During the code review several issues related to the MSIx vectors, and message ID mappings and calculations were detected:

1. Possible overflows for queue to message ID translations.
2. Slightly convoluted code related to the queue to message ID translations found in some places
3. Initial calculations of amount of possible queues had a minor bug
4. Handling of the corner case when only one vector is available and configuration interrupt and queue interrupt will use the same vector
5. Possible improvements for mass vCPU cases and CPU hot plug 

The Epic created as a placeholder for dev upstream work and QE work needed for the downstream.

Current upstream related PRs:

• viostor: Fix device configuration updates:  https://github.com/virtio-win/kvm-guest-drivers-windows/pull/1520 
• [viostor] Limit num_cpus and max_cpus to MAX_CPU https://github.com/virtio-win/kvm-guest-drivers-windows/pull/1518
• viostor: Cleanups around msix_one_vector and MessageId https://github.com/virtio-win/kvm-guest-drivers-windows/pull/1503
• [viostor] Use ULONG for num_queues https://github.com/virtio-win/kvm-guest-drivers-windows/pull/1502
• viostor: Fix MessageId update in MESSAGENUMBER_TO_QUEUE() and related cleanup https://github.com/virtio-win/kvm-guest-drivers-windows/pull/1495