Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Tue, 20 Oct 2020 16:07:32 -0400
From:   Steven Rostedt <rostedt@goodmis.org>
To:     Thomas Gleixner <tglx@linutronix.de>
Cc:     LKML <linux-kernel@vger.kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@kernel.org>,
        Juri Lelli <juri.lelli@redhat.com>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Dietmar Eggemann <dietmar.eggemann@arm.com>,
        Ben Segall <bsegall@google.com>, Mel Gorman <mgorman@suse.de>,
        Daniel Bristot de Oliveira <bristot@redhat.com>
Subject: Re: sched: Reenable interrupts in do sched_yield()
Message-ID: <20201020160732.5f8fc24e@oasis.local.home>
In-Reply-To: <87o8kw93n4.fsf@nanos.tec.linutronix.de>
References: <87r1pt7y5c.fsf@nanos.tec.linutronix.de>
        <20201020113830.378b4a4c@gandalf.local.home>
        <87o8kw93n4.fsf@nanos.tec.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Tue, 20 Oct 2020 20:02:55 +0200
Thomas Gleixner <tglx@linutronix.de> wrote:

> On Tue, Oct 20 2020 at 11:38, Steven Rostedt wrote:
> > On Tue, 20 Oct 2020 16:46:55 +0200
> > Thomas Gleixner <tglx@linutronix.de> wrote:
> >  
> >> -	/*
> >> -	 * Since we are going to call schedule() anyway, there's
> >> -	 * no need to preempt or enable interrupts:  
> >
> > I think the above comment still makes sense, just needs to be tweeked:
> >
> > 	/*
> > 	 * Since we are going to call schedule() anyway, there's
> > 	 * no need to allow preemption after releasing the rq lock.  
> >> -	 */  
> >
> > Especially, since we are now enabling interrupts, which is likely to
> > trigger a preemption.  
> 
> sched_preempt_enable_no_resched() still enables preemption. It just
> avoids the check. And it still allows preemption when an interrupt
> triggering preemption happens between sched_preempt_enable_no_resched()
> and __schedule() disabling preemption/interrupts.
> 
> So no, your new variant is just differently bogus and misleading.

What I wrote wasn't exactly what I meant. What I meant to have:

	/*
	 * Since we are going to call schedule() anyways, there's
	 * no need to do the preemption check when the rq_lock is released.
	 */

That is, to document why we have the preempt_disable() before the unlock:

	preempt_disable();
	rq_unlock_irq(rq, &rf);
	sched_preempt_enable_no_resched();


-- Steve