Re: [PATCH 2/3] rtmutex: deboost priority conditionally when rt-mutex unlock

From: Steven Rostedt <rostedt@goodmis.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Alex Shi <alex.shi@linaro.org>,
	mingo@redhat.com, corbet@lwn.net,
	"open list:LOCKING PRIMITIVES" <linux-kernel@vger.kernel.org>,
	Sebastian Siewior <bigeasy@linutronix.de>,
	Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH 2/3] rtmutex: deboost priority conditionally when rt-mutex unlock
Date: Thu, 13 Apr 2017 12:09:25 -0400	[thread overview]
Message-ID: <20170413120925.76e22c03@gandalf.local.home> (raw)
In-Reply-To: <20170413143952.ty4dw5mvyfnhaqgh@hirez.programming.kicks-ass.net>

On Thu, 13 Apr 2017 16:39:52 +0200
Peter Zijlstra <peterz@infradead.org> wrote:

> On Thu, Apr 13, 2017 at 10:02:53PM +0800, Alex Shi wrote:
> >  	/*
> > +	 * 'current' release this lock, so 'current' should be a higher prio
> > +	 * task than the next top waiter, unless the current prio was gotten
> > +	 * from this top waiter, iff so, we need to deboost 'current' after
> > +	 * the lock release.
> > +	 */
> > +	if (current->prio == waiter->prio)
> > +		deboost = true;  
> 
> This is wrong.

The comment is, especially that "iff". What if current and waiter
happen to have the same priority? Then it too doesn't need to be
deboosted.

But that said, we currently perform the deboost unconditionally. I
can't think of a case where current->prio != waiter->prio where we
should perform the deboost, because current->prio should always be <=
waiter->prio (where lower prio means higher priority). Maybe I'm missing
something.

-- Steve