On Sat, Aug 07, 2021, Paolo Bonzini wrote: > This patch started as a way to make kvm_mmu_hugepage_adjust a bit simpler, > in preparation for switching it to struct kvm_page_fault, but it does > fix a microscopic bug in zapping collapsible PTEs. I think this also fixes a bug where userspace backs guest memory with a 1gb hugepage but only assigns a subset of the page to the guest. 1gb pages would be disallowed by the memslot, but not 2mb. kvm_mmu_max_mapping_level() would fall through to the host_pfn_mapping_level() logic, see the 1gb huge, and map the whole thing into the guest. I can't imagine any userspace would actually do something like that, but the failure mode is serious enough that it warrants a Fixes: + Cc: stable@.
