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 S1751350AbcGMVBf (ORCPT <rfc822;w@1wt.eu>);
	Wed, 13 Jul 2016 17:01:35 -0400
Received: from mail-yw0-f171.google.com ([209.85.161.171]:34124 "EHLO
	mail-yw0-f171.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751276AbcGMVB0 (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Wed, 13 Jul 2016 17:01:26 -0400
Date: Wed, 13 Jul 2016 17:01:14 -0400
From: Tejun Heo <tj@kernel.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: John Stultz <john.stultz@linaro.org>, Ingo Molnar <mingo@redhat.com>,
        lkml <linux-kernel@vger.kernel.org>,
        Dmitry Shmidt <dimitrysh@google.com>,
        Rom Lemarchand <romlem@google.com>, Colin Cross <ccross@google.com>,
        Todd Kjos <tkjos@google.com>, Oleg Nesterov <oleg@redhat.com>,
        "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Subject: Re: Severe performance regression w/ 4.4+ on Android due to cgroup
 locking changes
Message-ID: <20160713210114.GE29670@mtj.duckdns.org>
References: <CALAqxLV2H06M2AxP2kK7pGHjLgQAaGdHB9HJEGZjH_Rouy4WhQ@mail.gmail.com>
 <20160713182102.GJ4065@mtj.duckdns.org>
 <20160713183347.GK4065@mtj.duckdns.org>
 <CALAqxLU76kVm8b5GNnSk25-Xk6D1efiWtwNJbY1th_pcN9Ks=g@mail.gmail.com>
 <20160713201823.GB29670@mtj.duckdns.org>
 <20160713202657.GW30154@twins.programming.kicks-ass.net>
 <20160713203944.GC29670@mtj.duckdns.org>
 <20160713205102.GZ30909@twins.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20160713205102.GZ30909@twins.programming.kicks-ass.net>
User-Agent: Mutt/1.6.1 (2016-04-27)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Wed, Jul 13, 2016 at 10:51:02PM +0200, Peter Zijlstra wrote:
> On Wed, Jul 13, 2016 at 04:39:44PM -0400, Tejun Heo wrote:
> So, IIRC, the trade-off is a full memory barrier in read_lock and
> read_unlock() vs sync_sched() in write.
>
> Full memory barriers are expensive and while the combined cost might
> well exceed the cost of the sync_sched() it doesn't suffer the latency
> issues.

Given the way read side is used for percpu_rwsem, full memory barrier
on reader side shouldn't matter at all.  The paths are not *that* hot.

> Not sure if we can frob the two in a single codebase, but I can have a
> poke if Oleg or Paul doesn't beat me to it.

At the simplest, it can be rwsem equivalence of lglock.

-- 
tejun