Re: [patch 12/13] KVM: x86: trap invlpg

[Avi Kivity <avi@qumranet.com>]

Marcelo Tosatti wrote:
> With pages out of sync invlpg needs to be trapped. For now simply nuke
> the entry.
>
>   

We could kvm_mmu_pte_write() it, with some modifications.  Probably 
worth it for the guest breaking cow.


>  	[EXIT_REASON_MSR_WRITE]               = handle_wrmsr,
>  	[EXIT_REASON_PENDING_INTERRUPT]       = handle_interrupt_window,
>  	[EXIT_REASON_HLT]                     = handle_halt,
> +	[EXIT_REASON_INVLPG]		      = handle_invlpg,
>  	[EXIT_REASON_VMCALL]                  = handle_vmcall,
>  	[EXIT_REASON_TPR_BELOW_THRESHOLD]     = handle_tpr_below_threshold,
>  	[EXIT_REASON_APIC_ACCESS]             = handle_apic_access,

We shouldn't do that if ept or npt is enabled.

> Index: kvm/arch/x86/kvm/mmu.c
> ===================================================================
> --- kvm.orig/arch/x86/kvm/mmu.c
> +++ kvm/arch/x86/kvm/mmu.c
> @@ -889,6 +889,12 @@ static int nonpaging_sync_page(struct kv
>  	return 1;
>  }
>  
> +static void nonpaging_invlpg(struct kvm_vcpu *vcpu, gva_t gva)
> +{
> +	/* should never happen */
> +	WARN_ON(1);
> +}
>   

Nevertheless, invlpg is legal in real mode.

-- 
error compiling committee.c: too many arguments to function