Received-SPF: pass (google.com: domain of linux-kernel+bounces-153745-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Message-ID: <0bb540291542fd37b2d0f6d39ebf1c9f35fa1ac3.camel@infradead.org>
Subject: Re: [PATCH 02/10] KVM: x86: Improve accuracy of KVM clock when TSC
 scaling is in force
From: David Woodhouse <dwmw2@infradead.org>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: kvm@vger.kernel.org, Jonathan Corbet <corbet@lwn.net>, Sean
 Christopherson <seanjc@google.com>, Thomas Gleixner <tglx@linutronix.de>,
 Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>, Dave Hansen
 <dave.hansen@linux.intel.com>, x86@kernel.org, "H. Peter Anvin"
 <hpa@zytor.com>,  Paul Durrant <paul@xen.org>, Shuah Khan
 <shuah@kernel.org>, linux-doc@vger.kernel.org, 
 linux-kernel@vger.kernel.org, linux-kselftest@vger.kernel.org, Oliver Upton
 <oliver.upton@linux.dev>, Marcelo Tosatti <mtosatti@redhat.com>, 
 jalliste@amazon.co.uk, sveith@amazon.de
Date: Mon, 22 Apr 2024 17:44:31 +0100
In-Reply-To: <CABgObfbK0aqqmAz7Z2efX4hNf7WRBYpoJ1a07oKMZdFXS2r0+g@mail.gmail.com>
References: <20240418193528.41780-1-dwmw2@infradead.org>
	 <20240418193528.41780-3-dwmw2@infradead.org>
	 <CABgObfa0j34iEh81hhd7-t7ZM1GKAsvJb5xP6EoD2-c-8TnPqQ@mail.gmail.com>
	 <a35e69e07b9cd297dac9993c886667add144e833.camel@infradead.org>
	 <CABgObfbK0aqqmAz7Z2efX4hNf7WRBYpoJ1a07oKMZdFXS2r0+g@mail.gmail.com>
Content-Type: multipart/signed; micalg="sha-256"; protocol="application/pkcs7-signature";
	boundary="=-kMnqTVDkh4tbIOg2M2ET"
User-Agent: Evolution 3.44.4-0ubuntu2 
Precedence: bulk
MIME-Version: 1.0


--=-kMnqTVDkh4tbIOg2M2ET
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Mon, 2024-04-22 at 17:54 +0200, Paolo Bonzini wrote:
> On Mon, Apr 22, 2024 at 5:39=E2=80=AFPM David Woodhouse <dwmw2@infradead.=
org> wrote:
>=20
> > > ... especially considering that you did use a 64-bit integer here
> > > (though---please use u64 not uint64_t; and BTW if you want to add a
> > > patch to change kvm_get_time_scale() to u64, please do.
> >=20
> > Meh, I'm used to programming in C. Yes, I *am* old enough to have been
> > doing this since the last decade of the 1900s, but it *has* been a long
> > time since 1999, and my fingers have learned :)
>=20
> Oh, I am on the same page (working on both QEMU and Linux, adapting my
> muscle memory to the context sucks) but u64/s64 is the preferred
> spelling and I have been asked to use them before.

Ever heard of Jury Nullification... ? :)

> > Heh, looks like it was you who made it uint64_t, in 2016. In a commit
> > (3ae13faac) which said "Prepare for improving the precision in the next
> > patch"... which never came, AFAICT?
>=20
> Yes, it was posted as
> https://lore.kernel.org/lkml/1454944711-33022-5-git-send-email-pbonzini@r=
edhat.com/
> but not committed.

Ah, thanks. So that isn't about arithmetic precision, but about dealing
with the mess we had where the KVM clock was afflicted by NTP skew.

We live in a much saner world now where it's simply based on the guest
TSC (or, in pathological cases, the host's CLOCK_MONOTONIC_RAW.=20

> As an aside, we discovered later that the patch you list as "Fixes"
> fixed another tricky bug: before, kvmclock could jump if the TSC is
> set within the 250 ppm tolerance that does not activate TSC scaling.
> This is possible after a first live migration, and then the second
> live migration used the guest TSC frequency *that userspace desired*
> instead of the *actual* TSC frequency.

Given our saner world where the KVM clock now *isn't* adjusted for NTP
skew, that "bug" was probably better left unfixed. In fact, I may give
some thought to reverting commit 78db6a503796 completely.

Perhaps we should call that "definition E". I think we're up to five
now? But no, let's not add historical ones to the taxonomy :)

I believe Jack's KVM_SET_CLOCK_GUEST fixes the fundamental issue there
of the clock jumping on migration. It's just a special case of the
breakage in KVM_REQ_MASTERCLOCK_UPDATE, where the KVM clock has happily
been running as a direct function of the guest TSC, and when we yank it
back to some other definition when we create a new masterclock
reference point.

With KVM_SET_CLOCK_GUEST, the KVM clock is restored as a function of
the guest TSC, rather than=C2=A0based on realtime/CLOCK_MONOTONIC_RAW/etc.

So even though a new masterclock reference is taken in the new kernel
(and the scaling factors in the pvclock may be slightly different, as
we discussed in the comment you responded to), the ka->kvmclock_offset
is adjusted so that the value of the KVM clock at *that* moment when
the new reference is taken, is precisely what it would have been under
the old kvmclock regime for the contemporary guest TSC value.


> Before:
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 this_tsc_khz =3D __this_cpu_re=
ad(cpu_tsc_khz);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (unlikely(vcpu->hw_tsc_khz =
!=3D this_tsc_khz)) {
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 tgt_tsc_khz =3D vcpu->virtual_tsc_khz;
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 kvm_get_time_scale(NSEC_PER_SEC / 1000, tgt_tsc_khz,
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 &vcpu->h=
v_clock.tsc_shift,
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 &vcpu->h=
v_clock.tsc_to_system_mul);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 vcpu->hw_tsc_khz =3D this_tsc_khz;
>=20
> After:
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 tgt_tsc_khz =3D __this_cpu_rea=
d(cpu_tsc_khz);
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 // tgt_tsc_khz unchanged becau=
se TSC scaling was not enabled
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 tgt_tsc_khz =3D kvm_scale_tsc(=
v, tgt_tsc_khz);
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (unlikely(vcpu->hw_tsc_khz =
!=3D tgt_tsc_khz)) {
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 kvm_get_time_scale(NSEC_PER_SEC / 1000, tgt_tsc_khz,
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 &vcpu->h=
v_clock.tsc_shift,
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 &vcpu->h=
v_clock.tsc_to_system_mul);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 vcpu->hw_tsc_khz =3D tgt_tsc_khz;
>=20
> So in the first case kvm_get_time_scale uses vcpu->virtual_tsc_khz, in
> the second case it uses __this_cpu_read(cpu_tsc_khz).
>=20
> This then caused a mismatch between the actual guest frequency and
> what is used by kvm_guest_time_update, which only becomes visible when
> migration resets the clock with KVM_GET/SET_CLOCK. KVM_GET_CLOCK
> returns what _should have been_ the same value read by the guest, but
> it's wrong.

Hm? Until I fixed it in this series, KVM_GET_CLOCK didn't even *try*
scaling via the guest's TSC frequency, did it? It just converted from
the *host* TSC to nanoseconds (since master_kernel_now) directly. Which
was even *more* broken :)


--=-kMnqTVDkh4tbIOg2M2ET
Content-Type: application/pkcs7-signature; name="smime.p7s"
Content-Disposition: attachment; filename="smime.p7s"
Content-Transfer-Encoding: base64
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--=-kMnqTVDkh4tbIOg2M2ET--