| Message ID | 20200615115952.GA224592@fuller.cnet (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | KVM: x86: allow TSC to differ by NTP correction bounds without TSC scaling | expand |
On 15/06/20 13:59, Marcelo Tosatti wrote: > The Linux TSC calibration procedure is subject to small variations > (its common to see +-1 kHz difference between reboots on a given CPU, for example). > > So migrating a guest between two hosts with identical processor can fail, in case > of a small variation in calibrated TSC between them. > > Allow a conservative 250ppm error between host TSC and VM TSC frequencies, > rather than requiring an exact match. NTP daemon in the guest can > correct this difference. > > Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com> This is the userspace commit message. Can you resend with a better commit message that actually matches what the patch does and explains why the userspace patch is not enough? Also you should explain what happens with new userspace and old kernel. Thanks, Paolo > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index 3156e25..39a6664 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -1772,6 +1772,8 @@ static int set_tsc_khz(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale) > > /* TSC scaling supported? */ > if (!kvm_has_tsc_control) { > + if (!scale) > + return 0; > if (user_tsc_khz > tsc_khz) { > vcpu->arch.tsc_catchup = 1; > vcpu->arch.tsc_always_catchup = 1; > @@ -4473,7 +4475,8 @@ long kvm_arch_vcpu_ioctl(struct file *filp, > r = -EINVAL; > user_tsc_khz = (u32)arg; > > - if (user_tsc_khz >= kvm_max_guest_tsc_khz) > + if (kvm_has_tsc_control && > + user_tsc_khz >= kvm_max_guest_tsc_khz) > goto out; > > if (user_tsc_khz == 0) >
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 3156e25..39a6664 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -1772,6 +1772,8 @@ static int set_tsc_khz(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale) /* TSC scaling supported? */ if (!kvm_has_tsc_control) { + if (!scale) + return 0; if (user_tsc_khz > tsc_khz) { vcpu->arch.tsc_catchup = 1; vcpu->arch.tsc_always_catchup = 1; @@ -4473,7 +4475,8 @@ long kvm_arch_vcpu_ioctl(struct file *filp, r = -EINVAL; user_tsc_khz = (u32)arg; - if (user_tsc_khz >= kvm_max_guest_tsc_khz) + if (kvm_has_tsc_control && + user_tsc_khz >= kvm_max_guest_tsc_khz) goto out; if (user_tsc_khz == 0)
The Linux TSC calibration procedure is subject to small variations (its common to see +-1 kHz difference between reboots on a given CPU, for example). So migrating a guest between two hosts with identical processor can fail, in case of a small variation in calibrated TSC between them. Allow a conservative 250ppm error between host TSC and VM TSC frequencies, rather than requiring an exact match. NTP daemon in the guest can correct this difference. Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>