Re: [PATCH v2 1/2] x86/cpu, kvm: Use CPU capabilities for CPUID[0x80000021].EAX

[Borislav Petkov]

On Wed, Nov 23, 2022 at 06:04:48PM -0600, Kim Phillips wrote:
> The AMD Zen4 Automatic IBRS feature bit resides in the 0x80000021 leaf,
> for which there is already support for exposing Zen3 bits to the guest.
> 
> Add AMD AutoIBRS feature bit support, including for the other bits,
> using scattered/synthetic bits.
> 
> Add the corresponding word to KVM's feature machinery so that AutoIBRS
> gets advertized into the guest too.
> 
> Co-developed-by: Babu Moger <Babu.Moger@xxxxxxx>

verify_tags: WARNING: Co-developed-by Babu Moger <Babu.Moger@xxxxxxx> hasn't signed off on the patch!

> Co-developed-by: Borislav Petkov <bp@xxxxxxx>
> Signed-off-by: Kim Phillips <kim.phillips@xxxxxxx>

...

> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> index c92c49a0b35b..61cd33a848cc 100644
> --- a/arch/x86/kvm/cpuid.c
> +++ b/arch/x86/kvm/cpuid.c
> @@ -730,6 +730,25 @@ void kvm_set_cpu_caps(void)
>  		0 /* SME */ | F(SEV) | 0 /* VM_PAGE_FLUSH */ | F(SEV_ES) |
>  		F(SME_COHERENT));
>  
> +	/*
> +	 * Pass down these bits:
> +	 *    EAX      0      NNDBP, Processor ignores nested data breakpoints
> +	 *    EAX      2      LAS, LFENCE always serializing
> +	 *    EAX      6      NSCB, Null selector clear base
> +	 *    EAX      8      Automatic IBRS
> +	 *
> +	 * Other defined bits are for MSRs that KVM does not expose:
> +	 *   EAX      3      SPCL, SMM page configuration lock
> +	 *   EAX      13     PCMSR, Prefetch control MSR
> +	 */
> +	kvm_cpu_cap_init_scattered(CPUID_8000_0021_EAX,
> +				   SF(NO_NESTED_DATA_BP) | SF(LFENCE_RDTSC) |
> +				   SF(NULL_SEL_CLR_BASE) | SF(AUTOIBRS));
> +	if (static_cpu_has(X86_FEATURE_LFENCE_RDTSC))
> +		kvm_cpu_cap_set(X86_FEATURE_LFENCE_RDTSC);
> +	if (!static_cpu_has_bug(X86_BUG_NULL_SEG))
> +		kvm_cpu_cap_set(X86_FEATURE_NULL_SEL_CLR_BASE);

So this looks backwards:

if X86_FEATURE_NULL_SEL_CLR_BASE is set, then X86_BUG_NULL_SEG should
not be.

Which means, you'd have to update check_null_seg_clears_base() too.

Which means, you should make the X86_FEATURE_NULL_SEL_CLR_BASE bit
addition a separate patch because this one is clearly doing too many
things at once.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette