Virtualization.framework VM execution ownership and crash reclamation on Intel macOS Monterey

[jegheloo]

I'm investigating a lifecycle/security property of Apple's Virtualization.framework on Intel macOS Monterey 12.7.x, and UTM seems like one of the communities with the most practical experience with this backend.
The specific question is:
If a process creates and owns a running VZVirtualMachine, and that process is terminated abruptly — for example by crash or SIGKILL — is the underlying VM/vCPU execution guaranteed to terminate as part of that process's reclamation?
I'm specifically trying to understand the relationship between:

• the process holding the VZVirtualMachine object,
• the com.apple.Virtualization.VirtualMachine / XPC backend process,
• and the actual Hypervisor-backed guest execution.
  I'm not asking about normal shutdown via stop() or graceful cleanup. The requirement is stronger: after the execution-owning userspace process has been destroyed, guest computation must not be able to continue indefinitely as an independent execution domain.
  Target environment:
• Intel x86_64
• macOS Monterey 12.7.x
• Virtualization.framework
• Linux guest
• abrupt owner-process termination / crash reclamation
  Has anyone working with UTM's Apple Virtualization backend tested this directly on Intel Monterey, or observed what happens to the Virtualization.framework XPC/backend process after the owning process is killed?
  I'm especially interested in:
• whether the backend process survives,
• whether it becomes orphaned / adopted by launchd,
• whether guest vCPU execution can continue,
• and whether there is any reliable diagnostic way to map a VM instance to the process actually executing it.
  Apple Developer Forums thread with the full context:
  https://developer.apple.com/forums/thread/848542
  Any firsthand observations, logs, reproducible tests, or pointers to relevant UTM code paths would be very helpful.