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:   Wed, 19 Oct 2022 12:23:15 +0100
From:   Qais Yousef <qyousef@layalina.io>
To:     John Stultz <jstultz@google.com>
Cc:     Alexander Gordeev <agordeev@linux.ibm.com>,
        LKML <linux-kernel@vger.kernel.org>,
        Connor O'Brien <connoro@google.com>,
        John Dias <joaodias@google.com>, Rick Yiu <rickyiu@google.com>,
        John Kacur <jkacur@redhat.com>,
        Qais Yousef <qais.yousef@arm.com>,
        Chris Redpath <chris.redpath@arm.com>,
        Abhijeet Dharmapurikar <adharmap@quicinc.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>,
        Juri Lelli <juri.lelli@redhat.com>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Dietmar Eggemann <dietmar.eggemann@arm.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Thomas Gleixner <tglx@linutronix.de>, kernel-team@android.com,
        "J . Avila" <elavila@google.com>
Subject: Re: [PATCH RFC v4 2/3] sched: Avoid placing RT threads on cores
 handling long softirqs
Message-ID: <20221019112315.7d6wyc7d7dgzwsch@airbuntu>
References: <20221003232033.3404802-1-jstultz@google.com>
 <20221003232033.3404802-3-jstultz@google.com>
 <Y01NPB4sa8Z98ntK@li-4a3a4a4c-28e5-11b2-a85c-a8d192c6f089.ibm.com>
 <CANDhNCoTGNANDnOQ6touhreG_UEn1-N4T4BktWFxpLPSWVfrdA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <CANDhNCoTGNANDnOQ6touhreG_UEn1-N4T4BktWFxpLPSWVfrdA@mail.gmail.com>
Precedence: bulk

On 10/17/22 20:42, John Stultz wrote:

> > > +     return ret;
> > > +}
> > > +#else
> > > +#define __use_softirq_opt 0
> > > +static bool cpu_busy_with_softirqs(int cpu)
> > > +{
> > > +     return false;
> > > +}
> > > +#endif /* CONFIG_RT_SOFTIRQ_OPTIMIZATION */
> > > +
> > > +static bool rt_task_fits_cpu(struct task_struct *p, int cpu)
> >
> > To me, the new name is unfortunate, since it strips a notion
> > of the reason. Instead, "CPU un/fits, because of capacity" it
> > reads as "CPU un/fits, because of ..." what?
> 
> As with all complicated questions: "It depends" :) On the kernel
> config specifically.
> 
> The earlier version of the patch had:
> rt_task_fits_capacity_and_may_preempt() but Qais suggested it be
> switched to the generic "fits_cpu" as it would depend on the kernel
> config as to which aspect of "fit" was being considered.
> 
> I guess it could be  rt_task_fits_capacity_and_softirq_free() ?

My line of thinking is that we have multiple reasons which could potentially
come and go. Having a simple generic name is future proof to the details of the
reason.

For example, in the future we might find a better way to handle the softirq
conflict and remove this reason; or we might find a new 'reason' is required.
Documenting that in the function header (if required, I see the one line
implementation is self descriptive so far) rather than in name made more sense
to me, hence the suggestion.

I am already aware of another new reason required to handle thermal pressure
when checking for capacity [1]. The discussion has stalled, but the problem is
still there and we'd need to address it. So I expect this code will be massaged
somewhat in the near future.

Sticking to rt_task_fits_cpu() will reduce the churn IMHO. But if you really
prefer something else, works for me :-)

[1] https://lore.kernel.org/lkml/20220514235513.jm7ul2y6uddj6eh2@airbuntu/


Cheers

--
Qais Yousef