Machine hangs with 6.1.32 (possible regression w/ Hyper-V?)Hi

Discussions about using Linux guests in VirtualBox.
Post Reply
SCAI_Andre
Posts: 3
Joined: 17. Apr 2012, 10:00

Machine hangs with 6.1.32 (possible regression w/ Hyper-V?)Hi

Post by SCAI_Andre »

Hi,

it seems that there is an issue with guest hang ups not only with Windows guests, as also reported here viewtopic.php?f=2&t=105026

I have Linux (XUbuntu 21.04) guest, which does not work on 6.1.32, but works fine on 6.1.30.

When starting w/ 6.1.32 on a MSI Notebook running Win10 Home, the box either locks up after showing the text login prompt (but before launching the X-Windows session) or (on a host machine Lenovo Thinkpad P17 Gen 2, running Win10Pro) the guest launches, but under heavy load the machine starts to perform very badly until locking up.

The attached log file is from the MSI system. I left the VM open in its stuck state for an hour or two, to see if it would recover. It didn't.

Do you see anything useful in the logs?
Attachments
xxx-dev-21.04-lts-2022-02-15-13-15-13.log.zip
(34.91 KiB) Downloaded 6 times
scottgus1
Site Moderator
Posts: 20965
Joined: 30. Dec 2009, 20:14
Primary OS: MS Windows 10
VBox Version: PUEL
Guest OSses: Windows, Linux

Re: Machine hangs with 6.1.32 (possible regression w/ Hyper-V?)Hi

Post by scottgus1 »

The hangup issue with 6.1.32 is actually in the Windows host, not dependent on the guest OS.

Your log does not show the long time offset statistic that has been fingered as the 6.1.32 bug indicator. However, host OS Hyper-V is enabled, and your host just may not be able to run Virtualbox under it. See HMR3Init: Attempting fall back to NEM (Hyper-V is active).
klaus
Oracle Corporation
Posts: 1115
Joined: 10. May 2007, 14:57

Re: Machine hangs with 6.1.32 (possible regression w/ Hyper-V?)Hi

Post by klaus »

Doubt that the issue with freezing VMs in 6.1.32 in 'turtle mode' is 100% connected to the "long time offset statistic" which has been found to have at least a significant correlation. By now we actually have tracked down the cause, and it's an issue with interrupt delivery. A true regression in 6.1.32 from optimizing some APIC handling code, losing one bit. Resulting in Windows Hypervisor doing what it is asked to do but not what is the right thing.
fth0
Volunteer
Posts: 5668
Joined: 14. Feb 2019, 03:06
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Linux, Windows 10, ...
Location: Germany

Re: Machine hangs with 6.1.32 (possible regression w/ Hyper-V?)Hi

Post by fth0 »

Please try the VirtualBox test build 6.1.35r140972 (or newer) and report back, including a VBox.log file from a complete VM run if possible.
Post Reply