linux-kernel - Re: [PATCH] KVM: nSVM: Additions to optimizing L12 to L2 vmcb.save copies

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]

Message-ID: <YFoSj9xmCzHe0/X7@google.com>
Date:   Tue, 23 Mar 2021 16:08:47 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     Cathy Avery <cavery@...hat.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
        pbonzini@...hat.com, vkuznets@...hat.com, wei.huang2@....com
Subject: Re: [PATCH] KVM: nSVM: Additions to optimizing L12 to L2 vmcb.save
 copies

On Wed, Mar 17, 2021, Cathy Avery wrote:
> Extend using the vmcb12 control clean field to determine which
> vmcb12.save registers were marked dirty in order to minimize
> register copies by including the CR bit.
> 
> This patch also fixes the init of last_vmcb12_gpa by using an invalid
> physical address instead of 0.
> 
> Tested:
> kvm-unit-tests
> kvm selftests
> Fedora L1 L2
> 
> Signed-off-by: Cathy Avery <cavery@...hat.com>
> ---
>  arch/x86/kvm/svm/nested.c | 9 ++++++---
>  arch/x86/kvm/svm/svm.c    | 2 +-
>  2 files changed, 7 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
> index 8523f60adb92..6f9a40e002bc 100644
> --- a/arch/x86/kvm/svm/nested.c
> +++ b/arch/x86/kvm/svm/nested.c
> @@ -449,9 +449,12 @@ static void nested_vmcb02_prepare_save(struct vcpu_svm *svm, struct vmcb *vmcb12
>  	}
>  
>  	kvm_set_rflags(&svm->vcpu, vmcb12->save.rflags | X86_EFLAGS_FIXED);
> -	svm_set_efer(&svm->vcpu, vmcb12->save.efer);
> -	svm_set_cr0(&svm->vcpu, vmcb12->save.cr0);
> -	svm_set_cr4(&svm->vcpu, vmcb12->save.cr4);
> +
> +	if (unlikely(new_vmcb12 || vmcb_is_dirty(vmcb12, VMCB_CR))) {
> +		svm_set_efer(&svm->vcpu, vmcb12->save.efer);
> +		svm_set_cr0(&svm->vcpu, vmcb12->save.cr0);
> +		svm_set_cr4(&svm->vcpu, vmcb12->save.cr4);

This doesn't seem correct.  Regardless of when vmcb12 was last touched, KVM
still needs to set L2's state in vcpu->arch and set the correct MMU context.

> +	}
>  
>  	svm->vcpu.arch.cr2 = vmcb12->save.cr2;
>  
> diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> index 271196400495..41f5cd1009ca 100644
> --- a/arch/x86/kvm/svm/svm.c
> +++ b/arch/x86/kvm/svm/svm.c
> @@ -1234,7 +1234,7 @@ static void init_vmcb(struct kvm_vcpu *vcpu)
>  	svm->asid = 0;
>  
>  	svm->nested.vmcb12_gpa = 0;
> -	svm->nested.last_vmcb12_gpa = 0;
> +	svm->nested.last_vmcb12_gpa = -1;

INVALID_PAGE would be even better.

>  	vcpu->arch.hflags = 0;
>  
>  	if (!kvm_pause_in_guest(vcpu->kvm)) {
> -- 
> 2.26.2
>