[PATCH] KVM: x86: allow TSC to differ by NTP correction bounds without TSC scaling

[PATCH] KVM: x86: allow TSC to differ by NTP correction bounds without TSC scaling

[Marcelo Tosatti]

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>

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)

Re: [PATCH] KVM: x86: allow TSC to differ by NTP correction bounds without TSC scaling

[Paolo Bonzini]

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)
> 

[PATCH v2] KVM: x86: allow TSC to differ by NTP correction bounds without TSC scaling

[Marcelo Tosatti]

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.

Without TSC scaling, the current kernel interface will either return an error
(if user_tsc_khz <= tsc_khz) or enable TSC catchup mode.

This change enables the following TSC tolerance check to
accept KVM_SET_TSC_KHZ within tsc_tolerance_ppm (which is 250ppm by default).

        /*
         * Compute the variation in TSC rate which is acceptable
         * within the range of tolerance and decide if the
         * rate being applied is within that bounds of the hardware
         * rate.  If so, no scaling or compensation need be done.
         */
        thresh_lo = adjust_tsc_khz(tsc_khz, -tsc_tolerance_ppm);
        thresh_hi = adjust_tsc_khz(tsc_khz, tsc_tolerance_ppm);
        if (user_tsc_khz < thresh_lo || user_tsc_khz > thresh_hi) {
                pr_debug("kvm: requested TSC rate %u falls outside tolerance [%u,%u]\n", user_tsc_khz, thresh_lo, thresh_hi);
                use_scaling = 1;
        }

NTP daemon in the guest can correct this difference (NTP can correct upto 500ppm).

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

---

v2: improve changelog (Paolo Bonzini)

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)

Re: [PATCH v2] KVM: x86: allow TSC to differ by NTP correction bounds without TSC scaling

[Paolo Bonzini]

On 16/06/20 13:47, Marcelo Tosatti wrote:
> v2: improve changelog (Paolo Bonzini)
> 
> 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;

This cannot happen, there is an "if (!scale) right above.

>  		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)
> 

Queued this second hunk.

Paolo