On Monday 22 Feb 2021 at 15:58:56 (+0000), Quentin Perret wrote:
> But in any case, if we're going to address this, I'm still not sure this
> patch will be what we want. As per my first comment we need to keep the
> frequency estimation right.
Totally untested, but I think in principle you would like something like
the snippet below. Would that work?
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 04a3ce20da67..6594d875c6ac 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -6534,8 +6534,13 @@ compute_energy(struct task_struct *p, int dst_cpu,
struct perf_domain *pd)
* its pd list and will not be accounted by compute_energy().
*/
for_each_cpu_and(cpu, pd_mask, cpu_online_mask) {
- unsigned long cpu_util, util_cfs = cpu_util_next(cpu, p,
dst_cpu);
+ unsigned long util_freq = cpu_util_next(cpu, p, dst_cpu);
+ unsigned long util_running = cpu_util_without(cpu, p);
struct task_struct *tsk = cpu == dst_cpu ? p : NULL;
+ unsigned long cpu_util;
+
+ if (cpu == dst_cpu)
+ util_running += task_util_est();
/*
* Busy time computation: utilization clamping is not
@@ -6543,7 +6548,7 @@ compute_energy(struct task_struct *p, int dst_cpu, struct
perf_domain *pd)
* is already enough to scale the EM reported power
* consumption at the (eventually clamped) cpu_capacity.
*/
- sum_util += schedutil_cpu_util(cpu, util_cfs, cpu_cap,
+ sum_util += schedutil_cpu_util(cpu, util_running, cpu_cap,
ENERGY_UTIL, NULL);
/*
@@ -6553,7 +6558,7 @@ compute_energy(struct task_struct *p, int dst_cpu, struct
perf_domain *pd)
* NOTE: in case RT tasks are running, by default the
* FREQUENCY_UTIL's utilization can be max OPP.
*/
- cpu_util = schedutil_cpu_util(cpu, util_cfs, cpu_cap,
+ cpu_util = schedutil_cpu_util(cpu, util_freq, cpu_cap,
FREQUENCY_UTIL, tsk);
max_util = max(max_util, cpu_util);
}
