On Tue, Feb 27, 2024 at 06:41:32PM -0800, Sean Christopherson wrote:
> Exit to userspace with -EFAULT / KVM_EXIT_MEMORY_FAULT if a private fault
> triggers emulation of any kind, as KVM doesn't currently support emulating
> access to guest private memory. Practically speaking, private faults and
> emulation are already mutually exclusive, but there are edge cases upon
> edge cases where KVM can return RET_PF_EMULATE, and adding one last check
> to harden against weird, unexpected combinations is inexpensive.
>
> Suggested-by: Yan Zhao <yan.y.zhao@xxxxxxxxx>
> Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
> ---
> arch/x86/kvm/mmu/mmu.c | 8 --------
> arch/x86/kvm/mmu/mmu_internal.h | 13 +++++++++++++
> 2 files changed, 13 insertions(+), 8 deletions(-)
>
> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index e4cc7f764980..e2fd74e06ff8 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
> @@ -4309,14 +4309,6 @@ static inline u8 kvm_max_level_for_order(int order)
> return PG_LEVEL_4K;
> }
>
> -static void kvm_mmu_prepare_memory_fault_exit(struct kvm_vcpu *vcpu,
> - struct kvm_page_fault *fault)
> -{
> - kvm_prepare_memory_fault_exit(vcpu, fault->gfn << PAGE_SHIFT,
> - PAGE_SIZE, fault->write, fault->exec,
> - fault->is_private);
> -}
> -
> static int kvm_faultin_pfn_private(struct kvm_vcpu *vcpu,
> struct kvm_page_fault *fault)
> {
> diff --git a/arch/x86/kvm/mmu/mmu_internal.h b/arch/x86/kvm/mmu/mmu_internal.h
> index 0669a8a668ca..0eea6c5a824d 100644
> --- a/arch/x86/kvm/mmu/mmu_internal.h
> +++ b/arch/x86/kvm/mmu/mmu_internal.h
> @@ -279,6 +279,14 @@ enum {
> RET_PF_SPURIOUS,
> };
>
> +static inline void kvm_mmu_prepare_memory_fault_exit(struct kvm_vcpu *vcpu,
> + struct kvm_page_fault *fault)
> +{
> + kvm_prepare_memory_fault_exit(vcpu, fault->gfn << PAGE_SHIFT,
> + PAGE_SIZE, fault->write, fault->exec,
> + fault->is_private);
> +}
> +
> static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
> u32 err, bool prefetch, int *emulation_type)
> {
> @@ -320,6 +328,11 @@ static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
> else
> r = vcpu->arch.mmu->page_fault(vcpu, &fault);
>
> + if (r == RET_PF_EMULATE && fault.is_private) {
Should we just check VM type + RET_PF_EMULATE, and abort?
If r is RET_PF_EMULATE, and fault is caused by accesing a shared address,
the emulation code could still meet error if guest page table pages are in
private memory, right?
> + kvm_mmu_prepare_memory_fault_exit(vcpu, &fault);
> + return -EFAULT;
> + }
> +
> if (fault.write_fault_to_shadow_pgtable && emulation_type)
> *emulation_type |= EMULTYPE_WRITE_PF_TO_SP;
>
> --
> 2.44.0.278.ge034bb2e1d-goog
>
