Figure out what cr8 should do in AccessVpState

[cperezvargas]

AccessVpState stubs out cr8 read/writes. Figure out if/how we should wire this up. Supposedly it's only used for the instruction emulator, which doesn't need cr8.

These error tracing calls will be downgraded to trace. There's also updates to rflags in this path

[cperezvargas]

For CR8 @chris-oo thinks when we fix the instruction emulator to only grab state it needs, this goes away, but we need to confirm

[smalis-msft]

That work is happening in #482

[babayet2]

@chris-oo I don't think this is connected to the emulator, am I missing something?

The emulator only reads registers via the EmulatorSupport implementation, which in turn reads the entry state / vmcs. The AccessVpState implementation reads registers in the same way that the EmulatorSupport implementation does, but the EmulatorSupport refactor does not impact AccessVpState.

As far as I can tell, the AccessVpState is used only in vp initialization (no cr8 access), and for hardware breakpoint support. I'd imagine at least in the hardware bp context we'd want all registers reported accurately: am I correct that this cr8 read needs to be wired up, not dropped?

*for translation registers, we also use the implementation of VirtualRegister

[chris-oo]

Yes I think this is separate, and yes we probably want to wire it up so it reads the correct values for hardware breakpoint/gdbstub support. It's not pressing as I'm not sure if that even works today, but we should fix it.

[cperezvargas]

Moving to TDX GA milestone after reassessment with @chris-oo

[chris-oo]

This was implemented in main by steven as part of the vsm work.