linux-kernel - Re: [PATCH v10 21/39] KVM: nSVM: hyper-v: Enable L2 TLB flush

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <YyuCpelHQa00qNNF@google.com>
Date:   Wed, 21 Sep 2022 21:31:17 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     Vitaly Kuznetsov <vkuznets@...hat.com>
Cc:     kvm@...r.kernel.org, Paolo Bonzini <pbonzini@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Michael Kelley <mikelley@...rosoft.com>,
        Siddharth Chandrasekaran <sidcha@...zon.de>,
        Yuan Yao <yuan.yao@...ux.intel.com>,
        Maxim Levitsky <mlevitsk@...hat.com>,
        linux-hyperv@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v10 21/39] KVM: nSVM: hyper-v: Enable L2 TLB flush

On Wed, Sep 21, 2022, Vitaly Kuznetsov wrote:
> diff --git a/arch/x86/kvm/svm/hyperv.h b/arch/x86/kvm/svm/hyperv.h
> index dd2e393f84a0..7b01722838bf 100644
> --- a/arch/x86/kvm/svm/hyperv.h
> +++ b/arch/x86/kvm/svm/hyperv.h
> @@ -33,6 +33,9 @@ struct hv_enlightenments {
>   */
>  #define VMCB_HV_NESTED_ENLIGHTENMENTS VMCB_SW
>  
> +#define HV_SVM_EXITCODE_ENL 0xF0000000
> +#define HV_SVM_ENL_EXITCODE_TRAP_AFTER_FLUSH   (1)

Same as the synthetic VMX exit reason, these should go in hyperv-tlfs.h.  Keeping
these out of KVM also helps avoid the need for svm/hyperv.h.

https://learn.microsoft.com/en-us/virtualization/hyper-v-on-windows/tlfs/nested-virtualization#synthetic-vm-exit

> +
>  static inline void nested_svm_hv_update_vm_vp_ids(struct kvm_vcpu *vcpu)
>  {
>  	struct vcpu_svm *svm = to_svm(vcpu);
> @@ -48,6 +51,33 @@ static inline void nested_svm_hv_update_vm_vp_ids(struct kvm_vcpu *vcpu)
>  	hv_vcpu->nested.vp_id = hve->hv_vp_id;
>  }
>  
> +static inline bool

Strongly prefer 'int' with 0/-errno over a boolean.  Hrm, maybe add a prep patch
to convert kvm_hv_get_assist_page() to return 0/-errno?  That way this can still
return kvm_hv_get_assist_page() directly.

> nested_svm_hv_update_vp_assist(struct kvm_vcpu *vcpu)

Maybe s/update/verify since there isn't a true update anywhere?

> +{
> +	if (!to_hv_vcpu(vcpu))

This check isn't necessary, it's covered by kvm_hv_assist_page_enabled().

> +		return true;
> +
> +	if (!kvm_hv_assist_page_enabled(vcpu))
> +		return true;
> +
> +	return kvm_hv_get_assist_page(vcpu);

As mentioned earlier, I think this belongs in arch/x86/kvm/hyperv.h.