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;
From:   Vitaly Kuznetsov <vkuznets@redhat.com>
To:     Tom Lendacky <thomas.lendacky@amd.com>
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Sean Christopherson <sean.j.christopherson@intel.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>,
        Brijesh Singh <brijesh.singh@amd.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        "H. Peter Anvin" <hpa@zytor.com>, kvm@vger.kernel.org,
        x86@kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] KVM: SVM: Add a dedicated INVD intercept routine
In-Reply-To: <16f36f9a51608758211c54564cd17c8b909372f1.1600892859.git.thomas.lendacky@amd.com>
References: <16f36f9a51608758211c54564cd17c8b909372f1.1600892859.git.thomas.lendacky@amd.com>
Date:   Thu, 24 Sep 2020 15:58:24 +0200
Message-ID: <87zh5fcm4f.fsf@vitty.brq.redhat.com>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

Tom Lendacky <thomas.lendacky@amd.com> writes:

> From: Tom Lendacky <thomas.lendacky@amd.com>
>
> The INVD instruction intercept performs emulation. Emulation can't be done
> on an SEV guest because the guest memory is encrypted.
>
> Provide a dedicated intercept routine for the INVD intercept. Within this
> intercept routine just skip the instruction for an SEV guest, since it is
> emulated as a NOP anyway.
>
> Fixes: 1654efcbc431 ("KVM: SVM: Add KVM_SEV_INIT command")
> Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
> ---
>  arch/x86/kvm/svm/svm.c | 13 ++++++++++++-
>  1 file changed, 12 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> index c91acabf18d0..332ec4425d89 100644
> --- a/arch/x86/kvm/svm/svm.c
> +++ b/arch/x86/kvm/svm/svm.c
> @@ -2183,6 +2183,17 @@ static int iret_interception(struct vcpu_svm *svm)
>  	return 1;
>  }
>  
> +static int invd_interception(struct vcpu_svm *svm)
> +{
> +	/*
> +	 * Can't do emulation on an SEV guest and INVD is emulated
> +	 * as a NOP, so just skip the instruction.
> +	 */
> +	return (sev_guest(svm->vcpu.kvm))
> +		? kvm_skip_emulated_instruction(&svm->vcpu)
> +		: kvm_emulate_instruction(&svm->vcpu, 0);
> +}
> +
>  static int invlpg_interception(struct vcpu_svm *svm)
>  {
>  	if (!static_cpu_has(X86_FEATURE_DECODEASSISTS))
> @@ -2774,7 +2785,7 @@ static int (*const svm_exit_handlers[])(struct vcpu_svm *svm) = {
>  	[SVM_EXIT_RDPMC]			= rdpmc_interception,
>  	[SVM_EXIT_CPUID]			= cpuid_interception,
>  	[SVM_EXIT_IRET]                         = iret_interception,
> -	[SVM_EXIT_INVD]                         = emulate_on_interception,
> +	[SVM_EXIT_INVD]                         = invd_interception,
>  	[SVM_EXIT_PAUSE]			= pause_interception,
>  	[SVM_EXIT_HLT]				= halt_interception,
>  	[SVM_EXIT_INVLPG]			= invlpg_interception,

Out of pure curiosity,

does it sill make sense to intercept INVD when we just skip it? Would it
rather make sense to disable INVD intercept for SEV guests completely?

-- 
Vitaly