Re: [RFC PATCH 0/3] KVM: Dirty page logging for guest_memfd-only memslots

[Sean Christopherson <seanjc@google.com>]

On Fri, Aug 14, 2026, Alexandru Elisei wrote:
> On Fri, Aug 14, 2026 at 06:03:54AM -0700, Sean Christopherson wrote:
> > > That would be in contrast to user space having to guess that page migration on
> > > the current hardware with the current guest_memfd implementation does not
> > > support page migration, to then disable exactly that.
> 
> If guest_memfd picks up support for another feature that might lead to
> memory being unmapped from stage 2 without userspace initiating it, would
> you be ok with that feature being gated by another guest_memfd flag?

Yes, absolutely.  FWIW, if I get my way and reclaim is driven by userspace, then
hopefully such a feature never comes along.  But if guest_memfd does gain some
feature that triggers invalidations, then yeah, at the very least it would need
to have some knowledge of whether or not the associated VM supports said feature.

> I get the feeling we're all on the same page, that guest_memfd will support
> a mode where memory isn't unmapped from stage 2, and right now the
> discussion is more about how that will be implemented.

Ya.