On Mon, 22 Feb 2021 04:58:41 +0000,
pnagar@codeaurora.org wrote:

On 2021-02-17 15:45, Marc Zyngier wrote:

[...]

quoted

+1 on that. Even if, as I suspect, this is targeting some unspecified
hypervisor that is not KVM, the first course of action should be for
this to be implemented in the kernel's own hypervisor first so that
anyone can review understand what is at play.

Thanks,

         M.

Thank you for your comments. The key value add of the feature is a
third party independent entity keeping a watch on crucial kernel
assets, such that in case the kernel itself is compromised, still,
the protection can remain intact.  Can this be achieved if the
implementation is done in KVM? I've limited knowledge of KVM
currently, can surely look into more details for a better
understanding.

[+Quentin]

KVM/arm64 doesn't currently support Stage-2 mappings on the host side,
but there are patches[1] on the list that implement this
functionality, and that I'm hoping to get in 5.13 (no pressure,
Quentin... ;-).

This could also be implemented with the current KVM code though, as a
PV service to guests, and I'd suggest looking into that as an initial
approach.

Thanks,

	M.

[1] https://lore.kernel.org/r/20210108121524.656872-1-qperret@google.com (local)

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