linux-kernel - Re: [PATCH v3 1/3] KVM: SEV-ES: Prevent MSR access post VMSA encryption

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] [day] [month] [year] [list]

Date: Tue, 4 Jun 2024 18:10:02 -0500
From: Michael Roth <michael.roth@....com>
To: Ravi Bangoria <ravi.bangoria@....com>
CC: Paolo Bonzini <pbonzini@...hat.com>, <seanjc@...gle.com>,
	<nikunj.dadhania@....com>, <thomas.lendacky@....com>, <tglx@...utronix.de>,
	<mingo@...hat.com>, <bp@...en8.de>, <dave.hansen@...ux.intel.com>,
	<x86@...nel.org>, <hpa@...or.com>, <pankaj.gupta@....com>,
	<kvm@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
	<santosh.shukla@....com>
Subject: Re: [PATCH v3 1/3] KVM: SEV-ES: Prevent MSR access post VMSA
 encryption

On Wed, May 29, 2024 at 04:14:10PM +0530, Ravi Bangoria wrote:
> On 5/28/2024 10:01 PM, Paolo Bonzini wrote:
> > On 5/23/24 14:18, Ravi Bangoria wrote:
> >> From: Nikunj A Dadhania <nikunj@....com>
> >>
> >> KVM currently allows userspace to read/write MSRs even after the VMSA is
> >> encrypted. This can cause unintentional issues if MSR access has side-
> >> effects. For ex, while migrating a guest, userspace could attempt to
> >> migrate MSR_IA32_DEBUGCTLMSR and end up unintentionally disabling LBRV on
> >> the target. Fix this by preventing access to those MSRs which are context
> >> switched via the VMSA, once the VMSA is encrypted.
> >>
> >> Suggested-by: Sean Christopherson <seanjc@...gle.com>
> >> Signed-off-by: Nikunj A Dadhania <nikunj@....com>
> >> Signed-off-by: Ravi Bangoria <ravi.bangoria@....com>
> >> ---
> >>   arch/x86/kvm/svm/svm.c | 18 ++++++++++++++++++
> >>   1 file changed, 18 insertions(+)
> >>
> >> diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> >> index 3d0549ca246f..489b0183f37d 100644
> >> --- a/arch/x86/kvm/svm/svm.c
> >> +++ b/arch/x86/kvm/svm/svm.c
> >> @@ -2834,10 +2834,24 @@ static int svm_get_msr_feature(struct kvm_msr_entry *msr)
> >>       return 0;
> >>   }
> >>   +static bool
> >> +sev_es_prevent_msr_access(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> >> +{
> >> +    return sev_es_guest(vcpu->kvm) &&
> >> +           vcpu->arch.guest_state_protected &&
> >> +           svm_msrpm_offset(msr_info->index) != MSR_INVALID &&
> >> +           !msr_write_intercepted(vcpu, msr_info->index);
> >> +}
> >> +
> >>   static int svm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> >>   {
> >>       struct vcpu_svm *svm = to_svm(vcpu);
> >>   +    if (sev_es_prevent_msr_access(vcpu, msr_info)) {
> >> +        msr_info->data = 0;
> >> +        return 0;
> > 
> > This should return -EINVAL, not 0.  Likewise below in svm_set_msr().
> 
> Sure.

One consequence of this change is that older VMMs that might still call
svm_get_msr()/svm_set_msr() for SEV-ES guests.

Newer VMMs that are aware of KVM_SEV_INIT2 however are already aware of
the stricter limitations of what vCPU state can be sync'd during
guest run-time, so newer QEMU for instance will work both for legacy
KVM_SEV_ES_INIT interface as well as KVM_SEV_INIT2.

So when using KVM_SEV_INIT2 it's okay to assume userspace can deal with
-EINVAL, whereas for legacy KVM_SEV_ES_INIT we sort of have to assume the
VMM does not have the necessary changes to deal with -EINVAL, so in that
case it's probably more appropriate to return 0 and just silently noop.

We had a similar situations with stricter limitations on fpstate sync'ing
for KVM_SEV_INIT2 and that was the approach taken there:

  https://lore.kernel.org/kvm/ZfRhu0GVjWeAAJMB@google.com/

so I'll submit a patch that takes the same approach.

-Mike

> 
> Thanks,
> Ravi