If CPU don't have support hardware virtualization, RHEL KVM will not run. But virt-host-validate just warn as "performance will be significantly limited."
I think if virt-host-validate shows "Check firmware settings to enable VT-x/AMD-V", it will be good.
(In reply to Kazuo Moriwaka from comment #1)
> I think if virt-host-validate shows "Check firmware settings to enable
> VT-x/AMD-V", it will be good.
I guess the virt-host-validate does not have downstream hacks, so if there is missing VT-X/AMD-V cpu feature it just warns about it even though QEMU's TCG mode should not be supported in RHEL. Jirka, any idea?
That's correct, virt-host-validate does not contain any downstream hacks.
However, the bug description is not correct. While KVM will not be usable, one
can still create QEMU virtual machines using TCG (domain type "qemu" in
libvirt vs. "kvm"). And in that case, performance will indeed be significantly
limited. I believe such usage is also officially unsupported, but I don't
really see anything wrong with what virt-host-validate says.
I agree with virt-host-validate's message is not wrong as general purpose diagnose message.
I suggest adding a note or sentence after procedure 3 to complement virt-host-validate's message. Such as "To avoid unpractical low performance, RHEL QEMU will fail to start if CPU's hardware virtualization support is missing."
There isn't a clear statement about the requirements of CPU's HVM support in this document. It's not a problem (all recent CPU has HVM support), but adding this note would decrease # of support cases from nested virtualization users.
It's very clear description and I'm OK with this fix.