Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230704075054.3344915-1-stevensd@google.com> <20230704075054.3344915-6-stevensd@google.com>
 <20230705101800.ut4c6topn6ylwczs@linux.intel.com> <CAD=HUj41PAKC0x+c3zWAr-aCm59K7hs2zRh1uWs9778_Mai4UA@mail.gmail.com>
 <20230706155805.GD3894444@ls.amr.corp.intel.com> <CAD=HUj6GiK3TSSe7UY8C2Jd+3tjZNBa-TLgk-UodyL=E+qKavg@mail.gmail.com>
 <20230710163448.GE3894444@ls.amr.corp.intel.com>
In-Reply-To: <20230710163448.GE3894444@ls.amr.corp.intel.com>
From:   David Stevens <stevensd@chromium.org>
Date:   Tue, 11 Jul 2023 11:59:08 +0900
Message-ID: <CAD=HUj6ZEzVEFgn_J_9EPNMj5i-N=MSc0xZF8FweqrgTxUks0g@mail.gmail.com>
Subject: Re: [PATCH v7 5/8] KVM: x86/mmu: Don't pass FOLL_GET to __kvm_follow_pfn
To:     Isaku Yamahata <isaku.yamahata@gmail.com>
Cc:     Yu Zhang <yu.c.zhang@linux.intel.com>,
        Sean Christopherson <seanjc@google.com>,
        Marc Zyngier <maz@kernel.org>,
        Michael Ellerman <mpe@ellerman.id.au>,
        Peter Xu <peterx@redhat.com>,
        linux-arm-kernel@lists.infradead.org, kvmarm@lists.linux.dev,
        linux-kernel@vger.kernel.org, linuxppc-dev@lists.ozlabs.org,
        kvm@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Jul 11, 2023 at 1:34=E2=80=AFAM Isaku Yamahata <isaku.yamahata@gmai=
l.com> wrote:
>
> On Fri, Jul 07, 2023 at 10:35:02AM +0900,
> David Stevens <stevensd@chromium.org> wrote:
>
> > > > > > diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> > > > > > index e44ab512c3a1..b1607e314497 100644
> > > > > > --- a/arch/x86/kvm/mmu/mmu.c
> > > > > > +++ b/arch/x86/kvm/mmu/mmu.c
> > > > >
> > > > > ...
> > > > >
> > > > > > @@ -2937,6 +2943,7 @@ static int mmu_set_spte(struct kvm_vcpu *=
vcpu, struct kvm_memory_slot *slot,
> > > > > >       bool host_writable =3D !fault || fault->map_writable;
> > > > > >       bool prefetch =3D !fault || fault->prefetch;
> > > > > >       bool write_fault =3D fault && fault->write;
> > > > > > +     bool is_refcounted =3D !fault || fault->is_refcounted_pag=
e;
> > > > >
> > > > > Just wonder, what if a non-refcounted page is prefetched?  Or is =
it possible in
> > > > > practice?
> > > >
> > > > Prefetching is still done via gfn_to_page_many_atomic, which sets
> > > > FOLL_GET. That's fixable, but it's not something this series curren=
tly
> > > > does.
> > >
> > > So if we prefetch a page, REFCOUNTED bit is cleared unconditionally w=
ith this
> > > hunk.  kvm_set_page_{dirty, accessed} won't be called as expected for=
 prefetched
> > > spte.  If I read the patch correctly, REFCOUNTED bit in SPTE should r=
epresent
> > > whether the corresponding page is ref-countable or not, right?
> > >
> > > Because direct_pte_prefetch_many() is for legacy KVM MMU and FNAME(pr=
efetch_pte)
> > > is shadow paging, we need to test it with legacy KVM MMU or shadow pa=
ging to hit
> > > the issue, though.
> > >
> >
> > direct_pte_prefetch_many and prefetch_gpte both pass NULL for the
> > fault parameter, so is_refcounted will evaluate to true. So the spte's
> > refcounted bit will get set in that case.
>
> Oops, my bad.  My point is "unconditionally".  Is the bit always set for
> non-refcountable pages?  Or non-refcountable pages are not prefeched?

The bit is never set for non-refcounted pages, and is always set for
refcounted pages. The current series never prefetches non-refcounted
pages, since it continues to use the gfn_to_page_many_atomic API.

-David