Re: [PATCH] KVM: arm64: Drop mte_allowed check during memslot creation

[Oliver Upton <oliver.upton@linux.dev>]

On Mon, Feb 24, 2025 at 05:23:38PM +0000, Marc Zyngier wrote:
> On Mon, 24 Feb 2025 16:44:06 +0000, Aneesh Kumar K.V <aneesh.kumar@kernel.org> wrote:
> > What if we trigger a memory fault exit with the TAGACCESS flag, allowing
> > the VMM to use the GPA to retrieve additional details and print extra
> > information to aid in analysis? BTW, we will do this on the first fault
> > in cacheable, non-tagged memory even if there is no tagaccess in that
> > region. This can be further improved using the NoTagAccess series I
> > posted earlier, which ensures the memory fault exit occurs only on
> > actual tag access
> > 
> > Something like below?
> 
> Something like that, only with:
> 
> - a capability informing userspace of this behaviour
> 
> - a per-VM (or per-VMA) flag as a buy-in for that behaviour
> 
> - the relaxation is made conditional on the memslot not being memory
> (i.e. really MMIO-only).

I pretty much agree with you here but I think the flag ought to be a
per-memslot thing (rather than VMA or VM). Rather than open up the
entire memory attributes space to userspace we could just have a flag to
prevent cacheable mappings for the memslot.

Similar to how MTE is enforced today, we can have a shared check between
memslot creation && the abort path that'd require VM_MTE_ALLOWED for any
'cacheable memslot'. Failing memslot creation still is the clearest
signal of misuse to the VMM, IMO.

Thanks,
Oliver