On Mon, 2021-05-10 at 14:44 +0000, Stamatis, Ilias wrote:
> On Mon, 2021-05-10 at 16:53 +0300, Maxim Levitsky wrote:
> > On Thu, 2021-05-06 at 10:32 +0000, ilstam@xxxxxxxxxxx wrote:
> > > From: Ilias Stamatis <ilstam@xxxxxxxxxx>
> > >
> > > When L2 is entered we need to "merge" the TSC multiplier and TSC
> > > offset
> > > values of VMCS01 and VMCS12 and store the result into the current
> > > VMCS02.
> > >
> > > The 02 values are calculated using the following equations:
> > > offset_02 = ((offset_01 * mult_12) >> 48) + offset_12
> > > mult_02 = (mult_01 * mult_12) >> 48
> >
> > I would mention that 48 is kvm_tsc_scaling_ratio_frac_bits instead.
> > Also maybe add the common code in a separate patch?
>
> Right, will do.
>
> > > Signed-off-by: Ilias Stamatis <ilstam@xxxxxxxxxx>
> > > ---
> > > arch/x86/include/asm/kvm_host.h | 1 +
> > > arch/x86/kvm/vmx/nested.c | 26 ++++++++++++++++++++++----
> > > arch/x86/kvm/x86.c | 25 +++++++++++++++++++++++++
> > > 3 files changed, 48 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/arch/x86/include/asm/kvm_host.h
> > > b/arch/x86/include/asm/kvm_host.h
> > > index cdddbf0b1177..e7a1eb36f95a 100644
> > > --- a/arch/x86/include/asm/kvm_host.h
> > > +++ b/arch/x86/include/asm/kvm_host.h
> > > @@ -1780,6 +1780,7 @@ void kvm_define_user_return_msr(unsigned
> > > index, u32 msr);
> > > int kvm_set_user_return_msr(unsigned index, u64 val, u64 mask);
> > >
> > > u64 kvm_scale_tsc(struct kvm_vcpu *vcpu, u64 tsc, bool l1);
> > > +u64 kvm_compute_02_tsc_offset(u64 l1_offset, u64 l2_multiplier, u64
> > > l2_offset);
> > > u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc);
> > >
> > > unsigned long kvm_get_linear_rip(struct kvm_vcpu *vcpu);
> > > diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> > > index bced76637823..a1bf28f33837 100644
> > > --- a/arch/x86/kvm/vmx/nested.c
> > > +++ b/arch/x86/kvm/vmx/nested.c
> > > @@ -3353,8 +3353,22 @@ enum nvmx_vmentry_status
> > > nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu,
> > > }
> > >
> > > enter_guest_mode(vcpu);
> > > - if (vmcs12->cpu_based_vm_exec_control &
> > > CPU_BASED_USE_TSC_OFFSETTING)
> > > - vcpu->arch.tsc_offset += vmcs12->tsc_offset;
> > > +
> > > + if (vmcs12->cpu_based_vm_exec_control &
> > > CPU_BASED_USE_TSC_OFFSETTING) {
> > > + if (vmcs12->secondary_vm_exec_control &
> > > SECONDARY_EXEC_TSC_SCALING) {
> > > + vcpu->arch.tsc_offset =
> > > kvm_compute_02_tsc_offset(
> > > + vcpu->arch.l1_tsc_offset,
> > > + vmcs12->tsc_multiplier,
> > > + vmcs12->tsc_offset);
> > > +
> > > + vcpu->arch.tsc_scaling_ratio =
> > > mul_u64_u64_shr(
> > > + vcpu->arch.tsc_scaling_ratio,
> > > + vmcs12->tsc_multiplier,
> > > + kvm_tsc_scaling_ratio_frac_bit
> > > s);
> > > + } else {
> > > + vcpu->arch.tsc_offset += vmcs12->tsc_offset;
> > > + }
> > > + }
> > >
> > > if (prepare_vmcs02(vcpu, vmcs12, &entry_failure_code)) {
> > > exit_reason.basic = EXIT_REASON_INVALID_STATE;
> > > @@ -4454,8 +4468,12 @@ void nested_vmx_vmexit(struct kvm_vcpu *vcpu,
> > > u32 vm_exit_reason,
> > > if (nested_cpu_has_preemption_timer(vmcs12))
> > > hrtimer_cancel(&to_vmx(vcpu)-
> > > > nested.preemption_timer);
> > >
> > > - if (vmcs12->cpu_based_vm_exec_control &
> > > CPU_BASED_USE_TSC_OFFSETTING)
> > > - vcpu->arch.tsc_offset -= vmcs12->tsc_offset;
> > > + if (vmcs12->cpu_based_vm_exec_control &
> > > CPU_BASED_USE_TSC_OFFSETTING) {
> > > + vcpu->arch.tsc_offset = vcpu->arch.l1_tsc_offset;
> > > +
> > > + if (vmcs12->secondary_vm_exec_control &
> > > SECONDARY_EXEC_TSC_SCALING)
> > > + vcpu->arch.tsc_scaling_ratio = vcpu-
> > > > arch.l1_tsc_scaling_ratio;
> > > + }
> > >
> > > if (likely(!vmx->fail)) {
> > > sync_vmcs02_to_vmcs12(vcpu, vmcs12);
> > > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> > > index 26a4c0f46f15..87deb119c521 100644
> > > --- a/arch/x86/kvm/x86.c
> > > +++ b/arch/x86/kvm/x86.c
> > > @@ -2266,6 +2266,31 @@ static u64 kvm_compute_tsc_offset(struct
> > > kvm_vcpu *vcpu, u64 target_tsc)
> > > return target_tsc - tsc;
> > > }
> > >
> > > +/*
> > > + * This function computes the TSC offset that is stored in VMCS02
> > > when entering
> > > + * L2 by combining the offset and multiplier values of VMCS01 and
> > > VMCS12.
> > > + */
> > > +u64 kvm_compute_02_tsc_offset(u64 l1_offset, u64 l2_multiplier, u64
> > > l2_offset)
> > > +{
> > > + u64 offset;
> > > +
> > > + /*
> > > + * The L1 offset is interpreted as a signed number by the CPU
> > > and can
> > > + * be negative. So we extract the sign before the
> > > multiplication and
> > > + * put it back afterwards if needed.
> >
> > If I understand correctly the reason for sign extraction is that we
> > don't
> > have mul_s64_u64_shr. Maybe we should add it?
> >
>
> Initially I added a mul_s64_s64_shr and used that.
>
> You can take a look
> here:
>
> https://github.com/ilstam/linux/commit/7bc0b75bb7b8d9bcecf41e2e1e1936880d3b93d1
>
>
> I added mul_s64_s64_shr instead of mul_s64_u64_shr thinking that a)
> perhaps it would be more useful as a general-purpose function and b)
> even though the multiplier is unsigned in reality it's caped so it
> should be safe to cast it to signed.
Actually is it? Intel's PRM I think defines TSC multiplier to be full
64 bit. It is shifted right after multiplication, making the multiplier
be effictively a fixed point 16.48 number but since multiplication is done
first, it will still have to use full 64 bit value.
>
> But then I realized that
> mul_u64_u64_shr that we already have is enough and I can just simply re-
> use that.
>
> How would a mul_s64_u64_shr be implemented?
I'll say even if it is implemented by just extracting the sign as you did
and then calling mul_u64_u64_shr,
it could still be a bit better documentation wise.
>
> I think in the end
> we need to decide if we want to do signed or unsigned multiplication.
> And depending on this we need to do a signed or unsigned right shift
> accordingly.
I agree.
>
> So if we did have a mul_s64_u64_shr wouldn't it need to cast
> either of the arguments and then use mul_u64_u64 (or mul_s64_s64) being
> just a wrapper essentially? I don't think that we can guarantee that
> casting either of the arguments is always safe if we want to advertise
> this as a general purpose function.
Note that I don't have a strong opinion on this whole thing, so if you feel
that it should be left as is, I don't mind.
>
> Thanks for the reviews!
No problem!
Best regards,
Maxim Levitsky
>
> Ilias
>
> > The pattern of (A * B) >> shift appears many times in TSC scaling.
> >
> > So instead of this function maybe just use something like that?
> >
> > merged_offset = l2_offset + mul_s64_u64_shr ((s64) l1_offset,
> > l2_multiplier,
> > kvm_tsc_scaling_ratio_fra
> > c_bits)
> >
> > Or another idea:
> >
> > How about
> >
> > u64 __kvm_scale_tsc_value(u64 value, u64 multiplier) {
> > return mul_u64_u64_shr(value,
> > multiplier,
> > kvm_tsc_scaling_ratio_frac_bits);
> > }
> >
> >
> > and
> >
> > s64 __kvm_scale_tsc_offset(u64 value, u64 multiplier)
> > {
> > return mul_s64_u64_shr((s64)value,
> > multiplier,
> > kvm_tsc_scaling_ratio_frac_bits);
> > }
> >
> > And then use them in the code.
> >
> > Overall though the code *looks* correct to me
> > but I might have missed something.
> >
> > Best regards,
> > Maxim Levitsky
> >
> >
> > > + */
> > > + offset = mul_u64_u64_shr(abs((s64) l1_offset),
> > > + l2_multiplier,
> > > + kvm_tsc_scaling_ratio_frac_bits);
> > > +
> > > + if ((s64) l1_offset < 0)
> > > + offset = -((s64) offset);
> > > +
> > > + offset += l2_offset;
> > > + return offset;
> > > +}
> > > +EXPORT_SYMBOL_GPL(kvm_compute_02_tsc_offset);
> > > +
> > > u64 kvm_read_l1_tsc(struct kvm_vcpu *vcpu, u64 host_tsc)
> > > {
> > > return vcpu->arch.l1_tsc_offset + kvm_scale_tsc(vcpu,
> > > host_tsc, true);
