From mboxrd@z Thu Jan  1 00:00:00 1970
From: Joerg Roedel <joerg.roedel@amd.com>
Subject: Re: KVM: SVM: force new asid on vcpu migration
Date: Tue, 21 Jul 2009 17:56:59 +0200
Message-ID: <20090721155659.GJ25756@amd.com>
References: <20090721154745.GA4188@amt.cnet>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Cc: kvm <kvm@vger.kernel.org>
To: Marcelo Tosatti <mtosatti@redhat.com>
Return-path: <kvm-owner@vger.kernel.org>
Received: from tx2ehsobe002.messaging.microsoft.com ([65.55.88.12]:12682 "EHLO
	TX2EHSOBE004.bigfish.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1755116AbZGUP5V (ORCPT <rfc822;kvm@vger.kernel.org>);
	Tue, 21 Jul 2009 11:57:21 -0400
Content-Disposition: inline
In-Reply-To: <20090721154745.GA4188@amt.cnet>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On Tue, Jul 21, 2009 at 12:47:45PM -0300, Marcelo Tosatti wrote:
> 
> If a migrated vcpu matches the asid_generation value of the target pcpu,
> there will be no TLB flush via TLB_CONTROL_FLUSH_ALL_ASID.
> 
> The check for vcpu.cpu in pre_svm_run is meaningless since svm_vcpu_load 
> already updated it on schedule in.
> 
> Such vcpu will VMRUN with stale TLB entries.
> 
> Based on original patch from Joerg Roedel (http://patchwork.kernel.org/patch/10021/)
> 
> Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

Acked-by: Joerg Roedel <joerg.roedel@amd.com>

> 
> diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
> index 18085d3..90fe88f 100644
> --- a/arch/x86/kvm/svm.c
> +++ b/arch/x86/kvm/svm.c
> @@ -739,6 +739,7 @@ static void svm_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
>  		svm->vmcb->control.tsc_offset += delta;
>  		vcpu->cpu = cpu;
>  		kvm_migrate_timers(vcpu);
> +		svm->asid_generation = 0;
>  	}
>  
>  	for (i = 0; i < NR_HOST_SAVE_USER_MSRS; i++)
> @@ -1071,7 +1072,6 @@ static void new_asid(struct vcpu_svm *svm, struct svm_cpu_data *svm_data)
>  		svm->vmcb->control.tlb_ctl = TLB_CONTROL_FLUSH_ALL_ASID;
>  	}
>  
> -	svm->vcpu.cpu = svm_data->cpu;
>  	svm->asid_generation = svm_data->asid_generation;
>  	svm->vmcb->control.asid = svm_data->next_asid++;
>  }
> @@ -2320,8 +2320,8 @@ static void pre_svm_run(struct vcpu_svm *svm)
>  	struct svm_cpu_data *svm_data = per_cpu(svm_data, cpu);
>  
>  	svm->vmcb->control.tlb_ctl = TLB_CONTROL_DO_NOTHING;
> -	if (svm->vcpu.cpu != cpu ||
> -	    svm->asid_generation != svm_data->asid_generation)
> + 	/* FIXME: handle wraparound of asid_generation */
> +	if (svm->asid_generation != svm_data->asid_generation)
>  		new_asid(svm, svm_data);
>  }
>  
> 

--