arm: Cpu invariant scheduler load-tracking and capacity support

Provides the scheduler with a cpu scaling correction factor for more
accurate load-tracking and cpu capacity handling.

The Energy Model (EM) (in fact the capacity value of the last element
of the capacity states vector of the core (MC) level sched_group_energy
structure) is used instead of the arm arch specific cpu_efficiency and
dtb property 'clock-frequency' values as the source for this cpu
scaling factor.

The cpu capacity value depends on the micro-architecture and the
maximum frequency of the cpu.

The maximum frequency part should not be confused with the frequency
invariant scheduler load-tracking support which deals with frequency
related scaling due to DFVS functionality.

Signed-off-by: Juri Lelli <juri.lelli@arm.com>
Signed-off-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
This commit is contained in:
Dietmar Eggemann
2015-07-10 13:57:19 +01:00
committed by Leo Yan
parent f0f739d887
commit c7aeeb88c7

View File

@@ -153,6 +153,8 @@ static void __init parse_dt_topology(void)
}
static const struct sched_group_energy * const cpu_core_energy(int cpu);
/*
* Look for a customed capacity of a CPU in the cpu_capacity table during the
* boot. The update of all CPUs is in O(n^2) for heteregeneous system but the
@@ -160,10 +162,14 @@ static void __init parse_dt_topology(void)
*/
static void update_cpu_capacity(unsigned int cpu)
{
if (!cpu_capacity(cpu))
return;
unsigned long capacity = SCHED_CAPACITY_SCALE;
set_capacity_scale(cpu, cpu_capacity(cpu) / middle_capacity);
if (cpu_core_energy(cpu)) {
int max_cap_idx = cpu_core_energy(cpu)->nr_cap_states - 1;
capacity = cpu_core_energy(cpu)->cap_states[max_cap_idx].cap;
}
set_capacity_scale(cpu, capacity);
pr_info("CPU%u: update cpu_capacity %lu\n",
cpu, arch_scale_cpu_capacity(NULL, cpu));