linux-kernel - Re: [PATCH v3 04/12] KVM: X86: Add a ratio parameter to kvm_scale

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YKvKwpRP6UcftcnQ@google.com>
Date:   Mon, 24 May 2021 15:48:18 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     Paolo Bonzini <pbonzini@...hat.com>
Cc:     Ilias Stamatis <ilstam@...zon.com>, kvm@...r.kernel.org,
        linux-kernel@...r.kernel.org, mlevitsk@...hat.com,
        vkuznets@...hat.com, wanpengli@...cent.com, jmattson@...gle.com,
        joro@...tes.org, zamsden@...il.com, mtosatti@...hat.com,
        dwmw@...zon.co.uk
Subject: Re: [PATCH v3 04/12] KVM: X86: Add a ratio parameter to
 kvm_scale_tsc()

On Mon, May 24, 2021, Paolo Bonzini wrote:
> On 21/05/21 12:24, Ilias Stamatis wrote:
> > @@ -3537,10 +3539,14 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> >   		 * return L1's TSC value to ensure backwards-compatible
> >   		 * behavior for migration.
> >   		 */
> > -		u64 tsc_offset = msr_info->host_initiated ? vcpu->arch.l1_tsc_offset :
> > -							    vcpu->arch.tsc_offset;
> > -
> > -		msr_info->data = kvm_scale_tsc(vcpu, rdtsc()) + tsc_offset;
> > +		if (msr_info->host_initiated)
> > +			msr_info->data = kvm_scale_tsc(
> > +				vcpu, rdtsc(), vcpu->arch.l1_tsc_scaling_ratio
> > +				) + vcpu->arch.l1_tsc_offset;
> 
> Better indentation:
> 
> 	msr_info->data = vcpu->arch.l1_tsc_offset +
> 		kvm_scale_tsc(vcpu, rdtsc(),

I like this more, but it's still a bit ugly.

> 			      vcpu->arch.tsc_scaling_ratio);

Wrong pairing of ratio and offset (not necessarily what you queued, obviously).

> 
> Same below.

Alternatively, what about something like this?

		u64 offset, ratio;

		if (msr_info->host_initiated) {
			offset = vcpu->arch.l1_tsc_offset;
			ratio = vcpu->arch.l1_tsc_scaling_ratio;
		} else {
			offset = vcpu->arch.tsc_offset;
			ratio = vcpu->arch.tsc_scaling_ratio;
		}
		msr_info->data = kvm_scale_tsc(vcpu, rdtsc(), ratio) + offset;

or an option that would require additional refactoring:

		struct kvm_guest_tsc *tsc;

		tsc = msr_info->host_initiated ? &vcpu->arch.l1_tsc :
						 &vcpu->arch.current_tsc;

		msr_info->data = tsc->offset +
				 kvm_scale_tsc(vcpu, rdtsc(), tsc->scaling_ratio);


> 
> Paolo
> 
> > +		else
> > +			msr_info->data = kvm_scale_tsc(
> > +				vcpu, rdtsc(), vcpu->arch.tsc_scaling_ratio
> > +				) + vcpu->arch.tsc_offset;
> >   		break;
> >   	}
> >   	case MSR_MTRRcap:
> > 
>