linux-kernel - Re: [PATCH v9 09/15] sched: Introduce sched_energy

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 for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20181122091715.6kvhcbgoqrd54nbs@queper01-lin>
Date:   Thu, 22 Nov 2018 09:17:16 +0000
From:   Quentin Perret <quentin.perret@....com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     rjw@...ysocki.net, linux-kernel@...r.kernel.org,
        linux-pm@...r.kernel.org, gregkh@...uxfoundation.org,
        mingo@...hat.com, dietmar.eggemann@....com,
        morten.rasmussen@....com, chris.redpath@....com,
        patrick.bellasi@....com, valentin.schneider@....com,
        vincent.guittot@...aro.org, thara.gopinath@...aro.org,
        viresh.kumar@...aro.org, tkjos@...gle.com, joel@...lfernandes.org,
        smuckle@...gle.com, adharmap@...eaurora.org,
        skannan@...eaurora.org, pkondeti@...eaurora.org,
        juri.lelli@...hat.com, edubezval@...il.com,
        srinivas.pandruvada@...ux.intel.com, currojerez@...eup.net,
        javi.merino@...nel.org
Subject: Re: [PATCH v9 09/15] sched: Introduce sched_energy_present static key

On Wednesday 21 Nov 2018 at 15:14:45 (+0000), Quentin Perret wrote:
> Hi Peter,
> 
> On Wednesday 21 Nov 2018 at 14:08:22 (+0100), Peter Zijlstra wrote:
> > On Mon, Nov 19, 2018 at 02:18:51PM +0000, Quentin Perret wrote:
> > 
> > > +static void sched_energy_start(int ndoms_new, cpumask_var_t doms_new[])
> > > +{
> > > +	/*
> > > +	 * The conditions for EAS to start are checked during the creation of
> > > +	 * root domains. If one of them meets all conditions, it will have a
> > > +	 * non-null list of performance domains.
> > > +	 */
> > > +	while (ndoms_new) {
> > > +		if (cpu_rq(cpumask_first(doms_new[ndoms_new - 1]))->rd->pd)
> > > +			goto enable;
> > > +		ndoms_new--;
> > > +	}
> > 
> > That seems quite ugly; can't you simply return a bool from
> > build_perf_domains() ?
> > 
> > Something like the below, but less fugly ;-)
> > 
> > --- a/kernel/sched/topology.c
> > +++ b/kernel/sched/topology.c
> > @@ -299,7 +299,7 @@ static void destroy_perf_domain_rcu(stru
> >  #define EM_MAX_COMPLEXITY 2048
> >  
> >  extern struct cpufreq_governor schedutil_gov;
> > -static void build_perf_domains(const struct cpumask *cpu_map)
> > +static bool build_perf_domains(const struct cpumask *cpu_map)
> >  {
> >  	int i, nr_pd = 0, nr_cs = 0, nr_cpus = cpumask_weight(cpu_map);
> >  	struct perf_domain *pd = NULL, *tmp;
> > @@ -365,7 +365,7 @@ static void build_perf_domains(const str
> >  	if (tmp)
> >  		call_rcu(&tmp->rcu, destroy_perf_domain_rcu);
> >  
> > -	return;
> > +	return !!pd;
> >  
> >  free:
> >  	free_pd(pd);
> > @@ -373,6 +373,8 @@ static void build_perf_domains(const str
> >  	rcu_assign_pointer(rd->pd, NULL);
> >  	if (tmp)
> >  		call_rcu(&tmp->rcu, destroy_perf_domain_rcu);
> > +
> > +	return false;
> >  }
> >  #else
> >  static void free_pd(struct perf_domain *pd) { }
> > @@ -2173,6 +2175,9 @@ void partition_sched_domains(int ndoms_n
> >  	}
> >  
> >  #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
> > +	{
> > +	bool has_eas = false;
> 
> That one wants to be declared '__maybe_unused' at the top of
> partition_sched_domains() perhaps ? Just to avoid the { } section.
> Hopefully the compiler will be smart enough to not use stack space for
> it if not needed.
> 
> > +
> >  	/* Build perf. domains: */
> >  	for (i = 0; i < ndoms_new; i++) {
> >  		for (j = 0; j < n && !sched_energy_update; j++) {
> > @@ -2181,10 +2186,13 @@ void partition_sched_domains(int ndoms_n
> >  				goto match3;
> 
> And we want to do 'has_eas = true' just before 'goto match3' here.
> Otherwise we'll end up disabling EAS if we hit 'goto match3' for all
> root domains.
> 
> >  		}
> >  		/* No match - add perf. domains for a new rd */
> > -		build_perf_domains(doms_new[i]);
> > +		has_eas |= build_perf_domains(doms_new[i]);
> >  match3:
> >  		;
> >  	}
> > +
> > +	sched_energy_set(has_eas);
> > +	}
> >  #endif
> >  
> >  	/* Remember the new sched domains: */
> 
> Other than that I guess that should work OK :-)

So I actually just came across your patch that does
static_branch_{inc,dec} on sched_smt_present. Maybe I could do something
similar here ?

How about something like the below (totally untested):

diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
index 8f9dfea60344..10ac32a0dc2e 100644
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -298,6 +298,7 @@ static void destroy_perf_domain_rcu(struct rcu_head *rp)
 
        pd = container_of(rp, struct perf_domain, rcu);
        free_pd(pd);
+       static_branch_dec_cpuslocked(&sched_energy_present);
 }
 
 static void sched_energy_start(int ndoms_new, cpumask_var_t doms_new[])
@@ -421,6 +422,8 @@ static void build_perf_domains(const struct cpumask *cpu_map)
        /* Attach the new list of performance domains to the root domain. */
        tmp = rd->pd;
        rcu_assign_pointer(rd->pd, pd);
+       static_branch_inc_cpuslocked(&sched_energy_present);
+
        if (tmp)
                call_rcu(&tmp->rcu, destroy_perf_domain_rcu);
 
@@ -2246,7 +2249,6 @@ void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
 match3:
                ;
        }
-       sched_energy_start(ndoms_new, doms_new);
 #endif
 
        /* Remember the new sched domains: */