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;
Date:   Mon, 8 Mar 2021 12:43:48 -0800
From:   Sean Christopherson <seanjc@google.com>
To:     Paolo Bonzini <pbonzini@redhat.com>
Cc:     linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
        vkuznets@redhat.com, mlevitsk@redhat.com,
        Jim Mattson <jmattson@google.com>
Subject: Re: [PATCH 03/28] KVM: nSVM: inject exceptions via
 svm_check_nested_events
Message-ID: <YEaMhHG7ylvTpoYD@google.com>
References: <YELdblXaKBTQ4LGf@google.com>
 <fc2b0085-eb0f-dbab-28c2-a244916c655f@redhat.com>
 <YEZUhbBtNjWh0Zka@google.com>
 <006be822-697e-56d5-84a7-fa51f5087a34@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <006be822-697e-56d5-84a7-fa51f5087a34@redhat.com>
Precedence: bulk

On Mon, Mar 08, 2021, Paolo Bonzini wrote:
> On 08/03/21 17:44, Sean Christopherson wrote:
> > VMCALL is also probably ok
> > in most scenarios, but patching L2's code from L0 KVM is sketchy.
> 
> I agree that patching is sketchy and I'll send a patch.  However...
> 
> > > The same is true for the VMware #GP interception case.
> > 
> > I highly doubt that will ever work out as intended for the modified IO #GP
> > behavior.  The only way emulating #GP in L2 is correct if L1 wants to pass
> > through the capabilities to L2, i.e. the I/O access isn't intercepted by L1.
> > That seems unlikely.
> 
> ... not all hypervisors trap everything.  In particular in this case the
> VMCS12 I/O permission bitmap should be consulted (which we do in
> vmx_check_intercept_io), but if the I/O is not trapped by L1 it should
> bypass the IOPL and TSS-bitmap checks in my opinion.

I agree, _if_ it's not trapped.  But bypassing the checks when it is trapped is
clearly wrong.