From mboxrd@z Thu Jan  1 00:00:00 1970
From: Mike Galbraith <bitbucket@online.de>
Subject: Re: [PATCH RT]  rt,ipc,sem: fix -rt livelock
Date: Fri, 13 Sep 2013 05:20:21 +0200
Message-ID: <1379042421.5455.47.camel@marge.simpson.net>
References: <20130819213324.405942342@goodmis.org>
	 <1377842245.5422.7.camel@marge.simpson.net>
	 <1377873580.5422.25.camel@marge.simpson.net>
	 <1377926850.6234.22.camel@marge.simpson.net>
	 <1378451338.5434.91.camel@marge.simpson.net>
	 <1378794614.6046.37.camel@marge.simpson.net>
	 <52307814.4020104@colorfullife.com>
	 <1378971628.10696.88.camel@marge.simpson.net>
	 <20130912151547.389673d2@gandalf.local.home>
Mime-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Cc: Manfred Spraul <manfred@colorfullife.com>,
	linux-rt-users <linux-rt-users@vger.kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
	Peter Zijlstra <a.p.zijlstra@chello.nl>
To: Steven Rostedt <rostedt@goodmis.org>
Return-path: <linux-rt-users-owner@vger.kernel.org>
Received: from moutng.kundenserver.de ([212.227.17.10]:49899 "EHLO
	moutng.kundenserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751533Ab3IMDU0 (ORCPT
	<rfc822;linux-rt-users@vger.kernel.org>);
	Thu, 12 Sep 2013 23:20:26 -0400
In-Reply-To: <20130912151547.389673d2@gandalf.local.home>
Sender: linux-rt-users-owner@vger.kernel.org
List-ID: <linux-rt-users.vger.kernel.org>

On Thu, 2013-09-12 at 15:15 -0400, Steven Rostedt wrote: 
> On Thu, 12 Sep 2013 09:40:28 +0200
> Mike Galbraith <bitbucket@online.de> wrote:
> 
> 
> > If maintainers are interested, I can add that and clean up hammer marks
> > so it can be applied in lieu of the existing rt wakeup patch instead of
> > being crammed in on top of it, or they can trivially do that themselves,
> > both options presuming your (implied) approval.
> > 
> 
> Mike,
> 
> What's the final patch for this?

Both loop removal patch or something like it (but prettier), and
Manfred's wakeup scheme to prevent the locked array lock being being
seen and reacted to by wakees, who can then contend on that lock,
defeating the whole purpose for a while.

> Could you post another one with what you believe is the correct fix.

As noted, I _think_ they're both needed.  Manfred's wakeup scheme cures
a trigger, but not the bug methinks (non-deterministic trylock loop).
Would be nice if it turned out that it cured all possible livelocks, but
I don't think it does.  Even if it did, crawling over a large array
always taking and releasing every lock will suck, which is why I did the
look at it before taking it thing.  That ain't pretty, but it's less
painful.

-Mike