Re: [PATCH v3] KVM: VMX: Execute WBINVD to keep data consistency with assigned devices

[Avi Kivity <avi@redhat.com>]

On 06/28/2010 06:36 AM, Sheng Yang wrote:
> Some guest device driver may leverage the "Non-Snoop" I/O, and explicitly
> WBINVD or CLFLUSH to a RAM space. Since migration may occur before WBINVD or
> CLFLUSH, we need to maintain data consistency either by:
>    

Don't we always force enable snooping?  Or is that only for the 
processor, and you're worried about devices?

> 1: flushing cache (wbinvd) when the guest is scheduled out if there is no
> wbinvd exit, or
> 2: execute wbinvd on all dirty physical CPUs when guest wbinvd exits.
>
>
>    


>   	/* fields used by HYPER-V emulation */
>   	u64 hv_vapic;
> +
> +	cpumask_t wbinvd_dirty_mask;
>   };
>
>    

Need alloc_cpumask_var()/free_cpumask_var() for very large hosts.

>
> +static void wbinvd_ipi(void *garbage)
> +{
> +	wbinvd();
> +}
>    

Like Jan mentioned, this is quite heavy.  What about a clflush() loop 
instead?  That may take more time, but at least it's preemptible.  Of 
course, it isn't preemptible in an IPI.

> +
>   void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
>   {
> +	/* Address WBINVD may be executed by guest */
> +	if (vcpu->kvm->arch.iommu_domain) {
> +		if (kvm_x86_ops->has_wbinvd_exit())
> +			cpu_set(cpu, vcpu->arch.wbinvd_dirty_mask);
> +		else if (vcpu->cpu != -1)
> +			smp_call_function_single(vcpu->cpu,
> +					wbinvd_ipi, NULL, 1);
>    

Is there any point to doing this if !has_wbinvd_exit()?  The vcpu might 
not have migrated in time, so the cache is flushed too late.

> +	}
> +
>   	kvm_x86_ops->vcpu_load(vcpu, cpu);
>   	if (unlikely(per_cpu(cpu_tsc_khz, cpu) == 0)) {
>   		unsigned long khz = cpufreq_quick_get(cpu);
> @@ -3650,6 +3664,21 @@ int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address)
>   	return X86EMUL_CONTINUE;
>   }
>
> +int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu)
> +{
> +	if (!vcpu->kvm->arch.iommu_domain)
> +		return X86EMUL_CONTINUE;
> +
> +	if (kvm_x86_ops->has_wbinvd_exit()) {
> +		smp_call_function_many(&vcpu->arch.wbinvd_dirty_mask,
> +				wbinvd_ipi, NULL, 1);
> +		cpus_clear(vcpu->arch.wbinvd_dirty_mask);
>    

Race - a migration may set a new bit in wbinvd_dirty_mask after the 
s_c_f_m().

However, it's probably benign, since we won't be entering the guest in 
that period.

> +	} else
> +		wbinvd();
> +	return X86EMUL_CONTINUE;
> +}
> +EXPORT_SYMBOL_GPL(kvm_emulate_wbinvd);
> +
>   int emulate_clts(struct kvm_vcpu *vcpu)
>   {
>   	kvm_x86_ops->set_cr0(vcpu, kvm_read_cr0_bits(vcpu, ~X86_CR0_TS));
>    


-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.