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: <20200827161237.889877377@infradead.org> <20200827161754.535381269@infradead.org>
In-Reply-To: <20200827161754.535381269@infradead.org>
From:   Lai Jiangshan <jiangshanlai+lkml@gmail.com>
Date:   Fri, 28 Aug 2020 12:03:54 +0800
Message-ID: <CAJhGHyBBqdD7cpxDbRL3myGF7924EDrp0_-RLEd8m10dQGFXzA@mail.gmail.com>
Subject: Re: [RFC][PATCH 6/7] freelist: Lock less freelist
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     LKML <linux-kernel@vger.kernel.org>,
        Masami Hiramatsu <mhiramat@kernel.org>, Eddy_Wu@trendmicro.com,
        X86 ML <x86@kernel.org>, davem@davemloft.net,
        Steven Rostedt <rostedt@goodmis.org>,
        naveen.n.rao@linux.ibm.com, anil.s.keshavamurthy@intel.com,
        linux-arch@vger.kernel.org, cameron@moodycamel.com,
        Oleg Nesterov <oleg@redhat.com>, Will Deacon <will@kernel.org>,
        "Paul E. McKenney" <paulmck@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Aug 28, 2020 at 12:23 AM Peter Zijlstra <peterz@infradead.org> wrote:

> +static inline void __freelist_add(struct freelist_node *node, struct freelist_head *list)
> +{
> +       /*
> +        * Since the refcount is zero, and nobody can increase it once it's
> +        * zero (except us, and we run only one copy of this method per node at
> +        * a time, i.e. the single thread case), then we know we can safely


> +
> +               /*
> +                * OK, the head must have changed on us, but we still need to decrement
> +                * the refcount we increased.
> +                */
> +               refs = atomic_fetch_add(-1, &prev->refs);
> +               if (refs == REFS_ON_FREELIST + 1)
> +                       __freelist_add(prev, list);

I'm curious whether it is correct to just set the prev->refs to zero and return
@prev? So that it can remove an unneeded "add()&get()" pair (although in
an unlikely branch) and __freelist_add() can be folded into freelist_add()
for tidier code.

Thanks
Lai.