Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] KVM: X86: Fix async pf caused null-ptr-deref
To:     Vitaly Kuznetsov <vkuznets@redhat.com>,
        Wanpeng Li <kernellwp@gmail.com>
Cc:     Sean Christopherson <sean.j.christopherson@intel.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>,
        Vivek Goyal <vgoyal@redhat.com>, linux-kernel@vger.kernel.org,
        kvm@vger.kernel.org
References: <1593426391-8231-1-git-send-email-wanpengli@tencent.com>
 <877dvqc7cs.fsf@vitty.brq.redhat.com>
From:   Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <f9b06428-51c3-09af-48cc-d378182916fd@redhat.com>
Date:   Mon, 29 Jun 2020 15:59:25 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.6.0
MIME-Version: 1.0
In-Reply-To: <877dvqc7cs.fsf@vitty.brq.redhat.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 29/06/20 15:46, Vitaly Kuznetsov wrote:
>> +	if (!lapic_in_kernel(vcpu))
>> +		return 1;
>> +
> I'm not sure how much we care about !lapic_in_kernel() case but this
> change should be accompanied with userspace changes to not expose
> KVM_FEATURE_ASYNC_PF_INT or how would the guest know that writing a
> legitimate value will result in #GP?

Almost any pv feature is broken with QEMU if kernel_irqchip=off.  I
wouldn't bother and I am seriously thinking of dropping all support for
that, including:

- just injecting #UD for MOV from/to CR8 unless lapic_in_kernel()

- make KVM_INTERRUPT fail unless irqchip_in_kernel(), so that
KVM_INTERRUPT is only used to inject EXTINT with kernel_irqchip=split

Paolo

> Alternatively, we may just return '0' here: guest will be able to check
> what's in the MSR to see if the feature was enabled. Normally, guests
> shouldn't care about this but maybe there are cases when they do?
>