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 S938683AbdAKQ6C (ORCPT <rfc822;w@1wt.eu>);
        Wed, 11 Jan 2017 11:58:02 -0500
Received: from mail.fireflyinternet.com ([109.228.58.192]:52390 "EHLO
        fireflyinternet.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org
        with ESMTP id S937856AbdAKQ6B (ORCPT
        <rfc822;linux-kernel@vger.kernel.org>);
        Wed, 11 Jan 2017 11:58:01 -0500
X-Default-Received-SPF: pass (skip=forwardok (res=PASS)) x-ip-name=78.156.65.138;
Date: Wed, 11 Jan 2017 16:57:32 +0000
From: Chris Wilson <chris@chris-wilson.co.uk>
To: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org, Waiman Long <Waiman.Long@hpe.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        "Paul E . McKenney" <paulmck@linux.vnet.ibm.com>,
        Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH] locking/mutex: Clear mutex-handoff flag on interrupt
Message-ID: <20170111165731.GB16278@nuc-i3427.alporthouse.com>
References: <20161007150211.196801561@infradead.org>
 <20170109115203.2750-1-chris@chris-wilson.co.uk>
 <20170111164302.GA3051@twins.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20170111164302.GA3051@twins.programming.kicks-ass.net>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Wed, Jan 11, 2017 at 05:43:02PM +0100, Peter Zijlstra wrote:
> On Mon, Jan 09, 2017 at 11:52:03AM +0000, Chris Wilson wrote:
> > If we abort the mutex_lock() due to an interrupt, or other error from
> 
> s/interrupt/signal/, right?

Yes. EINTR is ingrained.
 
> > ww_mutex, we need to relinquish the handoff flag if we applied it.
> > Otherwise, we may cause missed wakeups as the current owner may try to
> > handoff to a new thread that is not expecting the handoff and so sleep
> > thinking the lock is already claimed (and since the owner unlocked there
> > may never be a new wakeup).
> 
> Isn't that the exact same scenario as Nicolai fixed here:
> 
>   http://lkml.kernel.org/r/1482346000-9927-3-git-send-email-nhaehnle@gmail.com
> 
> Did you, like Nicolai, find this by inspection, or can you reproduce?

Looks like it should be. It takes about 20 minutes of running a stress
test, but it is very reliably hit on the current kernel.

> FWIW, I have the below patch that should also solve this problem afaict.

Thanks, I shall see if makes my machines happy.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre