From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1756386AbbE3Ut6 (ORCPT <rfc822;w@1wt.eu>);
	Sat, 30 May 2015 16:49:58 -0400
Received: from mx1.redhat.com ([209.132.183.28]:53165 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1750856AbbE3Utu (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Sat, 30 May 2015 16:49:50 -0400
Date: Sat, 30 May 2015 22:49:00 +0200
From: Oleg Nesterov <oleg@redhat.com>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>,
        Paul McKenney <paulmck@linux.vnet.ibm.com>, Tejun Heo <tj@kernel.org>,
        Ingo Molnar <mingo@redhat.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        der.herr@hofr.at, Davidlohr Bueso <dave@stgolabs.net>
Subject: Re: [RFC][PATCH 5/5] percpu-rwsem: Optimize readers and reduce
	global impact
Message-ID: <20150530204900.GA17772@redhat.com>
References: <20150526114356.609107918@infradead.org> <20150526120215.042527659@infradead.org> <20150529194534.GA31860@redhat.com> <20150529200932.GA2516@redhat.com> <CA+55aFxVtg7i9Hqg7a4MyQQs8XGH6QUNr06oj87Pf0tG_e1xvQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CA+55aFxVtg7i9Hqg7a4MyQQs8XGH6QUNr06oj87Pf0tG_e1xvQ@mail.gmail.com>
User-Agent: Mutt/1.5.18 (2008-05-17)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 05/29, Linus Torvalds wrote:
>
> On Fri, May 29, 2015 at 1:09 PM, Oleg Nesterov <oleg@redhat.com> wrote:
> >
> > Doesn't it need mb() before "state = readers_slow" to ensure
> > "release" semantics?
>
> Please don't do any mb() at all. Just do smp_store_release() if
> release semantics is what you want.

Agreed!

> (And try to make sure to pair it with smp_load_acquire() on the other
> side for things to make sense)

Hmm. I think you are right...

__percpu_down_read() lacks another mb() after the "state != BLOCK"
check for the same reason, and we can use smp_load_acquire(state)
instead.

Oleg.