Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 21 Jul 2021 12:08:18 +0100
Message-ID: <87a6mf3owt.wl-maz@kernel.org>
From:   Marc Zyngier <maz@kernel.org>
To:     Sergey Senozhatsky <senozhatsky@chromium.org>
Cc:     Joel Fernandes <joelaf@google.com>, Will Deacon <will@kernel.org>,
        Suleiman Souhlal <suleiman@google.com>,
        linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu,
        linux-kernel@vger.kernel.org,
        virtualization@lists.linux-foundation.org
Subject: Re: [PATCHv2 4/4] arm64: add host pv-vcpu-state support
In-Reply-To: <YPf5MFV99zPdTu/U@google.com>
References: <20210709043713.887098-1-senozhatsky@chromium.org>
        <20210709043713.887098-5-senozhatsky@chromium.org>
        <874kcz33g5.wl-maz@kernel.org>
        <CAJWu+oqCyj3H0=1KNo3c+crdcktYinFoTQJ5jHgU8gjeF4d2yA@mail.gmail.com>
        <87h7go2h69.wl-maz@kernel.org>
        <YPf5MFV99zPdTu/U@google.com>
User-Agent: Wanderlust/2.15.9 (Almost Unreal) SEMI-EPG/1.14.7 (Harue)
 FLIM-LB/1.14.9 (=?UTF-8?B?R29qxY0=?=) APEL-LB/10.8 EasyPG/1.0.0 Emacs/27.1
 (x86_64-pc-linux-gnu) MULE/6.0 (HANACHIRUSATO)
MIME-Version: 1.0 (generated by SEMI-EPG 1.14.7 - "Harue")
Content-Type: text/plain; charset=US-ASCII
Precedence: bulk

On Wed, 21 Jul 2021 11:38:40 +0100,
Sergey Senozhatsky <senozhatsky@chromium.org> wrote:
> 
> On (21/07/21 09:40), Marc Zyngier wrote:
> > > 
> > > Can that be cured by just checking vcpu->preempted before calling
> > > kvm_update_vcpu_preempted() ?
> > 
> > It isn't obvious to me that this is the right thing to do.
> > vcpu->preempted is always updated on sched-out from the preempt
> > notifier if the vcpu was on the run-queue, so my guess is that it will
> > always be set when switching to another task.
> > 
> > What you probably want is to check whether the vcpu is blocked by
> > introspecting the wait-queue with:
> > 
> > 	scuwait_active(kvm_arch_vcpu_get_wait(vcpu)
> > 
> > which will tell you whether you are blocking or not. We are already
> > using a similar construct for arming a background timer in this case.
> 
> Can we examine if vcpu->run->exit_reason == WFE/WFI and avoid setting
> preempted state if so?

We never go back to userspace for WFI/WFE, so no reason to populate
the run structure.

Checking for the blocked state is the right thing to do, and we
already have the primitive for this. Just use it.

	M.

-- 
Without deviation from the norm, progress is not possible.