Received-SPF: pass (google.com: domain of linux-kernel+bounces-147365-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Date: Tue, 16 Apr 2024 11:19:38 -0700
In-Reply-To: <20240215160136.1256084-2-alejandro.j.jimenez@oracle.com>
Precedence: bulk
Mime-Version: 1.0
References: <20240215160136.1256084-1-alejandro.j.jimenez@oracle.com> <20240215160136.1256084-2-alejandro.j.jimenez@oracle.com>
Message-ID: <Zh7BOkOf0i_KZVNO@google.com>
Subject: Re: [RFC 1/3] x86: KVM: stats: Add a stat to report status of APICv inhibition
From: Sean Christopherson <seanjc@google.com>
To: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Cc: kvm@vger.kernel.org, pbonzini@redhat.com, linux-kernel@vger.kernel.org, 
	joao.m.martins@oracle.com, boris.ostrovsky@oracle.com, mark.kanda@oracle.com, 
	suravee.suthikulpanit@amd.com, mlevitsk@redhat.com
Content-Type: text/plain; charset="us-ascii"

On Thu, Feb 15, 2024, Alejandro Jimenez wrote:
> The inhibition status of APICv can currently be checked using the
> 'kvm_apicv_inhibit_changed' tracepoint, but this is not accessible if
> tracefs is not available (e.g. kernel lockdown, non-root user). Export
> inhibition status as a binary stat that can be monitored from userspace
> without elevated privileges.
> 
> Signed-off-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
> ---
>  arch/x86/include/asm/kvm_host.h |  1 +
>  arch/x86/kvm/x86.c              | 10 +++++++++-
>  2 files changed, 10 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index ad5319a503f0..9b960a523715 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -1524,6 +1524,7 @@ struct kvm_vm_stat {
>  	u64 nx_lpage_splits;
>  	u64 max_mmu_page_hash_collisions;
>  	u64 max_mmu_rmap_size;
> +	u64 apicv_inhibited;

Tracking the negative is odd, i.e. if we add a stat, KVM should probably track
if APICv is fully enabled, not if it's inhibited.

This also should be a boolean, not a u64.  Precisely enumerating _why_ APICv is
inhibited is firmly in debug territory, i.e. not in scope for "official" stats.

Oh, and this should be a per-vCPU stat, not a VM-wide stat.

As for whether or not we should add a stat for this, I'm leaning towards "yes".
APICv can have such a profound impact on performance (and functionality) that
definitively knowing that it's enabled seems justified.