On Mon, 2026-10-12 at 14:23 -0700, Jim Mattson wrote: > > I think there's a window here in which another vCPU can consume > HV_TSC_PAGE_GUEST_CHANGED before kvm_set_clock_and_tsc() steps the > clocks. If TSC emulation control is enabled, > kvm_hv_request_tsc_page_update() won't invalidate the resulting stale > TSC page. > > > + kvm_set_clock_and_tsc(vcpu->kvm, in.reference_time_in_100ns * 100, > > in.tsc); > > When stepping the reference time forward, don't we have to recalculate > the hrtimer deadlines associated with armed synthetic timers? > Otherwise, those hrtimers will fire late.
My understanding is that Windows only does this at resume time when nothing else is running, so it doesn't matter much. Either way, I don't think we want *any* of this in the kernel; we should just pass the hypercall up to userspace and let userspace handle it. There's a commit somewhere in https://git.infradead.org/?p=users/dwmw2/linux.git;a=shortlog;h=refs/heads/vsm which does that; I'm working on cleaning that lot up to be postable this week. (s/linux/kvm-unit-tests/ and s/linux/qemu/ for the other parts, if you want an early preview that actually boots Windows guests under KVM with VSM support).
smime.p7s
Description: S/MIME cryptographic signature
Amazon Development Centre (London) Ltd. Registered in England and Wales with registration number 04543232 with its registered office at 1 Principal Place, Worship Street, London EC2A 2FA, United Kingdom.
