Re: [PATCH v4 1/3] KVM: X86: Extend KVM_SET_VCPU_EVENTS to inject a SHUTDOWN event

[Sean Christopherson <seanjc@google.com>]

On Thu, Mar 10, 2022, Chenyi Qiang wrote:
> In some fatal case, the target vcpu would run into unexpected behavior
> and should get shutdown (e.g. VM context is corrupted and not valid in
> VMCS). User space would be informed in such case. To kill the target
> vcpu, extend KVM_SET_VCPU_EVENTS ioctl to inject a synthesized SHUTDOWN
> event with a new bit set in flags field. KVM would accordingly make
> KVM_REQ_TRIPLE_FAULT request to trigger the real shutdown exit. Noting
> that the KVM_REQ_TRIPLE_FAULT request also applies to the nested case,
> so that only the target L2 vcpu would be killed.
> 
> Signed-off-by: Chenyi Qiang <chenyi.qiang@intel.com>
> ---
>  Documentation/virt/kvm/api.rst  | 3 +++
>  arch/x86/include/uapi/asm/kvm.h | 1 +
>  arch/x86/kvm/x86.c              | 6 +++++-
>  3 files changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
> index 691ff84444bd..d1971ef613e7 100644
> --- a/Documentation/virt/kvm/api.rst
> +++ b/Documentation/virt/kvm/api.rst
> @@ -1241,6 +1241,9 @@ can be set in the flags field to signal that the
>  exception_has_payload, exception_payload, and exception.pending fields
>  contain a valid state and shall be written into the VCPU.
>  
> +KVM_VCPUEVENT_SHUTDOWN can be set in flags field to synthesize a SHUTDOWN
> +event for a vcpu from user space.
> +
>  ARM/ARM64:
>  ^^^^^^^^^^
>  
> diff --git a/arch/x86/include/uapi/asm/kvm.h b/arch/x86/include/uapi/asm/kvm.h
> index bf6e96011dfe..44757bd6122d 100644
> --- a/arch/x86/include/uapi/asm/kvm.h
> +++ b/arch/x86/include/uapi/asm/kvm.h
> @@ -325,6 +325,7 @@ struct kvm_reinject_control {
>  #define KVM_VCPUEVENT_VALID_SHADOW	0x00000004
>  #define KVM_VCPUEVENT_VALID_SMM		0x00000008
>  #define KVM_VCPUEVENT_VALID_PAYLOAD	0x00000010
> +#define KVM_VCPUEVENT_SHUTDOWN		0x00000020
>  
>  /* Interrupt shadow states */
>  #define KVM_X86_SHADOW_INT_MOV_SS	0x01
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 4fa4d8269e5b..53c8592066c8 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -4903,7 +4903,8 @@ static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu,
>  			      | KVM_VCPUEVENT_VALID_SIPI_VECTOR
>  			      | KVM_VCPUEVENT_VALID_SHADOW
>  			      | KVM_VCPUEVENT_VALID_SMM
> -			      | KVM_VCPUEVENT_VALID_PAYLOAD))
> +			      | KVM_VCPUEVENT_VALID_PAYLOAD
> +			      | KVM_VCPUEVENT_SHUTDOWN))
>  		return -EINVAL;
>  
>  	if (events->flags & KVM_VCPUEVENT_VALID_PAYLOAD) {
> @@ -4976,6 +4977,9 @@ static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu,
>  		}
>  	}
>  
> +	if (events->flags & KVM_VCPUEVENT_SHUTDOWN)
> +		kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);

Huh.  I think we need to make this bidirection and add it to get_vcpu_events()
as well, and treat it as a bug fix.  In direct triple fault cases, i.e. hardware
detected and morphed to VM-Exit, KVM will never lose the triple fault.  But for
triple faults sythesized by KVM, e.g. the RSM path or nested_vmx_abort(), if KVM
exits to userspace before the request is serviced, userspace could migrate the
VM and lose the triple fault.