Message-ID: <d3cc3cd1a90d7ee9a31e40fbe2db9f3f338d5004.camel@redhat.com>
Subject: Re: [PATCH 2/2] KVM: x86: Forbid KVM_SET_CPUID{,2} after KVM_RUN
From:   Maxim Levitsky <mlevitsk@redhat.com>
To:     Sean Christopherson <seanjc@google.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Igor Mammedov <imammedo@redhat.com>, kvm@vger.kernel.org,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>, linux-kernel@vger.kernel.org
Date:   Thu, 13 Jan 2022 18:30:49 +0200
In-Reply-To: <YeBSrcNawgzvTzQ6@google.com>
References: <877dbbq5om.fsf@redhat.com>
         <5505d731-cf87-9662-33f3-08844d92877c@redhat.com>
         <20220111090022.1125ffb5@redhat.com> <87fsptnjic.fsf@redhat.com>
         <50136685-706e-fc6a-0a77-97e584e74f93@redhat.com>
         <87bl0gnfy5.fsf@redhat.com>
         <7e7c7e22f8b1b1695d26d9e19a767b87c679df93.camel@redhat.com>
         <87zgnzn1nr.fsf@redhat.com>
         <6ae7e64c53727f9f00537d787e9612c292c4e244.camel@redhat.com>
         <87wnj3n0k0.fsf@redhat.com> <YeBSrcNawgzvTzQ6@google.com>
Content-Type: text/plain; charset="UTF-8"
User-Agent: Evolution 3.36.5 (3.36.5-2.fc32) 
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Thu, 2022-01-13 at 16:26 +0000, Sean Christopherson wrote:
> On Thu, Jan 13, 2022, Vitaly Kuznetsov wrote:
> > Maxim Levitsky <mlevitsk@redhat.com> writes:
> > > For my nested AVIC work I would really want the APIC ID of a VCPU to be read-only
> > > and be equal to vcpu_id.
> > > 
> > 
> > Doesn't APIC ID have topology encoded in it?
> 
> Yeah, APIC IDs are derived from the topology.  From the SDM (this doesn't
> talk about core/SMT info, but that's included as well):
> 
>   The hardware assigned APIC ID is based on system topology and includes encoding
>   for socket position and cluster information.
> 
> The SDM also says:
> 
>   Some processors permit software to modify the APIC ID. However, the ability of
>   software to modify the APIC ID is processor model specific.
> 
> So I _think_ we could define KVM behavior to ignore writes from the _guest_, but
> the APIC_ID == vcpu_id requirement won't fly as userspace expects to be able to
> stuff virtual toplogy info into the APIC ID.
> 
That is a very good piece of information! Thanks!

Best regards,
	Maxim Levitsky