Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 11 Jul 2023 11:19:26 -0500
From:   David Vernet <void@manifault.com>
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     linux-kernel@vger.kernel.org, mingo@redhat.com,
        juri.lelli@redhat.com, vincent.guittot@linaro.org,
        dietmar.eggemann@arm.com, rostedt@goodmis.org, bsegall@google.com,
        mgorman@suse.de, bristot@redhat.com, vschneid@redhat.com,
        gautham.shenoy@amd.com, kprateek.nayak@amd.com, aaron.lu@intel.com,
        clm@meta.com, tj@kernel.org, roman.gushchin@linux.dev,
        kernel-team@meta.com
Subject: Re: [PATCH v2 4/7] sched/fair: Add SHARED_RUNQ sched feature and
 skeleton calls
Message-ID: <20230711161926.GA389526@maniforge>
References: <20230710200342.358255-1-void@manifault.com>
 <20230710200342.358255-5-void@manifault.com>
 <20230711094547.GE3062772@hirez.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230711094547.GE3062772@hirez.programming.kicks-ass.net>
User-Agent: Mutt/2.2.10 (2023-03-25)
Precedence: bulk

On Tue, Jul 11, 2023 at 11:45:47AM +0200, Peter Zijlstra wrote:
> On Mon, Jul 10, 2023 at 03:03:39PM -0500, David Vernet wrote:
> 
> > @@ -11843,6 +11871,9 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> >  	if (!cpu_active(this_cpu))
> >  		return 0;
> >  
> > +	if (sched_feat(SHARED_RUNQ) && shared_runq_pick_next_task(this_rq, rf))
> > +		return -1;
> > +
> 
> Next patch implements shared_runq_pick_next_task() with the same return
> values as newidle_balance(), which would seem to suggest the following
> instead:
> 
> 	if (sched_feat(SHARED_RUNQ)) {
> 		pulled_task = shared_runq_pick_next_task(this_rq, rf);
> 		if (pulled_task)
> 			return pulled_task;
> 	}

Yep, that's cleaner.

> Additionally, I think we wants something like:
> 
> 	sd = rcu_dereference_check_sched_domain(this_rq->sd);
> 	if (sched_feat(SHARED_RUNQ)) {
> 		... /* see above */
> 
> 		sd = rcu_dereference(*this_cpu_ptr(&sd_llc));
> 		sd = sd->parent;
> 	}
> 
> to ensure we skip <=LLC domains, since those have already been covered
> by this shiny new thing, no?

Indeed, another nice optimization. I'll incorporate both of these into
the next version.