Virtual Machine Introspection (of QEMU) with rr

VMI (Virtual Machine Introspection) is a term (among other things) for inspecting the state of virtual machines meaningfully from an external point of view. (see e.g. https://libvmi.com , https://drakvuf.com/ , https://en.wikipedia.org/wiki/Virtual_machine_introspection)

Given that rr enables reversible debugging of QEMU, it would be very Cool and Good (TM) if there was a way to use it to directly inspect guest state without forcing the user to handle stepping through the QEMU layer as well. It has really been bothering me for a while that this does not seem to be a thing, because it sounds so good. :)

What would it take to do something like this?

I'm (edit:) NOT well versed in these things, however:

I imagine the minimum intuitive requirements for this, depending on the information desired (several levels could be supported) are

backend:

• the ability to understand the exposed guest machine model, for low level debugging (hardware threads, etc)
• the ability to understand the guest OS (processes, kernel debugging)

frontend:

• possibly, exposing user-facing access methods to the above functionality

If this was possible, it might also be reasonably generalized to other situations where a general reverse execution framework is useful to implement a more domain specific debugger.

No files, tests, or entry points are named. Start by surveying rr's existing QEMU and record/replay integration, then define the supported guest-state levels and user-facing access needed; done would require an agreed design and implementation scope for guest introspection.