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:   Thu, 17 Dec 2020 11:48:23 +0100
From:   Peter Zijlstra <peterz@infradead.org>
To:     "Paul E. McKenney" <paulmck@kernel.org>
Cc:     Thomas Gleixner <tglx@linutronix.de>,
        Marco Elver <elver@google.com>,
        LKML <linux-kernel@vger.kernel.org>,
        kasan-dev <kasan-dev@googlegroups.com>,
        Ingo Molnar <mingo@kernel.org>,
        Frederic Weisbecker <frederic@kernel.org>,
        Will Deacon <will@kernel.org>,
        Naresh Kamboju <naresh.kamboju@linaro.org>,
        syzbot+23a256029191772c2f02@syzkaller.appspotmail.com,
        syzbot+56078ac0b9071335a745@syzkaller.appspotmail.com,
        syzbot+867130cb240c41f15164@syzkaller.appspotmail.com
Subject: Re: [patch 3/3] tick: Annotate tick_do_timer_cpu data races
Message-ID: <20201217104823.GU3040@hirez.programming.kicks-ass.net>
References: <20201206211253.919834182@linutronix.de>
 <20201206212002.876987748@linutronix.de>
 <20201207120943.GS3021@hirez.programming.kicks-ass.net>
 <87y2i94igo.fsf@nanos.tec.linutronix.de>
 <CANpmjNNQiTbnkkj+ZHS5xxQuQfnWN_JGwSnN-_xqfa=raVrXHQ@mail.gmail.com>
 <20201207194406.GK2657@paulmck-ThinkPad-P72>
 <20201208081129.GQ2414@hirez.programming.kicks-ass.net>
 <20201208150309.GP2657@paulmck-ThinkPad-P72>
 <873606tx1c.fsf@nanos.tec.linutronix.de>
 <20201216211931.GL2657@paulmck-ThinkPad-P72>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20201216211931.GL2657@paulmck-ThinkPad-P72>
Precedence: bulk

On Wed, Dec 16, 2020 at 01:19:31PM -0800, Paul E. McKenney wrote:
> Given that there is no optimization potential, then the main reason to use
> data_race() instead of *_ONCE() is to prevent KCSAN from considering the
> accesses when looking for data races.  But that is mostly for debugging
> accesses, in cases when these accesses are not really part of the
> concurrent algorithm.
> 
> So if I understand the situation correctly, I would be using *ONCE().

Huh, what, why?

The code doesn't need READ_ONCE(), it merely wants to tell kasan that
the race it observes is fine and as to please shut up.

IOW data_race() is accurate and right.