Received-SPF: pass (google.com: domain of linux-kernel+bounces-131836-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Date: Thu, 4 Apr 2024 09:38:01 -0700
In-Reply-To: <6d24f7b1-a271-43de-ba94-847a3c9e0d8c@linux.intel.com>
Precedence: bulk
Mime-Version: 1.0
References: <20240228024147.41573-1-seanjc@google.com> <20240228024147.41573-2-seanjc@google.com>
 <2441bd2a-a0b1-5184-bac8-28c94b151c93@amd.com> <6d24f7b1-a271-43de-ba94-847a3c9e0d8c@linux.intel.com>
Message-ID: <Zg7XaYwmFbZUljfi@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: Binbin Wu <binbin.wu@linux.intel.com>
Cc: Pankaj Gupta <pankaj.gupta@amd.com>, Paolo Bonzini <pbonzini@redhat.com>, kvm@vger.kernel.org, 
	linux-kernel@vger.kernel.org, Yan Zhao <yan.y.zhao@intel.com>, 
	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="utf-8"
Content-Transfer-Encoding: quoted-printable

On Tue, Mar 12, 2024, Binbin Wu wrote:
>=20
> On 3/7/2024 8:52 PM, Gupta, Pankaj wrote:
> > > Exit to userspace with -EFAULT / KVM_EXIT_MEMORY_FAULT if a private f=
ault
> > > triggers emulation of any kind, as KVM doesn't currently support
> > > emulating access to guest private memory.=C2=A0 Practically speaking,=
 private
> > > faults and emulation are already mutually exclusive, but there are ed=
ge
> > > cases upon edge cases where KVM can return RET_PF_EMULATE, and adding=
 one
> > > last check
> >=20
> > edge cases upon edge cases?
> >=20
> > Just curious about the details of the edge cases scenarios?
>=20
> Same question out of curiosity.

Accesses that hit the APIC-access page and gfns that are write-tracked, are=
 the
two most likely candidates.  Even plain old emulated MMIO falls into this b=
ucket,
e.g. if KVM botched things and generated a RSVD fault on a private mapping.=
  I'll
reword that line to

  faults and emulation are already mutually exclusive, but there are many
  flows that can result in KVM returning RET_PF_EMULATE, and adding one las=
t
  check to harden against weird, unexpected combinations and/or KVM bugs is
  inexpensive.

to make it sound less dramatic/hand-wavy.