Screen lock in Windows 10 Host makes the Linux Guest unresponsive (regression)

[MichalBrz]

After upgrading from 6.1.30 to 6.1.34, I've started to notice that my Centos 7.8 Guest goes unresponsive on regular basis.

I've been able to pinpoint it to the screen lock in the Windows 10 21H2 Host. Whenever I unlock the screen the VBoxHeadless.exe process uses 100% of all of the cores available to the Guest machine. Guest is unreachable neither by ssh nor by GUI.

This happens regardless of the graphic display used, and also in the headless start. Also changing hypervisor between the Hyper-V and KVM does not make a difference.

VirtualBox 6.1.36 behaves the same as 6.1.34, however downgrading to the 6.1.30 solves the issue.

Logs do not contain anything except repeating pattern of: 00:02:22.924401 VMMDev: vmmDevHeartbeatFlatlinedTimer: Guest seems to be unresponsive. Last heartbeat received 4 seconds ago 00:02:24.627881 VMMDev: GuestHeartBeat: Guest is alive (gone 5 713 045 344 ns) 00:03:09.694062 VMMDev: vmmDevHeartbeatFlatlinedTimer: Guest seems to be unresponsive. Last heartbeat received 4 seconds ago 00:03:15.648850 VMMDev: GuestHeartBeat: Guest is alive (gone 9 990 252 305 ns) 00:03:30.642448 VMMDev: vmmDevHeartbeatFlatlinedTimer: Guest seems to be unresponsive. Last heartbeat received 4 seconds ago 00:03:30.792413 VMMDev: GuestHeartBeat: Guest is alive (gone 4 159 194 451 ns) 00:03:47.088470 VMMDev: vmmDevHeartbeatFlatlinedTimer: Guest seems to be unresponsive. Last heartbeat received 4 seconds ago 00:03:52.836984 VMMDev: GuestHeartBeat: Guest is alive (gone 9 762 953 270 ns) 00:03:56.874575 VMMDev: vmmDevHeartbeatFlatlinedTimer: Guest seems to be unresponsive. Last heartbeat received 4 seconds ago 00:03:57.771949 VMMDev: GuestHeartBeat: Guest is alive (gone 4 934 966 056 ns) 00:04:01.784622 VMMDev: vmmDevHeartbeatFlatlinedTimer: Guest seems to be unresponsive. Last heartbeat received 4 seconds ago

[MichalBrz]

example vbox.log