linux-kernel - Re: [PATCH] KVM: x86: init x2apic

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: <20190724164826.GC25376@linux.intel.com>
Date:   Wed, 24 Jul 2019 09:48:26 -0700
From:   Sean Christopherson <sean.j.christopherson@...el.com>
To:     luferry@....com
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        "H. Peter Anvin" <hpa@...or.com>, x86@...nel.org,
        kvm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] KVM: x86: init x2apic_enabled() once

On Tue, Jul 23, 2019 at 09:06:08PM +0800, luferry@....com wrote:
> From: luferry <luferry@....com>
> 
> x2apic_eanbled() costs about 200 cycles
> when guest trigger halt pretty high, pi ops in hotpath
> 
> Signed-off-by: luferry <luferry@....com>
> ---
>  arch/x86/kvm/vmx/vmx.c | 10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index d98eac371c0a..e17dbf011e47 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -186,6 +186,8 @@ static DEFINE_STATIC_KEY_FALSE(vmx_l1d_should_flush);
>  static DEFINE_STATIC_KEY_FALSE(vmx_l1d_flush_cond);
>  static DEFINE_MUTEX(vmx_l1d_flush_mutex);
>  
> +static int __read_mostly host_x2apic_enabled;
> +
>  /* Storage for pre module init parameter parsing */
>  static enum vmx_l1d_flush_state __read_mostly vmentry_l1d_flush_param = VMENTER_L1D_FLUSH_AUTO;
>  
> @@ -1204,7 +1206,7 @@ static void vmx_vcpu_pi_load(struct kvm_vcpu *vcpu, int cpu)
>  
>  		dest = cpu_physical_id(cpu);
>  
> -		if (x2apic_enabled())
> +		if (host_x2apic_enabled)

Instead of caching x2apic_enabled(), this can be:

	if (x2apic_supported() && x2apic_mode)

which will get compiled out if CONFIG_X86_X2APIC=n.

It's quite suprising (to me) that 2apic_enabled() reads the MSR in the
first place, but at a glance the other users of x2apic_enabled() do need
to query the MSR and/or may be called before x2apic_mode is set.

>  			new.ndst = dest;
>  		else
>  			new.ndst = (dest << 8) & 0xFF00;
> @@ -7151,7 +7153,7 @@ static void __pi_post_block(struct kvm_vcpu *vcpu)
>  
>  		dest = cpu_physical_id(vcpu->cpu);
>  
> -		if (x2apic_enabled())
> +		if (host_x2apic_enabled)
>  			new.ndst = dest;
>  		else
>  			new.ndst = (dest << 8) & 0xFF00;
> @@ -7221,7 +7223,7 @@ static int pi_pre_block(struct kvm_vcpu *vcpu)
>  		 */
>  		dest = cpu_physical_id(vcpu->pre_pcpu);
>  
> -		if (x2apic_enabled())
> +		if (host_x2apic_enabled)
>  			new.ndst = dest;
>  		else
>  			new.ndst = (dest << 8) & 0xFF00;
> @@ -7804,6 +7806,8 @@ static int __init vmx_init(void)
>  	}
>  #endif
>  
> +	host_x2apic_enabled = x2apic_enabled();
> +
>  	r = kvm_init(&vmx_x86_ops, sizeof(struct vcpu_vmx),
>  		     __alignof__(struct vcpu_vmx), THIS_MODULE);
>  	if (r)
> -- 
> 2.14.1.40.g8e62ba1
> 
>