linux-kernel - Re: [PATCH v2 4/4] KVM: SVM: Treat all "skip" emulation for SEV guests as outright failures

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: <9dddb2ef-e021-087d-f0ea-9e0e3d8843b9@amd.com>
Date:   Fri, 25 Aug 2023 08:43:50 -0500
From:   Tom Lendacky <thomas.lendacky@....com>
To:     Sean Christopherson <seanjc@...gle.com>,
        Paolo Bonzini <pbonzini@...hat.com>
Cc:     kvm@...r.kernel.org, linux-kernel@...r.kernel.org,
        Wu Zongyo <wuzongyo@...l.ustc.edu.cn>
Subject: Re: [PATCH v2 4/4] KVM: SVM: Treat all "skip" emulation for SEV
 guests as outright failures

On 8/24/23 20:36, Sean Christopherson wrote:
> Treat EMULTYPE_SKIP failures on SEV guests as unhandleable emulation
> instead of simply resuming the guest, and drop the hack-a-fix which
> effects that behavior for the INT3/INTO injection path.  If KVM can't
> skip an instruction for which KVM has already done partial emulation,
> resuming the guest is undesirable as doing so may corrupt guest state.
> 
> Signed-off-by: Sean Christopherson <seanjc@...gle.com>
> ---
>   arch/x86/kvm/svm/svm.c | 12 +-----------
>   arch/x86/kvm/x86.c     |  9 +++++++--
>   2 files changed, 8 insertions(+), 13 deletions(-)
> 
> diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> index 39ce680013c4..fc2cd5585349 100644
> --- a/arch/x86/kvm/svm/svm.c
> +++ b/arch/x86/kvm/svm/svm.c
> @@ -364,8 +364,6 @@ static void svm_set_interrupt_shadow(struct kvm_vcpu *vcpu, int mask)
>   		svm->vmcb->control.int_state |= SVM_INTERRUPT_SHADOW_MASK;
>   
>   }
> -static int svm_check_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type,
> -					 void *insn, int insn_len);
>   
>   static int __svm_skip_emulated_instruction(struct kvm_vcpu *vcpu,
>   					   bool commit_side_effects)
> @@ -386,14 +384,6 @@ static int __svm_skip_emulated_instruction(struct kvm_vcpu *vcpu,
>   	}
>   
>   	if (!svm->next_rip) {
> -		/*
> -		 * FIXME: Drop this when kvm_emulate_instruction() does the
> -		 * right thing and treats "can't emulate" as outright failure
> -		 * for EMULTYPE_SKIP.
> -		 */
> -		if (svm_check_emulate_instruction(vcpu, EMULTYPE_SKIP, NULL, 0) != X86EMUL_CONTINUE)
> -			return 0;
> -
>   		if (unlikely(!commit_side_effects))
>   			old_rflags = svm->vmcb->save.rflags;
>   
> @@ -4752,7 +4742,7 @@ static int svm_check_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type,
>   	 */
>   	if (unlikely(!insn)) {
>   		if (emul_type & EMULTYPE_SKIP)
> -			return X86EMUL_RETRY_INSTR;
> +			return X86EMUL_UNHANDLEABLE;

Trying to follow this, bear with me...

This results in an "emulation failure" which fills out all the KVM 
userspace exit information in prepare_emulation_failure_exit(). But 
because of the return 0 in handle_emulation_failure(), in the end this 
ends up just acting like the first patch because we exit out 
svm_update_soft_interrupt_rip() early and the instruction just gets retried?

Thanks,
Tom

>   
>   		kvm_queue_exception(vcpu, UD_VECTOR);
>   		return X86EMUL_PROPAGATE_FAULT;
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index f897d582d560..1f4a8fbc5390 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -8858,8 +8858,13 @@ int x86_emulate_instruction(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
>   	bool writeback = true;
>   
>   	r = kvm_check_emulate_insn(vcpu, emulation_type, insn, insn_len);
> -	if (r != X86EMUL_CONTINUE)
> -		return 1;
> +	if (r != X86EMUL_CONTINUE) {
> +		if (r == X86EMUL_RETRY_INSTR || r == X86EMUL_PROPAGATE_FAULT)
> +			return 1;
> +
> +		WARN_ON_ONCE(r != X86EMUL_UNHANDLEABLE);
> +		return handle_emulation_failure(vcpu, emulation_type);
> +	}
>   
>   	vcpu->arch.l1tf_flush_l1d = true;
>