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 S941063AbcJ2DVl (ORCPT <rfc822;w@1wt.eu>);
        Fri, 28 Oct 2016 23:21:41 -0400
Received: from merlin.infradead.org ([205.233.59.134]:48704 "EHLO
        merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1751223AbcJ2DVk (ORCPT
        <rfc822;linux-kernel@vger.kernel.org>);
        Fri, 28 Oct 2016 23:21:40 -0400
Date: Sat, 29 Oct 2016 05:21:26 +0200
From: Peter Zijlstra <peterz@infradead.org>
To: Tejun Heo <htejun@fb.com>
Cc: Tejun Heo <tj@kernel.org>, torvalds@linux-foundation.org,
        akpm@linux-foundation.org, mingo@redhat.com, axboe@kernel.dk,
        tytso@mit.edu, jack@suse.com, adilger.kernel@dilger.ca,
        linux-ext4@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        linux-kernel@vger.kernel.org, kernel-team@fb.com, mingbo@fb.com
Subject: Re: [PATCH 1/4] sched: move IO scheduling accounting from
 io_schedule_timeout() to __schedule()
Message-ID: <20161029032126.GH3568@worktop.programming.kicks-ass.net>
References: <1477673892-28940-1-git-send-email-tj@kernel.org>
 <1477673892-28940-2-git-send-email-tj@kernel.org>
 <20161028182712.GM3142@twins.programming.kicks-ass.net>
 <20161028190702.GL3157@twins.programming.kicks-ass.net>
 <20161028191231.GA29044@htj.duckdns.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20161028191231.GA29044@htj.duckdns.org>
User-Agent: Mutt/1.5.22.1 (2013-10-16)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Fri, Oct 28, 2016 at 03:12:32PM -0400, Tejun Heo wrote:
> Hello, Peter.
> 
> On Fri, Oct 28, 2016 at 09:07:02PM +0200, Peter Zijlstra wrote:
> > One alternative is to inherit the iowait state of the task we block on.
> > That'll not get rid of the branches much, but it will remove the new
> > mutex APIs.
> 
> Yeah, thought about that briefly but we don't necessarily track mutex

This one I actually fixed and should be in -next. And it would be
sufficient to cover the use case here.

> or other synchronization construct owners, things get gnarly with
> rwsems (the inode ones sometimes end up in a similar situation), and
> we'll probably end up dealing with some surprising propagations down
> the line.

rwsems could be done for writers only.