linux-kernel - Re: [PATCH 1/3] sched/rt: cpupri_find: implement fallback mechanism for !fit case

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 18 Feb 2020 17:27:46 +0000
From:   Qais Yousef <qais.yousef@....com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Dietmar Eggemann <dietmar.eggemann@....com>,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Pavan Kondeti <pkondeti@...eaurora.org>,
        Juri Lelli <juri.lelli@...hat.com>,
        Vincent Guittot <vincent.guittot@...aro.org>,
        Ben Segall <bsegall@...gle.com>, Mel Gorman <mgorman@...e.de>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/3] sched/rt: cpupri_find: implement fallback mechanism
 for !fit case

On 02/18/20 11:46, Steven Rostedt wrote:
> On Mon, 17 Feb 2020 23:45:49 +0000
> Qais Yousef <qais.yousef@....com> wrote:
> 
> > --- a/kernel/sched/rt.c
> > +++ b/kernel/sched/rt.c
> > @@ -14,6 +14,8 @@ static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun);
> > 
> >  struct rt_bandwidth def_rt_bandwidth;
> > 
> > +typedef bool (*fitness_fn_t)(struct task_struct *p, int cpu);
> > +
> >  static enum hrtimer_restart sched_rt_period_timer(struct hrtimer *timer)
> >  {
> >         struct rt_bandwidth *rt_b =
> > @@ -1708,6 +1710,7 @@ static int find_lowest_rq(struct task_struct *task)
> >         struct cpumask *lowest_mask = this_cpu_cpumask_var_ptr(local_cpu_mask);
> >         int this_cpu = smp_processor_id();
> >         int cpu      = task_cpu(task);
> > +       fitness_fn_t fitness_fn;
> > 
> >         /* Make sure the mask is initialized first */
> >         if (unlikely(!lowest_mask))
> > @@ -1716,8 +1719,17 @@ static int find_lowest_rq(struct task_struct *task)
> >         if (task->nr_cpus_allowed == 1)
> >                 return -1; /* No other targets possible */
> > 
> > +       /*
> > +        * Help cpupri_find avoid the cost of looking for a fitting CPU when
> > +        * not really needed.
> > +        */
> > +       if (static_branch_unlikely(&sched_asym_cpucapacity))
> > +               fitness_fn = rt_task_fits_capacity;
> > +       else
> > +               fitness_fn = NULL;
> > +
> >         if (!cpupri_find(&task_rq(task)->rd->cpupri, task, lowest_mask,
> > -                        rt_task_fits_capacity))
> > +                        fitness_fn))
> >                 return -1; /* No targets found */
> > 
> >         /*
> 
> 
> If we are going to use static branches, then lets just remove the
> parameter totally. That is, make two functions (with helpers), where
> one needs this fitness function the other does not.
> 
> 	if (static_branch_unlikely(&sched_asym_cpu_capacity))
> 		ret = cpupri_find_fitness(...);
> 	else
> 		ret = cpupri_find(...);
> 
> 	if (!ret)
> 		return -1;
> 
> Something like that?

Is there any implication on code generation here?

I like my flavour better tbh. But I don't mind refactoring the function out if
it does make it more readable.

Thanks

--
Qais Yousef