Hi Lorenzo, On Wed, 23 Sep 2026 16:15:43 +0100, "Lorenzo Stoakes (ARM)" <[email protected]> wrote: > This series implements the KVM stage 2 page table pre-faulting feature for > arm64.
Retested as follows (let's see if this one sticks ;) ): QEMU/arm64 -cpu max: - guests boot on VHE, nVHE and protected-mode hosts, protected guests included - pre_fault_memory_test (all guest modes, anonymous, THP, hugetlb and shmem backing) and nv_pre_fault_memory_test pass - pre-faulting an uninitialised vCPU returns -ENOEXEC - a failed guest_memfd pre-fault leaves kvm_run's exit_reason alone - dirty_log_test, demand_paging_test, mmu_stress_test and stage2_block_transitions also pass Apple M4 (nVHE, 4K and 16K pages): - pre_fault_memory_test passes on the five guest modes available there, with the same backings - in protected mode (4K pages), the cap isn't advertised, the ioctl returns -EOPNOTSUPP before and after vCPU init, and guests boot Tested-by: Fuad Tabba <[email protected]> Cheers, /fuad
