Re: [PATCH v6 00/36] Cleaning up the KVM clock mess

On Sat, Aug 01, 2026, David Woodhouse wrote:
> On Fri, 2026-07-31 at 13:09 -0700, Sean Christopherson wrote:
> > Applied patch 1 to kvm-x86 clocks.
> >
> > [01/36] KVM: x86/xen: Do not corrupt KVM clock in kvm_xen_shared_info_init()
> >         https://github.com/kvm-x86/linux/commit/3d4b20b5a7df
> 
> Thanks. Could the clocks branch be based on something that includes the
> timekeeping work from the timers-ptp-2026-06-13 merge (in v7.2-rc1)?

Yeah, I can rebase onto a 7.2-rcN.  Y'all are likely the only people that care
about the above commit, so a late rebase isn't a big deal.

There's basically zero chance I'll get the entire series applied for 7.3, but
"Use ktime_get_snapshot_id() for master clock" is at least in striking distance,
so there's no reason not to allow for the possibility.

> The later parts of the series depend on ktime_get_snapshot_id() and the
> reworked struct system_time_snapshot from there. Your kvm-x86/next
> branch does have it; I'll keep my WIP kvmclock8 branch based on that
> for now as I address the other comments.
> 
> 
> *   2d6d57f889f3 Merge tag 'timers-ptp-2026-06-13' of tip
> |\
> | * bc484a509673 ptp: vmclock: Use hw_cycles from snapshot for precise TSC pairing
> | ...
> | * ca1ec8bfac8c timekeeping: Add clocksource read_snapshot() method and hw_cycles to snapshot
> | ...
> | * ef22786707e3 timekeeping: Use system_time_snapshot::systime/monoraw instead of ::real/raw
> | * eba302268a01 timekeeping: Provide ktime_get_snapshot_id()