Received-SPF: pass (google.com: domain of linux-kernel+bounces-131844-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Thu, 4 Apr 2024 09:45:27 -0700
In-Reply-To: <ZeqSncClqOQqCO41@yzhao56-desk.sh.intel.com>
Precedence: bulk
Mime-Version: 1.0
References: <20240228024147.41573-1-seanjc@google.com> <20240228024147.41573-2-seanjc@google.com>
 <ZeqSncClqOQqCO41@yzhao56-desk.sh.intel.com>
Message-ID: <Zg7ZJwQ0nOqwwmQI@google.com>
Subject: Re: [PATCH 01/16] KVM: x86/mmu: Exit to userspace with -EFAULT if
 private fault hits emulation
From: Sean Christopherson <seanjc@google.com>
To: Yan Zhao <yan.y.zhao@intel.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>, kvm@vger.kernel.org, linux-kernel@vger.kernel.org, 
	Isaku Yamahata <isaku.yamahata@intel.com>, Michael Roth <michael.roth@amd.com>, 
	Yu Zhang <yu.c.zhang@linux.intel.com>, Chao Peng <chao.p.peng@linux.intel.com>, 
	Fuad Tabba <tabba@google.com>, David Matlack <dmatlack@google.com>
Content-Type: text/plain; charset="us-ascii"

On Fri, Mar 08, 2024, Yan Zhao wrote:
> On Tue, Feb 27, 2024 at 06:41:32PM -0800, Sean Christopherson wrote:
> > @@ -320,6 +328,11 @@ static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
> >  	else
> >  		r = vcpu->arch.mmu->page_fault(vcpu, &fault);
> >  
> > +	if (r == RET_PF_EMULATE && fault.is_private) {
> Should we just check VM type + RET_PF_EMULATE, and abort?

No, the goal here is purely to ensure that emulation is never triggered for
private memory.  Guarding against attempting emulation for a VM type that doesn't
support emulation at all is something different.

And more concretely, as of this commit, all VM types that support private memory
(i.e. SW_PROTECTED_VM) support emulation, just not for private memory.

> If r is RET_PF_EMULATE, and fault is caused by accesing a shared address,
> the emulation code could still meet error if guest page table pages are in
> private memory, right?

Yes, which is why I squeezed in a documentation update for v6.8 to make it super
clear that SW_PROTECTED_VM is a development vehicle, i.e. that trying to use it
to run a real VM is all but guaranteed to cause explosions.