CPU microcode in a VM

unnilquadium · 2024-08-26 13:43:25

I'm installing arch linux as a guest OS with libvirt on a host (also) arch OS.
CPU cores are passed with host-passthrough.

According to this page on the wiki, the microcode should only be installed on the host OS and not on the guest.

As I am not absolutely sure if this is correct, can someone confirm this?

Thank you.

loqs · 2024-08-26 14:05:59

unnilquadium wrote:

As I am not absolutely sure if this is correct, can someone confirm this?

What is the basis for your doubt?

unnilquadium · 2024-08-26 22:49:02

loqs wrote:

What is the basis for your doubt?

Does it apply under every circumstance?
According to openstack, host-passthrough:

...gives absolutely best performance, and can be important to some apps which check low level CPU details...

Since it talks about low-level access to the CPU, I'm only looking for confirmation.

loqs · 2024-08-26 23:05:00

I would expect the guest kernel to detect virtualization and not attempt to install the microcode and if it did attempt it to fail.

Arch Linux