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;
MIME-Version: 1.0
References: <20211020110638.797389-1-pbonzini@redhat.com> <YW/61zpycsD8/z4g@hirez.programming.kicks-ass.net>
 <659fbd82-7d18-0e76-6fe4-b311897b4ae0@redhat.com>
In-Reply-To: <659fbd82-7d18-0e76-6fe4-b311897b4ae0@redhat.com>
From:   Lai Jiangshan <jiangshanlai+lkml@gmail.com>
Date:   Fri, 22 Oct 2021 12:11:55 +0800
Message-ID: <CAJhGHyALnn2EmPDXS3hyt1sgTH8vu=yeLo+N5meg7PZg7WrFQw@mail.gmail.com>
Subject: Re: [PATCH] rcuwait: do not enter RCU protection unless a wakeup is needed
To:     Paolo Bonzini <pbonzini@redhat.com>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        LKML <linux-kernel@vger.kernel.org>, kvm@vger.kernel.org,
        Davidlohr Bueso <dave@stgolabs.net>,
        Oleg Nesterov <oleg@redhat.com>,
        Ingo Molnar <mingo@kernel.org>,
        "Paul E . McKenney" <paulmck@linux.vnet.ibm.com>,
        Wanpeng Li <wanpengli@tencent.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Oct 20, 2021 at 7:39 PM Paolo Bonzini <pbonzini@redhat.com> wrote:
>
> On 20/10/21 13:17, Peter Zijlstra wrote:
> > AFAICT, rcu_read_lock() for PREEMPT_RCU is:
> >
> >    WRITE_ONCE(current->rcu_read_lock_nesting, READ_ONCE(current->rcu_read_lock_nesting) + 1);
> >    barrier();
>
> rcu_read_unlock() is the expensive one if you need to go down
> rcu_read_unlock_special().
>

If "actual likelihood of a wakeup is very low." as stated in the changelog,
the likelihood of rcu_read_unlock_special() is also very low.

rcu_read_lock() for PREEMPT_RCU is a function call, is it relevant?

(It is possible to remove the function call if the include-hell can
be resolved or remove the function call via LTO or just remove the
function call in X86 via percpu.)

Thanks
Lai

> Paolo
>
> > Paul?
>