Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <547250051f1578b7ddf60311be46b3eb7990ccc6.camel@redhat.com>
Subject: Re: [PATCH v2 02/24] KVM: VMX: Drop bits 31:16 when shoving
 exception error code into VMCS
From:   Maxim Levitsky <mlevitsk@redhat.com>
To:     Sean Christopherson <seanjc@google.com>,
        Paolo Bonzini <pbonzini@redhat.com>
Cc:     kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
        Jim Mattson <jmattson@google.com>,
        Oliver Upton <oupton@google.com>,
        Peter Shier <pshier@google.com>
Date:   Mon, 18 Jul 2022 16:04:04 +0300
In-Reply-To: <20220715204226.3655170-3-seanjc@google.com>
References: <20220715204226.3655170-1-seanjc@google.com>
         <20220715204226.3655170-3-seanjc@google.com>
Content-Type: text/plain; charset="UTF-8"
User-Agent: Evolution 3.40.4 (3.40.4-5.fc34) 
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

On Fri, 2022-07-15 at 20:42 +0000, Sean Christopherson wrote:
> Deliberately truncate the exception error code when shoving it into the
> VMCS (VM-Entry field for vmcs01 and vmcs02, VM-Exit field for vmcs12).
> Intel CPUs are incapable of handling 32-bit error codes and will never
> generate an error code with bits 31:16, but userspace can provide an
> arbitrary error code via KVM_SET_VCPU_EVENTS.  Failure to drop the bits
> on exception injection results in failed VM-Entry, as VMX disallows
> setting bits 31:16.  Setting the bits on VM-Exit would at best confuse
> L1, and at worse induce a nested VM-Entry failure, e.g. if L1 decided to
> reinject the exception back into L2.
> 
> Cc: stable@vger.kernel.org
> Signed-off-by: Sean Christopherson <seanjc@google.com>
> Reviewed-by: Jim Mattson <jmattson@google.com>
> ---
>  arch/x86/kvm/vmx/nested.c |  9 ++++++++-
>  arch/x86/kvm/vmx/vmx.c    | 11 ++++++++++-
>  2 files changed, 18 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index 8c2c81406248..05c34a72c266 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -3822,7 +3822,14 @@ static void nested_vmx_inject_exception_vmexit(struct kvm_vcpu *vcpu,
>         u32 intr_info = nr | INTR_INFO_VALID_MASK;
>  
>         if (vcpu->arch.exception.has_error_code) {
> -               vmcs12->vm_exit_intr_error_code = vcpu->arch.exception.error_code;
> +               /*
> +                * Intel CPUs will never generate an error code with bits 31:16
> +                * set, and more importantly VMX disallows setting bits 31:16
> +                * in the injected error code for VM-Entry.  Drop the bits to
> +                * mimic hardware and avoid inducing failure on nested VM-Entry
> +                * if L1 chooses to inject the exception back to L2.

Very small nitpick:
I think I would still prefer to have a mention that AMD CPUs can have error code > 16 bit,
The above comment kind of implies this, but it would be a bit more clear, but I don't
have a strong preference on this.


> +                */
> +               vmcs12->vm_exit_intr_error_code = (u16)vcpu->arch.exception.error_code;
>                 intr_info |= INTR_INFO_DELIVER_CODE_MASK;
>         }
>  
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index b0cc911a8f6f..d2b3d30d6afb 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -1621,7 +1621,16 @@ static void vmx_queue_exception(struct kvm_vcpu *vcpu)
>         kvm_deliver_exception_payload(vcpu);
>  
>         if (has_error_code) {
> -               vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, error_code);
> +               /*
> +                * Despite the error code being architecturally defined as 32
> +                * bits, and the VMCS field being 32 bits, Intel CPUs and thus
> +                * VMX don't actually supporting setting bits 31:16.  Hardware
> +                * will (should) never provide a bogus error code, but KVM's
> +                * ABI lets userspace shove in arbitrary 32-bit values.  Drop
> +                * the upper bits to avoid VM-Fail, losing information that
> +                * does't really exist is preferable to killing the VM.
> +                */
> +               vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, (u16)error_code);
>                 intr_info |= INTR_INFO_DELIVER_CODE_MASK;
>         }
>  


Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>

Best regards,
 Maxim Levitsky