Re: [PATCH v10 08/11] x86/tdx: Wire up KVM hypercalls

[Thomas Gleixner <tglx@linutronix.de>]

On Fri, Oct 08 2021 at 22:37, Kuppuswamy Sathyanarayanan wrote:
> From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
>  
>  #ifdef CONFIG_KVM_GUEST
> @@ -32,6 +34,10 @@ static inline bool kvm_check_and_clear_guest_paused(void)
>  static inline long kvm_hypercall0(unsigned int nr)
>  {
>  	long ret;
> +
> +	if (cc_platform_has(CC_ATTR_GUEST_TDX))
> +		return tdx_kvm_hypercall(nr, 0, 0, 0, 0);

So if TDX is not enabled in Kconfig this cannot be optimized out unless
CC_PLATFORM is disabled as well. But what's worse is that every
hypercall needs to call into cc_platform_has().

None of the hypercalls is used before the early TDX detection. So we can
simply use

       if (cpu_feature_enabled(X86_FEATURE_TDX_GUEST))

here, right? Then you add X86_FEATURE_TDX_GUEST to the disabled feature
bits correctly and all of the above is solved.

Hmm?
 
> +#if defined(CONFIG_KVM_GUEST) && defined(CONFIG_INTEL_TDX_GUEST)
> +static inline long tdx_kvm_hypercall(unsigned int nr, unsigned long p1,
> +				     unsigned long p2, unsigned long p3,
> +				     unsigned long p4)
> +{
> +	struct tdx_hypercall_output out;
> +	u64 err;
> +
> +	err = __tdx_hypercall(TDX_HYPERCALL_VENDOR_KVM, nr, p1, p2,
> +			      p3, p4, &out);
> +
> +	/*
> +	 * Non zero return value means buggy TDX module (which is fatal).
> +	 * So use BUG_ON() to panic.
> +	 */
> +	BUG_ON(err);
> +
> +	return out.r10;
> +}

Can we make that a proper exported function (instead of
tdx_kvm_hypercall) so we don't end up with the very same code inlined
all over the place?

Thanks,

        tglx