Re: [PATCH v11 21/21] KVM: selftests: Add TDX lifecycle test

[Ira Weiny <ira.weiny@intel.com>]

Sean Christopherson wrote:
> On Thu, Sep 25, 2025, Sagi Shahar wrote:
> > diff --git a/tools/testing/selftests/kvm/lib/x86/tdx/tdx_util.c b/tools/testing/selftests/kvm/lib/x86/tdx/tdx_util.c
> > index 53cfadeff8de..714413e062fd 100644
> > --- a/tools/testing/selftests/kvm/lib/x86/tdx/tdx_util.c
> > +++ b/tools/testing/selftests/kvm/lib/x86/tdx/tdx_util.c
> > @@ -328,3 +328,21 @@ void vm_tdx_finalize(struct kvm_vm *vm)
> >  	load_td_private_memory(vm);
> >  	vm_tdx_vm_ioctl(vm, KVM_TDX_FINALIZE_VM, 0, NULL);
> >  }
> > +
> > +struct kvm_vm *vm_tdx_create_with_one_vcpu(void *guest_code,
> > +					   struct kvm_vcpu **vcpu)
> > +{
> > +	struct vm_shape shape = {
> > +		.mode = VM_MODE_DEFAULT,
> > +		.type = KVM_X86_TDX_VM,
> > +	};
> > +	struct kvm_vm *vm;
> > +	struct kvm_vcpu *vcpus[1];
> > +
> > +	vm = __vm_create_with_vcpus(shape, 1, 0, guest_code, vcpus);
> > +	*vcpu = vcpus[0];
> > +
> > +	vm_tdx_finalize(vm);
> > +
> > +	return vm;
> > +}
> 
> Rather than add a full wrapper, and duplicate all of vm_sev_create_with_one_vcpu(),
> we should just add macros to convert a type to a shape.
> 
> E.g. with this, you can simply add:
> 
>   #define VM_SHAPE_TDX	VM_TYPE(KVM_X86_TDX_VM)
> 
> And coupled with Ira's suggestion regarding vm_tdx_finalize(), there should be
> no need for vm_tdx_create_with_one_vcpu().

All sounds reasonable to me but some questions/comments below.


[snip]

> diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h
> index af52cd938b50..af0b53987c06 100644
> --- a/tools/testing/selftests/kvm/include/kvm_util.h
> +++ b/tools/testing/selftests/kvm/include/kvm_util.h
> @@ -210,6 +210,20 @@ kvm_static_assert(sizeof(struct vm_shape) == sizeof(uint64_t));
>  	shape;					\
>  })
>  
> +#define __VM_TYPE(__mode, __type)		\
> +({						\
> +	struct vm_shape shape = {		\
> +		.mode = (__mode),		\
> +		.type = (__type)		\
> +	};					\
> +						\
> +	shape;					\
> +})
> +
> +#define VM_TYPE(__type)				\
> +	__VM_TYPE(VM_MODE_DEFAULT, __type)

We already have VM_SHAPE()?  Why do we need this as well?

> +
> +
>  #if defined(__aarch64__)
>  
>  extern enum vm_guest_mode vm_mode_default;
> diff --git a/tools/testing/selftests/kvm/include/x86/processor.h b/tools/testing/selftests/kvm/include/x86/processor.h
> index 51cd84b9ca66..dd21e11e1908 100644
> --- a/tools/testing/selftests/kvm/include/x86/processor.h
> +++ b/tools/testing/selftests/kvm/include/x86/processor.h
> @@ -362,6 +362,10 @@ static inline unsigned int x86_model(unsigned int eax)
>  	return ((eax >> 12) & 0xf0) | ((eax >> 4) & 0x0f);
>  }
>  
> +#define VM_SHAPE_SEV		VM_TYPE(KVM_X86_SEV_VM)
> +#define VM_SHAPE_SEV_ES		VM_TYPE(KVM_X86_SEV_ES_VM)
> +#define VM_SHAPE_SNP		VM_TYPE(KVM_X86_SNP_VM)

FWIW I think the SEV bits should be pulled apart from the TDX bits and the
TDX bits squashed back into this series with the SEV as a per-cursor patch.

Ira

[snip]