Re: [PATCH v2] x86/sgx: Allow exposing EDECCSSA user leaf function to KVM guest

["Huang, Kai" <kai.huang@intel.com>]

On Thu, 2022-08-25 at 15:19 +0000, Sean Christopherson wrote:
> On Thu, Aug 25, 2022, Jarkko Sakkinen wrote:
> > Nit: shouldn't be this be x86/kvm?
> 
> Heh, no, because x86/kvm is the scope for Linux running as a KVM guest, i.e. for
> changes to arch/x86/kernel/kvm.c.
> 
> But yeah, "KVM: x86:" or maybe even "KVM: VMX:" would be preferable given that all
> of the meaningful changes are KVM specific.
> 
> > On Thu, Aug 18, 2022 at 02:38:29PM +1200, Kai Huang wrote:
> > > The new Asynchronous Exit (AEX) notification mechanism (AEX-notify)
> > > allows one enclave to receive a notification in the ERESUME after the
> > > enclave exit due to an AEX.  EDECCSSA is a new SGX user leaf function
> > > (ENCLU[EDECCSSA]) to facilitate the AEX notification handling.  The new
> > > EDECCSSA is enumerated via CPUID(EAX=0x12,ECX=0x0):EAX[11].
> > > 
> > > Besides Allowing reporting the new AEX-notify attribute to KVM guests,
> > > also allow reporting the new EDECCSSA user leaf function to KVM guests
> > > so the guest can fully utilize the AEX-notify mechanism.
> > > 
> > > Similar to existing X86_FEATURE_SGX1 and X86_FEATURE_SGX2, introduce a
> > > new scattered X86_FEATURE_SGX_EDECCSSA bit for the new EDECCSSA, and
> > > report it in KVM's supported CPUIDs so the userspace hypervisor (i.e.
> > > Qemu) can enable it for the guest.
> 
> Silly nit, but I'd prefer to leave off the "so the userspace hypervisor ... can
> enable it for the guest".  Userspace doesn't actually need to wait for KVM enabling.
> As noted below, KVM doesn't need to do anything extra, and KVM _can't_ prevent the
> guest from using EDECCSSA.

Indeed KVM cannot prevent.

> 
> > > Note there's no additional enabling work required to allow guest to use
> > > the new EDECCSSA.  KVM is not able to trap ENCLU anyway.
> 
> And maybe call out that the KVM "enabling" is not strictly necessary?  And note
> that there's a virtualization hole?  E.g.
> 
>   Note, no additional KVM enabling is required to allow the guest to use
>   EDECCSSA, it's impossible to trap ENCLU (without completely preventing the
>   guest from using SGX).  Advertise EDECCSSA as supported purely so that
>   userspace doesn't need to special case EDECCSSA, i.e. doesn't need to
>   manually check host CPUID.
> 
>   The inability to trap ENCLU also means that KVM can't prevent the guest
>   from using EDECCSSA, but that virtualization hole is benign as far as KVM
>   is concerned.  EDECCSSA is simply a fancy way to modify internal enclave
>   state.

Thanks.  Will use above.


-- 
Thanks,
-Kai