Re: [PATCH v12 04/24] KVM: x86: Allow KVM master clock mode when TSCs are offset from each other

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Low] KVM: x86: missing masterclock update on error in
  kvm_arch_vcpu_postcreate()
--

commit 63e87344888586250d53904cdb6246fdd7dee791
Author: David Woodhouse <dwmw@amazon.co.uk>

KVM: x86: Allow KVM master clock mode when TSCs are offset from each other

This commit tracks vCPU TSC frequency matching separately from TSC offset
matching so that master clock mode remains enabled when vCPUs have
matching frequencies but different TSC offsets, while requiring matched
offsets for PVCLOCK_TSC_STABLE_BIT.

> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index d29e5bc75f7a5..73015825f2188 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c

[ ... ]

> @@ -9568,6 +9636,7 @@ void kvm_arch_vcpu_postcreate(struct kvm_vcpu *vcpu)
>  
>  		raw_spin_lock_irqsave(&ka->tsc_write_lock, flags);
>  		WRITE_ONCE(ka->all_vcpus_matched_tsc, false);
> +		ka->all_vcpus_matched_freq = false;
>  		raw_spin_unlock_irqrestore(&ka->tsc_write_lock, flags);
>  		return;
>  	}

[Severity: Low]
Does this error path also need to call kvm_update_masterclock(vcpu->kvm)
when mutex_lock_killable(&vcpu->mutex) is interrupted?

Without kvm_update_masterclock(), pvclock_update_vm_gtod_copy() does not
run, so ka->use_master_clock stays true while the new unsynchronized vCPU
is runnable, and already-running vCPUs do not receive a
KVM_REQ_CLOCK_UPDATE broadcast to clear PVCLOCK_TSC_STABLE_BIT in their
guest pvclock structures.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261109012112.3415328-1-dwmw2@infradead.org?part=4