| Message ID | 20190208100554.32196-11-patrick.bellasi@arm.com (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
| Series | Add utilization clamping support | expand |
Hi Patrick, On Friday 08 Feb 2019 at 10:05:49 (+0000), Patrick Bellasi wrote: > The Energy Aware Scheduler (AES) estimates the energy impact of waking s/AES/EAS > up a task on a given CPU. This estimation is based on: > a) an (active) power consumptions defined for each CPU frequency s/consumptions/consumption > b) an estimation of which frequency will be used on each CPU > c) an estimation of the busy time (utilization) of each CPU > > Utilization clamping can affect both b) and c) estimations. A CPU is > expected to run: > - on an higher than required frequency, but for a shorter time, in case > its estimated utilization will be smaller then the minimum utilization s/then/than > enforced by uclamp > - on a smaller than required frequency, but for a longer time, in case > its estimated utilization is bigger then the maximum utilization s/then/than > enforced by uclamp > > While effects on busy time for both boosted/capped tasks are already > considered by compute_energy(), clamping effects on frequency selection > are currently ignored by that function. > > Fix it by considering how CPU clamp values will be affected by a > task waking up and being RUNNABLE on that CPU. > > Do that by refactoring schedutil_freq_util() to take an additional > task_struct* which allows EAS to evaluate the impact on clamp values of > a task being eventually queued in a CPU. Clamp values are applied to the > RT+CFS utilization only when a FREQUENCY_UTIL is required by > compute_energy(). > > Do note that switching from ENERGY_UTIL to FREQUENCY_UTIL in the > computation of cpu_util signal implies that we are more likely to > estimate the higherst OPP when a RT task is running in another CPU of s/higherst/highest > the same performance domain. This can have an impact on energy > estimation but: > - it's not easy to say which approach is better, since it quite likely > depends on the use case > - the original approach could still be obtained by setting a smaller > task-specific util_min whenever required > > Since we are at that: > - rename schedutil_freq_util() into schedutil_cpu_util(), > since it's not only used for frequency selection. > - use "unsigned int" instead of "unsigned long" whenever the tracked > utilization value is not expected to overflow 32bit. We use unsigned long all over the place right ? All the task_util*() functions return unsigned long, the capacity-related functions too, and util_avg is an unsigned long in sched_avg. So I'm not sure if we want to do this TBH. > Signed-off-by: Patrick Bellasi <patrick.bellasi@arm.com> > Cc: Ingo Molnar <mingo@redhat.com> > Cc: Peter Zijlstra <peterz@infradead.org> > Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> > > --- > Changes in v7: > Message-ID: <20190122151404.5rtosic6puixado3@queper01-lin> > - add a note on side-effects due to the usage of FREQUENCY_UTIL for > performance domain frequency estimation > - add a similer note to this changelog > --- > kernel/sched/cpufreq_schedutil.c | 18 ++++++++------- > kernel/sched/fair.c | 39 +++++++++++++++++++++++++++----- > kernel/sched/sched.h | 18 ++++----------- > 3 files changed, 48 insertions(+), 27 deletions(-) > > diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c > index 70a8b87fa29c..fdad719fca8b 100644 > --- a/kernel/sched/cpufreq_schedutil.c > +++ b/kernel/sched/cpufreq_schedutil.c > @@ -195,10 +195,11 @@ static unsigned int get_next_freq(struct sugov_policy *sg_policy, > * based on the task model parameters and gives the minimal utilization > * required to meet deadlines. > */ > -unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, > - unsigned long max, enum schedutil_type type) > +unsigned int schedutil_cpu_util(int cpu, unsigned int util_cfs, > + unsigned int max, enum schedutil_type type, > + struct task_struct *p) > { > - unsigned long dl_util, util, irq; > + unsigned int dl_util, util, irq; > struct rq *rq = cpu_rq(cpu); > > if (!IS_BUILTIN(CONFIG_UCLAMP_TASK) && > @@ -229,7 +230,7 @@ unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, > */ > util = util_cfs + cpu_util_rt(rq); > if (type == FREQUENCY_UTIL) > - util = uclamp_util(rq, util); > + util = uclamp_util_with(rq, util, p); > > dl_util = cpu_util_dl(rq); > > @@ -283,13 +284,14 @@ unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, > static unsigned long sugov_get_util(struct sugov_cpu *sg_cpu) > { > struct rq *rq = cpu_rq(sg_cpu->cpu); > - unsigned long util = cpu_util_cfs(rq); > - unsigned long max = arch_scale_cpu_capacity(NULL, sg_cpu->cpu); > + unsigned int util_cfs = cpu_util_cfs(rq); > + unsigned int cpu_cap = arch_scale_cpu_capacity(NULL, sg_cpu->cpu); Do you really need this one ? What's wrong with 'max' :-) ? > - sg_cpu->max = max; > + sg_cpu->max = cpu_cap; > sg_cpu->bw_dl = cpu_bw_dl(rq); > > - return schedutil_freq_util(sg_cpu->cpu, util, max, FREQUENCY_UTIL); > + return schedutil_cpu_util(sg_cpu->cpu, util_cfs, cpu_cap, > + FREQUENCY_UTIL, NULL); > } > > /** > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c > index 8c0aa76af90a..f6b0808e01ad 100644 > --- a/kernel/sched/fair.c > +++ b/kernel/sched/fair.c > @@ -6453,11 +6453,20 @@ static unsigned long cpu_util_next(int cpu, struct task_struct *p, int dst_cpu) > static long > compute_energy(struct task_struct *p, int dst_cpu, struct perf_domain *pd) > { > - long util, max_util, sum_util, energy = 0; > + unsigned int max_util, cfs_util, cpu_util, cpu_cap; > + unsigned long sum_util, energy = 0; > int cpu; > > for (; pd; pd = pd->next) { > + struct cpumask *pd_mask = perf_domain_span(pd); > + > + /* > + * The energy model mandate all the CPUs of a performance s/mandate/mandates > + * domain have the same capacity. > + */ > + cpu_cap = arch_scale_cpu_capacity(NULL, cpumask_first(pd_mask)); > max_util = sum_util = 0; > + > /* > * The capacity state of CPUs of the current rd can be driven by > * CPUs of another rd if they belong to the same performance > @@ -6468,11 +6477,29 @@ compute_energy(struct task_struct *p, int dst_cpu, struct perf_domain *pd) > * it will not appear in its pd list and will not be accounted > * by compute_energy(). > */ > - for_each_cpu_and(cpu, perf_domain_span(pd), cpu_online_mask) { > - util = cpu_util_next(cpu, p, dst_cpu); > - util = schedutil_energy_util(cpu, util); > - max_util = max(util, max_util); > - sum_util += util; > + for_each_cpu_and(cpu, pd_mask, cpu_online_mask) { > + cfs_util = cpu_util_next(cpu, p, dst_cpu); > + > + /* > + * Busy time computation: utilization clamping is not > + * required since the ratio (sum_util / cpu_capacity) > + * is already enough to scale the EM reported power > + * consumption at the (eventually clamped) cpu_capacity. > + */ > + sum_util += schedutil_cpu_util(cpu, cfs_util, cpu_cap, > + ENERGY_UTIL, NULL); > + > + /* > + * Performance domain frequency: utilization clamping > + * must be considered since it affects the selection > + * of the performance domain frequency. > + * NOTE: in case RT tasks are running, by default the > + * FREQUENCY_UTIL's utilization can be max OPP. > + */ > + cpu_util = schedutil_cpu_util(cpu, cfs_util, cpu_cap, > + FREQUENCY_UTIL, > + cpu == dst_cpu ? p : NULL); > + max_util = max(max_util, cpu_util); > } > > energy += em_pd_energy(pd->em_pd, max_util, sum_util); > diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h > index de181b8a3a2a..b9acef080d99 100644 > --- a/kernel/sched/sched.h > +++ b/kernel/sched/sched.h > @@ -2335,6 +2335,7 @@ static inline unsigned long capacity_orig_of(int cpu) > #endif > > #ifdef CONFIG_CPU_FREQ_GOV_SCHEDUTIL > + > /** > * enum schedutil_type - CPU utilization type Since you're using this enum unconditionally in fair.c, you should to move it out of the #ifdef CONFIG_CPU_FREQ_GOV_SCHEDUTIL block, I think. > * @FREQUENCY_UTIL: Utilization used to select frequency > @@ -2350,15 +2351,9 @@ enum schedutil_type { > ENERGY_UTIL, > }; > > -unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, > - unsigned long max, enum schedutil_type type); > - > -static inline unsigned long schedutil_energy_util(int cpu, unsigned long cfs) > -{ > - unsigned long max = arch_scale_cpu_capacity(NULL, cpu); > - > - return schedutil_freq_util(cpu, cfs, max, ENERGY_UTIL); > -} > +unsigned int schedutil_cpu_util(int cpu, unsigned int util_cfs, > + unsigned int max, enum schedutil_type type, > + struct task_struct *p); > > static inline unsigned long cpu_bw_dl(struct rq *rq) > { > @@ -2387,10 +2382,7 @@ static inline unsigned long cpu_util_rt(struct rq *rq) > return READ_ONCE(rq->avg_rt.util_avg); > } > #else /* CONFIG_CPU_FREQ_GOV_SCHEDUTIL */ > -static inline unsigned long schedutil_energy_util(int cpu, unsigned long cfs) > -{ > - return cfs; > -} > +#define schedutil_cpu_util(cpu, util_cfs, max, type, p) 0 > #endif > > #ifdef CONFIG_HAVE_SCHED_AVG_IRQ > -- > 2.20.1 > Thanks, Quentin
On 06-Mar 17:21, Quentin Perret wrote: [...] > > Since we are at that: > > - rename schedutil_freq_util() into schedutil_cpu_util(), > > since it's not only used for frequency selection. > > - use "unsigned int" instead of "unsigned long" whenever the tracked > > utilization value is not expected to overflow 32bit. > > We use unsigned long all over the place right ? All the task_util*() > functions return unsigned long, the capacity-related functions too, and > util_avg is an unsigned long in sched_avg. So I'm not sure if we want to > do this TBH. For utilization we never need more then an "unsigned int" as storage class. Even at RQ level, 32bits allows +4mln tasks. However we started with long and keep going on with that, this was just an attempt to incrementally fix that whenever we do some changes or we add some new code. But, perhaps a single whole sale update patch would fit better this job in case we really wanna do it at some point. I'll drop this change in v8 and keep this patch focused on functional bits, don't want to risk to sidetrack the discussion again. [...] > > @@ -283,13 +284,14 @@ unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, > > static unsigned long sugov_get_util(struct sugov_cpu *sg_cpu) > > { > > struct rq *rq = cpu_rq(sg_cpu->cpu); > > - unsigned long util = cpu_util_cfs(rq); > > - unsigned long max = arch_scale_cpu_capacity(NULL, sg_cpu->cpu); > > + unsigned int util_cfs = cpu_util_cfs(rq); > > + unsigned int cpu_cap = arch_scale_cpu_capacity(NULL, sg_cpu->cpu); > > Do you really need this one ? What's wrong with 'max' :-) ? Being a pretty "generic" and thus confusing name is not enough? :) Anyway, same reasoning as above and same conclusions: I'll drop the renaming so that we don't sidetrack the discussion on v8. [...] > > diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h > > index de181b8a3a2a..b9acef080d99 100644 > > --- a/kernel/sched/sched.h > > +++ b/kernel/sched/sched.h > > @@ -2335,6 +2335,7 @@ static inline unsigned long capacity_orig_of(int cpu) > > #endif > > > > #ifdef CONFIG_CPU_FREQ_GOV_SCHEDUTIL > > + > > /** > > * enum schedutil_type - CPU utilization type > > Since you're using this enum unconditionally in fair.c, you should to > move it out of the #ifdef CONFIG_CPU_FREQ_GOV_SCHEDUTIL block, I think. > > > * @FREQUENCY_UTIL: Utilization used to select frequency > > @@ -2350,15 +2351,9 @@ enum schedutil_type { > > ENERGY_UTIL, > > }; Good point, will do! Cheers, Patrick
diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c index 70a8b87fa29c..fdad719fca8b 100644 --- a/kernel/sched/cpufreq_schedutil.c +++ b/kernel/sched/cpufreq_schedutil.c @@ -195,10 +195,11 @@ static unsigned int get_next_freq(struct sugov_policy *sg_policy, * based on the task model parameters and gives the minimal utilization * required to meet deadlines. */ -unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, - unsigned long max, enum schedutil_type type) +unsigned int schedutil_cpu_util(int cpu, unsigned int util_cfs, + unsigned int max, enum schedutil_type type, + struct task_struct *p) { - unsigned long dl_util, util, irq; + unsigned int dl_util, util, irq; struct rq *rq = cpu_rq(cpu); if (!IS_BUILTIN(CONFIG_UCLAMP_TASK) && @@ -229,7 +230,7 @@ unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, */ util = util_cfs + cpu_util_rt(rq); if (type == FREQUENCY_UTIL) - util = uclamp_util(rq, util); + util = uclamp_util_with(rq, util, p); dl_util = cpu_util_dl(rq); @@ -283,13 +284,14 @@ unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, static unsigned long sugov_get_util(struct sugov_cpu *sg_cpu) { struct rq *rq = cpu_rq(sg_cpu->cpu); - unsigned long util = cpu_util_cfs(rq); - unsigned long max = arch_scale_cpu_capacity(NULL, sg_cpu->cpu); + unsigned int util_cfs = cpu_util_cfs(rq); + unsigned int cpu_cap = arch_scale_cpu_capacity(NULL, sg_cpu->cpu); - sg_cpu->max = max; + sg_cpu->max = cpu_cap; sg_cpu->bw_dl = cpu_bw_dl(rq); - return schedutil_freq_util(sg_cpu->cpu, util, max, FREQUENCY_UTIL); + return schedutil_cpu_util(sg_cpu->cpu, util_cfs, cpu_cap, + FREQUENCY_UTIL, NULL); } /** diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 8c0aa76af90a..f6b0808e01ad 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -6453,11 +6453,20 @@ static unsigned long cpu_util_next(int cpu, struct task_struct *p, int dst_cpu) static long compute_energy(struct task_struct *p, int dst_cpu, struct perf_domain *pd) { - long util, max_util, sum_util, energy = 0; + unsigned int max_util, cfs_util, cpu_util, cpu_cap; + unsigned long sum_util, energy = 0; int cpu; for (; pd; pd = pd->next) { + struct cpumask *pd_mask = perf_domain_span(pd); + + /* + * The energy model mandate all the CPUs of a performance + * domain have the same capacity. + */ + cpu_cap = arch_scale_cpu_capacity(NULL, cpumask_first(pd_mask)); max_util = sum_util = 0; + /* * The capacity state of CPUs of the current rd can be driven by * CPUs of another rd if they belong to the same performance @@ -6468,11 +6477,29 @@ compute_energy(struct task_struct *p, int dst_cpu, struct perf_domain *pd) * it will not appear in its pd list and will not be accounted * by compute_energy(). */ - for_each_cpu_and(cpu, perf_domain_span(pd), cpu_online_mask) { - util = cpu_util_next(cpu, p, dst_cpu); - util = schedutil_energy_util(cpu, util); - max_util = max(util, max_util); - sum_util += util; + for_each_cpu_and(cpu, pd_mask, cpu_online_mask) { + cfs_util = cpu_util_next(cpu, p, dst_cpu); + + /* + * Busy time computation: utilization clamping is not + * required since the ratio (sum_util / cpu_capacity) + * is already enough to scale the EM reported power + * consumption at the (eventually clamped) cpu_capacity. + */ + sum_util += schedutil_cpu_util(cpu, cfs_util, cpu_cap, + ENERGY_UTIL, NULL); + + /* + * Performance domain frequency: utilization clamping + * must be considered since it affects the selection + * of the performance domain frequency. + * NOTE: in case RT tasks are running, by default the + * FREQUENCY_UTIL's utilization can be max OPP. + */ + cpu_util = schedutil_cpu_util(cpu, cfs_util, cpu_cap, + FREQUENCY_UTIL, + cpu == dst_cpu ? p : NULL); + max_util = max(max_util, cpu_util); } energy += em_pd_energy(pd->em_pd, max_util, sum_util); diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index de181b8a3a2a..b9acef080d99 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -2335,6 +2335,7 @@ static inline unsigned long capacity_orig_of(int cpu) #endif #ifdef CONFIG_CPU_FREQ_GOV_SCHEDUTIL + /** * enum schedutil_type - CPU utilization type * @FREQUENCY_UTIL: Utilization used to select frequency @@ -2350,15 +2351,9 @@ enum schedutil_type { ENERGY_UTIL, }; -unsigned long schedutil_freq_util(int cpu, unsigned long util_cfs, - unsigned long max, enum schedutil_type type); - -static inline unsigned long schedutil_energy_util(int cpu, unsigned long cfs) -{ - unsigned long max = arch_scale_cpu_capacity(NULL, cpu); - - return schedutil_freq_util(cpu, cfs, max, ENERGY_UTIL); -} +unsigned int schedutil_cpu_util(int cpu, unsigned int util_cfs, + unsigned int max, enum schedutil_type type, + struct task_struct *p); static inline unsigned long cpu_bw_dl(struct rq *rq) { @@ -2387,10 +2382,7 @@ static inline unsigned long cpu_util_rt(struct rq *rq) return READ_ONCE(rq->avg_rt.util_avg); } #else /* CONFIG_CPU_FREQ_GOV_SCHEDUTIL */ -static inline unsigned long schedutil_energy_util(int cpu, unsigned long cfs) -{ - return cfs; -} +#define schedutil_cpu_util(cpu, util_cfs, max, type, p) 0 #endif #ifdef CONFIG_HAVE_SCHED_AVG_IRQ
The Energy Aware Scheduler (AES) estimates the energy impact of waking up a task on a given CPU. This estimation is based on: a) an (active) power consumptions defined for each CPU frequency b) an estimation of which frequency will be used on each CPU c) an estimation of the busy time (utilization) of each CPU Utilization clamping can affect both b) and c) estimations. A CPU is expected to run: - on an higher than required frequency, but for a shorter time, in case its estimated utilization will be smaller then the minimum utilization enforced by uclamp - on a smaller than required frequency, but for a longer time, in case its estimated utilization is bigger then the maximum utilization enforced by uclamp While effects on busy time for both boosted/capped tasks are already considered by compute_energy(), clamping effects on frequency selection are currently ignored by that function. Fix it by considering how CPU clamp values will be affected by a task waking up and being RUNNABLE on that CPU. Do that by refactoring schedutil_freq_util() to take an additional task_struct* which allows EAS to evaluate the impact on clamp values of a task being eventually queued in a CPU. Clamp values are applied to the RT+CFS utilization only when a FREQUENCY_UTIL is required by compute_energy(). Do note that switching from ENERGY_UTIL to FREQUENCY_UTIL in the computation of cpu_util signal implies that we are more likely to estimate the higherst OPP when a RT task is running in another CPU of the same performance domain. This can have an impact on energy estimation but: - it's not easy to say which approach is better, since it quite likely depends on the use case - the original approach could still be obtained by setting a smaller task-specific util_min whenever required Since we are at that: - rename schedutil_freq_util() into schedutil_cpu_util(), since it's not only used for frequency selection. - use "unsigned int" instead of "unsigned long" whenever the tracked utilization value is not expected to overflow 32bit. Signed-off-by: Patrick Bellasi <patrick.bellasi@arm.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> --- Changes in v7: Message-ID: <20190122151404.5rtosic6puixado3@queper01-lin> - add a note on side-effects due to the usage of FREQUENCY_UTIL for performance domain frequency estimation - add a similer note to this changelog --- kernel/sched/cpufreq_schedutil.c | 18 ++++++++------- kernel/sched/fair.c | 39 +++++++++++++++++++++++++++----- kernel/sched/sched.h | 18 ++++----------- 3 files changed, 48 insertions(+), 27 deletions(-)