Linux Documentation
 help / color / mirror / Atom feed
From: David Woodhouse <dwmw2@infradead.org>
To: Paolo Bonzini <pbonzini@redhat.com>,
	Jonathan Corbet <corbet@lwn.net>,
	Shuah Khan <skhan@linuxfoundation.org>,
	Sean Christopherson <seanjc@google.com>,
	Thomas Gleixner <tglx@kernel.org>, 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>,
	Vitaly Kuznetsov <vkuznets@redhat.com>,
	Juergen Gross <jgross@suse.com>,
	Boris Ostrovsky <boris.ostrovsky@oracle.com>,
	David Woodhouse <dwmw2@infradead.org>,
	Paul Durrant <paul@xen.org>, Jonathan Cameron <jic23@kernel.org>,
	Sascha Bischoff <Sascha.Bischoff@arm.com>,
	Marc Zyngier <maz@kernel.org>, Joey Gouly <joey.gouly@arm.com>,
	Jack Allister <jalliste@amazon.com>,
	Dongli Zhang <dongli.zhang@oracle.com>,
	joe.jin@oracle.com, kvm@vger.kernel.org,
	linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org,
	xen-devel@lists.xenproject.org, linux-kselftest@vger.kernel.org
Subject: [PATCH v4 22/30] KVM: x86: Allow KVM master clock mode when TSCs are offset from each other
Date: Sat,  9 May 2026 23:46:48 +0100	[thread overview]
Message-ID: <20260509224824.3264567-23-dwmw2@infradead.org> (raw)
In-Reply-To: <20260509224824.3264567-1-dwmw2@infradead.org>

From: David Woodhouse <dwmw@amazon.co.uk>

There is no reason why the KVM clock cannot be in masterclock mode when
the TSCs are not in sync, as long as they are at the same *frequency*.

Running at a different frequency would lead to a systemic skew between
the clock(s) as observed by different vCPUs due to arithmetic precision
in the scaling. So that should indeed force the clock to be based on the
host's CLOCK_MONOTONIC_RAW instead of being in masterclock mode where it
is defined by the guest TSC.

But when the vCPUs merely have a different TSC *offset*, that's not a
problem. The offset is applied to that vCPU's kvmclock->tsc_timestamp
field, and it all comes out in the wash.

Track frequency matching separately from full TSC matching. Use
frequency match for master clock eligibility, and full TSC match
(including offset) only for PVCLOCK_TSC_STABLE_BIT, which tells the
guest it is safe to skip cross-vCPU monotonicity enforcement.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
---
 arch/x86/include/asm/kvm_host.h |  1 +
 arch/x86/kvm/x86.c              | 27 +++++++++++++++++++++------
 2 files changed, 22 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index eb81f90284ba..c770c63087cb 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -1493,6 +1493,7 @@ struct kvm_arch {
 	u64 cur_tsc_offset;
 	u64 cur_tsc_generation;
 	bool all_vcpus_matched_tsc;
+	bool all_vcpus_matched_freq;
 	int nr_vcpus_matched_tsc;
 
 	u32 default_tsc_khz;
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index b74fd8b088ad..d36d03b8268e 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2647,13 +2647,22 @@ static void kvm_track_tsc_matching(struct kvm_vcpu *vcpu, bool new_generation)
 	struct pvclock_gtod_data *gtod = &pvclock_gtod_data;
 
 	/*
-	 * To use the masterclock, the host clocksource must be based on TSC
-	 * and all vCPUs must have matching TSCs.
+	 * Track whether all vCPUs have matching TSC offsets (for
+	 * PVCLOCK_TSC_STABLE_BIT) and matching frequencies (for
+	 * master clock eligibility).
 	 */
 	ka->all_vcpus_matched_tsc = (ka->nr_vcpus_matched_tsc + 1 ==
 				     atomic_read(&vcpu->kvm->online_vcpus));
+	if (ka->all_vcpus_matched_tsc)
+		ka->all_vcpus_matched_freq = true;
 
-	bool use_master_clock = ka->all_vcpus_matched_tsc &&
+	/*
+	 * To use the masterclock, the host clocksource must be based on TSC
+	 * and all vCPUs must have matching TSC *frequency*. Different offsets
+	 * are fine — each vCPU's pvclock has its own tsc_timestamp that
+	 * accounts for its offset.
+	 */
+	bool use_master_clock = ka->all_vcpus_matched_freq &&
 				gtod_is_based_on_tsc(gtod->clock.vclock_mode);
 
 	/*
@@ -2817,7 +2826,13 @@ static void __kvm_synchronize_tsc(struct kvm_vcpu *vcpu, u64 offset, u64 tsc,
 	 * Track the TSC frequency, scaling ratio, and offset for the current
 	 * generation. These are used to detect matching TSC writes and to
 	 * compute the guest TSC from the host clock.
+	 *
+	 * If the frequency changed, master clock mode can no longer be used
+	 * since the kvmclock scaling factors differ between vCPUs.
 	 */
+	if (vcpu->arch.virtual_tsc_khz != kvm->arch.cur_tsc_khz)
+		kvm->arch.all_vcpus_matched_freq = false;
+
 	kvm->arch.cur_tsc_khz = vcpu->arch.virtual_tsc_khz;
 	kvm->arch.cur_tsc_scaling_ratio = vcpu->arch.l1_tsc_scaling_ratio;
 
@@ -3178,7 +3193,7 @@ static void pvclock_update_vm_gtod_copy(struct kvm *kvm)
 	bool host_tsc_clocksource, vcpus_matched;
 
 	lockdep_assert_held(&kvm->arch.tsc_write_lock);
-	vcpus_matched = ka->all_vcpus_matched_tsc;
+	vcpus_matched = ka->all_vcpus_matched_freq;
 
 	/*
 	 * If the host uses TSC clock, then passthrough TSC as stable
@@ -3513,7 +3528,7 @@ int kvm_guest_time_update(struct kvm_vcpu *v)
 
 	/* If the host uses TSC clocksource, then it is stable */
 	hv_clock.flags = 0;
-	if (use_master_clock)
+	if (use_master_clock && ka->all_vcpus_matched_tsc)
 		hv_clock.flags |= PVCLOCK_TSC_STABLE_BIT;
 
 	if (vcpu->pv_time.active) {
@@ -6340,7 +6355,7 @@ static int kvm_vcpu_ioctl_get_clock_guest(struct kvm_vcpu *v, void __user *argp)
 
 	hv_clock.tsc_shift = vcpu->pvclock_tsc_shift;
 	hv_clock.tsc_to_system_mul = vcpu->pvclock_tsc_mul;
-	hv_clock.flags = PVCLOCK_TSC_STABLE_BIT;
+	hv_clock.flags = ka->all_vcpus_matched_tsc ? PVCLOCK_TSC_STABLE_BIT : 0;
 
 	if (copy_to_user(argp, &hv_clock, sizeof(hv_clock)))
 		return -EFAULT;
-- 
2.51.0


  parent reply	other threads:[~2026-05-09 22:49 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-09 22:46 [PATCH v4] 00/30] Cleaning up the KVM clock mess David Woodhouse
2026-05-09 22:46 ` [PATCH v4 01/30] KVM: x86/xen: Do not corrupt KVM clock in kvm_xen_shared_info_init() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 02/30] KVM: x86: Improve accuracy of KVM clock when TSC scaling is in force David Woodhouse
2026-05-09 22:46 ` [PATCH v4 03/30] UAPI: x86: Move pvclock-abi to UAPI for x86 platforms David Woodhouse
2026-05-09 22:46 ` [PATCH v4 04/30] KVM: x86: Add KVM_[GS]ET_CLOCK_GUEST for accurate KVM clock migration David Woodhouse
2026-05-09 22:46 ` [PATCH v4 05/30] KVM: selftests: Add KVM/PV clock selftest to prove timer correction David Woodhouse
2026-05-09 22:46 ` [PATCH v4 06/30] KVM: x86: Explicitly disable TSC scaling without CONSTANT_TSC David Woodhouse
2026-05-09 22:46 ` [PATCH v4 07/30] KVM: x86: Add KVM_VCPU_TSC_SCALE and fix the documentation on TSC migration David Woodhouse
2026-05-09 22:46 ` [PATCH v4 08/30] KVM: x86: Avoid NTP frequency skew for KVM clock on 32-bit host David Woodhouse
2026-05-09 22:46 ` [PATCH v4 09/30] KVM: x86: WARN if kvm_get_walltime_and_clockread() fails unexpectedly David Woodhouse
2026-05-09 22:46 ` [PATCH v4 10/30] KVM: x86: Fold __get_kvmclock() into get_kvmclock() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 11/30] KVM: x86: Add WARN and restructure get_kvmclock() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 12/30] KVM: x86: Use get_kvmclock_base_ns() as fallback in get_kvmclock() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 13/30] KVM: x86: Fix KVM clock precision in get_kvmclock() with TSC scaling David Woodhouse
2026-05-09 22:46 ` [PATCH v4 14/30] KVM: x86: Use get_kvmclock() in kvm_get_wall_clock_epoch() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 15/30] KVM: x86: Fix compute_guest_tsc() to handle negative time deltas David Woodhouse
2026-05-09 22:46 ` [PATCH v4 16/30] KVM: x86: Restructure kvm_guest_time_update() for TSC upscaling David Woodhouse
2026-05-09 22:46 ` [PATCH v4 17/30] KVM: x86: Simplify and comment kvm_get_time_scale() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 18/30] KVM: x86: Remove implicit rdtsc() from kvm_compute_l1_tsc_offset() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 19/30] KVM: x86: Improve synchronization in kvm_synchronize_tsc() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 20/30] KVM: x86: Kill last_tsc_{nsec,write,offset} fields David Woodhouse
2026-05-09 22:46 ` [PATCH v4 21/30] KVM: x86: Replace nr_vcpus_matched_tsc count with all_vcpus_matched_tsc bool David Woodhouse
2026-05-09 22:46 ` David Woodhouse [this message]
2026-05-09 22:46 ` [PATCH v4 23/30] KVM: x86: Factor out kvm_use_master_clock() David Woodhouse
2026-05-09 22:46 ` [PATCH v4 24/30] KVM: x86: Avoid gratuitous global clock updates David Woodhouse
2026-05-09 22:46 ` [PATCH v4 25/30] KVM: x86/xen: Prevent runstate times from becoming negative David Woodhouse
2026-05-09 22:46 ` [PATCH v4 26/30] KVM: x86: Avoid redundant masterclock updates from multiple vCPUs David Woodhouse
2026-05-09 22:46 ` [PATCH v4 27/30] KVM: x86: Add KVM_VCPU_TSC_EFFECTIVE_FREQ attribute David Woodhouse
2026-05-09 22:46 ` [PATCH v4 28/30] KVM: x86: Remove runtime Xen TSC frequency CPUID update David Woodhouse
2026-05-09 22:46 ` [PATCH v4 29/30] x86/kvm: Obtain TSC frequency from CPUID if present David Woodhouse
2026-05-09 22:46 ` [PATCH v4 30/30] x86/xen: " David Woodhouse
2026-05-10 20:56 ` [PATCH v4 33/30] KVM: selftests: Add Xen runstate migration test David Woodhouse
2026-05-10 20:58 ` [PATCH v4 31/30] KVM: selftests: Add Xen/generic CPUID timing leaf test David Woodhouse
2026-05-10 21:05 ` [PATCH v4 32/30] KVM: x86: Re-synchronize TSC after KVM_SET_TSC_KHZ David Woodhouse

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260509224824.3264567-23-dwmw2@infradead.org \
    --to=dwmw2@infradead.org \
    --cc=Sascha.Bischoff@arm.com \
    --cc=boris.ostrovsky@oracle.com \
    --cc=bp@alien8.de \
    --cc=corbet@lwn.net \
    --cc=dave.hansen@linux.intel.com \
    --cc=dongli.zhang@oracle.com \
    --cc=hpa@zytor.com \
    --cc=jalliste@amazon.com \
    --cc=jgross@suse.com \
    --cc=jic23@kernel.org \
    --cc=joe.jin@oracle.com \
    --cc=joey.gouly@arm.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-kselftest@vger.kernel.org \
    --cc=maz@kernel.org \
    --cc=mingo@redhat.com \
    --cc=paul@xen.org \
    --cc=pbonzini@redhat.com \
    --cc=seanjc@google.com \
    --cc=skhan@linuxfoundation.org \
    --cc=tglx@kernel.org \
    --cc=vkuznets@redhat.com \
    --cc=x86@kernel.org \
    --cc=xen-devel@lists.xenproject.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox