Windows Guests (XP, Vista, 7 32 or 64 bit) cause non-recoverable (reboot only) machine freeze (hang) with AMD VT extensions enabled.

Does turning off 3d support in the VM settings help? If it does, then try to find updates for your NVidia drivers.

3D support does not seem to make any difference in my case.

Ditto on shutdown freeze with XP

I have a similar problem. My host system is a Shuttle X48, 8GB Ram, Intel Core 2 Quad Q9650 CPU. Ubuntu 64-bit host. Hardware virtualization support is enabled in the BIOS. Whenever VT-x is enabled, the entire system instantly freezes with a black screen as soon as Virtualbox starts. If in the CPU options in the BIOS, I allow code space memory to not be write protected, then the same thing happens, except the screen doesn't go black. I have tried various versions of Nvidia drivers, VBox from V2.2 and now 3.02, I upgraded from Jaunty 9.04 to 9.10, nothing has any effect on the problem. I upgraded the BIOS as well. I am pretty sure that back < v2.0 of virtual box it would work. I Have tried enabling/disabling the 3D options. I typically have about 1GB of ram assigned to the guest, with 64MB or 128MB of ram for the video display.

Getting Worse!

Well 3.0.0 and 3.0.2 only worsened the situation. The 32 bit XP was very stable on this platform with 2.2.4 Guest Additions. Reinstalling 3.0.0 Guest additions returned the hardware virtualization freeze.

In addition in the 3.0.x series 64 bit XP and Win 7 refuse to run at all.

I tried adding "nohz=off" to the end of the "kernel" line in "/boot/grub/menu.lst" but that did absolutely nothing.

Interestingly enough this problem is specific to Acer 4530 hardware and Jaunty. Tried running VBox 3.0.2 PUEL on a AMD Athlon X2 5200 desktop system running Ubuntu 8.10 x64. X64 XP Guest and Win 7 x64 ran like a charm. Very Stable. Soon to upgrade desktop machine to Jaunty, will update with observations when it happens.

Is there any difference if you disable nested paging for the VM?

I confirm this bug. Machine: Acer Asprire 4530 Host: Fedora 11 x86_64 Virtualbox version: 3.0.2 and 3.0.4

I've tried windows xp and vista, 32 bit and 64 bit guest. All have freeze my host.

adhisimon: VBox.log please. If you don't have one, please tell me which AMD CPU you have.

Update. The problem continues on the 3.0.6 beta1. 32 bit XP runs untill total collapse either under Photoshop, or on shutdown signal (AMD-V, and Nested Paging, or without Nested Paging). 64 bit XP runs VDI's created on other systems, but cannot setup or run a 64 bit setup program. Additionally sending shutdown signal to a 64 bit xp vm causes system wide freeze, and irreparable damage to a VDI file (BSOD, FAT destruction, etc).

On machine that has a different amd cpu (VBOX 3.0.4), I am experiencing BSOD on XP 64 after serious system update or software install. Turning of the host (not loging out) and restarting vm seems to solve that problem by itself.

Let me know what info do you guys need to rectify the problem.

tiras_dude: try to disable 3d support in the VM settings. If that doesn't help, then please attach a log of the VM with 3.0.6 PUEL (please don't use your own compiled build to rule out any problems there).

No 3d with AMD-V

Hello,

Still the same issue. I attached the VBox-3.0.6-PUEL-XPsp3-Jaunty-2-6-28-25-generic.log. Died within minutes of boot.

BTW running 3d without AMD-V works ok, except for Solidworks 2008 (video issues, but no system stability problems).

Still can not run 64 bit windows successfully at all.

Sounds like a BIOS problem to me. Try to find an update. There are known errata for AMD cpus and BIOSes must be updated to implement workarounds. For a similar ticket see #4988.

A good thing to check is if there are similar reports for Windows 7/XP Compatibility mode & VirtualPC for the same hardware.

sandervl73 - I would tend to disagree

1. I have the newest BIOS from acer on the machine.

2. I have tested vmware and parallels on the same machine and had no stability issues. VBox is much faster.

3. 3.0.8 is a whole lot more stable. I am cautiously going to say that XP32 issue is gone.

XP 64 (and the same for linuxes amd64) do occasionally freeze but this is an exception rather than the rule as previously.

In addition, I have been testing ubuntu karmic with 2.6.31 kernel with Vbox 3.0.8 and have not noticed any issues at all yet. Keeping my fingers crossed.

If the improvement continues in 3.1.0 and above, this would be the most awesome virtualization product out there. (And if we could run snow leopard as guest (as on can do now in VMWare) with additions, without telling mr. jobs there will be no equal to this product =)

VMWare nor Parallels use AMD-V by default as far as I know.

There were no fixes in 3.0.8 to address your issue, so something else might have contributed to the improvement. Perhaps the Ubuntu update, but it sounds like the problem is still there.

sandervl73 - I may have spoken too soon, the new Jaunty -16 kernel is unstable again with VBox.

Do you have links to known errata for AMD Hyper-V cpu issue so I can forward the info to ACER to see if they can get a bug fix bios release for our machine?

Now, I don't know who should be fixing this (amd is definitely not doing it =(

But this document has something that I think is related http://support.amd.com/us/Processor_TechDocs/41788.pdf

In Particular:

"246 Breakpoint Due to An Instruction That Has an Interrupt Shadow May Be Delivered to the Hypervisor

Description

A #DB exception occurring in guest mode may be delivered in the host context under the following conditions: • A trap-type #DB exception is generated in guest mode during execution of an instruction with an interrupt shadow, and • The instruction that generated the exception is immediately followed by an instruction resulting in #VMEXIT.

Potential Effect on System Unpredictable results due to an unexpected #DB exception.

Suggested Workaround The hypervisor should have a valid interrupt gate in the IDT of the #DB handler entry and the handler must be able to determine that this event has occurred. If the event is detected, the handler should execute an IRET back to the hypervisor; one method that could be used to evaluate for this condition is to compare the RIP pushed on the stack to the RIP of the instruction following VMRUN, if they are equivalent then this event has occurred.

Fix Planned No"

Now given that our hypervison is VBox, may be somebody from V-Box team can implement this adjustment in to the code in the future release.

If it's not the area for v-box please advise and I will go bug Acer or Ubuntu kernel team.

problem continues 3.0.8, 3.0.10, 3.1.0 Beta 1

Can you find anything with google about your machine & Windows 7 (xp comp mode) or Virtual PC?

nothing on the net, but I am trying to install win7 rc to see if i get the same problems in virtual pc.

why in the world can you not run windows from a usb drive? arrrr. hate that company.

I have installed windows 7 x64 with xp mode / virtual pc. So far no issues, installing same soft as in my xp guest as of the moment. Is there any particular tests you want me to run while this thing is installed? I will keep this os on my machine untill tomorow, please let me know asap.

How did you use the xp mode? I had to install an additional piece of software (from MSDN) here. Only then will it show up in the menu and you can start the XP VM.

Here is an update:

To get to use xp vm I had to go here: http://www.microsoft.com/windows/virtual-pc/download.aspx and install both the xp-vm and virtual pc (the first is the actual xp vm build and the later being the hypervisor as I understand).

I set the memory and everything else to mach as closely to my VBox as I could.

Eventually I did manage to break it to get consistent BSOD's. By default XP-VM does not shut down, but hibernates. As soon as I forced it to shudtown, I got BSOD. I am attaching windows dump data, I don't know if it will help or not.

Called Acer, and forced them to post a BIOS update to version 1.3336 on their website: http://global-download.acer.com/GDFiles%5CBIOS/BIOS/BIOS_Acer_3336_A_A.zip?acerid=633940463055803203&Step1=Notebook&Step2=Aspire&Step3=Aspire%204530&OS=V10&LC=en&BC=Acer&SC=PA_7 (If the link does not work go www.support.acer.com and use driver download script)

Unfortunately, update did not solve the problem. I will be trying to talk to level II techs today, once they wake up, but so far I am dealing with level I and it is terrible.

If you have any suggestions on what I should tell them, please post, I still have the Win 7 installed, and probably keep it untill the weekend or untill issue is resolved.

file 1 of the windows crash report

2nd file of windows crash report

copy of the windows crash report dialog window

Photo of the BSOD

Got through to level II tech at ACER today. Highlighted the AMD errata and our forum thread there. The issue has been forward to engineering. Hopefully the answer will come sooner than later.

sandervl73 - if you know what needs to be done by ACER tech, please post here, my guess they will be reading this soon.

Cheers

Guess Acer does not care about hardware support: because virtual pc is not supposed to run on windows home premium (original os of the Acer 4530) they will not provide support for it, or issue a bios patch.

Got to love cop-outs. "And its also not a hardware issue."

Anyway, if any other 4530 owners want to bug Acer, you can reference case 1-5N1FZA, but I am really disappointed in such a stance from the company.

sandervl73 - if your team can do something on your side to resolve it, you would be awesome, otherwise I guess I can forget about hardware virtualization on this system.

sandrvl, got an e-mail from the Acer tech guy, guess they do have some good people working there, he sent me this link:

http://www.faultwire.com/solutions_bsod/A_clock_interrupt_was_not_received_on_a_secondary_0x00000101-1254.html (in particular the slew of links to msnd issues at the bottom). Apparetly some of the similar bsod's are indeed software related. I am digging through to see if there is anything particular to amd, so far mostly intel xeon issues.

Just an FYI

BSOD 0x101 is just a generic one. It would be interesting if there are more reports from 4530 owners about similar problems.

#4988 is a similar issue with a Dell laptop for instance. (and there are a lot of reports on the internet/dell support forums) Not quite the same CPU as you've got, but it looks like a similar issue.

There are known errata for certain AMD CPU models that require BIOS workarounds. That's all I can say unfortunately (NDA).

From my end without access to such a machine it's rather difficult to provide further assistance.

sandrvl73 are you US based? If so, I am willing to send you this thing in the name of science =) for a week or so. If this is acceptable, e-mail me privately just add hotmail to the user name for e-mail.

We're all in Europe, so that would be difficult (customs, danger to hardware) unfortunately. Thanks for the offer though.

Do you have the same stability issues with linux guests (e.g. DSL bootable iso)?

The issue is pretty much for windows (xp,7) guests (and select sofware [photoshop cs3 is notorious for killing the guest and host {zoom, clone tool}]/happens always/, or operations (installation, shutdown, file transfer)/happens ocasionally, except for windows7 x64 install wich fails all the time/. Also, the install environment on xp64 likes to take out the system.

Unix derivatives: Linux (Ubuntu, Suse, Fedora), BSD, or Solaris guests run fine (x32/x64) exept for Moblin and OSX. I do get a freeze on start-up onece in a blue moon, but that is usualy due to faulty settings or oddball occurence (can't replicate it).

Well, I suppose I don't really have much to add that is different. Same system, same problems.

*nix based guest installs seem to work fine. Windows based guest installs, well I can't actually get it installed. XP freezes my entire box (archlinux x86_64) when it copies the install files to start the install.

I wish I found this before I wiped my box thinking it was the encrypted drive setup I did. Atleast now I know I can redo that.

I read one guy who said he installed even a newer bios than was available on Acer's site and still is having the problems. I actually haven't updated my bios and was wandering which one I should update to?

I've had the same problem here for a while. Same machine, except I am running a windows 7 x64 host and the linux guest causes a bsod on the host. Running 1.334 bios here. 1.336 seemed to have some issues. I wonder if a software workaround could be found, as the issue seems related to the bios and acer doesn't seemed to be all that inclined to figure out why someone might want to use the AMD VT and fix it for them. XP Mode completely kills my system, usually before it even gets to boot, though I did get it to go to the desktop once before dying. Also older versions of vbox seemed more stable on this machine, with the BSOD only occuring when IOAPIC was enabled. Now just enabling AMD VT and nothing else at all on a single core will cause a pretty immediate BSOD, so clearly stability is going south. I am using the latest stable 3.14 version. I just reinstalled to Windows 7 Ultimate from the RC and was pretty disappointed that XP Mode had this problem and then really sad that vbox was the same way. Clearly it is the hardware and not windows 7 or anything since it does the same thing on linux hosts. I guess you will just have to accept that the acer 4530 is pretty broken when it comes to virtualization and hope that perhaps acer will actually get around to implementing it properly, which may or may not be possible to be honest, since they may have taken shortcuts, hence the incompatibilities. I've submitted a few bug reports on this and this one is actually the most active I've seen. It seems the turion processors have a commonality in some bioses (dell, acer) that keep their VT feature from working properly. There is a working dell bios for one of the machines, so maybe some crazy bios hacker could digg into it a bit and figure out exactly what they changed.......

Same machine same issue, oddly I had it working then It just freezes, I have tried different versions of 7 and cant run any in a vm in linux, it will freeze everything and will make me do a force shutdown to go back to normal.