Re: [PATCH v6 4/7] KVM: x86: hyperv: keep track of mismatched VP indexes

From: Roman Kagan <rkagan@virtuozzo.com>
To: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: kvm@vger.kernel.org, "Paolo Bonzini" <pbonzini@redhat.com>,
	"Radim Krčmář" <rkrcmar@redhat.com>,
	"K. Y. Srinivasan" <kys@microsoft.com>,
	"Haiyang Zhang" <haiyangz@microsoft.com>,
	"Stephen Hemminger" <sthemmin@microsoft.com>,
	"Michael Kelley (EOSG)" <Michael.H.Kelley@microsoft.com>,
	"Mohammed Gamal" <mmorsy@redhat.com>,
	"Cathy Avery" <cavery@redhat.com>,
	"Wanpeng Li" <wanpeng.li@hotmail.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v6 4/7] KVM: x86: hyperv: keep track of mismatched VP indexes
Date: Thu, 27 Sep 2018 10:59:12 +0300	[thread overview]
Message-ID: <20180927075911.GB4186@rkaganb.sw.ru> (raw)
In-Reply-To: <20180926170259.29796-5-vkuznets@redhat.com>

On Wed, Sep 26, 2018 at 07:02:56PM +0200, Vitaly Kuznetsov wrote:
> In most common cases VP index of a vcpu matches its vcpu index. Userspace
> is, however, free to set any mapping it wishes and we need to account for
> that when we need to find a vCPU with a particular VP index. To keep search
> algorithms optimal in both cases introduce 'num_mismatched_vp_indexes'
> counter showing how many vCPUs with mismatching VP index we have. In case
> the counter is zero we can assume vp_index == vcpu_idx.
> 
> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> ---
>  arch/x86/include/asm/kvm_host.h |  3 +++
>  arch/x86/kvm/hyperv.c           | 26 +++++++++++++++++++++++---
>  2 files changed, 26 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 09b2e3e2cf1b..711f79f1b5e6 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -781,6 +781,9 @@ struct kvm_hv {
>  	u64 hv_reenlightenment_control;
>  	u64 hv_tsc_emulation_control;
>  	u64 hv_tsc_emulation_status;
> +
> +	/* How many vCPUs have VP index != vCPU index */
> +	atomic_t num_mismatched_vp_indexes;
>  };
>  
>  enum kvm_irqchip_mode {
> diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
> index c8764faf783b..6a19c8e3c432 100644
> --- a/arch/x86/kvm/hyperv.c
> +++ b/arch/x86/kvm/hyperv.c
> @@ -1045,11 +1045,31 @@ static int kvm_hv_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host)
>  	struct kvm_vcpu_hv *hv_vcpu = &vcpu->arch.hyperv;
>  
>  	switch (msr) {
> -	case HV_X64_MSR_VP_INDEX:
> -		if (!host || (u32)data >= KVM_MAX_VCPUS)
> +	case HV_X64_MSR_VP_INDEX: {
> +		struct kvm_hv *hv = &vcpu->kvm->arch.hyperv;
> +		int vcpu_idx = kvm_vcpu_get_idx(vcpu);
> +		u32 new_vp_index = (u32)data;
> +
> +		if (!host || new_vp_index >= KVM_MAX_VCPUS)
>  			return 1;
> -		hv_vcpu->vp_index = (u32)data;
> +
> +		if (new_vp_index == hv_vcpu->vp_index)
> +			return 0;
> +
> +		/*
> +		 * VP index is changing, increment num_mismatched_vp_indexes in
> +		 * case it was equal to vcpu_idx before; on the other hand, if
> +		 * the new VP index matches vcpu_idx num_mismatched_vp_indexes
> +		 * needs to be decremented.

It may be worth mentioning that the initial balance is provided by
kvm_hv_vcpu_postcreate setting vp_index = vcpu_idx.

> +		 */
> +		if (hv_vcpu->vp_index == vcpu_idx)
> +			atomic_inc(&hv->num_mismatched_vp_indexes);
> +		else if (new_vp_index == vcpu_idx)
> +			atomic_dec(&hv->num_mismatched_vp_indexes);

> +
> +		hv_vcpu->vp_index = new_vp_index;
>  		break;
> +	}
>  	case HV_X64_MSR_VP_ASSIST_PAGE: {
>  		u64 gfn;
>  		unsigned long addr;

Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>