Re: [PATCH kvmtool 0/3] arm64: Nested virtualization support

[Marc Zyngier <maz@kernel.org>]

Hi Andre,

On Fri, 20 Jun 2025 11:44:51 +0100,
Andre Przywara <andre.przywara@arm.com> wrote:
> 
> Thanks to the imperturbable efforts from Marc, arm64 support for nested
> virtualization has now reached the mainline kernel, which means the
> respective kvmtool support should now be ready as well.

Thanks for pushing this stuff out.

> 
> Patch 1 updates the kernel headers, to get the new EL2 capability, and
> the VGIC device control to setup the maintenance IRQ.
> Patch 2 introduces the new "--nested" command line option, to let the
> VCPUs start in EL2. To allow KVM guests running in such a guest, we also
> need VGIC support, which patch 3 allows by setting the maintenance IRQ.
> 
> Tested on the FVP (with some good deal of patience), and some commercial
> (non-fruity) hardware, down to a guest's guest's guest.
> 
> Cheers,
> Andre
> 
> P.S.: Marc: I saw the other patches in your kernel.org repo, do we need any
> of them - HYP timer IRQ, E2H0, counter offset?

Yes, please. They are very much necessary, and should serve as a
template for other VMMs (exposing all the interrupts is required, the
counter offset is necessary to test things resembling live migration,
and the e2h0 selection to run nVHE.

You can probably ignore the virtio patch for now, as this needs to be
properly debugged,

> I guess E2H0 for fruity hardware, what about the others?

The other way around. The '--e2h0' option forces the use of
HCR_EL2.NV1. Rotten fruits can't use NV1 (they actually can, but the
EL2 S1 PTW is fscked, so we hide it from KVM). However, other
implementations do have proper NV1 support, and that option is
extremely useful to boot a nVHE hypervisor.

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.