On Fri, May 10, 2024, Michael Roth wrote:
> Implement a platform hook to do the work of restoring the direct map
> entries of gmem-managed pages and transitioning the corresponding RMP
> table entries back to the default shared/hypervisor-owned state.
...
> +void sev_gmem_invalidate(kvm_pfn_t start, kvm_pfn_t end)
> +{
> + kvm_pfn_t pfn;
> +
> + pr_debug("%s: PFN start 0x%llx PFN end 0x%llx\n", __func__, start, end);
> +
> + for (pfn = start; pfn < end;) {
> + bool use_2m_update = false;
> + int rc, rmp_level;
> + bool assigned;
> +
> + rc = snp_lookup_rmpentry(pfn, &assigned, &rmp_level);
> + if (WARN_ONCE(rc, "SEV: Failed to retrieve RMP entry for PFN 0x%llx error %d\n",
> + pfn, rc))
> + goto next_pfn;
This is comically trivial to hit, as it fires when running guest_memfd_test on a
!SNP host. Presumably the correct fix is to simply do nothing for !sev_snp_guest(),
but that's easier said than done due to the lack of a @kvm in .gmem_invalidate().
That too is not a big fix, but that's beside the point. IMO, the fact that I'm
the first person to (completely inadvertantly) hit this rather basic bug is a
good hint that we should wait until 6.11 to merge SNP support.
