and hyperthreading).

I say this, because I've noticed most times, when I get the kernel bug, I notice the line prior mentions SMP, as well as I use a P4 with HT, and have noticed most everyone else is using Dual Core. I'm assuming that when the kernel probes for CPU features, VirtualBox simply passes it through to the real hardware, but VirtualBox will not "emulate" SMP, thus causing the error. This would also explain why many people seem to find enabling VT fixes it, as the "emulated" CPU is more compatible.

The "final straw" that has me believing this, however, (note, I'm using Debian and not Ubuntu here), when I installed 2.6.18 for 686, it ran fine, however upon upgrading to 2.6.26, I have the panic (I'm assuming the SMP feature would have been added between), as well when I decided to.... "sidegrade" to 2.6.26 for 486, it booted just fine (I'm assuming, since no 80486 processor had HT or multiple cores or any feature similar to that, it doesn't even attempt such features).

[Saivann Carignan (oxmosys)]

Leann Ogasawara : You might be interested by comment #1 of upstream bug http://www.virtualbox.org/ticket/1875#comment:1 . It refers to a Linux git revision that might have an impact on this bug.

[Luiz Capitulino (lcapitulino)]

I can confirm that reverting commit e587cadd8f47e202a30712e2906a65a0606d5865 (as pointed out in the upstream bug report) fixes the problem for me, in Mandriva.

It is also important to say that I have ran git bisect between versions 2.6.25 and 2.6.26-rc1 (see comment #73) and it confirms that the commit above is the culprit.

I will also revert that commit from latest 2.6.27-rc3 just to be sure that it really fixes the problem.

Now the hard part: we have to investigate whether this is a kernel bug (which, for some reason only happens on virtualbox) or whether it is a virtualbox bug.

[Luiz Capitulino (lcapitulino)]

Some news.

The original reporter of the Mandriva's bug has added a comment saying that the problem also happens with QEMU. After investigating it for a while I have found that, indeed, QEMU also OOPSes but with a different behaivor. For example, QEMU works with latest 2.6.27-rc3, while virtualbox does not and both get OOPSes with 2.6.26-rc1 and reverting the DEBUG_RODATA commit makes no difference in QEMU, it still get OOPSes.

But the important news is that I have reported this upstream and there is a thread going on :)

http://lkml.org/lkml/2008/8/20/359

[Luiz Capitulino (lcapitulino)]

Looks like it works if you pass 'noreplace-paravirt' in the command-line,
something like this:

boot: linux vga=0 noreplace-paravirt

Could someone please try it and report the results?

[Cleber Santz (clebersantz)]

Addinding 'noreplace-paravirt' work`s for me. very slow to boot, but without panic.

[Stefan Bader (smb)]

Works for me, too.

[Christoph Korn (c-korn)]

Running very slow, but no kernel panic.

[Christophe Charlot (c-charlot)]

Same thing here, booting without kernel panic when adding 'noreplace-paravirt'

[costales (costales)]

I can confirmed this bug with beta 4 of Intrepid Ibex.
Version No OEM Virtualbox.

[Colin Ian King (colin-king)]

Thanks Luiz for reporting this upstream and informing us of the on-going discussion. Just to add:

Intrepid Alpha-4 2.6.26-5 - confirmed
and also 2.6.27-rc4 - confirmed

Works with noreplace-paravirt on Intrepid Alpha 4 2.6.26-5, and 2.6.27-rc4.

[Colin Ian King (colin-king)]

I'm still unsure if this is a bug in the kernel or in VirtualBox. I've added a post to the virtualbox ticket for this bug as I have some better understanding of where it is going wrong.

http://www.virtualbox.org/ticket/1875

Colin

[Christoph Korn (c-korn)]

The bug also exists in the 2.6.27 kernel that was included into intrepid.

[Leann Ogasawara (leannogasawara)]

The Ubuntu Kernel Team is planning to move to the 2.6.27 kernel for the upcoming Intrepid Ibex 8.10 release. As a result, the kernel team would appreciate it if you could please test this newer 2.6.27 Ubuntu kernel. There are one of two ways you should be able to test:

1) If you are comfortable installing packages on your own, the linux-image-2.6.27-* package is currently available for you to install and test.

--or--

2) The upcoming Alpha5 for Intrepid Ibex 8.10 will contain this newer 2.6.27 Ubuntu kernel. Alpha5 is set to be released Thursday Sept 4. Please watch http://www.ubuntu.com/testing for Alpha5 to be announced. You should then be able to test via a LiveCD.

Please let us know immediately if this newer 2.6.27 kernel resolves the bug reported here or if the issue remains. More importantly, please open a new bug report for each new bug/regression introduced by the 2.6.27 kernel and tag the bug report with 'linux-2.6.27'. Also, please specifically note if the issue does or does not appear in the 2.6.26 kernel. Thanks again, we really appreicate your help and feedback.

[Luiz Capitulino (lcapitulino)]

This bug is also present in current Linus tree.

FYI, I have reverted the patch which is triggering the bug from Mandriva kernel. Have a look at the Mandriva's ticket for more information.

[Colin Ian King (colin-king)]

Hi Luiz,

does reverting e587cadd8f47e202a30712e2906a65a0606d5865 in Linus' current tree resolve the issue? - it seemed to me that the bug is probably a problem VirtualBox rather kernel specific although it's still difficult to pin point what is going wrong. Reverting that particular patch does not trip the bug for the Mandriva kernel does may be just a workaround to a more fundamental problem - I will tinker with our 2.6.27 kernel a little more and see what gives.

Incidentally, by skipping the para virtualisation patching in vprintk() we don't get this bug occurring on 2.6.26/27.

Colin

[Christoph Korn (c-korn)]

After reversing the commit e587cadd8f47e202a30712e2906a65a0606d5865 (partially) and
rebuilding the kernel I now can boot intrepid inside VirtualBox 1.6.4 without having VT-x enabled
or 'noreplace-paravirt' boot option set.

Attached is the patch I applied on the intrepid kernel sources.
http://packages.ubuntu.com/intrepid/linux-image-2.6.27-1-generic

As you can see I was not able to reverse the whole commit, because the code has changed
at some points and I have no idea what I actually changed there ;-)
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=e587cadd8f47e202a30712e2906a65a0606d5865

It looks like these spin_unlock() and spin_lock() functions have been replaced by
mutex_unlock() and mutex_lock(). So I did not change these lines.

[Christoph Korn (c-korn)]

The sources are being compiled in my PPA for public testing.
https://launchpad.net/~c-korn/+archive

[Colin Ian King (colin-king)]

Christoph,

Thanks for the reverse patch - from this I was able to reduce it down to a few changes which seems to fix the problem with my VirtualBox test environment.

commit fa921c78c5cd2000a6ea2bd10298118fbf076107
Author: Colin Ian King <email address hidden>
Date: Fri Aug 29 08:57:04 2008 +0000

    UBUNTU: SAUCE: fix kernel oops in VirtualBox during paravirt patching
    Bug: 246067

    Paravirtualisation patching of kernel trips a kernel oops when
    running in VirtualBox. Turning off interrupts around the
    entire paravirtualisation patching process avoids patching
    active code inside an interrupt.

    Signed-off-by: Colin Ian King <email address hidden>

[Christoph Korn (c-korn)]

Colin,
I am glad my revert patch helped you.

Could you attach the diff of your patch so I can
built this patched kernel?

[Colin Ian King (colin-king)]

Christoph, find attached, and also at:

http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-intrepid.git;a=commit;h=fa921c78c5cd2000a6ea2bd10298118fbf076107

[Christoph Korn (c-korn)]

Colin,
I can confirm that this commit fixes the bug.

The new kernel version with this patch applied
is available in my PPA:
https://launchpad.net/~c-korn/+archive

[Oliver Gerlich (ogerlich)]

Hint: until an updated kernel is out, a good workaround seems to be to add noreplace-paravirt to kernel command line. This is noted in the LKML thread (http://lkml.org/lkml/2008/8/22/213). The normal 2.6.27 Intrepid kernel boots up fine in VB with that option, and I didn't notice any drawbacks yet.

[Christoph Korn (c-korn)]

This option also has been mentioned in this bug report.

It has the disadvantage to make intrepid inside the virtual
machine very slow.

[Carroarmato0 (carroarmato0)]

I have found a possible workaround. The workaround works only depending on the amount of luck you have:

You can try this only if you were able to install Intrepid Alpha 4 and are only experiencing the kernel problem with Init being killed.

***Workaround***

The goal here it to at least get a terminal working and update everything from there, hoping that it will install the latest kernel.

Before the system can boot, press "ESC" in order to access Grub.
Edit the default kernel by pressing "e", select the kernel boot line and press "e" again.

By the end of the line you should change the "ro" into "rw" and adding the number "2" at the very end of the line.

Confirm by pressing "Enter" and press "b" on that line to boot the kernel with those options.

If everything goes right, you should be presented with a prompt and a screen with strange behavior like not displaying everything.

If you have enough command line experience, you should be able to hit some commands and now what you are doing. This might take some tries to figure out what you have done.

You'll be presented with the prompt asking you for your login and password. Wait a few seconds in between entering your login and password.

If it worked you should be able to see the licensing things of Ubuntu, possibly scrambled all over the screen.

Next you'll want to update and upgrade everything:

sudo apt-get update && sudo apt-get upgrade

If this worked you'll see some activity on the screen (scrambled like in my case) and the VirtualBox icons representing your harddrive and network will be flashing, indicating that it's downloading and installing the packages.

When you think that everything is done upgrading, mostly when VirtualBox's status icons isn't showing any activity for some time. Then it's safe to restart.

shutdown -h now

*** End Workaround ***

Hope this worked for you!

[joenix (woutersj)]

I can confirm that the kernel in Christopher Korn's PPA works.

To me the easiest workaround currently would be to install the system using the noreplace-paravirt kernel option and then to download and install the patched kernel.

[Carroarmato0 (carroarmato0)]

Note about the workaround:

Forgot to mention that you should also do a:

sudo apt-get dist-upgrade

which will actually perform the entire upgrade to the new kernel

[Carroarmato0 (carroarmato0)]

I'm sorry to report that although my workaround was able to get the system to be updated to the latest packages and kernel, even with the kernel 2.6.27-2 the system miserably fails to load al together. The kernel generates a trace a7919e7f17c0a725 message.

[Luiz Capitulino (lcapitulino)]

Yes, this also affects latest 2.6.27-rc kernels.

And Colin should be careful when applying his patch in current 2.6.27 because mainline now uses a mutex to protect some code regions.

[Scott Wegner (swegner)]

Sun released VirtualBox 2.0 today, which is a major version update. Amongst other things, the changelog cites "VMM: VT-x fixes". Has anybody upgraded and given it a try?

[Vadim Peretokin (vperetokin)]

Upgraded, still got a kernel panic:
http://img502.imageshack.us/img502/1898/screenshot6qt9.png

[Cleber Santz (clebersantz)]

For me too, the error still persists

[TAC one (tacone)]

On Thu, Sep 4, 2008 at 8:15 PM, Cleber Santz <email address hidden> wrote:
> For me too, the error still persists

The error persists for me too.

[Christoph Korn (c-korn)]

This bug is milestoned for intrepid-alpha-6.

[pitit.atchoum (pitit-atchoum)]

it works with Virtualbox 2.0 :)

[TAC one (tacone)]

On Sat, Sep 6, 2008 at 7:34 PM, pitit.atchoum <email address hidden> wrote:
> it works with Virtualbox 2.0 :)

Not for me. Virtualbox 2.0 + Intrepid Alpha 5 + virtualization
disabled in the bios == same bug as before.

[Trip McKay (trip-mckay-deactivatedaccount)]

confirmed

did 3 flawless boots: it works with Virtualbox 2.0 if VT-X/AMD-V is enabled (only)

[Christoph Korn (c-korn)]

It does not.

[Christoph Korn (c-korn)]

It also worked with VT-X enabled in prior versions of virtualbox.

This is no progress.

[Trip McKay (trip-mckay-deactivatedaccount)]

more info:

tested with kernels 2.6.27-2, 2.6.27-1 and with 2.6.26-5 and always booted with VT-X/AMD-V enabled

[Vadim Peretokin (vperetokin)]

It worked when I enabled the extension, but not without it. Ubuntu 8.04
works fine with that disabled fyi.

[Snavely (tompv11)]

To all:
Ubuntu 8.04 host
Alpha's 1-5 have the same kernel panic problem using VMware Player 2.0.5

[Jacopo Moronato (jmoronat)]

Same kernel panic after upgrade to Alpha 5 & VirtualBox 2.0 .

[Dimitris Kalamaras (dimitris-kalamaras)]

Trying to upgrade from Ubuntu Hardy to Intrepid Ibex (on VirtualBox OSE 1.6.2 and Debian Sid/linux 2.6.26), here.
Initially I got the same kernel panic error, but "noreplace-paravirt" boot option did the trick.
Now I can boot with or w/ any ACPI enabled in the Advanced settings of VirtualBox.

[Cyphase (cyphase)]

I was able to get Alpha 5 working in Virtualbox 2.0. I used the alternate installer instead of the desktop CD. Of course, I used the noreplace-paravirt kernel option. After installing, I tried booting with noreplace-paravirt. It booted, but X wouldn't respond to my keyboard or mouse. I went to a virtual terminal and tried to add the c-korn ppa ( https://launchpad.net/~c-korn/+archive ), but when I ran apt-get update, it couldn't resolve any domain names. The network interface wasn't working. I manually downloaded the patched kernel from http://ppa.launchpad.net/c-korn/ubuntu/pool/main/l/linux/ and put it on an iso, then mounted it in Virtualbox. After installing it via dpkg, I restarted the VM. It booted without a problem. After that I used Synaptic to force the versions of all the kernel packages to the c-korn repository. This seems like it should work for everyone.

[Ulrich Hobelmann (u-hobelmann)]

"noreplace-paravirt" is the mot du jour. I tried the latest Alpha 5 desktop CD (i386) in install mode with that option, and at least it booted so far. It seems very very very sluggish (compared to previous Ubuntu-versions-in-virtualbox) and is I might abort the install as I'm short on time, but it seems to work.

[Saivann Carignan (oxmosys)]

Last comment on upstream VirtualBox bug says that this was virtualbox problem, and that this issue is fixed with upcoming 2.0.2 release. http://www.virtualbox.org/ticket/1875

[Christoph Korn (c-korn)]

If the bug is a VirtualBox bug is the kernel patch being reverted?

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 2.6.27-3.4

---------------
linux (2.6.27-3.4) intrepid; urgency=low

  [ Colin Ian King ]

  * SAUCE: fix kernel oops in VirtualBox during paravirt patching
    - LP: #246067
  * SAUCE: qc-usb: Enable Logitech QuickCam Messenger
    - LP: #209901
  * SAUCE: appleir: Enable driver for new MacBook Pro
    - LP: #157919

  [ Tim Gardner ]

  * Enabled CONFIG_DEBUG_RODATA=y

  [ Upstream Kernel Changes ]

  * Revert "ALSA: hda - Added model selection for iMac 24""
  * Revert "x86: fix HPET regression in 2.6.26 versus 2.6.25, check hpet
    against BAR, v3"
  * Revert "[ARM] use the new byteorder headers"
  * Revert "mac80211: Use IWEVASSOCREQIE instead of IWEVCUSTOM"
  * Revert "crypto: camellia - Use kernel-provided bitops, unaligned access
    helpers"
  * svcrdma: Fix race between svc_rdma_recvfrom thread and the dto_tasklet
  * sched, cpuset: rework sched domains and CPU hotplug handling (v4)
  * ACPI: Fix now signed module parameter.
  * ACPI: Change package length error to warning
  * ACPI: Fix now signed module parameter.
  * ACPI: Fix typo in "Disable MWAIT via DMI on broken Compal board"
  * acpi: add checking for NULL early param
  * UBIFS: fix zero-length truncations
  * Input: bcm5974 - add maintainer entry
  * sh64: re-add the __strnlen_user() prototype
  * sh: fix ptrace_64.c:user_disable_single_step()
  * PNPACPI: ignore the producer/consumer bit for extended IRQ descriptors
  * UBIFS: always read hashed-key nodes under TNC mutex
  * UBIFS: allow for racing between GC and TNC
  * [CIFS] Fix plaintext authentication
  * sparc32: Implement smp_call_function_single().
  * sh: crash kernel resource fix
  * sh: fix kexec entry point for crash kernels
  * sh: fix platform_resource_setup_memory() section mismatch
  * sh: update Migo-R defconfig
  * sh: update AP325RXA defconfig
  * sh: fix semtimedop syscall
  * cifs: fix O_APPEND on directio mounts
  * [CIFS] update cifs change log
  * [CIFS] Turn off Unicode during session establishment for plaintext
    authentication
  * ACPI: thinkpad-acpi: wan radio control is not experimental
  * sparc: Fix resource flags for PCI children in OF device tree.
  * remove blk_register_filter and blk_unregister_filter in gendisk
  * ALSA: oxygen: fix distorted output on AK4396-based cards
  * ipv6: When we droped a packet, we should return NET_RX_DROP instead of
    0
  * pkt_sched: Fix locking of qdisc_root with qdisc_root_sleeping_lock()
  * net: Unbreak userspace usage of linux/mroute.h
  * Don't trigger softlockup detector on network fs blocked tasks
  * Resource handling: add 'insert_resource_expand_to_fit()' function
  * sparc64: setup_valid_addr_bitmap_from_pavail() should be __init
  * UBIFS: do not update min_idx_lebs in stafs
  * UBIFS: push empty flash hack down
  * UBIFS: remove incorrect index space check
  * UBIFS: improve statfs reporting
  * UBIFS: fix assertion
  * UBIFS: add forgotten gc_idx_lebs component
  * UBIFS: introduce LEB overhead
  * UBIFS: improve statfs reporting even more
  * UBIFS: fill f_fsid
  * drm/radeon: downgrade debug message from info to debug.
  * Remove invalidate_partition call from do_md_stop.
  * Fix problem with wai...

[Trip McKay (trip-mckay-deactivatedaccount)]

Booted up-to-date Intrepid with 2.6.27-3.4 successfull 3 times without any problems to desktop with Virtualbox 2.0.0 and VT-X/AMD-V disabled. The third time it fell back to text-output doing a fsck check but continued without any problems.
I do have to say that with VT-X/AMD-V (which my CPU supports) enabled it feels somehow faster...

[TAC one (tacone)]

On Fri, Sep 12, 2008 at 9:28 AM, Trip McKay <email address hidden> wrote:
> Booted up-to-date Intrepid with 2.6.27-3.4 successfull 3 times without any problems to desktop with Virtualbox 2.0.0 and VT-X/AMD-V disabled. The third time it fell back to text-output doing a fsck check but continued without any problems.
> I do have to say that with VT-X/AMD-V (which my CPU supports) enabled it feels somehow faster...

This night I booted with an alpha5 livecd using noreplace-paravirt,
chrooted my existing (virtual) alpha5 Hd installation, and installed
the 2.6.27-3 image. After rebooting it worked but, as the livecd
itself, feels very sluggish. For every thing I do I have to wait for
the screen to redraw (and it does *slowly*).

I am using Virtualbox 2.0 (1.6 had the same problem, though).

[Pelládi Gábor (pelladigabor)]

I'm asking again what Christoph Korn asked. If the bug is caused by VirtualBox, and it will be fixed in 2.0.2, then why is the kernel patch necessary? It does help with VBox <= 2.0.0, but only makes things slower with VBox 2.0.2 (currently under development).
So is the custom kernel patch going to be reverted or not?

[Oliver Gerlich (ogerlich)]

Does the patch really make things slower with newer VBox version?

Anyway, it is nice to be able to run Intrepid at normal speed also in a non-bleeding-edge Vbox installation, so I'd appreciate if the patch is kept.

[Christoph Korn (c-korn)]

VirtualBox 2.0.2 fixes the bug.

It also runs with the unpatched kernel.

[Tristan Wibberley (tristan-wibberley)]

For me also, VB2.0.2 + alpha5 boots straight away. Although VB2.0.2 and alpha3 dropped to textmode once and panicked a second time.

[TAC one (tacone)]

On Sat, Sep 13, 2008 at 1:58 AM, Tristan Wibberley
<email address hidden> wrote:
> For me also, VB2.0.2 + alpha5 boots straight away. Although VB2.0.2 and
> alpha3 dropped to textmode once and panicked a second time.
>
> --
> Kernel panic during boot in VirtualBox with kernel 2.6.26.*-generic
> https://bugs.launchpad.net/bugs/246067
> You received this bug notification because you are a direct subscriber
> of the bug.

About the slow screen refresh, I installed Vbox 2.02, then downgraded
to 2.6 kernel and everything seems back to normal. 2.7 is still
sluggish.

[Saivann Carignan (oxmosys)]

Colin King : This bug is now fixed in virtualbox, should you consider reverting the kernel to his initial state?

[Óscar Rodríguez Ríos (ingorr01)]

Same problem, kernel panic.

Config:
HOST:
Ubuntu Hardy 8.04.1 amd84 updated.
Intel Core 2 Duo Q6600
8GB RAM

GUEST:
Ubuntu Intrepid Alpha 4 ISO i386 Desktop CD
1GB RAM
8GB Virtual Disk

Best regards.

[Christoph Korn (c-korn)]

Which VirtualBox version do you use?

[Óscar Rodríguez Ríos (ingorr01)]

Virtualbox OSE 1.5.6 (Hardy)

[Christoph Korn (c-korn)]

The bug is fixed in 2.0.2.

http://www.virtualbox.org/wiki/Downloads

[Christophe Charlot (c-charlot)]

I have virtualbox 2.0.2.
Intrepid is starting correctly with kernel 2.6.27.2 but I have again kernel panic with latest kernel 2.6.27.3

[Christoph Korn (c-korn)]

I cannot confirm that.

Is it exactly the same panic as before?

[Saivann Carignan (oxmosys)]

I also can't reproduce any kernel panic with 2.0.2, VT-x enabled or not with latest kernel.

[Christophe Charlot (c-charlot)]

I had kernel panics (not the same as initially) with kernel 2.6.26-3 yesterday, but now it's booting fine...

[Christoph Korn (c-korn)]

I reverted the virtualbox patch from latest intrepid kernel.

https://launchpad.net/~c-korn/+archive

For those who experience speed problems with the patched kernel.

[Aspa (b-karthiksankar)]

I too got the same error

[Michael Rooney (mrooney)]

Did you mean to change the status? It is not good to change them without explanation.

[Janus (reslayer-mail)]

Does anybody know if this actual for 4.2.0 kernel (far more recent than the one described in the OP)?
Because I ran into this after some googling while having the same issue on rather fresh Ubuntu.

[Janus (reslayer-mail)]

Actually both of them show the same error:
4.2.0.16-generic
4.2.0-30-generic