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 S932697AbcIAK3d (ORCPT <rfc822;w@1wt.eu>);
        Thu, 1 Sep 2016 06:29:33 -0400
Received: from bombadil.infradead.org ([198.137.202.9]:51747 "EHLO
        bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S932298AbcIAK3c (ORCPT
        <rfc822;linux-kernel@vger.kernel.org>);
        Thu, 1 Sep 2016 06:29:32 -0400
Date: Thu, 1 Sep 2016 12:29:25 +0200
From: Peter Zijlstra <peterz@infradead.org>
To: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: linux-kernel@vger.kernel.org,
        Giovanni Gherdovich <ggherdovich@suse.cz>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Mel Gorman <mgorman@techsingularity.net>,
        Mike Galbraith <mgalbraith@suse.de>,
        Paolo Bonzini <pbonzini@redhat.com>, Rik van Riel <riel@redhat.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Wanpeng Li <wanpeng.li@hotmail.com>, Ingo Molnar <mingo@kernel.org>
Subject: Re: [PATCH 1/3] sched/cputime: Improve scalability of
 times()/clock_gettime() on 32 bit cpus
Message-ID: <20160901102925.GR10153@twins.programming.kicks-ass.net>
References: <1472722064-7151-1-git-send-email-sgruszka@redhat.com>
 <1472722064-7151-2-git-send-email-sgruszka@redhat.com>
 <20160901094906.GP10153@twins.programming.kicks-ass.net>
 <20160901100733.GA6388@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20160901100733.GA6388@redhat.com>
User-Agent: Mutt/1.5.23.1 (2014-03-12)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Thu, Sep 01, 2016 at 12:07:34PM +0200, Stanislaw Gruszka wrote:
> On Thu, Sep 01, 2016 at 11:49:06AM +0200, Peter Zijlstra wrote:
> > You're now making rather hot paths slower to benefit a rather slow path,
> > that too is backwards.
> 
> Ok, you have right, I made update_curr() slower (a bit I think, since
> this new seqcount primitive should be in the same cache line as other
> things).

seqcount adds 2 smp_wmb(), which on ARM, are not free (it is possible to
do with just 1 FWIW).

> But do we don't care about inconsistency of accessing of 64 bit variable
> on 32 bit processors (see patch 3) ? I know this is unlikely scenario
> to get inconsistency, but I assume it's still possible, or not?

Its actually quite possible. We've observed it a fair few times. 64bit
variables are 2 32bit stores/loads and getting interleaved data is quite
possible.

> If not, I can get rid of read_sum_exec_runtime() and just read
> sum_exec_runtime without task_rq_lock() protection on 
> thread_group_cputime() . That would make the benchmark happy. 

I think this benchmark is misguided. Just accept that O(nr_threads) is
expensive, same with process wide itimer, just don't use them when you
care about performance.