1 CPU frequency and voltage scaling code in the Linux(TM) kernel
4 L i n u x C P U F r e q
9 Dominik Brodowski <linux@brodo.de>
10 David Kimdon <dwhedon@debian.org>
14 Clock scaling allows you to change the clock speed of the CPUs on the
15 fly. This is a nice method to save battery power, because the lower
16 the clock speed, the less power the CPU consumes.
21 1. CPUFreq core and interfaces
23 3. CPUFreq Table Generation with Operating Performance Point (OPP)
25 1. General Information
26 =======================
28 The CPUFreq core code is located in drivers/cpufreq/cpufreq.c. This
29 cpufreq code offers a standardized interface for the CPUFreq
30 architecture drivers (those pieces of code that do actual
31 frequency transitions), as well as to "notifiers". These are device
32 drivers or other part of the kernel that need to be informed of
33 policy changes (ex. thermal modules like ACPI) or of all
34 frequency changes (ex. timing code) or even need to force certain
35 speed limits (like LCD drivers on ARM architecture). Additionally, the
36 kernel "constant" loops_per_jiffy is updated on frequency changes
39 Reference counting is done by cpufreq_get_cpu and cpufreq_put_cpu,
40 which make sure that the cpufreq processor driver is correctly
41 registered with the core, and will not be unloaded until
42 cpufreq_put_cpu is called.
47 CPUFreq notifiers conform to the standard kernel notifier interface.
48 See linux/include/linux/notifier.h for details on notifiers.
50 There are two different CPUFreq notifiers - policy notifiers and
54 2.1 CPUFreq policy notifiers
55 ----------------------------
57 These are notified when a new policy is intended to be set. Each
58 CPUFreq policy notifier is called three times for a policy transition:
60 1.) During CPUFREQ_ADJUST all CPUFreq notifiers may change the limit if
61 they see a need for this - may it be thermal considerations or
64 2.) During CPUFREQ_INCOMPATIBLE only changes may be done in order to avoid
67 3.) And during CPUFREQ_NOTIFY all notifiers are informed of the new policy
68 - if two hardware drivers failed to agree on a new policy before this
69 stage, the incompatible hardware shall be shut down, and the user
72 The phase is specified in the second argument to the notifier.
74 The third argument, a void *pointer, points to a struct cpufreq_policy
75 consisting of five values: cpu, min, max, policy and max_cpu_freq. min
76 and max are the lower and upper frequencies (in kHz) of the new
77 policy, policy the new policy, cpu the number of the affected CPU; and
78 max_cpu_freq the maximum supported CPU frequency. This value is given
79 for informational purposes only.
82 2.2 CPUFreq transition notifiers
83 --------------------------------
85 These are notified twice when the CPUfreq driver switches the CPU core
86 frequency and this change has any external implications.
88 The second argument specifies the phase - CPUFREQ_PRECHANGE or
91 The third argument is a struct cpufreq_freqs with the following
93 cpu - number of the affected CPU
97 3. CPUFreq Table Generation with Operating Performance Point (OPP)
98 ==================================================================
99 For details about OPP, see Documentation/power/opp.txt
101 dev_pm_opp_init_cpufreq_table - cpufreq framework typically is initialized with
102 cpufreq_frequency_table_cpuinfo which is provided with the list of
103 frequencies that are available for operation. This function provides
104 a ready to use conversion routine to translate the OPP layer's internal
105 information about the available frequencies into a format readily
106 providable to cpufreq.
108 WARNING: Do not use this function in interrupt context.
114 r = dev_pm_opp_init_cpufreq_table(dev, &freq_table);
116 cpufreq_frequency_table_cpuinfo(policy, freq_table);
117 /* Do other things */
120 NOTE: This function is available only if CONFIG_CPU_FREQ is enabled in
121 addition to CONFIG_PM_OPP.
123 dev_pm_opp_free_cpufreq_table - Free up the table allocated by dev_pm_opp_init_cpufreq_table