Nits on the shortlog...
Please don't capitalize x86, spell out "and" instead of using an ampersand (though
I think it can be omitted entirely), and since there are plenty of chars left, call
out that this is adding/extending KVM's ABI, e.g. it's not obvious from the shortlog
where/when the save+restore happens.
KVM: x86: Extend KVM_{G,S}ET_VCPU_EVENTS to support pending triple fault
On Thu, Apr 21, 2022, Chenyi Qiang wrote:
> For the triple fault 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.
>
> Add the support to save and restore the triple fault event from
> userspace. Introduce a new event KVM_VCPUEVENT_VALID_TRIPLE_FAULT in
> get/set_vcpu_events to track the triple fault request.
>
> Note that in the set_vcpu_events path, userspace is able to set/clear
> the triple fault request through triple_fault_pending field.
>
> Signed-off-by: Chenyi Qiang <chenyi.qiang@xxxxxxxxx>
> ---
> Documentation/virt/kvm/api.rst | 7 +++++++
> arch/x86/include/uapi/asm/kvm.h | 4 +++-
> arch/x86/kvm/x86.c | 15 +++++++++++++--
> 3 files changed, 23 insertions(+), 3 deletions(-)
>
> diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
> index 72183ae628f7..e09ce3cb49c5 100644
> --- a/Documentation/virt/kvm/api.rst
> +++ b/Documentation/virt/kvm/api.rst
> @@ -1150,6 +1150,9 @@ The following bits are defined in the flags field:
> fields contain a valid state. This bit will be set whenever
> KVM_CAP_EXCEPTION_PAYLOAD is enabled.
>
> +- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the
> + triple_fault_pending field contains a valid state.
> +
> ARM64:
> ^^^^^^
>
> @@ -1245,6 +1248,10 @@ 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_VALID_TRIPLE_FAULT can be set in flags field to signal that
> +the triple_fault_pending field contains a valid state and shall be written
> +into the VCPU.
> +
> ARM64:
> ^^^^^^
>
> diff --git a/arch/x86/include/uapi/asm/kvm.h b/arch/x86/include/uapi/asm/kvm.h
> index 21614807a2cb..fd083f6337af 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_VALID_TRIPLE_FAULT 0x00000020
>
> /* Interrupt shadow states */
> #define KVM_X86_SHADOW_INT_MOV_SS 0x01
> @@ -359,7 +360,8 @@ struct kvm_vcpu_events {
> __u8 smm_inside_nmi;
> __u8 latched_init;
> } smi;
> - __u8 reserved[27];
> + __u8 triple_fault_pending;
What about writing this as
struct {
__u8 pending;
} triple_fault;
to match the other events? It's kinda silly, but I find it easier to visually
identify the various events that are handled by kvm_vcpu_events.
> + __u8 reserved[26];
> __u8 exception_has_payload;
> __u64 exception_payload;
> };
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index ab336f7c82e4..c8b9b0bc42aa 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -4911,9 +4911,12 @@ static void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu,
> !!(vcpu->arch.hflags & HF_SMM_INSIDE_NMI_MASK);
> events->smi.latched_init = kvm_lapic_latched_init(vcpu);
>
> + events->triple_fault_pending = kvm_test_request(KVM_REQ_TRIPLE_FAULT, vcpu);
> +
> events->flags = (KVM_VCPUEVENT_VALID_NMI_PENDING
> | KVM_VCPUEVENT_VALID_SHADOW
> - | KVM_VCPUEVENT_VALID_SMM);
> + | KVM_VCPUEVENT_VALID_SMM
> + | KVM_VCPUEVENT_VALID_TRIPLE_FAULT);
Does setting KVM_VCPUEVENT_VALID_TRIPLE_FAULT need to be guarded with a capability,
a la KVM_CAP_EXCEPTION_PAYLOAD, so that migrating from a new KVM to an old KVM doesn't
fail? Seems rather pointless since the VM is likely hosed either way...
> if (vcpu->kvm->arch.exception_payload_enabled)
> events->flags |= KVM_VCPUEVENT_VALID_PAYLOAD;
