Re: [PULL 00/19] KVM: Add AMD Secure Nested Paging (SEV-SNP) Hypervisor Support

[Sean Christopherson]

On Sun, May 12, 2024, Paolo Bonzini wrote:
> On Sun, May 12, 2024 at 9:14 AM Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote:
> >
> > On Fri, May 10, 2024 at 11:17 PM Michael Roth <michael.roth@amdcom> wrote:
> > >
> > > Hi Paolo,
> > >
> > > This pull request contains v15 of the KVM SNP support patchset[1] along
> > > with fixes and feedback from you and Sean regarding PSC request processing,
> > > fast_page_fault() handling for SNP/TDX, and avoiding uncessary
> > > PSMASH/zapping for KVM_EXIT_MEMORY_FAULT events. It's also been rebased
> > > on top of kvm/queue (commit 1451476151e0), and re-tested with/without
> > > 2MB gmem pages enabled.
> >
> > Pulled into kvm-coco-queue, thanks (and sorry for the sev_complete_psc
> > mess up - it seemed too good to be true that the PSC changes were all
> > fine...).
> 
> ... and there was a missing signoff in "KVM: SVM: Add module parameter
> to enable SEV-SNP" so I ended up not using the pull request. But it
> was still good to have it because it made it simpler to double check
> what you tested vs. what I applied.
> 
> Also I have already received the full set of pull requests for
> submaintainers, so I put it in kvm/next.  It's not impossible that it
> ends up in the 6.10 merge window, so I might as well give it a week or
> two in linux-next.

I certainly don't object to getting coverage in linux-next, but unless we have a
very good reason to push for 6.10, which doesn't seem to be the case, my strong
preference is to wait until 6.11 for the official merge.  I haven't had a chance
to look at v15, and at a quick glance, the SNP_EXTENDED_GUEST_REQUEST support in
particular still looks kludgy.  In general, this all feels very rushed.