Re: [PATCH 2/5] KVM: VMX: Add guest physical address check in EPT violation and misconfig

[Jim Mattson <jmattson@google.com>]

On Thu, Feb 27, 2020 at 9:23 AM Mohammed Gamal <mgamal@redhat.com> wrote:
>
> Check guest physical address against it's maximum physical memory. If
Nit: "its," without an apostrophe.

> the guest's physical address exceeds the maximum (i.e. has reserved bits
> set), inject a guest page fault with PFERR_RSVD_MASK.
>
> Signed-off-by: Mohammed Gamal <mgamal@redhat.com>
> ---
>  arch/x86/kvm/vmx/vmx.c | 13 +++++++++++++
>  1 file changed, 13 insertions(+)
>
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index 63aaf44edd1f..477d196aa235 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -5162,6 +5162,12 @@ static int handle_ept_violation(struct kvm_vcpu *vcpu)
>         gpa = vmcs_read64(GUEST_PHYSICAL_ADDRESS);
>         trace_kvm_page_fault(gpa, exit_qualification);
>
> +       /* Check if guest gpa doesn't exceed physical memory limits */
> +       if (gpa >= (1ull << cpuid_maxphyaddr(vcpu))) {
> +               kvm_inject_rsvd_bits_pf(vcpu, gpa);

Even if PFERR_RSVD_MASK is set in the page fault error code, shouldn't
we set still conditionally set:
    PFERR_WRITE_MASK - for an attempted write
    PFERR_USER_MASK - for a usermode access
    PFERR_FETCH_MASK - for an instruction fetch

> +               return 1;
> +       }
> +
>         /* Is it a read fault? */
>         error_code = (exit_qualification & EPT_VIOLATION_ACC_READ)
>                      ? PFERR_USER_MASK : 0;
> @@ -5193,6 +5199,13 @@ static int handle_ept_misconfig(struct kvm_vcpu *vcpu)
>          * nGPA here instead of the required GPA.
>          */
>         gpa = vmcs_read64(GUEST_PHYSICAL_ADDRESS);
> +
> +       /* Check if guest gpa doesn't exceed physical memory limits */
> +       if (gpa >= (1ull << cpuid_maxphyaddr(vcpu))) {
> +               kvm_inject_rsvd_bits_pf(vcpu, gpa);

And here as well?

> +               return 1;
> +       }
> +
>         if (!is_guest_mode(vcpu) &&
>             !kvm_io_bus_write(vcpu, KVM_FAST_MMIO_BUS, gpa, 0, NULL)) {
>                 trace_kvm_fast_mmio(gpa);
> --
> 2.21.1
>