Energy Model of devices¶
1. Overview¶
The Energy Model (EM) framework serves as an interface between drivers knowingthe power consumed by devices at various performance levels, and the kernelsubsystems willing to use that information to make energy-aware decisions.
The source of the information about the power consumed by devices can vary greatlyfrom one platform to another. These power costs can be estimated usingdevicetree data in some cases. In others, the firmware will know better.Alternatively, userspace might be best positioned. And so on. In order to avoideach and every client subsystem to re-implement support for each and everypossible source of information on its own, the EM framework intervenes as anabstraction layer which standardizes the format of power cost tables in thekernel, hence enabling to avoid redundant work.
The figure below depicts an example of drivers (Arm-specific here, but theapproach is applicable to any architecture) providing power costs to the EMframework, and interested clients reading the data from it:
+---------------+ +-----------------+ +---------------+| Thermal (IPA) | | Scheduler (EAS) | | Other |+---------------+ +-----------------+ +---------------+ | | em_cpu_energy() | | | em_cpu_get() | +---------+ | +---------+ | | | v v v +---------------------+ | Energy Model | | Framework | +---------------------+ ^ ^ ^ | | | em_dev_register_perf_domain() +----------+ | +---------+ | | | +---------------+ +---------------+ +--------------+ | cpufreq-dt | | arm_scmi | | Other | +---------------+ +---------------+ +--------------+ ^ ^ ^ | | | +--------------+ +---------------+ +--------------+ | Device Tree | | Firmware | | ? | +--------------+ +---------------+ +--------------+
In case of CPU devices the EM framework manages power cost tables per‘performance domain’ in the system. A performance domain is a group of CPUswhose performance is scaled together. Performance domains generally have a1-to-1 mapping with CPUFreq policies. All CPUs in a performance domain arerequired to have the same micro-architecture. CPUs in different performancedomains can have different micro-architectures.
2. Core APIs¶
2.1 Config options¶
CONFIG_ENERGY_MODEL must be enabled to use the EM framework.
2.2 Registration of performance domains¶
Drivers are expected to register performance domains into the EM framework bycalling the following API:
int em_dev_register_perf_domain(struct device *dev, unsigned int nr_states, struct em_data_callback *cb, cpumask_t *cpus);
Drivers must provide a callback function returning <frequency, power> tuplesfor each performance state. The callback function provided by the driver is freeto fetch data from any relevant location (DT, firmware, …), and by any meandeemed necessary. Only for CPU devices, drivers must specify the CPUs of theperformance domains using cpumask. For other devices than CPUs the lastargument must be set to NULL.See Section 3. for an example of driver implementing thiscallback, and kernel/power/energy_model.c for further documentation on thisAPI.
2.3 Accessing performance domains¶
There are two API functions which provide the access to the energy model:em_cpu_get() which takes CPU id as an argument and em_pd_get() with devicepointer as an argument. It depends on the subsystem which interface it isgoing to use, but in case of CPU devices both functions return the sameperformance domain.
Subsystems interested in the energy model of a CPU can retrieve it using theem_cpu_get() API. The energy model tables are allocated once upon creation ofthe performance domains, and kept in memory untouched.
The energy consumed by a performance domain can be estimated using theem_cpu_energy() API. The estimation is performed assuming that the schedutilCPUfreq governor is in use in case of CPU device. Currently this calculation isnot provided for other type of devices.
More details about the above APIs can be found in include/linux/energy_model.h.
3. Example driver¶
This section provides a simple example of a CPUFreq driver registering aperformance domain in the Energy Model framework using the (fake) ‘foo’protocol. The driver implements an est_power() function to be provided to theEM framework:
-> drivers/cpufreq/foo_cpufreq.c01 static int est_power(unsigned long *mW, unsigned long *KHz,02 struct device *dev)03 {04 long freq, power;0506 /* Use the 'foo' protocol to ceil the frequency */07 freq = foo_get_freq_ceil(dev, *KHz);08 if (freq < 0);09 return freq;1011 /* Estimate the power cost for the dev at the relevant freq. */12 power = foo_estimate_power(dev, freq);13 if (power < 0);14 return power;1516 /* Return the values to the EM framework */17 *mW = power;18 *KHz = freq;1920 return 0;21 }2223 static int foo_cpufreq_init(struct cpufreq_policy *policy)24 {25 struct em_data_callback em_cb = EM_DATA_CB(est_power);26 struct device *cpu_dev;27 int nr_opp, ret;2829 cpu_dev = get_cpu_device(cpumask_first(policy->cpus));3031 /* Do the actual CPUFreq init work ... */32 ret = do_foo_cpufreq_init(policy);33 if (ret)34 return ret;3536 /* Find the number of OPPs for this policy */37 nr_opp = foo_get_nr_opp(policy);3839 /* And register the new performance domain */40 em_dev_register_perf_domain(cpu_dev, nr_opp, &em_cb, policy->cpus);4142 return 0;43 }