The shortlog is very misleading. KVM already supports 5-level paging for traditional shadow paging. This is specifically for shadowing nNPT, and it's specifically for shadow everything _except_ 5-level nNPT. Something like: KVM: x86/mmu: Support shadowing nNPT when 5-level paging is enabled in host On Wed, Aug 18, 2021, Wei Huang wrote: > When the 5-level page table CPU flag is exposed, KVM code needs to handle > this case by pointing mmu->root_hpa to a properly-constructed 5-level page > table. Similarly, this is wrong, or maybe just poorly worded. This has nothing to do with LA57 being exposed to the guest, it's purely the host using 5-level paging and NPT being enabled and exposed to L1.
