Rob Bradford

So I looked at *why* CpuManager needed the MemoryManager: 1.) The guest_memory for programming the GDT when booting a vCPU (not needed for resizing). Can be refactored to be past...

> @rbradford This sounds like a very good plan and indeed a much cleaner solution as compared to mine. Are you planning to continue working on this or should I...

You need to outline how this feature will work and how it will be structured. As we discussed earlier it will be need to be under a compile time feature...

> But according to SEV-SNP specification vCPUs must be created before mapping any memory to guest because usually the guest memory would be encrypted. This translates to creating CpuManager independent...

Seems reasonable to split that functionality.

> Found the issue. It was a problem w/ CloudHv build of UEFI firmware rather than a problem with cloud-hypervisor itself. What is the review process for cloud-hypervisor flavored edk2...

The new OVMF binary is generating a boot loop in the firmware for the Windows tests (https://cloud-hypervisor-jenkins.westus.cloudapp.azure.com/job/cloud-hypervisor/job/main/803/) so i've rolled it back. I think we need to adjust the OVMF...

I think it would be best to solve this with the event_monitor infrastructure that we already have?

> If I'm being honest, I don't even know what that is. I just did a brief search around and didn't find any documentation on it. Can you point me...

If you're not seeing an event it's possible that either a.) we're losing event messages due the process exiting or b.) we didn't add an event yet. Since events are...