Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.36 as permitted sender) client-ip=23.128.96.36;
Date:   Mon, 4 Dec 2023 07:55:45 -0800
In-Reply-To: <20231204074535.9567-1-likexu@tencent.com>
Mime-Version: 1.0
References: <20231204074535.9567-1-likexu@tencent.com>
Message-ID: <ZW32geNb18p9ibrR@google.com>
Subject: Re: [PATCH] KVM: x86/intr: Explicitly check NMI from guest to
 eliminate false positives
From:   Sean Christopherson <seanjc@google.com>
To:     Like Xu <like.xu.linux@gmail.com>
Cc:     Paolo Bonzini <pbonzini@redhat.com>, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="us-ascii"
Precedence: bulk

On Mon, Dec 04, 2023, Like Xu wrote:
> From: Like Xu <likexu@tencent.com>
> 
> Explicitly checking the source of external interrupt is indeed NMI and not
> other types in the kvm_arch_pmi_in_guest(), which prevents perf-kvm false
> positive samples generated after vm-exit but before kvm_before_interrupt()
> from being incorrectly labelled as guest samples:

...

> Fixes: 73cd107b9685 ("KVM: x86: Drop current_vcpu for kvm_running_vcpu + kvm_arch_vcpu variable")

The behavior is deliberate, and was added by commit dd60d217062f ("KVM: x86: Fix
perf timer mode IP reporting").  *If* we want to undo that, then the best "fix"
would be to effective reverting that commit by dropping the IRQ usage of
kvm_before_interrupt() and renaming the helpers kvm_{before,after}_nmi().  But
my understanding is that the behavior is necessary for select PMU usage.