[MODERATED] Re: [patch V10 00/10] Control knobs and Documentation 0

[Paolo Bonzini <pbonzini@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1665 bytes --]

On 13/07/2018 19:28, speck for Konrad Rzeszutek Wilk wrote:
> Perhaps:
> 
> diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
> index 0e75170..f03ec33 100644
> --- a/arch/x86/include/asm/msr-index.h
> +++ b/arch/x86/include/asm/msr-index.h
> @@ -70,6 +70,7 @@
>  #define MSR_IA32_ARCH_CAPABILITIES	0x0000010a
>  #define ARCH_CAP_RDCL_NO		(1 << 0)   /* Not susceptible to Meltdown */
>  #define ARCH_CAP_IBRS_ALL		(1 << 1)   /* Enhanced IBRS support */
> +#define ARCH_CAP_SKIP_L1DFL_VMENTRY	(1 << 3)   /* Skip L1DF on VMENTRY */

If this bit is set, KVM is effectively not vulnerable.  I just sent a
more complete follow-up, replacing what I had cooked up last Friday.

Thanks,

Paolo

>  #define ARCH_CAP_SSB_NO			(1 << 4)   /*
>  						    * Not susceptible to Speculative Store Bypass
>  						    * attack, so no Speculative Store Bypass
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index c5c0118..5209252 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -216,6 +216,15 @@ static int vmx_setup_l1d_flush(enum vmx_l1d_flush_state l1tf)
>  		l1tf_vmx_mitigation = VMENTER_L1D_FLUSH_EPT_DISABLED;
>  		return 0;
>  	}
> +	if (static_cpu_has(X86_FEATURE_HYPERVISOR) &&
> +	    static_cpu_has(X86_FEATURE_FLUSH_L1D) &&
> +	    boot_cpu_has(X86_FEATURE_ARCH_CAPABILITIES)) {
> +		u64 msr;
> +
> +		rdmsrl(MSR_IA32_ARCH_CAPABILITIES, msr);
> +		if (msr & ARCH_CAP_SKIP_L1DFL_VMENTRY)
> +			l1tf_vmx_mitigation = VMENTER_L1D_FLUSH_NESTED_VM;
> +	}
>  
>  	/* If set to auto use the default l1tf mitigation method */
>  	if (l1tf == VMENTER_L1D_FLUSH_AUTO) {
>