Re: [PATCH] KVM: x86: add hint to skip hidden rdpkru under kvm_load_host_xsave_state

From: Paolo Bonzini <pbonzini@redhat.com>
To: Jon Kohler <jon@nutanix.com>
Cc: Babu Moger <babu.moger@amd.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
	x86@kernel.org, "H. Peter Anvin" <hpa@zytor.com>,
	Sean Christopherson <seanjc@google.com>,
	Vitaly Kuznetsov <vkuznets@redhat.com>,
	Wanpeng Li <wanpengli@tencent.com>,
	Jim Mattson <jmattson@google.com>, Joerg Roedel <joro@8bytes.org>,
	Dave Hansen <dave.hansen@intel.com>,
	Fenghua Yu <fenghua.yu@intel.com>,
	Yu-cheng Yu <yu-cheng.yu@intel.com>,
	Tony Luck <tony.luck@intel.com>, Uros Bizjak <ubizjak@gmail.com>,
	Petteri Aimonen <jpa@git.mail.kapsi.fi>,
	Kan Liang <kan.liang@linux.intel.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Mike Rapoport <rppt@kernel.org>,
	Benjamin Thiel <b.thiel@posteo.de>,
	Fan Yang <Fan_Yang@sjtu.edu.cn>, Juergen Gross <jgross@suse.com>,
	Dave Jiang <dave.jiang@intel.com>,
	"Peter Zijlstra (Intel)" <peterz@infradead.org>,
	Ricardo Neri <ricardo.neri-calderon@linux.intel.com>,
	Arvind Sankar <nivedita@alum.mit.edu>,
	linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Subject: Re: [PATCH] KVM: x86: add hint to skip hidden rdpkru under kvm_load_host_xsave_state
Date: Fri, 7 May 2021 18:52:11 +0200	[thread overview]
Message-ID: <16af038e-5b30-2509-4e3b-3b3ed9d4b81e@redhat.com> (raw)
In-Reply-To: <20210507164456.1033-1-jon@nutanix.com>

Nice one.  The patch can be made simpler though (I think).

On 07/05/21 18:44, Jon Kohler wrote:
  @@ -122,7 +124,7 @@ static inline u32 rdpkru(void)
>   	return 0;
>   }
> 
> -static inline void __write_pkru(u32 pkru)
> +static inline void __write_pkru(u32 pkru, bool skip_comparison)
>   {
>   }
>   #endif
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index cebdaa1e3cf5..cd95adbd140c 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -912,10 +912,10 @@ void kvm_load_guest_xsave_state(struct kvm_vcpu *vcpu)
>   	}
> 
>   	if (static_cpu_has(X86_FEATURE_PKU) &&
> -	    (kvm_read_cr4_bits(vcpu, X86_CR4_PKE) ||
> -	     (vcpu->arch.xcr0 & XFEATURE_MASK_PKRU)) &&
> -	    vcpu->arch.pkru != vcpu->arch.host_pkru)
> -		__write_pkru(vcpu->arch.pkru);
> +	    vcpu->arch.pkru != vcpu->arch.host_pkru &&
> +	    ((vcpu->arch.xcr0 & XFEATURE_MASK_PKRU) ||
> +	     kvm_read_cr4_bits(vcpu, X86_CR4_PKE)))
> +		__write_pkru(vcpu->arch.pkru, false);

This can be optimized as well, can't it?  This means that the only case 
that needs the rdpkru is in switch_fpu_finish, and __write_pkru can be 
removed completely:

- do the rdpkru+wrpkru in switch_fpu_finish

- just use wrpkru in KVM

Paolo

>   }
>   EXPORT_SYMBOL_GPL(kvm_load_guest_xsave_state);
> 
> @@ -925,11 +925,11 @@ void kvm_load_host_xsave_state(struct kvm_vcpu *vcpu)
>   		return;
> 
>   	if (static_cpu_has(X86_FEATURE_PKU) &&
> -	    (kvm_read_cr4_bits(vcpu, X86_CR4_PKE) ||
> -	     (vcpu->arch.xcr0 & XFEATURE_MASK_PKRU))) {
> +	    ((vcpu->arch.xcr0 & XFEATURE_MASK_PKRU) ||
> +	     kvm_read_cr4_bits(vcpu, X86_CR4_PKE))) {
>   		vcpu->arch.pkru = rdpkru();
>   		if (vcpu->arch.pkru != vcpu->arch.host_pkru)
> -			__write_pkru(vcpu->arch.host_pkru);
> +			__write_pkru(vcpu->arch.host_pkru, true);
>   	}
> 
>   	if (kvm_read_cr4_bits(vcpu, X86_CR4_OSXSAVE)) {
> --
> 2.30.1 (Apple Git-130)
> 