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;
Date:   Thu, 22 Apr 2021 16:15:07 +0000
From:   Sean Christopherson <seanjc@google.com>
To:     Paolo Bonzini <pbonzini@redhat.com>
Cc:     Dave Hansen <dave.hansen@linux.intel.com>,
        Andy Lutomirski <luto@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Wei Huang <wei.huang2@amd.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org, Borislav Petkov <bp@suse.de>,
        Tom Lendacky <thomas.lendacky@amd.com>,
        Brijesh Singh <brijesh.singh@amd.com>
Subject: Re: [PATCH v5 03/15] KVM: SVM: Disable SEV/SEV-ES if NPT is disabled
Message-ID: <YIGhC/1vlIAZfwzm@google.com>
References: <20210422021125.3417167-1-seanjc@google.com>
 <20210422021125.3417167-4-seanjc@google.com>
 <5e8a2d7d-67de-eef4-ab19-33294920f50c@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <5e8a2d7d-67de-eef4-ab19-33294920f50c@redhat.com>
Precedence: bulk

On Thu, Apr 22, 2021, Paolo Bonzini wrote:
> On 22/04/21 04:11, Sean Christopherson wrote:
> > Disable SEV and SEV-ES if NPT is disabled.  While the APM doesn't clearly
> > state that NPT is mandatory, it's alluded to by:
> > 
> >    The guest page tables, managed by the guest, may mark data memory pages
> >    as either private or shared, thus allowing selected pages to be shared
> >    outside the guest.
> > 
> > And practically speaking, shadow paging can't work since KVM can't read
> > the guest's page tables.
> > 
> > Fixes: e9df09428996 ("KVM: SVM: Add sev module_param")
> > Cc: Brijesh Singh <brijesh.singh@amd.com
> > Cc: Tom Lendacky <thomas.lendacky@amd.com>
> > Signed-off-by: Sean Christopherson <seanjc@google.com>
> > ---
> >   arch/x86/kvm/svm/svm.c | 30 +++++++++++++++---------------
> >   1 file changed, 15 insertions(+), 15 deletions(-)
> > 
> > diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> > index fed153314aef..0e8489908216 100644
> > --- a/arch/x86/kvm/svm/svm.c
> > +++ b/arch/x86/kvm/svm/svm.c
> > @@ -970,7 +970,21 @@ static __init int svm_hardware_setup(void)
> >   		kvm_enable_efer_bits(EFER_SVME | EFER_LMSLE);
> >   	}
> > -	if (IS_ENABLED(CONFIG_KVM_AMD_SEV) && sev) {
> > +	/*
> > +	 * KVM's MMU doesn't support using 2-level paging for itself, and thus
> > +	 * NPT isn't supported if the host is using 2-level paging since host
> > +	 * CR4 is unchanged on VMRUN.
> > +	 */
> > +	if (!IS_ENABLED(CONFIG_X86_64) && !IS_ENABLED(CONFIG_X86_PAE))
> > +		npt_enabled = false;
> 
> Unrelated, but since you're moving this code: should we be pre-scient and
> tackle host 5-level paging as well?
> 
> Support for 5-level page tables on NPT is not hard to fix and could be
> tested by patching QEMU.  However, the !NPT case would also have to be fixed
> by extending the PDP and PML4 stacking trick to a PML5.

Isn't that backwards?  It's the nested NPT case that requires the stacking trick.
When !NPT is disabled in L0 KVM, 32-bit guests are run with PAE paging.  Maybe
I'm misunderstanding what you're suggesting.
 
> However, without real hardware to test on I'd be a bit wary to do it.
> Looking at 5-level EPT there might be other issues (e.g. what's the guest
> MAXPHYADDR) and I would prefer to see what AMD comes up with exactly in the
> APM.  So I would just block loading KVM on hypothetical AMD hosts with
> CR4.LA57=1.

Agreed, I think blocking KVM makes the most sense.