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;
Message-ID: <7458497a8694ba0fbabee28eabf557e6e4406fbe.camel@redhat.com>
Subject: Re: [PATCH] KVM: nSVM: Pull CS.Base from actual VMCB12 for soft
 int/ex re-injection
From:   Maxim Levitsky <mlevitsk@redhat.com>
To:     "Maciej S. Szmigiero" <mail@maciej.szmigiero.name>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Sean Christopherson <seanjc@google.com>
Cc:     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
Date:   Wed, 20 Jul 2022 11:43:55 +0300
In-Reply-To: <4caa0f67589ae3c22c311ee0e6139496902f2edc.1658159083.git.maciej.szmigiero@oracle.com>
References: <4caa0f67589ae3c22c311ee0e6139496902f2edc.1658159083.git.maciej.szmigiero@oracle.com>
Content-Type: text/plain; charset="UTF-8"
User-Agent: Evolution 3.40.4 (3.40.4-5.fc34) 
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

On Mon, 2022-07-18 at 17:47 +0200, Maciej S. Szmigiero wrote:
> From: "Maciej S. Szmigiero" <maciej.szmigiero@oracle.com>
> 
> enter_svm_guest_mode() first calls nested_vmcb02_prepare_control() to copy
> control fields from VMCB12 to the current VMCB, then
> nested_vmcb02_prepare_save() to perform a similar copy of the save area.
> 
> This means that nested_vmcb02_prepare_control() still runs with the
> previous save area values in the current VMCB so it shouldn't take the L2
> guest CS.Base from this area.
> 
> Explicitly pull CS.Base from the actual VMCB12 instead in
> enter_svm_guest_mode().
> 
> Granted, having a non-zero CS.Base is a very rare thing (and even
> impossible in 64-bit mode), having it change between nested VMRUNs is
> probably even rarer, but if it happens it would create a really subtle bug
> so it's better to fix it upfront.
> 
> Fixes: 6ef88d6e36c2 ("KVM: SVM: Re-inject INT3/INTO instead of retrying the instruction")
> Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
> ---
>  arch/x86/kvm/svm/nested.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
> index adf4120b05d90..23252ab821941 100644
> --- a/arch/x86/kvm/svm/nested.c
> +++ b/arch/x86/kvm/svm/nested.c
> @@ -639,7 +639,8 @@ static bool is_evtinj_nmi(u32 evtinj)
>  }
>  
>  static void nested_vmcb02_prepare_control(struct vcpu_svm *svm,
> -                                         unsigned long vmcb12_rip)
> +                                         unsigned long vmcb12_rip,
> +                                         unsigned long vmcb12_csbase)

Honestly I don't like that nested_vmcb02_prepare_control starts to grow its parameter list,
because it kind of defeats the purpose of vmcb12 cache we added back then.

I think that it is better to add csbase/rip to vmcb_save_area_cached,
but I am not 100% sure. What do you think?

Best regards,
	Maxim Levitsky


>  {
>         u32 int_ctl_vmcb01_bits = V_INTR_MASKING_MASK;
>         u32 int_ctl_vmcb12_bits = V_TPR_MASK | V_IRQ_INJECTION_BITS_MASK;
> @@ -711,7 +712,7 @@ static void nested_vmcb02_prepare_control(struct vcpu_svm *svm,
>         svm->nmi_l1_to_l2 = is_evtinj_nmi(vmcb02->control.event_inj);
>         if (is_evtinj_soft(vmcb02->control.event_inj)) {
>                 svm->soft_int_injected = true;
> -               svm->soft_int_csbase = svm->vmcb->save.cs.base;
> +               svm->soft_int_csbase = vmcb12_csbase;
>                 svm->soft_int_old_rip = vmcb12_rip;
>                 if (svm->nrips_enabled)
>                         svm->soft_int_next_rip = svm->nested.ctl.next_rip;
> @@ -800,7 +801,7 @@ int enter_svm_guest_mode(struct kvm_vcpu *vcpu, u64 vmcb12_gpa,
>         nested_svm_copy_common_state(svm->vmcb01.ptr, svm->nested.vmcb02.ptr);
>  
>         svm_switch_vmcb(svm, &svm->nested.vmcb02);
> -       nested_vmcb02_prepare_control(svm, vmcb12->save.rip);
> +       nested_vmcb02_prepare_control(svm, vmcb12->save.rip, vmcb12->save.cs.base);
>         nested_vmcb02_prepare_save(svm, vmcb12);
>  
>         ret = nested_svm_load_cr3(&svm->vcpu, svm->nested.save.cr3,
> @@ -1663,7 +1664,7 @@ static int svm_set_nested_state(struct kvm_vcpu *vcpu,
>         nested_copy_vmcb_control_to_cache(svm, ctl);
>  
>         svm_switch_vmcb(svm, &svm->nested.vmcb02);
> -       nested_vmcb02_prepare_control(svm, svm->vmcb->save.rip);
> +       nested_vmcb02_prepare_control(svm, svm->vmcb->save.rip, svm->vmcb->save.cs.base);
>  
>         /*
>          * While the nested guest CR3 is already checked and set by
>