On Mon, Mar 16, 2020 at 09:52:10PM -0700, Sean Christopherson wrote:
> From: Junaid Shahid <junaids@xxxxxxxxxx>
>
> Free all L2 (guest_mmu) roots when emulating INVVPID for L1 and EPT is
> disabled, as outstanding changes to the page tables managed by L1 need
> to be recognized.
>
> Similar to handle_invpcid() and handle_invept(), rely on
> kvm_mmu_free_roots() to do a remote TLB flush if necessary, e.g. if L1
> has never entered L2 then there is nothing to be done.
>
> Fixes: 5c614b3583e7b ("KVM: nVMX: nested VPID emulation")
> Signed-off-by: Junaid Shahid <junaids@xxxxxxxxxx>
> [sean: ported to upstream KVM, reworded the comment and changelog]
> Signed-off-by: Sean Christopherson <sean.j.christopherson@xxxxxxxxx>
> ---
> arch/x86/kvm/vmx/nested.c | 11 +++++++++++
> 1 file changed, 11 insertions(+)
>
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index 9624cea4ed9f..50bb7d8862aa 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -5250,6 +5250,17 @@ static int handle_invvpid(struct kvm_vcpu *vcpu)
> return kvm_skip_emulated_instruction(vcpu);
> }
>
> + /*
> + * Sync L2's shadow page tables if EPT is disabled, L1 is effectively
> + * invalidating linear mappings for L2 (tagged with L2's VPID). Sync
> + * all roots as VPIDs are not tracked in the MMU role.
> + *
> + * TODO: sync only the affected SPTEs for INVDIVIDUAL_ADDR.
> + */
> + if (!enable_ept)
> + kvm_mmu_free_roots(vcpu, &vcpu->arch.guest_mmu,
Aha! This is wrong. guest_mmu is only used if EPT is enabled. The silver
lining is that I have confirmed that this code is needed and works as
intended (when the correct mmu is nuked).
> + KVM_MMU_ROOTS_ALL);
> +
> return nested_vmx_succeed(vcpu);
> }
>
> --
> 2.24.1
>
