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 S1753813AbbG2UuY (ORCPT <rfc822;w@1wt.eu>);
	Wed, 29 Jul 2015 16:50:24 -0400
Received: from g2t2354.austin.hp.com ([15.217.128.53]:33435 "EHLO
	g2t2354.austin.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1752897AbbG2UuF (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Wed, 29 Jul 2015 16:50:05 -0400
Message-ID: <55B93C76.6030502@hp.com>
Date: Wed, 29 Jul 2015 16:49:58 -0400
From: Waiman Long <waiman.long@hp.com>
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:10.0.12) Gecko/20130109 Thunderbird/10.0.12
MIME-Version: 1.0
To: Davidlohr Bueso <dave@stgolabs.net>
CC: Peter Zijlstra <peterz@infradead.org>, Ingo Molnar <mingo@redhat.com>,
        Thomas Gleixner <tglx@linutronix.de>, "H. Peter Anvin" <hpa@zytor.com>,
        x86@kernel.org, linux-kernel@vger.kernel.org,
        Scott J Norton <scott.norton@hp.com>,
        Douglas Hatch <doug.hatch@hp.com>
Subject: Re: [PATCH v3 2/7] locking/pvqspinlock: Add pending bit support
References: <1437595962-21472-1-git-send-email-Waiman.Long@hp.com>  <1437595962-21472-3-git-send-email-Waiman.Long@hp.com>  <1437958584.25997.27.camel@stgolabs.net> <55B66ACB.6010702@hp.com> <1438025945.25997.46.camel@stgolabs.net>
In-Reply-To: <1438025945.25997.46.camel@stgolabs.net>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 07/27/2015 03:39 PM, Davidlohr Bueso wrote:
> On Mon, 2015-07-27 at 13:30 -0400, Waiman Long wrote:
>> The pending bit acts as a 1-slot waiting queue. So if the vCPU needs to
>> fall back to regular queuing, it needs to clear the bit.
> Right, that's what I thought. So you would also need to call
> clear_pending() as soon as the trylock/pending loop hits zero. Or am I
> missing something?
>
> /*
>   * trylock || pending
>   */
> for (;;) {
> 	...
> 	if (loop--<= 0) {
> 		clear_pending(lock);
> 		goto queue;
> 	}
> }
>
> Also, no need to check for negative loop, zero should be enough to
> breakout.

The while loop below can potentially make the loop variable become negative:


         /*
          * wait for in-progress pending->locked hand-overs
          */
         if (val == _Q_PENDING_VAL) {
                 while (((val = atomic_read(&lock->val)) == 
_Q_PENDING_VAL) &&
                         loop--)
                         cpu_relax();
         }

Cheers,
Longman