1 .. SPDX-License-Identifier: GPL-2.0
3 =======================
4 Energy Model of devices
5 =======================
10 The Energy Model (EM) framework serves as an interface between drivers knowing
11 the power consumed by devices at various performance levels, and the kernel
12 subsystems willing to use that information to make energy-aware decisions.
14 The source of the information about the power consumed by devices can vary greatly
15 from one platform to another. These power costs can be estimated using
16 devicetree data in some cases. In others, the firmware will know better.
17 Alternatively, userspace might be best positioned. And so on. In order to avoid
18 each and every client subsystem to re-implement support for each and every
19 possible source of information on its own, the EM framework intervenes as an
20 abstraction layer which standardizes the format of power cost tables in the
21 kernel, hence enabling to avoid redundant work.
23 The power values might be expressed in milli-Watts or in an 'abstract scale'.
24 Multiple subsystems might use the EM and it is up to the system integrator to
25 check that the requirements for the power value scale types are met. An example
26 can be found in the Energy-Aware Scheduler documentation
27 Documentation/scheduler/sched-energy.rst. For some subsystems like thermal or
28 powercap power values expressed in an 'abstract scale' might cause issues.
29 These subsystems are more interested in estimation of power used in the past,
30 thus the real milli-Watts might be needed. An example of these requirements can
31 be found in the Intelligent Power Allocation in
32 Documentation/driver-api/thermal/power_allocator.rst.
33 Kernel subsystems might implement automatic detection to check whether EM
34 registered devices have inconsistent scale (based on EM internal flag).
35 Important thing to keep in mind is that when the power values are expressed in
36 an 'abstract scale' deriving real energy in milli-Joules would not be possible.
38 The figure below depicts an example of drivers (Arm-specific here, but the
39 approach is applicable to any architecture) providing power costs to the EM
40 framework, and interested clients reading the data from it::
42 +---------------+ +-----------------+ +---------------+
43 | Thermal (IPA) | | Scheduler (EAS) | | Other |
44 +---------------+ +-----------------+ +---------------+
47 +---------+ | +---------+
50 +---------------------+
53 +---------------------+
55 | | | em_dev_register_perf_domain()
56 +----------+ | +---------+
58 +---------------+ +---------------+ +--------------+
59 | cpufreq-dt | | arm_scmi | | Other |
60 +---------------+ +---------------+ +--------------+
63 +--------------+ +---------------+ +--------------+
64 | Device Tree | | Firmware | | ? |
65 +--------------+ +---------------+ +--------------+
67 In case of CPU devices the EM framework manages power cost tables per
68 'performance domain' in the system. A performance domain is a group of CPUs
69 whose performance is scaled together. Performance domains generally have a
70 1-to-1 mapping with CPUFreq policies. All CPUs in a performance domain are
71 required to have the same micro-architecture. CPUs in different performance
72 domains can have different micro-architectures.
81 CONFIG_ENERGY_MODEL must be enabled to use the EM framework.
84 2.2 Registration of performance domains
85 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
87 Registration of 'advanced' EM
88 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
90 The 'advanced' EM gets it's name due to the fact that the driver is allowed
91 to provide more precised power model. It's not limited to some implemented math
92 formula in the framework (like it's in 'simple' EM case). It can better reflect
93 the real power measurements performed for each performance state. Thus, this
94 registration method should be preferred in case considering EM static power
95 (leakage) is important.
97 Drivers are expected to register performance domains into the EM framework by
98 calling the following API::
100 int em_dev_register_perf_domain(struct device *dev, unsigned int nr_states,
101 struct em_data_callback *cb, cpumask_t *cpus, bool milliwatts);
103 Drivers must provide a callback function returning <frequency, power> tuples
104 for each performance state. The callback function provided by the driver is free
105 to fetch data from any relevant location (DT, firmware, ...), and by any mean
106 deemed necessary. Only for CPU devices, drivers must specify the CPUs of the
107 performance domains using cpumask. For other devices than CPUs the last
108 argument must be set to NULL.
109 The last argument 'milliwatts' is important to set with correct value. Kernel
110 subsystems which use EM might rely on this flag to check if all EM devices use
111 the same scale. If there are different scales, these subsystems might decide
112 to: return warning/error, stop working or panic.
113 See Section 3. for an example of driver implementing this
114 callback, or Section 2.4 for further documentation on this API
116 Registration of 'simple' EM
117 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
119 The 'simple' EM is registered using the framework helper function
120 cpufreq_register_em_with_opp(). It implements a power model which is tight to
125 The EM which is registered using this method might not reflect correctly the
126 physics of a real device, e.g. when static power (leakage) is important.
129 2.3 Accessing performance domains
130 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
132 There are two API functions which provide the access to the energy model:
133 em_cpu_get() which takes CPU id as an argument and em_pd_get() with device
134 pointer as an argument. It depends on the subsystem which interface it is
135 going to use, but in case of CPU devices both functions return the same
138 Subsystems interested in the energy model of a CPU can retrieve it using the
139 em_cpu_get() API. The energy model tables are allocated once upon creation of
140 the performance domains, and kept in memory untouched.
142 The energy consumed by a performance domain can be estimated using the
143 em_cpu_energy() API. The estimation is performed assuming that the schedutil
144 CPUfreq governor is in use in case of CPU device. Currently this calculation is
145 not provided for other type of devices.
147 More details about the above APIs can be found in ``<linux/energy_model.h>``
151 2.4 Description details of this API
152 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
153 .. kernel-doc:: include/linux/energy_model.h
156 .. kernel-doc:: kernel/power/energy_model.c
163 The CPUFreq framework supports dedicated callback for registering
164 the EM for a given CPU(s) 'policy' object: cpufreq_driver::register_em().
165 That callback has to be implemented properly for a given driver,
166 because the framework would call it at the right time during setup.
167 This section provides a simple example of a CPUFreq driver registering a
168 performance domain in the Energy Model framework using the (fake) 'foo'
169 protocol. The driver implements an est_power() function to be provided to the
172 -> drivers/cpufreq/foo_cpufreq.c
174 01 static int est_power(unsigned long *mW, unsigned long *KHz,
175 02 struct device *dev)
179 06 /* Use the 'foo' protocol to ceil the frequency */
180 07 freq = foo_get_freq_ceil(dev, *KHz);
184 11 /* Estimate the power cost for the dev at the relevant freq. */
185 12 power = foo_estimate_power(dev, freq);
189 16 /* Return the values to the EM framework */
196 23 static void foo_cpufreq_register_em(struct cpufreq_policy *policy)
198 25 struct em_data_callback em_cb = EM_DATA_CB(est_power);
199 26 struct device *cpu_dev;
202 29 cpu_dev = get_cpu_device(cpumask_first(policy->cpus));
204 31 /* Find the number of OPPs for this policy */
205 32 nr_opp = foo_get_nr_opp(policy);
207 34 /* And register the new performance domain */
208 35 em_dev_register_perf_domain(cpu_dev, nr_opp, &em_cb, policy->cpus,
212 39 static struct cpufreq_driver foo_cpufreq_driver = {
213 40 .register_em = foo_cpufreq_register_em,