linux-kernel - Re: [PATCH v1] KVM: VMX: enable guest access to LMCE related MSRs

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]

Date:	Fri, 3 Jun 2016 17:34:33 +0200
From:	Radim Krčmář <rkrcmar@...hat.com>
To:	Haozhong Zhang <haozhong.zhang@...el.com>
Cc:	kvm@...r.kernel.org, Paolo Bonzini <pbonzini@...hat.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>,
	"H . Peter Anvin" <hpa@...or.com>, x86@...nel.org,
	linux-kernel@...r.kernel.org, Gleb Natapov <gleb@...nel.org>,
	Boris Petkov <bp@...e.de>, Tony Luck <tony.luck@...el.com>,
	Ashok Raj <ashok.raj@...el.com>,
	Andi Kleen <andi.kleen@...el.com>
Subject: Re: [PATCH v1] KVM: VMX: enable guest access to LMCE related MSRs

2016-06-03 14:08+0800, Haozhong Zhang:
> On Intel platforms, this patch adds LMCE to KVM MCE supported
> capabilities and handles guest access to LMCE related MSRs.
> 
> Signed-off-by: Ashok Raj <ashok.raj@...el.com>
> Signed-off-by: Haozhong Zhang <haozhong.zhang@...el.com>
> ---
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> @@ -2863,6 +2863,11 @@ static int vmx_get_vmx_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata)
>  	return 0;
>  }
>  
> +static inline bool vmx_feature_control_msr_required(struct kvm_vcpu *vcpu)

I'd call it "present", rather than "required".  SDM does so for other
MSRs and it is easier to understand in the condition that returns #GP if
this function is false.

> +{
> +	return nested_vmx_allowed(vcpu) || (vcpu->arch.mcg_cap & MCG_LMCE_P);
> +}
> @@ -2904,8 +2909,15 @@ static int vmx_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
>  	case MSR_IA32_FEATURE_CONTROL:
> -		if (!nested_vmx_allowed(vcpu))
> +		if (!vmx_feature_control_msr_required(vcpu))
>  			return 1;
>  		msr_info->data = to_vmx(vcpu)->nested.msr_ia32_feature_control;

(MSR_IA32_FEATURE_CONTROL does not depend only on nested anymore, so
 moving msr_ia32_feature_control from struct nested_vmx to struct
 vcpu_vmx would make sense.)

>  		break;
> @@ -2997,8 +3009,17 @@ static int vmx_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> +	case MSR_IA32_MCG_EXT_CTL:
> +		if (!(vcpu->arch.mcg_cap & MCG_LMCE_P) ||
> +		    !(to_vmx(vcpu)->nested.msr_ia32_feature_control &
> +		      FEATURE_CONTROL_LMCE))

(This check is used twice and could be given a name too,
 "vmx_mcg_ext_ctl_msr_present()"?)

> +			return 1;
> +		if (data && data != 0x1)

  (data & ~MCG_EXT_CTL_LMCE_EN)

is a clearer check for reserved bits.

> +			return -1;
> +		vcpu->arch.mcg_ext_ctl = data;
> +		break;
>  	case MSR_IA32_FEATURE_CONTROL:
> -		if (!nested_vmx_allowed(vcpu) ||
> +		if (!vmx_feature_control_msr_required(vcpu) ||
>  		    (to_vmx(vcpu)->nested.msr_ia32_feature_control &
>  		     FEATURE_CONTROL_LOCKED && !msr_info->host_initiated))
>  			return 1;

Does hardware throw #GP when FEATURE_CONTROL_LMCE is set without
MCG_LMCE_P?

(We could emulate that by having a mask of valid bits and also use that
 mask in place of vmx_feature_control_msr_required().  I don't think
 there is a reason to have only vmx_feature_control_msr_required() if
 the hardware can #GP on individual bits too.)

Thanks.