VirtualBox

Ticket #5883 (closed defect: fixed)

Opened 4 years ago

Last modified 4 years ago

BSOD KMODE_EXCEPTION_NOT_HANDLED - Windows 2000 guest on Virtualbox 3.1.x (PC & Mac hosts)

Reported by: msa.operations Owned by:
Priority: critical Component: other
Version: VirtualBox 3.1.2 Keywords:
Cc: Guest type: Windows
Host type: Mac OS X

Description

Windows 2000 guest fails to boot within VirtualBox 3.1.0 and 3.1.2 on both PC and Mac platforms.

This VM does boot and works correctly with VirtualBox 3.0.x - uninstalling 3.1.x and re-installing 3.0.x confirms this VM is still working correctly.

The same error occurs on both Windows (XP & 7) and Mac OS X (10.5.8). The VM has the latest 3.0.x tools installed within the guest OS.

Other Guest OS VMs are confirmed working on the same machines. Only Windows 2000 looks to be affected.

BSOD Error details:

*** STOP: 0x0000001E (0xC0000005, 0x8046ACC8, 0x00000000, 0x00000000) KMODE_EXCEPTION_NOT_HANDLED

*** Address 8046ACC8 base at 80400000, DateStamp 45ec3c8f - ntoskrnl.exe

Attachments

VBox.log Download (55.2 KB) - added by msa.operations 4 years ago.
VirtualBox Log File
Win2k-3.0.12-works.log Download (43.0 KB) - added by kyndigen 4 years ago.
Log of a working boot under 3.0.12
Win2k-3.1.6-BSOD.log Download (42.0 KB) - added by kyndigen 4 years ago.
Log of a boot that results in a BSOD under 3.1.6.

Change History

Changed 4 years ago by msa.operations

VirtualBox Log File

comment:1 Changed 4 years ago by msa.operations

Upon further searching, I found a similar problem posted by a Linux Host user in the Forums:

 http://forums.virtualbox.org/viewtopic.php?f=2&t=25998

*** STOP: 0x00000050 (0xF22DF000, 0X00000000, 0X804F5586, 0X00000000) PAGE_FAULT_IN_NONPAGED_AREA

*** Address 804f5586 base at 80400000 DateStamp 4a781s9a - ntoskrnl.exe

The suggested fix (Disable VT-x/AMD-V acceleration) also corrects my KMODE problem.

Should hardware acceleration work correctly for Windows 2000 guests? Or is this an OS specific problem? There is already 1 specific case workaround listed in the documentation for installing Windows 2000, perhaps this situation should also be noted?

comment:2 Changed 4 years ago by gilboa

I can confirm the report above. Fedora 12, x86_64, 2 x Xeon E5335. Windows 2K setup hangs mid-install if VT-d is enabled. Attempting to enable it while post-installing will result in a BSOD KMODE_EXCEPTION_NOT_HANDLED.

comment:3 Changed 4 years ago by cwaffles

I can confirm the same thing happens on Windows 7 x64 host with Windows 2000 SP4 with Rollup 1 as guest. This previously did not happen with 3.0.12

comment:4 follow-up: ↓ 5 Changed 4 years ago by netslayer

I'm having the same problem (PAGE_FAULT_IN_NONPAGED_AREA 0x00000050) but intermittently on Windows Server 2003 guest, Ubuntu Linux 9.10 host. This VM worked fine in VB3.0.x, upgraded to 3.1.2 r56127 today and it has become very unstable. Usually I can blue screen it in minutes of use while on the desktop doing a MSI install. Repro'd with Nested paging off, PAE/NX on or off and VT-X on. Confirmed disabling of VT-X returns stability.

Previous version it worked in virtualbox-3.0 3.0.12-54655_Ubuntu_jaunty 3.0.12-54655_Ubuntu_jaunty

New version it broke in virtualbox-3.1 3.1.2-56127_Ubuntu_jaunty 3.1.2-56127_Ubuntu_jaunty

comment:5 in reply to: ↑ 4 ; follow-up: ↓ 6 Changed 4 years ago by netslayer

Replying to netslayer:

I'm having the same problem (PAGE_FAULT_IN_NONPAGED_AREA 0x00000050) but intermittently on Windows Server 2003 guest, Ubuntu Linux 9.10 host. This VM worked fine in VB3.0.x, upgraded to 3.1.2 r56127 today and it has become very unstable. Usually I can blue screen it in minutes of use while on the desktop doing a MSI install. Repro'd with Nested paging off, PAE/NX on or off and VT-X on. Confirmed disabling of VT-X returns stability.

Previous version it worked in virtualbox-3.0 3.0.12-54655_Ubuntu_jaunty 3.0.12-54655_Ubuntu_jaunty

New version it broke in virtualbox-3.1 3.1.2-56127_Ubuntu_jaunty 3.1.2-56127_Ubuntu_jaunty

Spoke too soon about it being stable with VT-X disabled. Blue screened with MEMORY_MANAGEMENT stop code 0x0000001A (0x00041284, 0x77F3F001, 0x000000000, 0xC0502000). Also noticed I had the Jaunty version installed and I'm on Karmac. Re-installed with Karmac VB, repro'd again. I'm going back to 3.0.x

comment:6 in reply to: ↑ 5 Changed 4 years ago by netslayer

Replying to netslayer:

Replying to netslayer:

I'm having the same problem (PAGE_FAULT_IN_NONPAGED_AREA 0x00000050) but intermittently on Windows Server 2003 guest, Ubuntu Linux 9.10 host. This VM worked fine in VB3.0.x, upgraded to 3.1.2 r56127 today and it has become very unstable. Usually I can blue screen it in minutes of use while on the desktop doing a MSI install. Repro'd with Nested paging off, PAE/NX on or off and VT-X on. Confirmed disabling of VT-X returns stability.

Previous version it worked in virtualbox-3.0 3.0.12-54655_Ubuntu_jaunty 3.0.12-54655_Ubuntu_jaunty

New version it broke in virtualbox-3.1 3.1.2-56127_Ubuntu_jaunty 3.1.2-56127_Ubuntu_jaunty

Spoke too soon about it being stable with VT-X disabled. Blue screened with MEMORY_MANAGEMENT stop code 0x0000001A (0x00041284, 0x77F3F001, 0x000000000, 0xC0502000). Also noticed I had the Jaunty version installed and I'm on Karmac. Re-installed with Karmac VB, repro'd again. I'm going back to 3.0.x

I created bug id #5906 for my problem as I think it may not be the same

comment:7 follow-up: ↓ 9 Changed 4 years ago by kyndigen

I hit the exact same problem on boot of my Win2k guest. Here's the BSOD signature:

* STOP: 0x0000001E (0xC0000005,0x8046ACC8,0x00000000,0x00000000) KMODE_EXCEPTION_NOT_HANDLED * Address 8046ACC8 base at 80400000, DateStamp 4a781d9e - ntoskrnl.exe

I can run the same base VM image on two different boxes: a laptop and a desktop. The laptop runs both OSX and Win7, but the behavior is the same regardless of what host OS I use so I'm ignoring that variable.

Under 3.0.12, both machines work. Under 3.1.2, the desktop host works, but the laptop host crashes with the BSOD. The key difference is the desktop has a Core i7 CPU that supports nested paging, while the laptop doesn't (it's a MacBook Pro 5,1 which has a C2Duo-P8600). Turning off VT-x on the laptop eliminates the crash.

I don't have much installed on the VM, but I do have DaemonTools installed, which comes with the SPTD (SCSI PassThrough Direct) driver. If I boot into safe mode, that driver is not loaded and the guest boots fine. If I uninstall DaemonTools (thus removing SPTD.sys) the guest works fine. My guess is that the crash is happening during the load of SPTD.sys within the guest.

comment:8 Changed 4 years ago by vbrocks

windows guest on mac ox ...I am running mac ox, have installed virtualbox and then windows xp. It worked at first. Now I cannot power on windows. I get the sun virtualbox screen and then the windows screen and then the virtualbox screen comes back and shows that the windows is powered off. Where do I start to fix this problem. I did try to uninstall by putting the application into the trash. Then I reinstalled the virtualbox. When I opened the new virtualbox, I expected to see no windows installations but I do have the original installation and a second installation I tried when the problem began.

comment:9 in reply to: ↑ 7 Changed 4 years ago by msa.operations

Replying to kyndigen:

I hit the exact same problem on boot of my Win2k guest... Under 3.0.12, both machines work. Under 3.1.2, the desktop host works, but the laptop host crashes with the BSOD. The key difference is the desktop has a Core i7 CPU that supports nested paging, while the laptop doesn't (it's a MacBook Pro 5,1 which has a C2Duo-P8600). Turning off VT-x on the laptop eliminates the crash.

I don't have much installed on the VM, but I do have DaemonTools installed, which comes with the SPTD (SCSI PassThrough Direct) driver. If I boot into safe mode, that driver is not loaded and the guest boots fine. If I uninstall DaemonTools (thus removing SPTD.sys) the guest works fine. My guess is that the crash is happening during the load of SPTD.sys within the guest.

Interrestingly, I am also using a MacBook Pro (5,2 17" 2.66GHz C2D). However, I tested across a number of other PC laptop and desktop machines (P4 & C2D) and note the same BSOD behaviour in 3.1.x - and all work fine in 3.0.x. It would appear the change in VT-x functionality with 3.1 works correctly with i7 processors, but not 100% with earlier processors.

This bug is still outstanding, and will prevent me from upgrading until it is resolved.

comment:10 Changed 4 years ago by bradtem

Confirm. The puppy won't even boot in safe mode with vt-x enabled. You have to downgrade to one processor and no vt-x. In safe mode it does not BSOD, it just hangs. Win2K guest on kubuntu karmic amd64 host

comment:11 Changed 4 years ago by kyndigen

Still fails under 3.1.6.

comment:12 Changed 4 years ago by frank

Windows 2000 works fine here with VBox 3.1.6 (Linux host though). If your Windows 2000 guest boots fine with an older version of VBox but doesn't boot with VBox 3.1.6, please attach a VBox.log file for a VM session with the older VBox version and one VBox.log file for a VM session with VBox 3.1.6.

comment:13 Changed 4 years ago by borut

Windows 2000 SP4 guest works fine under VB 3.1.6 on Windows XP Home SP3 host too (AMD Athlon 64; no 2D/3D acceleration activated, no VT-x).

On the other hand, I haven't noticed any problems under VB 3.1.0 either (I skipped the VB versions in between and upgraded directly to VB 3.1.6, so can not tell about 3.1.2 and 3.1.4).

Changed 4 years ago by kyndigen

Log of a working boot under 3.0.12

Changed 4 years ago by kyndigen

Log of a boot that results in a BSOD under 3.1.6.

comment:14 Changed 4 years ago by msa.operations

My testing confirms this bug still exists.

PLEASE NOTE: I believe this bug is directly related to Hardware Acceleration (VT-x/AMD-V) and the Windows 2000 guest OS. Turning off the Hardware Virtualization options will cause the guest OS to boot correctly.

Test Conditions (Windows 2000 guest, VT-x/AMD-V enabled)
FAIL: Windows 7 Pro (64-bit), Intel Core 2 Duo P8600
FAIL: Mac OS X 10.6.2, Intel Core 2 Duo T9550
FAIL: Windows 7 Pro (32-bit), Intel Core 2 Duo T9550

Please note, all of the above environments test OK when using 3.0.x and FAIL using all 3.1.x versions.

comment:15 Changed 4 years ago by sandervl73

  • Status changed from new to closed
  • Resolution set to fixed

Try again with 3.2.0. One Win2k regression was fixed, so this is probably a duplicate.

comment:16 Changed 4 years ago by kyndigen

Confirm. My w2k image boots just fine on 3.2.0 with VT-x enabled.

Thanks for the fix!

Note: See TracTickets for help on using tickets.

www.oracle.com
ContactPrivacy policyTerms of Use