Received-SPF: pass (google.com: domain of linux-kernel+bounces-40577-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: Fri, 26 Jan 2024 11:34:30 -0800
In-Reply-To: <CAL715WKMpui=+U56Qc5AiuLhUw_g-bjvtN5OmVz_hGdJmF1M5g@mail.gmail.com>
Precedence: bulk
Mime-Version: 1.0
References: <20240124003858.3954822-1-mizhang@google.com> <20240124003858.3954822-2-mizhang@google.com>
 <ZbExcMMl-IAzJrfx@google.com> <CAAAPnDFAvJBuETUsBScX6WqSbf_j=5h_CpWwrPHwXdBxDg_LFQ@mail.gmail.com>
 <ZbGAXpFUso9JzIjo@google.com> <ZbGOK9m6UKkQ38bK@google.com>
 <ZbGUfmn-ZAe4lkiN@google.com> <ZbGn8lAj4XxiecFn@google.com>
 <ZbP7BTvdZ1-b3MmE@google.com> <CAL715WKMpui=+U56Qc5AiuLhUw_g-bjvtN5OmVz_hGdJmF1M5g@mail.gmail.com>
Message-ID: <ZbQJRtaxhpZR7ntT@google.com>
Subject: Re: [PATCH 1/2] KVM: x86/pmu: Reset perf_capabilities in vcpu to 0 if
 PDCM is disabled
From: Sean Christopherson <seanjc@google.com>
To: Mingwei Zhang <mizhang@google.com>
Cc: Frederick Mayle <fmayle@google.com>, Steven Moreland <smoreland@google.com>, 
	Aaron Lewis <aaronlewis@google.com>, Paolo Bonzini <pbonzini@redhat.com>, 
	"H. Peter Anvin" <hpa@zytor.com>, kvm@vger.kernel.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Jan 26, 2024, Mingwei Zhang wrote:
> +Frederick Mayle +Steven Moreland
>=20
> On Fri, Jan 26, 2024 at 10:33=E2=80=AFAM Sean Christopherson <seanjc@goog=
le.com> wrote:
> >
> > On Thu, Jan 25, 2024, Mingwei Zhang wrote:
> > > On Wed, Jan 24, 2024, Sean Christopherson wrote:
> > > > On Wed, Jan 24, 2024, Mingwei Zhang wrote:
> > > > > I think this makes a lot of confusions on migration where VMM on =
the source
> > > > > believes that a non-zero value from KVM_GET_MSRS is valid and the=
 VMM on the
> > > > > target will find it not true.
> > > >
> > > > Yes, but seeing a non-zero value is a KVM bug that should be fixed.
> > > >
> > > How about adding an entry in vmx_get_msr() for
> > > MSR_IA32_PERF_CAPABILITIES and check pmu_version? This basically pair=
s
> > > with the implementation in vmx_set_msr() for MSR_IA32_PERF_CAPABILITI=
ES.
> > > Doing so allows KVM_GET_MSRS return 0 for the MSR instead of returnin=
g
> > > the initial permitted value.
> >
> > Hrm, I don't hate it as a stopgap.  But if we are the only people that =
are affected,
> > because again I'm pretty sure QEMU is fine, I would rather we just fix =
things in
> > our VMM and/or internal kernel.
>=20
> It is not just QEMU. crossvm is another open source VMM that suffers
> from this one.

Does CrosVM support migration or some other form of save/restore (RR?)?  An=
d if
so, does CrosVM do that in conjunction with hiding the vPMU from the guest?

Because if not, then I think we can squeak by.