Re: [PATCH v8 05/17] KVM: x86: Avoid NTP frequency skew for KVM clock on 32-bit host

On Sat, Aug 08, 2026, David Woodhouse wrote:
> On Fri, 2026-08-07 at 18:26 +0100, David Woodhouse wrote:
> > (Home and on the phone after an insane week; full attention span
> > might be limited until Monday).
> 
> Hi, my name is Dave and I'm an addict...
> 
> > The only reason the ktime snapshot stuff is last in the series is
> > because I did that work on the kernel's core timekeeping two years
> > after the rest of the series, and it was still sitting at the tip
> > branch at the time so I didn't want to make it a hard dependency for
> > the kvmclock series. Now the timekeeping stuff is merged we can fold
> > it all into the kvmclock series however we like.
> 
> I think that goes for "Compute kvmclock base without pvclock_gtod_data"
> too. It *isn't* at the end of the series, but I'm fairly sure (from
> memory) that it's one of the newer commits, and I just shuffled it
> earlier in the series for very similar reasons to the ones you're
> looking at now.

That's good, because I came to same conclusion last week :-)

This is what I now have locally as "part 1", which I'm planning on posting today.
It'll be too late for 7.3, but if part 1 looks good to you, I'll throw them in an
unstable topic branch for 7.4 so that we can consider them "done" and work on the
remaining patches.

  KVM: x86: Update "last guest TSC" snapshot prior to enabling IRQs/preemption
  KVM: x86: Improve accuracy of KVM clock when TSC scaling is in force
  KVM: x86: Explicitly disable TSC scaling without CONSTANT_TSC
  KVM: x86: Activate master clock immediately on vCPU creation
  KVM: x86: Compute kvmclock base without pvclock_gtod_data
  KVM: x86: Avoid NTP frequency skew for KVM clock on 32-bit host
  KVM: x86: Drop unnecessary CPU pinning when computing/getting kvmclock
  KVM: x86: Move "no master clock" fallback from __get_kvmclock() to get_kvmclock()
  KVM: x86: Wrap all of __get_kvmclock_master_clock() with CONFIG_X86_64=y
  KVM: x86: Fall back to non-master-clock if clockread fails in get_kvmclock()
  KVM: x86: Fix KVM clock precision in get_kvmclock() with TSC scaling
  KVM: x86: Use get_kvmclock() in kvm_get_wall_clock_epoch()
  KVM: x86: Fix compute_guest_tsc() to handle negative time deltas
  KVM: x86: Disable preemption, not IRQs, when getting TSC+freq pair
  KVM: x86: Make master clock logic in guest PV clock updates 64-bit only
  KVM: x86: Upscale TSC to "now", not master clock when updating PV clocks
  KVM: x86: Simplify and comment kvm_get_time_scale()
  KVM: x86: Remove implicit rdtsc() from kvm_compute_l1_tsc_offset()
  KVM: x86: Use kernel timekeeping snapshots for getting kvmclock time since boot
  KVM: x86: Use kernel timekeeping snapshot for monotonic clock
  KVM: x86: Use kernel timekeeping snapshot to get walltime+TSC