Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 29 Jun 2023 09:50:34 -0700
In-Reply-To: <Y/ZFJfspU6L2RmQS@google.com>
Mime-Version: 1.0
References: <20230217231022.816138-1-seanjc@google.com> <20230217231022.816138-9-seanjc@google.com>
 <20230221152349.ulcjtbnvziair7ff@linux.intel.com> <20230221153306.qubx7tfmasnvodeu@linux.intel.com>
 <Y/VYN3n/lHePiDxM@google.com> <20230222064931.ppz6berhfr4edewf@linux.intel.com>
 <Y/ZFJfspU6L2RmQS@google.com>
Message-ID: <ZJ22Wts4WKKD19bN@google.com>
Subject: Re: [PATCH 08/12] KVM: nSVM: Use KVM-governed feature framework to
 track "vVM{SAVE,LOAD} enabled"
From:   Sean Christopherson <seanjc@google.com>
To:     Yu Zhang <yu.c.zhang@linux.intel.com>
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org, Maxim Levitsky <mlevitsk@redhat.com>
Content-Type: text/plain; charset="us-ascii"
Precedence: bulk

On Wed, Feb 22, 2023, Sean Christopherson wrote:
> +Maxim
> 
> On Wed, Feb 22, 2023, Yu Zhang wrote:
> I'll opportunistically massage the comment to make it more explicit about why
> VMLOAD needs to be intercepted.
>  
> That said, clearing the bits for this seems wrong.  That would corrupt the MSRs
> for 64-bit Intel guests.  The "target" of the fix was 32-bit L2s, i.e. I doubt
> anything would notice.
> 
>     This patch fixes nested migration of 32 bit nested guests, that was
>     broken because incorrect cached values of SYSENTER msrs were stored in
>     the migration stream if L1 changed these msrs with
>     vmload prior to L2 entry.

Aha!  Finally figured out what this code is doing.  KVM intercepts VMLOAD so that
KVM can correctly model the VMLOAD behavior of dropping bits 63:32, i.e. to clear
svm->sysenter_eip_hi and svm->sysenter_esp_hi.

So the code is correct.  I'll add this comment:

	/*
	 * Intercept VMLOAD if the vCPU mode is Intel in order to emulate that
	 * VMLOAD drops bits 63:32 of SYSENTER (ignoring the fact that exposing
	 * SVM on Intel is bonkers and extremely unlikely to work).
	 */