Re: [PATCH v2 10/50] KVM: VMX: Reset eVMCS controls in VP assist page during hardware disabling

[Vitaly Kuznetsov <vkuznets@redhat.com>]

Sean Christopherson <seanjc@google.com> writes:

> Reset the eVMCS controls in the per-CPU VP assist page during hardware
> disabling instead of waiting until kvm-intel's module exit.  The controls
> are activated if and only if KVM creates a VM, i.e. don't need to be
> reset if hardware is never enabled.
>
> Doing the reset during hardware disabling will naturally fix a potential
> NULL pointer deref bug once KVM disables CPU hotplug while enabling and
> disabling hardware (which is necessary to fix a variety of bugs).  If the
> kernel is running as the root partition, the VP assist page is unmapped
> during CPU hot unplug, and so KVM's clearing of the eVMCS controls needs
> to occur with CPU hot(un)plug disabled, otherwise KVM could attempt to
> write to a CPU's VP assist page after it's unmapped.
>
> Reported-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> Signed-off-by: Sean Christopherson <seanjc@google.com>
> ---
>  arch/x86/kvm/vmx/vmx.c | 50 +++++++++++++++++++++++++-----------------
>  1 file changed, 30 insertions(+), 20 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index cea8c07f5229..d85d175dca70 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -551,6 +551,33 @@ static int hv_enable_l2_tlb_flush(struct kvm_vcpu *vcpu)
>  	return 0;
>  }
>  
> +static void hv_reset_evmcs(void)
> +{
> +	struct hv_vp_assist_page *vp_ap;
> +
> +	if (!static_branch_unlikely(&enable_evmcs))
> +		return;
> +
> +	/*
> +	 * KVM should enable eVMCS if and only if all CPUs have a VP assist
> +	 * page, and should reject CPU onlining if eVMCS is enabled the CPU
> +	 * doesn't have a VP assist page allocated.
> +	 */
> +	vp_ap = hv_get_vp_assist_page(smp_processor_id());
> +	if (WARN_ON_ONCE(!vp_ap))
> +		return;
> +

In case my understanding is correct, this may actually get triggered
for Hyper-V root partition: vmx_hardware_disable() gets called from
kvm_dying_cpu() which has its own CPUHP_AP_KVM_STARTING stage. VP page
unmapping happens in hv_cpu_die() which uses generic CPUHP_AP_ONLINE_DYN
(happens first on CPU oflining AFAIR). I believe we need to introduce a
new CPUHP_AP_HYPERV_STARTING stage and put it before
CPUHP_AP_KVM_STARTING so it happens after it upon offlining.

The issue is likely theoretical as Hyper-V root partition is a very
special case, I'm not sure whether KVM is used there and whether CPU
offlining is possible. In any case, WARN_ON_ONCE() is much better than
NULL pointer dereference we have now :-)

> +	/*
> +	 * Reset everything to support using non-enlightened VMCS access later
> +	 * (e.g. when we reload the module with enlightened_vmcs=0)
> +	 */
> +	vp_ap->nested_control.features.directhypercall = 0;
> +	vp_ap->current_nested_vmcs = 0;
> +	vp_ap->enlighten_vmentry = 0;
> +}
> +
> +#else /* IS_ENABLED(CONFIG_HYPERV) */
> +static void hv_reset_evmcs(void) {}
>  #endif /* IS_ENABLED(CONFIG_HYPERV) */
>  
>  /*
> @@ -2496,6 +2523,8 @@ static void vmx_hardware_disable(void)
>  	if (cpu_vmxoff())
>  		kvm_spurious_fault();
>  
> +	hv_reset_evmcs();
> +
>  	intel_pt_handle_vmx(0);
>  }
>  
> @@ -8462,27 +8491,8 @@ static void vmx_exit(void)
>  	kvm_exit();
>  
>  #if IS_ENABLED(CONFIG_HYPERV)
> -	if (static_branch_unlikely(&enable_evmcs)) {
> -		int cpu;
> -		struct hv_vp_assist_page *vp_ap;
> -		/*
> -		 * Reset everything to support using non-enlightened VMCS
> -		 * access later (e.g. when we reload the module with
> -		 * enlightened_vmcs=0)
> -		 */
> -		for_each_online_cpu(cpu) {
> -			vp_ap =	hv_get_vp_assist_page(cpu);
> -
> -			if (!vp_ap)
> -				continue;
> -
> -			vp_ap->nested_control.features.directhypercall = 0;
> -			vp_ap->current_nested_vmcs = 0;
> -			vp_ap->enlighten_vmentry = 0;
> -		}
> -
> +	if (static_branch_unlikely(&enable_evmcs))
>  		static_branch_disable(&enable_evmcs);
> -	}
>  #endif
>  	vmx_cleanup_l1d_flush();

Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>

-- 
Vitaly