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 S1756166AbcIAVDM (ORCPT <rfc822;w@1wt.eu>);
        Thu, 1 Sep 2016 17:03:12 -0400
Received: from mx1.redhat.com ([209.132.183.28]:37176 "EHLO mx1.redhat.com"
        rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
        id S1756099AbcIAVDH (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
        Thu, 1 Sep 2016 17:03:07 -0400
Date: Thu, 1 Sep 2016 19:26:58 +0200
From: Oleg Nesterov <oleg@redhat.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@kernel.org>, Al Viro <viro@ZenIV.linux.org.uk>,
        Bart Van Assche <bvanassche@acm.org>,
        Johannes Weiner <hannes@cmpxchg.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Neil Brown <neilb@suse.de>, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/2] sched/wait: abort_exclusive_wait() should pass
        TASK_NORMAL to wake_up()
Message-ID: <20160901172658.GA14456@redhat.com>
References: <20160826124453.GA28894@redhat.com> <20160826124528.GA28904@redhat.com> <20160901113919.GI10138@twins.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20160901113919.GI10138@twins.programming.kicks-ass.net>
User-Agent: Mutt/1.5.18 (2008-05-17)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.30]); Thu, 01 Sep 2016 17:27:34 +0000 (UTC)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 09/01, Peter Zijlstra wrote:
>
> So mixing INTERRUPTIBLE and UNINTERRUPTIBLE and then not using
> TASK_NORMAL for wakeups is a mis-feature/abuse of waitqueues IMO.

Heh, agreed. When I was doing this fix I suddenly realize that I do
not understand why do we have, say, wake_up_interruptible().

I mean, I can't imagine the "real" use-case when you actually want
to wake up only the INTERRUPTIBLE tasks and leave the UNINTERRUPTIBLE
sleeping. Exclusive or not.

It seems that wake_up_interruptible() is mostly used simply because
the caller knows that UNINTERRUPTIBLE waiters are not possible, this
is often the case.

> @@ -67,6 +70,16 @@ static void __wake_up_common(wait_queue_head_t *q, unsigned int mode,
>  {
>  	wait_queue_t *curr, *next;
>
> +#ifdef CONFIG_DEBUG_WAITQUEUE
> +	if (q->state != -1) {
> +		/*
> +		 * WARN if we have INTERRUPTIBLE and UNINTERRUPTIBLE
> +		 * waiters and do not use TASK_NORMAL to wake.
> +		 */
> +		WARN_ON_ONCE(q->state != (mode & TASK_NORMAL));
> +	}
> +#endif

Yes, perhaps...

Actually, I think that TASK_NORMAL should be used even if wq mixes
UNINTERRUPTIBLE and KILLABLE waiters. The fact that TASK_KILLABLE
includes TASK_UNINTERRUPTIBLE is just "implementation detail" even
if I do not think this will be ever changed.

Oleg.