Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210219144632.2288189-1-david.edmondson@oracle.com>
 <20210219144632.2288189-2-david.edmondson@oracle.com> <YDWG51Io0VJEBHGg@google.com>
In-Reply-To: <YDWG51Io0VJEBHGg@google.com>
From:   Jim Mattson <jmattson@google.com>
Date:   Tue, 23 Feb 2021 15:13:54 -0800
Message-ID: <CALMp9eQ5HQqRRBu0HJbuTOJwKSUA950JWSHrLkXz7cHWKt+ymg@mail.gmail.com>
Subject: Re: [PATCH v2 1/3] KVM: x86: dump_vmcs should not assume
 GUEST_IA32_EFER is valid
To:     Sean Christopherson <seanjc@google.com>
Cc:     David Edmondson <david.edmondson@oracle.com>,
        LKML <linux-kernel@vger.kernel.org>,
        "H. Peter Anvin" <hpa@zytor.com>, Joerg Roedel <joro@8bytes.org>,
        "the arch/x86 maintainers" <x86@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        kvm list <kvm@vger.kernel.org>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        Vitaly Kuznetsov <vkuznets@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Feb 23, 2021 at 2:51 PM Sean Christopherson <seanjc@google.com> wrote:
>
> On Fri, Feb 19, 2021, David Edmondson wrote:
> > If the VM entry/exit controls for loading/saving MSR_EFER are either
> > not available (an older processor or explicitly disabled) or not
> > used (host and guest values are the same), reading GUEST_IA32_EFER
> > from the VMCS returns an inaccurate value.
> >
> > Because of this, in dump_vmcs() don't use GUEST_IA32_EFER to decide
> > whether to print the PDPTRs - do so if the EPT is in use and CR4.PAE
> > is set.
>
> This isn't necessarily correct either.  In a way, it's less correct as PDPTRs
> are more likely to be printed when they shouldn't, assuming most guests are
> 64-bit guests.  It's annoying to calculate the effective guest EFER, but so
> awful that it's worth risking confusion over PDTPRs.

I still prefer a dump_vmcs that always dumps every VMCS field. But if
you really want to skip printing the PDPTEs when they're irrelevant,
can't you just use the "IA-32e mode guest" VM-entry control as a proxy
for EFER.LMA?