1 .. SPDX-License-Identifier: GPL-2.0 1 .. SPDX-License-Identifier: GPL-2.0
2 .. include:: <isonum.txt> 2 .. include:: <isonum.txt>
3 3
4 .. |struct cpuidle_state| replace:: :c:type:`s 4 .. |struct cpuidle_state| replace:: :c:type:`struct cpuidle_state <cpuidle_state>`
5 .. |cpufreq| replace:: :doc:`CPU Performance S 5 .. |cpufreq| replace:: :doc:`CPU Performance Scaling <cpufreq>`
6 6
7 ======================== 7 ========================
8 CPU Idle Time Management 8 CPU Idle Time Management
9 ======================== 9 ========================
10 10
11 :Copyright: |copy| 2018 Intel Corporation 11 :Copyright: |copy| 2018 Intel Corporation
12 12
13 :Author: Rafael J. Wysocki <rafael.j.wysocki@in 13 :Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
14 14
15 15
16 Concepts 16 Concepts
17 ======== 17 ========
18 18
19 Modern processors are generally able to enter 19 Modern processors are generally able to enter states in which the execution of
20 a program is suspended and instructions belong 20 a program is suspended and instructions belonging to it are not fetched from
21 memory or executed. Those states are the *idl 21 memory or executed. Those states are the *idle* states of the processor.
22 22
23 Since part of the processor hardware is not us 23 Since part of the processor hardware is not used in idle states, entering them
24 generally allows power drawn by the processor 24 generally allows power drawn by the processor to be reduced and, in consequence,
25 it is an opportunity to save energy. 25 it is an opportunity to save energy.
26 26
27 CPU idle time management is an energy-efficien 27 CPU idle time management is an energy-efficiency feature concerned about using
28 the idle states of processors for this purpose 28 the idle states of processors for this purpose.
29 29
30 Logical CPUs 30 Logical CPUs
31 ------------ 31 ------------
32 32
33 CPU idle time management operates on CPUs as s 33 CPU idle time management operates on CPUs as seen by the *CPU scheduler* (that
34 is the part of the kernel responsible for the 34 is the part of the kernel responsible for the distribution of computational
35 work in the system). In its view, CPUs are *l 35 work in the system). In its view, CPUs are *logical* units. That is, they need
36 not be separate physical entities and may just 36 not be separate physical entities and may just be interfaces appearing to
37 software as individual single-core processors. 37 software as individual single-core processors. In other words, a CPU is an
38 entity which appears to be fetching instructio 38 entity which appears to be fetching instructions that belong to one sequence
39 (program) from memory and executing them, but 39 (program) from memory and executing them, but it need not work this way
40 physically. Generally, three different cases 40 physically. Generally, three different cases can be consider here.
41 41
42 First, if the whole processor can only follow 42 First, if the whole processor can only follow one sequence of instructions (one
43 program) at a time, it is a CPU. In that case 43 program) at a time, it is a CPU. In that case, if the hardware is asked to
44 enter an idle state, that applies to the proce 44 enter an idle state, that applies to the processor as a whole.
45 45
46 Second, if the processor is multi-core, each c 46 Second, if the processor is multi-core, each core in it is able to follow at
47 least one program at a time. The cores need n 47 least one program at a time. The cores need not be entirely independent of each
48 other (for example, they may share caches), bu 48 other (for example, they may share caches), but still most of the time they
49 work physically in parallel with each other, s 49 work physically in parallel with each other, so if each of them executes only
50 one program, those programs run mostly indepen 50 one program, those programs run mostly independently of each other at the same
51 time. The entire cores are CPUs in that case 51 time. The entire cores are CPUs in that case and if the hardware is asked to
52 enter an idle state, that applies to the core 52 enter an idle state, that applies to the core that asked for it in the first
53 place, but it also may apply to a larger unit 53 place, but it also may apply to a larger unit (say a "package" or a "cluster")
54 that the core belongs to (in fact, it may appl 54 that the core belongs to (in fact, it may apply to an entire hierarchy of larger
55 units containing the core). Namely, if all of 55 units containing the core). Namely, if all of the cores in the larger unit
56 except for one have been put into idle states 56 except for one have been put into idle states at the "core level" and the
57 remaining core asks the processor to enter an 57 remaining core asks the processor to enter an idle state, that may trigger it
58 to put the whole larger unit into an idle stat 58 to put the whole larger unit into an idle state which also will affect the
59 other cores in that unit. 59 other cores in that unit.
60 60
61 Finally, each core in a multi-core processor m 61 Finally, each core in a multi-core processor may be able to follow more than one
62 program in the same time frame (that is, each 62 program in the same time frame (that is, each core may be able to fetch
63 instructions from multiple locations in memory 63 instructions from multiple locations in memory and execute them in the same time
64 frame, but not necessarily entirely in paralle 64 frame, but not necessarily entirely in parallel with each other). In that case
65 the cores present themselves to software as "b 65 the cores present themselves to software as "bundles" each consisting of
66 multiple individual single-core "processors", 66 multiple individual single-core "processors", referred to as *hardware threads*
67 (or hyper-threads specifically on Intel hardwa 67 (or hyper-threads specifically on Intel hardware), that each can follow one
68 sequence of instructions. Then, the hardware 68 sequence of instructions. Then, the hardware threads are CPUs from the CPU idle
69 time management perspective and if the process 69 time management perspective and if the processor is asked to enter an idle state
70 by one of them, the hardware thread (or CPU) t 70 by one of them, the hardware thread (or CPU) that asked for it is stopped, but
71 nothing more happens, unless all of the other 71 nothing more happens, unless all of the other hardware threads within the same
72 core also have asked the processor to enter an 72 core also have asked the processor to enter an idle state. In that situation,
73 the core may be put into an idle state individ 73 the core may be put into an idle state individually or a larger unit containing
74 it may be put into an idle state as a whole (i 74 it may be put into an idle state as a whole (if the other cores within the
75 larger unit are in idle states already). 75 larger unit are in idle states already).
76 76
77 Idle CPUs 77 Idle CPUs
78 --------- 78 ---------
79 79
80 Logical CPUs, simply referred to as "CPUs" in 80 Logical CPUs, simply referred to as "CPUs" in what follows, are regarded as
81 *idle* by the Linux kernel when there are no t 81 *idle* by the Linux kernel when there are no tasks to run on them except for the
82 special "idle" task. 82 special "idle" task.
83 83
84 Tasks are the CPU scheduler's representation o 84 Tasks are the CPU scheduler's representation of work. Each task consists of a
85 sequence of instructions to execute, or code, 85 sequence of instructions to execute, or code, data to be manipulated while
86 running that code, and some context informatio 86 running that code, and some context information that needs to be loaded into the
87 processor every time the task's code is run by 87 processor every time the task's code is run by a CPU. The CPU scheduler
88 distributes work by assigning tasks to run to 88 distributes work by assigning tasks to run to the CPUs present in the system.
89 89
90 Tasks can be in various states. In particular 90 Tasks can be in various states. In particular, they are *runnable* if there are
91 no specific conditions preventing their code f 91 no specific conditions preventing their code from being run by a CPU as long as
92 there is a CPU available for that (for example 92 there is a CPU available for that (for example, they are not waiting for any
93 events to occur or similar). When a task beco 93 events to occur or similar). When a task becomes runnable, the CPU scheduler
94 assigns it to one of the available CPUs to run 94 assigns it to one of the available CPUs to run and if there are no more runnable
95 tasks assigned to it, the CPU will load the gi 95 tasks assigned to it, the CPU will load the given task's context and run its
96 code (from the instruction following the last 96 code (from the instruction following the last one executed so far, possibly by
97 another CPU). [If there are multiple runnable 97 another CPU). [If there are multiple runnable tasks assigned to one CPU
98 simultaneously, they will be subject to priori 98 simultaneously, they will be subject to prioritization and time sharing in order
99 to allow them to make some progress over time. 99 to allow them to make some progress over time.]
100 100
101 The special "idle" task becomes runnable if th 101 The special "idle" task becomes runnable if there are no other runnable tasks
102 assigned to the given CPU and the CPU is then 102 assigned to the given CPU and the CPU is then regarded as idle. In other words,
103 in Linux idle CPUs run the code of the "idle" 103 in Linux idle CPUs run the code of the "idle" task called *the idle loop*. That
104 code may cause the processor to be put into on 104 code may cause the processor to be put into one of its idle states, if they are
105 supported, in order to save energy, but if the 105 supported, in order to save energy, but if the processor does not support any
106 idle states, or there is not enough time to sp 106 idle states, or there is not enough time to spend in an idle state before the
107 next wakeup event, or there are strict latency 107 next wakeup event, or there are strict latency constraints preventing any of the
108 available idle states from being used, the CPU 108 available idle states from being used, the CPU will simply execute more or less
109 useless instructions in a loop until it is ass 109 useless instructions in a loop until it is assigned a new task to run.
110 110
111 111
112 .. _idle-loop: 112 .. _idle-loop:
113 113
114 The Idle Loop 114 The Idle Loop
115 ============= 115 =============
116 116
117 The idle loop code takes two major steps in ev 117 The idle loop code takes two major steps in every iteration of it. First, it
118 calls into a code module referred to as the *g 118 calls into a code module referred to as the *governor* that belongs to the CPU
119 idle time management subsystem called ``CPUIdl 119 idle time management subsystem called ``CPUIdle`` to select an idle state for
120 the CPU to ask the hardware to enter. Second, 120 the CPU to ask the hardware to enter. Second, it invokes another code module
121 from the ``CPUIdle`` subsystem, called the *dr 121 from the ``CPUIdle`` subsystem, called the *driver*, to actually ask the
122 processor hardware to enter the idle state sel 122 processor hardware to enter the idle state selected by the governor.
123 123
124 The role of the governor is to find an idle st 124 The role of the governor is to find an idle state most suitable for the
125 conditions at hand. For this purpose, idle st 125 conditions at hand. For this purpose, idle states that the hardware can be
126 asked to enter by logical CPUs are represented 126 asked to enter by logical CPUs are represented in an abstract way independent of
127 the platform or the processor architecture and 127 the platform or the processor architecture and organized in a one-dimensional
128 (linear) array. That array has to be prepared 128 (linear) array. That array has to be prepared and supplied by the ``CPUIdle``
129 driver matching the platform the kernel is run 129 driver matching the platform the kernel is running on at the initialization
130 time. This allows ``CPUIdle`` governors to be 130 time. This allows ``CPUIdle`` governors to be independent of the underlying
131 hardware and to work with any platforms that t 131 hardware and to work with any platforms that the Linux kernel can run on.
132 132
133 Each idle state present in that array is chara 133 Each idle state present in that array is characterized by two parameters to be
134 taken into account by the governor, the *targe 134 taken into account by the governor, the *target residency* and the (worst-case)
135 *exit latency*. The target residency is the m 135 *exit latency*. The target residency is the minimum time the hardware must
136 spend in the given state, including the time n 136 spend in the given state, including the time needed to enter it (which may be
137 substantial), in order to save more energy tha 137 substantial), in order to save more energy than it would save by entering one of
138 the shallower idle states instead. [The "dept 138 the shallower idle states instead. [The "depth" of an idle state roughly
139 corresponds to the power drawn by the processo 139 corresponds to the power drawn by the processor in that state.] The exit
140 latency, in turn, is the maximum time it will 140 latency, in turn, is the maximum time it will take a CPU asking the processor
141 hardware to enter an idle state to start execu 141 hardware to enter an idle state to start executing the first instruction after a
142 wakeup from that state. Note that in general 142 wakeup from that state. Note that in general the exit latency also must cover
143 the time needed to enter the given state in ca 143 the time needed to enter the given state in case the wakeup occurs when the
144 hardware is entering it and it must be entered 144 hardware is entering it and it must be entered completely to be exited in an
145 ordered manner. 145 ordered manner.
146 146
147 There are two types of information that can in 147 There are two types of information that can influence the governor's decisions.
148 First of all, the governor knows the time unti 148 First of all, the governor knows the time until the closest timer event. That
149 time is known exactly, because the kernel prog 149 time is known exactly, because the kernel programs timers and it knows exactly
150 when they will trigger, and it is the maximum 150 when they will trigger, and it is the maximum time the hardware that the given
151 CPU depends on can spend in an idle state, inc 151 CPU depends on can spend in an idle state, including the time necessary to enter
152 and exit it. However, the CPU may be woken up 152 and exit it. However, the CPU may be woken up by a non-timer event at any time
153 (in particular, before the closest timer trigg 153 (in particular, before the closest timer triggers) and it generally is not known
154 when that may happen. The governor can only s 154 when that may happen. The governor can only see how much time the CPU actually
155 was idle after it has been woken up (that time 155 was idle after it has been woken up (that time will be referred to as the *idle
156 duration* from now on) and it can use that inf 156 duration* from now on) and it can use that information somehow along with the
157 time until the closest timer to estimate the i 157 time until the closest timer to estimate the idle duration in future. How the
158 governor uses that information depends on what 158 governor uses that information depends on what algorithm is implemented by it
159 and that is the primary reason for having more 159 and that is the primary reason for having more than one governor in the
160 ``CPUIdle`` subsystem. 160 ``CPUIdle`` subsystem.
161 161
162 There are four ``CPUIdle`` governors available 162 There are four ``CPUIdle`` governors available, ``menu``, `TEO <teo-gov_>`_,
163 ``ladder`` and ``haltpoll``. Which of them is 163 ``ladder`` and ``haltpoll``. Which of them is used by default depends on the
164 configuration of the kernel and in particular 164 configuration of the kernel and in particular on whether or not the scheduler
165 tick can be `stopped by the idle loop <idle-cp 165 tick can be `stopped by the idle loop <idle-cpus-and-tick_>`_. Available
166 governors can be read from the :file:`availabl 166 governors can be read from the :file:`available_governors`, and the governor
167 can be changed at runtime. The name of the `` 167 can be changed at runtime. The name of the ``CPUIdle`` governor currently
168 used by the kernel can be read from the :file: 168 used by the kernel can be read from the :file:`current_governor_ro` or
169 :file:`current_governor` file under :file:`/sy 169 :file:`current_governor` file under :file:`/sys/devices/system/cpu/cpuidle/`
170 in ``sysfs``. 170 in ``sysfs``.
171 171
172 Which ``CPUIdle`` driver is used, on the other 172 Which ``CPUIdle`` driver is used, on the other hand, usually depends on the
173 platform the kernel is running on, but there a 173 platform the kernel is running on, but there are platforms with more than one
174 matching driver. For example, there are two d 174 matching driver. For example, there are two drivers that can work with the
175 majority of Intel platforms, ``intel_idle`` an 175 majority of Intel platforms, ``intel_idle`` and ``acpi_idle``, one with
176 hardcoded idle states information and the othe 176 hardcoded idle states information and the other able to read that information
177 from the system's ACPI tables, respectively. 177 from the system's ACPI tables, respectively. Still, even in those cases, the
178 driver chosen at the system initialization tim 178 driver chosen at the system initialization time cannot be replaced later, so the
179 decision on which one of them to use has to be 179 decision on which one of them to use has to be made early (on Intel platforms
180 the ``acpi_idle`` driver will be used if ``int 180 the ``acpi_idle`` driver will be used if ``intel_idle`` is disabled for some
181 reason or if it does not recognize the process 181 reason or if it does not recognize the processor). The name of the ``CPUIdle``
182 driver currently used by the kernel can be rea 182 driver currently used by the kernel can be read from the :file:`current_driver`
183 file under :file:`/sys/devices/system/cpu/cpui 183 file under :file:`/sys/devices/system/cpu/cpuidle/` in ``sysfs``.
184 184
185 185
186 .. _idle-cpus-and-tick: 186 .. _idle-cpus-and-tick:
187 187
188 Idle CPUs and The Scheduler Tick 188 Idle CPUs and The Scheduler Tick
189 ================================ 189 ================================
190 190
191 The scheduler tick is a timer that triggers pe 191 The scheduler tick is a timer that triggers periodically in order to implement
192 the time sharing strategy of the CPU scheduler 192 the time sharing strategy of the CPU scheduler. Of course, if there are
193 multiple runnable tasks assigned to one CPU at 193 multiple runnable tasks assigned to one CPU at the same time, the only way to
194 allow them to make reasonable progress in a gi 194 allow them to make reasonable progress in a given time frame is to make them
195 share the available CPU time. Namely, in roug 195 share the available CPU time. Namely, in rough approximation, each task is
196 given a slice of the CPU time to run its code, 196 given a slice of the CPU time to run its code, subject to the scheduling class,
197 prioritization and so on and when that time sl 197 prioritization and so on and when that time slice is used up, the CPU should be
198 switched over to running (the code of) another 198 switched over to running (the code of) another task. The currently running task
199 may not want to give the CPU away voluntarily, 199 may not want to give the CPU away voluntarily, however, and the scheduler tick
200 is there to make the switch happen regardless. 200 is there to make the switch happen regardless. That is not the only role of the
201 tick, but it is the primary reason for using i 201 tick, but it is the primary reason for using it.
202 202
203 The scheduler tick is problematic from the CPU 203 The scheduler tick is problematic from the CPU idle time management perspective,
204 because it triggers periodically and relativel 204 because it triggers periodically and relatively often (depending on the kernel
205 configuration, the length of the tick period i 205 configuration, the length of the tick period is between 1 ms and 10 ms).
206 Thus, if the tick is allowed to trigger on idl 206 Thus, if the tick is allowed to trigger on idle CPUs, it will not make sense
207 for them to ask the hardware to enter idle sta 207 for them to ask the hardware to enter idle states with target residencies above
208 the tick period length. Moreover, in that cas 208 the tick period length. Moreover, in that case the idle duration of any CPU
209 will never exceed the tick period length and t 209 will never exceed the tick period length and the energy used for entering and
210 exiting idle states due to the tick wakeups on 210 exiting idle states due to the tick wakeups on idle CPUs will be wasted.
211 211
212 Fortunately, it is not really necessary to all 212 Fortunately, it is not really necessary to allow the tick to trigger on idle
213 CPUs, because (by definition) they have no tas 213 CPUs, because (by definition) they have no tasks to run except for the special
214 "idle" one. In other words, from the CPU sche 214 "idle" one. In other words, from the CPU scheduler perspective, the only user
215 of the CPU time on them is the idle loop. Sin 215 of the CPU time on them is the idle loop. Since the time of an idle CPU need
216 not be shared between multiple runnable tasks, 216 not be shared between multiple runnable tasks, the primary reason for using the
217 tick goes away if the given CPU is idle. Cons 217 tick goes away if the given CPU is idle. Consequently, it is possible to stop
218 the scheduler tick entirely on idle CPUs in pr 218 the scheduler tick entirely on idle CPUs in principle, even though that may not
219 always be worth the effort. 219 always be worth the effort.
220 220
221 Whether or not it makes sense to stop the sche 221 Whether or not it makes sense to stop the scheduler tick in the idle loop
222 depends on what is expected by the governor. 222 depends on what is expected by the governor. First, if there is another
223 (non-tick) timer due to trigger within the tic 223 (non-tick) timer due to trigger within the tick range, stopping the tick clearly
224 would be a waste of time, even though the time 224 would be a waste of time, even though the timer hardware may not need to be
225 reprogrammed in that case. Second, if the gov 225 reprogrammed in that case. Second, if the governor is expecting a non-timer
226 wakeup within the tick range, stopping the tic 226 wakeup within the tick range, stopping the tick is not necessary and it may even
227 be harmful. Namely, in that case the governor 227 be harmful. Namely, in that case the governor will select an idle state with
228 the target residency within the time until the 228 the target residency within the time until the expected wakeup, so that state is
229 going to be relatively shallow. The governor 229 going to be relatively shallow. The governor really cannot select a deep idle
230 state then, as that would contradict its own e 230 state then, as that would contradict its own expectation of a wakeup in short
231 order. Now, if the wakeup really occurs short 231 order. Now, if the wakeup really occurs shortly, stopping the tick would be a
232 waste of time and in this case the timer hardw 232 waste of time and in this case the timer hardware would need to be reprogrammed,
233 which is expensive. On the other hand, if the 233 which is expensive. On the other hand, if the tick is stopped and the wakeup
234 does not occur any time soon, the hardware may 234 does not occur any time soon, the hardware may spend indefinite amount of time
235 in the shallow idle state selected by the gove 235 in the shallow idle state selected by the governor, which will be a waste of
236 energy. Hence, if the governor is expecting a 236 energy. Hence, if the governor is expecting a wakeup of any kind within the
237 tick range, it is better to allow the tick tri 237 tick range, it is better to allow the tick trigger. Otherwise, however, the
238 governor will select a relatively deep idle st 238 governor will select a relatively deep idle state, so the tick should be stopped
239 so that it does not wake up the CPU too early. 239 so that it does not wake up the CPU too early.
240 240
241 In any case, the governor knows what it is exp 241 In any case, the governor knows what it is expecting and the decision on whether
242 or not to stop the scheduler tick belongs to i 242 or not to stop the scheduler tick belongs to it. Still, if the tick has been
243 stopped already (in one of the previous iterat 243 stopped already (in one of the previous iterations of the loop), it is better
244 to leave it as is and the governor needs to ta 244 to leave it as is and the governor needs to take that into account.
245 245
246 The kernel can be configured to disable stoppi 246 The kernel can be configured to disable stopping the scheduler tick in the idle
247 loop altogether. That can be done through the 247 loop altogether. That can be done through the build-time configuration of it
248 (by unsetting the ``CONFIG_NO_HZ_IDLE`` config 248 (by unsetting the ``CONFIG_NO_HZ_IDLE`` configuration option) or by passing
249 ``nohz=off`` to it in the command line. In bo 249 ``nohz=off`` to it in the command line. In both cases, as the stopping of the
250 scheduler tick is disabled, the governor's dec 250 scheduler tick is disabled, the governor's decisions regarding it are simply
251 ignored by the idle loop code and the tick is 251 ignored by the idle loop code and the tick is never stopped.
252 252
253 The systems that run kernels configured to all 253 The systems that run kernels configured to allow the scheduler tick to be
254 stopped on idle CPUs are referred to as *tickl 254 stopped on idle CPUs are referred to as *tickless* systems and they are
255 generally regarded as more energy-efficient th 255 generally regarded as more energy-efficient than the systems running kernels in
256 which the tick cannot be stopped. If the give 256 which the tick cannot be stopped. If the given system is tickless, it will use
257 the ``menu`` governor by default and if it is 257 the ``menu`` governor by default and if it is not tickless, the default
258 ``CPUIdle`` governor on it will be ``ladder``. 258 ``CPUIdle`` governor on it will be ``ladder``.
259 259
260 260
261 .. _menu-gov: 261 .. _menu-gov:
262 262
263 The ``menu`` Governor 263 The ``menu`` Governor
264 ===================== 264 =====================
265 265
266 The ``menu`` governor is the default ``CPUIdle 266 The ``menu`` governor is the default ``CPUIdle`` governor for tickless systems.
267 It is quite complex, but the basic principle o 267 It is quite complex, but the basic principle of its design is straightforward.
268 Namely, when invoked to select an idle state f 268 Namely, when invoked to select an idle state for a CPU (i.e. an idle state that
269 the CPU will ask the processor hardware to ent 269 the CPU will ask the processor hardware to enter), it attempts to predict the
270 idle duration and uses the predicted value for 270 idle duration and uses the predicted value for idle state selection.
271 271
272 It first obtains the time until the closest ti 272 It first obtains the time until the closest timer event with the assumption
273 that the scheduler tick will be stopped. That 273 that the scheduler tick will be stopped. That time, referred to as the *sleep
274 length* in what follows, is the upper bound on 274 length* in what follows, is the upper bound on the time before the next CPU
275 wakeup. It is used to determine the sleep len 275 wakeup. It is used to determine the sleep length range, which in turn is needed
276 to get the sleep length correction factor. 276 to get the sleep length correction factor.
277 277
278 The ``menu`` governor maintains two arrays of 278 The ``menu`` governor maintains two arrays of sleep length correction factors.
279 One of them is used when tasks previously runn 279 One of them is used when tasks previously running on the given CPU are waiting
280 for some I/O operations to complete and the ot 280 for some I/O operations to complete and the other one is used when that is not
281 the case. Each array contains several correct 281 the case. Each array contains several correction factor values that correspond
282 to different sleep length ranges organized so 282 to different sleep length ranges organized so that each range represented in the
283 array is approximately 10 times wider than the 283 array is approximately 10 times wider than the previous one.
284 284
285 The correction factor for the given sleep leng 285 The correction factor for the given sleep length range (determined before
286 selecting the idle state for the CPU) is updat 286 selecting the idle state for the CPU) is updated after the CPU has been woken
287 up and the closer the sleep length is to the o 287 up and the closer the sleep length is to the observed idle duration, the closer
288 to 1 the correction factor becomes (it must fa 288 to 1 the correction factor becomes (it must fall between 0 and 1 inclusive).
289 The sleep length is multiplied by the correcti 289 The sleep length is multiplied by the correction factor for the range that it
290 falls into to obtain the first approximation o 290 falls into to obtain the first approximation of the predicted idle duration.
291 291
292 Next, the governor uses a simple pattern recog 292 Next, the governor uses a simple pattern recognition algorithm to refine its
293 idle duration prediction. Namely, it saves th 293 idle duration prediction. Namely, it saves the last 8 observed idle duration
294 values and, when predicting the idle duration 294 values and, when predicting the idle duration next time, it computes the average
295 and variance of them. If the variance is smal 295 and variance of them. If the variance is small (smaller than 400 square
296 milliseconds) or it is small relative to the a 296 milliseconds) or it is small relative to the average (the average is greater
297 that 6 times the standard deviation), the aver 297 that 6 times the standard deviation), the average is regarded as the "typical
298 interval" value. Otherwise, the longest of th 298 interval" value. Otherwise, the longest of the saved observed idle duration
299 values is discarded and the computation is rep 299 values is discarded and the computation is repeated for the remaining ones.
300 Again, if the variance of them is small (in th 300 Again, if the variance of them is small (in the above sense), the average is
301 taken as the "typical interval" value and so o 301 taken as the "typical interval" value and so on, until either the "typical
302 interval" is determined or too many data point 302 interval" is determined or too many data points are disregarded, in which case
303 the "typical interval" is assumed to equal "in 303 the "typical interval" is assumed to equal "infinity" (the maximum unsigned
304 integer value). The "typical interval" comput 304 integer value). The "typical interval" computed this way is compared with the
305 sleep length multiplied by the correction fact 305 sleep length multiplied by the correction factor and the minimum of the two is
306 taken as the predicted idle duration. 306 taken as the predicted idle duration.
307 307
308 Then, the governor computes an extra latency l 308 Then, the governor computes an extra latency limit to help "interactive"
309 workloads. It uses the observation that if th 309 workloads. It uses the observation that if the exit latency of the selected
310 idle state is comparable with the predicted id 310 idle state is comparable with the predicted idle duration, the total time spent
311 in that state probably will be very short and 311 in that state probably will be very short and the amount of energy to save by
312 entering it will be relatively small, so likel 312 entering it will be relatively small, so likely it is better to avoid the
313 overhead related to entering that state and ex 313 overhead related to entering that state and exiting it. Thus selecting a
314 shallower state is likely to be a better optio 314 shallower state is likely to be a better option then. The first approximation
315 of the extra latency limit is the predicted id 315 of the extra latency limit is the predicted idle duration itself which
316 additionally is divided by a value depending o 316 additionally is divided by a value depending on the number of tasks that
317 previously ran on the given CPU and now they a 317 previously ran on the given CPU and now they are waiting for I/O operations to
318 complete. The result of that division is comp 318 complete. The result of that division is compared with the latency limit coming
319 from the power management quality of service, 319 from the power management quality of service, or `PM QoS <cpu-pm-qos_>`_,
320 framework and the minimum of the two is taken 320 framework and the minimum of the two is taken as the limit for the idle states'
321 exit latency. 321 exit latency.
322 322
323 Now, the governor is ready to walk the list of 323 Now, the governor is ready to walk the list of idle states and choose one of
324 them. For this purpose, it compares the targe 324 them. For this purpose, it compares the target residency of each state with
325 the predicted idle duration and the exit laten 325 the predicted idle duration and the exit latency of it with the computed latency
326 limit. It selects the state with the target r 326 limit. It selects the state with the target residency closest to the predicted
327 idle duration, but still below it, and exit la 327 idle duration, but still below it, and exit latency that does not exceed the
328 limit. 328 limit.
329 329
330 In the final step the governor may still need 330 In the final step the governor may still need to refine the idle state selection
331 if it has not decided to `stop the scheduler t 331 if it has not decided to `stop the scheduler tick <idle-cpus-and-tick_>`_. That
332 happens if the idle duration predicted by it i 332 happens if the idle duration predicted by it is less than the tick period and
333 the tick has not been stopped already (in a pr 333 the tick has not been stopped already (in a previous iteration of the idle
334 loop). Then, the sleep length used in the pre 334 loop). Then, the sleep length used in the previous computations may not reflect
335 the real time until the closest timer event an 335 the real time until the closest timer event and if it really is greater than
336 that time, the governor may need to select a s 336 that time, the governor may need to select a shallower state with a suitable
337 target residency. 337 target residency.
338 338
339 339
340 .. _teo-gov: 340 .. _teo-gov:
341 341
342 The Timer Events Oriented (TEO) Governor 342 The Timer Events Oriented (TEO) Governor
343 ======================================== 343 ========================================
344 344
345 The timer events oriented (TEO) governor is an 345 The timer events oriented (TEO) governor is an alternative ``CPUIdle`` governor
346 for tickless systems. It follows the same bas 346 for tickless systems. It follows the same basic strategy as the ``menu`` `one
347 <menu-gov_>`_: it always tries to find the dee 347 <menu-gov_>`_: it always tries to find the deepest idle state suitable for the
348 given conditions. However, it applies a diffe 348 given conditions. However, it applies a different approach to that problem.
349 349
350 .. kernel-doc:: drivers/cpuidle/governors/teo. !! 350 First, it does not use sleep length correction factors, but instead it attempts
351 :doc: teo-description !! 351 to correlate the observed idle duration values with the available idle states
>> 352 and use that information to pick up the idle state that is most likely to
>> 353 "match" the upcoming CPU idle interval. Second, it does not take the tasks
>> 354 that were running on the given CPU in the past and are waiting on some I/O
>> 355 operations to complete now at all (there is no guarantee that they will run on
>> 356 the same CPU when they become runnable again) and the pattern detection code in
>> 357 it avoids taking timer wakeups into account. It also only uses idle duration
>> 358 values less than the current time till the closest timer (with the scheduler
>> 359 tick excluded) for that purpose.
>> 360
>> 361 Like in the ``menu`` governor `case <menu-gov_>`_, the first step is to obtain
>> 362 the *sleep length*, which is the time until the closest timer event with the
>> 363 assumption that the scheduler tick will be stopped (that also is the upper bound
>> 364 on the time until the next CPU wakeup). That value is then used to preselect an
>> 365 idle state on the basis of three metrics maintained for each idle state provided
>> 366 by the ``CPUIdle`` driver: ``hits``, ``misses`` and ``early_hits``.
>> 367
>> 368 The ``hits`` and ``misses`` metrics measure the likelihood that a given idle
>> 369 state will "match" the observed (post-wakeup) idle duration if it "matches" the
>> 370 sleep length. They both are subject to decay (after a CPU wakeup) every time
>> 371 the target residency of the idle state corresponding to them is less than or
>> 372 equal to the sleep length and the target residency of the next idle state is
>> 373 greater than the sleep length (that is, when the idle state corresponding to
>> 374 them "matches" the sleep length). The ``hits`` metric is increased if the
>> 375 former condition is satisfied and the target residency of the given idle state
>> 376 is less than or equal to the observed idle duration and the target residency of
>> 377 the next idle state is greater than the observed idle duration at the same time
>> 378 (that is, it is increased when the given idle state "matches" both the sleep
>> 379 length and the observed idle duration). In turn, the ``misses`` metric is
>> 380 increased when the given idle state "matches" the sleep length only and the
>> 381 observed idle duration is too short for its target residency.
>> 382
>> 383 The ``early_hits`` metric measures the likelihood that a given idle state will
>> 384 "match" the observed (post-wakeup) idle duration if it does not "match" the
>> 385 sleep length. It is subject to decay on every CPU wakeup and it is increased
>> 386 when the idle state corresponding to it "matches" the observed (post-wakeup)
>> 387 idle duration and the target residency of the next idle state is less than or
>> 388 equal to the sleep length (i.e. the idle state "matching" the sleep length is
>> 389 deeper than the given one).
>> 390
>> 391 The governor walks the list of idle states provided by the ``CPUIdle`` driver
>> 392 and finds the last (deepest) one with the target residency less than or equal
>> 393 to the sleep length. Then, the ``hits`` and ``misses`` metrics of that idle
>> 394 state are compared with each other and it is preselected if the ``hits`` one is
>> 395 greater (which means that that idle state is likely to "match" the observed idle
>> 396 duration after CPU wakeup). If the ``misses`` one is greater, the governor
>> 397 preselects the shallower idle state with the maximum ``early_hits`` metric
>> 398 (or if there are multiple shallower idle states with equal ``early_hits``
>> 399 metric which also is the maximum, the shallowest of them will be preselected).
>> 400 [If there is a wakeup latency constraint coming from the `PM QoS framework
>> 401 <cpu-pm-qos_>`_ which is hit before reaching the deepest idle state with the
>> 402 target residency within the sleep length, the deepest idle state with the exit
>> 403 latency within the constraint is preselected without consulting the ``hits``,
>> 404 ``misses`` and ``early_hits`` metrics.]
>> 405
>> 406 Next, the governor takes several idle duration values observed most recently
>> 407 into consideration and if at least a half of them are greater than or equal to
>> 408 the target residency of the preselected idle state, that idle state becomes the
>> 409 final candidate to ask for. Otherwise, the average of the most recent idle
>> 410 duration values below the target residency of the preselected idle state is
>> 411 computed and the governor walks the idle states shallower than the preselected
>> 412 one and finds the deepest of them with the target residency within that average.
>> 413 That idle state is then taken as the final candidate to ask for.
>> 414
>> 415 Still, at this point the governor may need to refine the idle state selection if
>> 416 it has not decided to `stop the scheduler tick <idle-cpus-and-tick_>`_. That
>> 417 generally happens if the target residency of the idle state selected so far is
>> 418 less than the tick period and the tick has not been stopped already (in a
>> 419 previous iteration of the idle loop). Then, like in the ``menu`` governor
>> 420 `case <menu-gov_>`_, the sleep length used in the previous computations may not
>> 421 reflect the real time until the closest timer event and if it really is greater
>> 422 than that time, a shallower state with a suitable target residency may need to
>> 423 be selected.
>> 424
352 425
353 .. _idle-states-representation: 426 .. _idle-states-representation:
354 427
355 Representation of Idle States 428 Representation of Idle States
356 ============================= 429 =============================
357 430
358 For the CPU idle time management purposes all 431 For the CPU idle time management purposes all of the physical idle states
359 supported by the processor have to be represen 432 supported by the processor have to be represented as a one-dimensional array of
360 |struct cpuidle_state| objects each allowing a 433 |struct cpuidle_state| objects each allowing an individual (logical) CPU to ask
361 the processor hardware to enter an idle state 434 the processor hardware to enter an idle state of certain properties. If there
362 is a hierarchy of units in the processor, one 435 is a hierarchy of units in the processor, one |struct cpuidle_state| object can
363 cover a combination of idle states supported b 436 cover a combination of idle states supported by the units at different levels of
364 the hierarchy. In that case, the `target resi 437 the hierarchy. In that case, the `target residency and exit latency parameters
365 of it <idle-loop_>`_, must reflect the propert 438 of it <idle-loop_>`_, must reflect the properties of the idle state at the
366 deepest level (i.e. the idle state of the unit 439 deepest level (i.e. the idle state of the unit containing all of the other
367 units). 440 units).
368 441
369 For example, take a processor with two cores i 442 For example, take a processor with two cores in a larger unit referred to as
370 a "module" and suppose that asking the hardwar 443 a "module" and suppose that asking the hardware to enter a specific idle state
371 (say "X") at the "core" level by one core will 444 (say "X") at the "core" level by one core will trigger the module to try to
372 enter a specific idle state of its own (say "M 445 enter a specific idle state of its own (say "MX") if the other core is in idle
373 state "X" already. In other words, asking for 446 state "X" already. In other words, asking for idle state "X" at the "core"
374 level gives the hardware a license to go as de 447 level gives the hardware a license to go as deep as to idle state "MX" at the
375 "module" level, but there is no guarantee that 448 "module" level, but there is no guarantee that this is going to happen (the core
376 asking for idle state "X" may just end up in t 449 asking for idle state "X" may just end up in that state by itself instead).
377 Then, the target residency of the |struct cpui 450 Then, the target residency of the |struct cpuidle_state| object representing
378 idle state "X" must reflect the minimum time t 451 idle state "X" must reflect the minimum time to spend in idle state "MX" of
379 the module (including the time needed to enter 452 the module (including the time needed to enter it), because that is the minimum
380 time the CPU needs to be idle to save any ener 453 time the CPU needs to be idle to save any energy in case the hardware enters
381 that state. Analogously, the exit latency par 454 that state. Analogously, the exit latency parameter of that object must cover
382 the exit time of idle state "MX" of the module 455 the exit time of idle state "MX" of the module (and usually its entry time too),
383 because that is the maximum delay between a wa 456 because that is the maximum delay between a wakeup signal and the time the CPU
384 will start to execute the first new instructio 457 will start to execute the first new instruction (assuming that both cores in the
385 module will always be ready to execute instruc 458 module will always be ready to execute instructions as soon as the module
386 becomes operational as a whole). 459 becomes operational as a whole).
387 460
388 There are processors without direct coordinati 461 There are processors without direct coordination between different levels of the
389 hierarchy of units inside them, however. In t 462 hierarchy of units inside them, however. In those cases asking for an idle
390 state at the "core" level does not automatical 463 state at the "core" level does not automatically affect the "module" level, for
391 example, in any way and the ``CPUIdle`` driver 464 example, in any way and the ``CPUIdle`` driver is responsible for the entire
392 handling of the hierarchy. Then, the definiti 465 handling of the hierarchy. Then, the definition of the idle state objects is
393 entirely up to the driver, but still the physi 466 entirely up to the driver, but still the physical properties of the idle state
394 that the processor hardware finally goes into 467 that the processor hardware finally goes into must always follow the parameters
395 used by the governor for idle state selection 468 used by the governor for idle state selection (for instance, the actual exit
396 latency of that idle state must not exceed the 469 latency of that idle state must not exceed the exit latency parameter of the
397 idle state object selected by the governor). 470 idle state object selected by the governor).
398 471
399 In addition to the target residency and exit l 472 In addition to the target residency and exit latency idle state parameters
400 discussed above, the objects representing idle 473 discussed above, the objects representing idle states each contain a few other
401 parameters describing the idle state and a poi 474 parameters describing the idle state and a pointer to the function to run in
402 order to ask the hardware to enter that state. 475 order to ask the hardware to enter that state. Also, for each
403 |struct cpuidle_state| object, there is a corr 476 |struct cpuidle_state| object, there is a corresponding
404 :c:type:`struct cpuidle_state_usage <cpuidle_s 477 :c:type:`struct cpuidle_state_usage <cpuidle_state_usage>` one containing usage
405 statistics of the given idle state. That info 478 statistics of the given idle state. That information is exposed by the kernel
406 via ``sysfs``. 479 via ``sysfs``.
407 480
408 For each CPU in the system, there is a :file:` 481 For each CPU in the system, there is a :file:`/sys/devices/system/cpu/cpu<N>/cpuidle/`
409 directory in ``sysfs``, where the number ``<N> 482 directory in ``sysfs``, where the number ``<N>`` is assigned to the given
410 CPU at the initialization time. That director 483 CPU at the initialization time. That directory contains a set of subdirectories
411 called :file:`state0`, :file:`state1` and so o 484 called :file:`state0`, :file:`state1` and so on, up to the number of idle state
412 objects defined for the given CPU minus one. 485 objects defined for the given CPU minus one. Each of these directories
413 corresponds to one idle state object and the l 486 corresponds to one idle state object and the larger the number in its name, the
414 deeper the (effective) idle state represented 487 deeper the (effective) idle state represented by it. Each of them contains
415 a number of files (attributes) representing th 488 a number of files (attributes) representing the properties of the idle state
416 object corresponding to it, as follows: 489 object corresponding to it, as follows:
417 490
418 ``above`` 491 ``above``
419 Total number of times this idle state 492 Total number of times this idle state had been asked for, but the
420 observed idle duration was certainly t 493 observed idle duration was certainly too short to match its target
421 residency. 494 residency.
422 495
423 ``below`` 496 ``below``
424 Total number of times this idle state 497 Total number of times this idle state had been asked for, but certainly
425 a deeper idle state would have been a 498 a deeper idle state would have been a better match for the observed idle
426 duration. 499 duration.
427 500
428 ``desc`` 501 ``desc``
429 Description of the idle state. 502 Description of the idle state.
430 503
431 ``disable`` 504 ``disable``
432 Whether or not this idle state is disa 505 Whether or not this idle state is disabled.
433 506
434 ``default_status`` 507 ``default_status``
435 The default status of this state, "ena 508 The default status of this state, "enabled" or "disabled".
436 509
437 ``latency`` 510 ``latency``
438 Exit latency of the idle state in micr 511 Exit latency of the idle state in microseconds.
439 512
440 ``name`` 513 ``name``
441 Name of the idle state. 514 Name of the idle state.
442 515
443 ``power`` 516 ``power``
444 Power drawn by hardware in this idle s 517 Power drawn by hardware in this idle state in milliwatts (if specified,
445 0 otherwise). 518 0 otherwise).
446 519
447 ``residency`` 520 ``residency``
448 Target residency of the idle state in 521 Target residency of the idle state in microseconds.
449 522
450 ``time`` 523 ``time``
451 Total time spent in this idle state by 524 Total time spent in this idle state by the given CPU (as measured by the
452 kernel) in microseconds. 525 kernel) in microseconds.
453 526
454 ``usage`` 527 ``usage``
455 Total number of times the hardware has 528 Total number of times the hardware has been asked by the given CPU to
456 enter this idle state. 529 enter this idle state.
457 530
458 ``rejected`` 531 ``rejected``
459 Total number of times a request to ent 532 Total number of times a request to enter this idle state on the given
460 CPU was rejected. 533 CPU was rejected.
461 534
462 The :file:`desc` and :file:`name` files both c 535 The :file:`desc` and :file:`name` files both contain strings. The difference
463 between them is that the name is expected to b 536 between them is that the name is expected to be more concise, while the
464 description may be longer and it may contain w 537 description may be longer and it may contain white space or special characters.
465 The other files listed above contain integer n 538 The other files listed above contain integer numbers.
466 539
467 The :file:`disable` attribute is the only writ 540 The :file:`disable` attribute is the only writeable one. If it contains 1, the
468 given idle state is disabled for this particul 541 given idle state is disabled for this particular CPU, which means that the
469 governor will never select it for this particu 542 governor will never select it for this particular CPU and the ``CPUIdle``
470 driver will never ask the hardware to enter it 543 driver will never ask the hardware to enter it for that CPU as a result.
471 However, disabling an idle state for one CPU d 544 However, disabling an idle state for one CPU does not prevent it from being
472 asked for by the other CPUs, so it must be dis 545 asked for by the other CPUs, so it must be disabled for all of them in order to
473 never be asked for by any of them. [Note that 546 never be asked for by any of them. [Note that, due to the way the ``ladder``
474 governor is implemented, disabling an idle sta 547 governor is implemented, disabling an idle state prevents that governor from
475 selecting any idle states deeper than the disa 548 selecting any idle states deeper than the disabled one too.]
476 549
477 If the :file:`disable` attribute contains 0, t 550 If the :file:`disable` attribute contains 0, the given idle state is enabled for
478 this particular CPU, but it still may be disab 551 this particular CPU, but it still may be disabled for some or all of the other
479 CPUs in the system at the same time. Writing 552 CPUs in the system at the same time. Writing 1 to it causes the idle state to
480 be disabled for this particular CPU and writin 553 be disabled for this particular CPU and writing 0 to it allows the governor to
481 take it into consideration for the given CPU a 554 take it into consideration for the given CPU and the driver to ask for it,
482 unless that state was disabled globally in the 555 unless that state was disabled globally in the driver (in which case it cannot
483 be used at all). 556 be used at all).
484 557
485 The :file:`power` attribute is not defined ver 558 The :file:`power` attribute is not defined very well, especially for idle state
486 objects representing combinations of idle stat 559 objects representing combinations of idle states at different levels of the
487 hierarchy of units in the processor, and it ge 560 hierarchy of units in the processor, and it generally is hard to obtain idle
488 state power numbers for complex hardware, so : 561 state power numbers for complex hardware, so :file:`power` often contains 0 (not
489 available) and if it contains a nonzero number 562 available) and if it contains a nonzero number, that number may not be very
490 accurate and it should not be relied on for an 563 accurate and it should not be relied on for anything meaningful.
491 564
492 The number in the :file:`time` file generally 565 The number in the :file:`time` file generally may be greater than the total time
493 really spent by the given CPU in the given idl 566 really spent by the given CPU in the given idle state, because it is measured by
494 the kernel and it may not cover the cases in w 567 the kernel and it may not cover the cases in which the hardware refused to enter
495 this idle state and entered a shallower one in 568 this idle state and entered a shallower one instead of it (or even it did not
496 enter any idle state at all). The kernel can 569 enter any idle state at all). The kernel can only measure the time span between
497 asking the hardware to enter an idle state and 570 asking the hardware to enter an idle state and the subsequent wakeup of the CPU
498 and it cannot say what really happened in the 571 and it cannot say what really happened in the meantime at the hardware level.
499 Moreover, if the idle state object in question 572 Moreover, if the idle state object in question represents a combination of idle
500 states at different levels of the hierarchy of 573 states at different levels of the hierarchy of units in the processor,
501 the kernel can never say how deep the hardware 574 the kernel can never say how deep the hardware went down the hierarchy in any
502 particular case. For these reasons, the only 575 particular case. For these reasons, the only reliable way to find out how
503 much time has been spent by the hardware in di 576 much time has been spent by the hardware in different idle states supported by
504 it is to use idle state residency counters in 577 it is to use idle state residency counters in the hardware, if available.
505 578
506 Generally, an interrupt received when trying t 579 Generally, an interrupt received when trying to enter an idle state causes the
507 idle state entry request to be rejected, in wh 580 idle state entry request to be rejected, in which case the ``CPUIdle`` driver
508 may return an error code to indicate that this 581 may return an error code to indicate that this was the case. The :file:`usage`
509 and :file:`rejected` files report the number o 582 and :file:`rejected` files report the number of times the given idle state
510 was entered successfully or rejected, respecti 583 was entered successfully or rejected, respectively.
511 584
512 .. _cpu-pm-qos: 585 .. _cpu-pm-qos:
513 586
514 Power Management Quality of Service for CPUs 587 Power Management Quality of Service for CPUs
515 ============================================ 588 ============================================
516 589
517 The power management quality of service (PM Qo 590 The power management quality of service (PM QoS) framework in the Linux kernel
518 allows kernel code and user space processes to 591 allows kernel code and user space processes to set constraints on various
519 energy-efficiency features of the kernel to pr 592 energy-efficiency features of the kernel to prevent performance from dropping
520 below a required level. 593 below a required level.
521 594
522 CPU idle time management can be affected by PM 595 CPU idle time management can be affected by PM QoS in two ways, through the
523 global CPU latency limit and through the resum 596 global CPU latency limit and through the resume latency constraints for
524 individual CPUs. Kernel code (e.g. device dri 597 individual CPUs. Kernel code (e.g. device drivers) can set both of them with
525 the help of special internal interfaces provid 598 the help of special internal interfaces provided by the PM QoS framework. User
526 space can modify the former by opening the :fi 599 space can modify the former by opening the :file:`cpu_dma_latency` special
527 device file under :file:`/dev/` and writing a 600 device file under :file:`/dev/` and writing a binary value (interpreted as a
528 signed 32-bit integer) to it. In turn, the re 601 signed 32-bit integer) to it. In turn, the resume latency constraint for a CPU
529 can be modified from user space by writing a s 602 can be modified from user space by writing a string (representing a signed
530 32-bit integer) to the :file:`power/pm_qos_res 603 32-bit integer) to the :file:`power/pm_qos_resume_latency_us` file under
531 :file:`/sys/devices/system/cpu/cpu<N>/` in ``s 604 :file:`/sys/devices/system/cpu/cpu<N>/` in ``sysfs``, where the CPU number
532 ``<N>`` is allocated at the system initializat 605 ``<N>`` is allocated at the system initialization time. Negative values
533 will be rejected in both cases and, also in bo 606 will be rejected in both cases and, also in both cases, the written integer
534 number will be interpreted as a requested PM Q 607 number will be interpreted as a requested PM QoS constraint in microseconds.
535 608
536 The requested value is not automatically appli 609 The requested value is not automatically applied as a new constraint, however,
537 as it may be less restrictive (greater in this 610 as it may be less restrictive (greater in this particular case) than another
538 constraint previously requested by someone els 611 constraint previously requested by someone else. For this reason, the PM QoS
539 framework maintains a list of requests that ha 612 framework maintains a list of requests that have been made so far for the
540 global CPU latency limit and for each individu 613 global CPU latency limit and for each individual CPU, aggregates them and
541 applies the effective (minimum in this particu 614 applies the effective (minimum in this particular case) value as the new
542 constraint. 615 constraint.
543 616
544 In fact, opening the :file:`cpu_dma_latency` s 617 In fact, opening the :file:`cpu_dma_latency` special device file causes a new
545 PM QoS request to be created and added to a gl 618 PM QoS request to be created and added to a global priority list of CPU latency
546 limit requests and the file descriptor coming 619 limit requests and the file descriptor coming from the "open" operation
547 represents that request. If that file descrip 620 represents that request. If that file descriptor is then used for writing, the
548 number written to it will be associated with t 621 number written to it will be associated with the PM QoS request represented by
549 it as a new requested limit value. Next, the 622 it as a new requested limit value. Next, the priority list mechanism will be
550 used to determine the new effective value of t 623 used to determine the new effective value of the entire list of requests and
551 that effective value will be set as a new CPU 624 that effective value will be set as a new CPU latency limit. Thus requesting a
552 new limit value will only change the real limi 625 new limit value will only change the real limit if the effective "list" value is
553 affected by it, which is the case if it is the 626 affected by it, which is the case if it is the minimum of the requested values
554 in the list. 627 in the list.
555 628
556 The process holding a file descriptor obtained 629 The process holding a file descriptor obtained by opening the
557 :file:`cpu_dma_latency` special device file co 630 :file:`cpu_dma_latency` special device file controls the PM QoS request
558 associated with that file descriptor, but it c 631 associated with that file descriptor, but it controls this particular PM QoS
559 request only. 632 request only.
560 633
561 Closing the :file:`cpu_dma_latency` special de 634 Closing the :file:`cpu_dma_latency` special device file or, more precisely, the
562 file descriptor obtained while opening it, cau 635 file descriptor obtained while opening it, causes the PM QoS request associated
563 with that file descriptor to be removed from t 636 with that file descriptor to be removed from the global priority list of CPU
564 latency limit requests and destroyed. If that 637 latency limit requests and destroyed. If that happens, the priority list
565 mechanism will be used again, to determine the 638 mechanism will be used again, to determine the new effective value for the whole
566 list and that value will become the new limit. 639 list and that value will become the new limit.
567 640
568 In turn, for each CPU there is one resume late 641 In turn, for each CPU there is one resume latency PM QoS request associated with
569 the :file:`power/pm_qos_resume_latency_us` fil 642 the :file:`power/pm_qos_resume_latency_us` file under
570 :file:`/sys/devices/system/cpu/cpu<N>/` in ``s 643 :file:`/sys/devices/system/cpu/cpu<N>/` in ``sysfs`` and writing to it causes
571 this single PM QoS request to be updated regar 644 this single PM QoS request to be updated regardless of which user space
572 process does that. In other words, this PM Qo 645 process does that. In other words, this PM QoS request is shared by the entire
573 user space, so access to the file associated w 646 user space, so access to the file associated with it needs to be arbitrated
574 to avoid confusion. [Arguably, the only legit 647 to avoid confusion. [Arguably, the only legitimate use of this mechanism in
575 practice is to pin a process to the CPU in que 648 practice is to pin a process to the CPU in question and let it use the
576 ``sysfs`` interface to control the resume late 649 ``sysfs`` interface to control the resume latency constraint for it.] It is
577 still only a request, however. It is an entry 650 still only a request, however. It is an entry in a priority list used to
578 determine the effective value to be set as the 651 determine the effective value to be set as the resume latency constraint for the
579 CPU in question every time the list of request 652 CPU in question every time the list of requests is updated this way or another
580 (there may be other requests coming from kerne 653 (there may be other requests coming from kernel code in that list).
581 654
582 CPU idle time governors are expected to regard 655 CPU idle time governors are expected to regard the minimum of the global
583 (effective) CPU latency limit and the effectiv 656 (effective) CPU latency limit and the effective resume latency constraint for
584 the given CPU as the upper limit for the exit 657 the given CPU as the upper limit for the exit latency of the idle states that
585 they are allowed to select for that CPU. They 658 they are allowed to select for that CPU. They should never select any idle
586 states with exit latency beyond that limit. 659 states with exit latency beyond that limit.
587 660
588 661
589 Idle States Control Via Kernel Command Line 662 Idle States Control Via Kernel Command Line
590 =========================================== 663 ===========================================
591 664
592 In addition to the ``sysfs`` interface allowin 665 In addition to the ``sysfs`` interface allowing individual idle states to be
593 `disabled for individual CPUs <idle-states-rep 666 `disabled for individual CPUs <idle-states-representation_>`_, there are kernel
594 command line parameters affecting CPU idle tim 667 command line parameters affecting CPU idle time management.
595 668
596 The ``cpuidle.off=1`` kernel command line opti 669 The ``cpuidle.off=1`` kernel command line option can be used to disable the
597 CPU idle time management entirely. It does no 670 CPU idle time management entirely. It does not prevent the idle loop from
598 running on idle CPUs, but it prevents the CPU 671 running on idle CPUs, but it prevents the CPU idle time governors and drivers
599 from being invoked. If it is added to the ker 672 from being invoked. If it is added to the kernel command line, the idle loop
600 will ask the hardware to enter idle states on 673 will ask the hardware to enter idle states on idle CPUs via the CPU architecture
601 support code that is expected to provide a def 674 support code that is expected to provide a default mechanism for this purpose.
602 That default mechanism usually is the least co 675 That default mechanism usually is the least common denominator for all of the
603 processors implementing the architecture (i.e. 676 processors implementing the architecture (i.e. CPU instruction set) in question,
604 however, so it is rather crude and not very en 677 however, so it is rather crude and not very energy-efficient. For this reason,
605 it is not recommended for production use. 678 it is not recommended for production use.
606 679
607 The ``cpuidle.governor=`` kernel command line 680 The ``cpuidle.governor=`` kernel command line switch allows the ``CPUIdle``
608 governor to use to be specified. It has to be 681 governor to use to be specified. It has to be appended with a string matching
609 the name of an available governor (e.g. ``cpui 682 the name of an available governor (e.g. ``cpuidle.governor=menu``) and that
610 governor will be used instead of the default o 683 governor will be used instead of the default one. It is possible to force
611 the ``menu`` governor to be used on the system 684 the ``menu`` governor to be used on the systems that use the ``ladder`` governor
612 by default this way, for example. 685 by default this way, for example.
613 686
614 The other kernel command line parameters contr 687 The other kernel command line parameters controlling CPU idle time management
615 described below are only relevant for the *x86 !! 688 described below are only relevant for the *x86* architecture and some of
616 to ``intel_idle`` affect Intel processors only !! 689 them affect Intel processors only.
617 690
618 The *x86* architecture support code recognizes 691 The *x86* architecture support code recognizes three kernel command line
619 options related to CPU idle time management: ` 692 options related to CPU idle time management: ``idle=poll``, ``idle=halt``,
620 and ``idle=nomwait``. The first two of them d 693 and ``idle=nomwait``. The first two of them disable the ``acpi_idle`` and
621 ``intel_idle`` drivers altogether, which effec 694 ``intel_idle`` drivers altogether, which effectively causes the entire
622 ``CPUIdle`` subsystem to be disabled and makes 695 ``CPUIdle`` subsystem to be disabled and makes the idle loop invoke the
623 architecture support code to deal with idle CP 696 architecture support code to deal with idle CPUs. How it does that depends on
624 which of the two parameters is added to the ke 697 which of the two parameters is added to the kernel command line. In the
625 ``idle=halt`` case, the architecture support c 698 ``idle=halt`` case, the architecture support code will use the ``HLT``
626 instruction of the CPUs (which, as a rule, sus 699 instruction of the CPUs (which, as a rule, suspends the execution of the program
627 and causes the hardware to attempt to enter th 700 and causes the hardware to attempt to enter the shallowest available idle state)
628 for this purpose, and if ``idle=poll`` is used 701 for this purpose, and if ``idle=poll`` is used, idle CPUs will execute a
629 more or less "lightweight" sequence of instruc 702 more or less "lightweight" sequence of instructions in a tight loop. [Note
630 that using ``idle=poll`` is somewhat drastic i 703 that using ``idle=poll`` is somewhat drastic in many cases, as preventing idle
631 CPUs from saving almost any energy at all may 704 CPUs from saving almost any energy at all may not be the only effect of it.
632 For example, on Intel hardware it effectively 705 For example, on Intel hardware it effectively prevents CPUs from using
633 P-states (see |cpufreq|) that require any numb 706 P-states (see |cpufreq|) that require any number of CPUs in a package to be
634 idle, so it very well may hurt single-thread c 707 idle, so it very well may hurt single-thread computations performance as well as
635 energy-efficiency. Thus using it for performa 708 energy-efficiency. Thus using it for performance reasons may not be a good idea
636 at all.] 709 at all.]
637 710
638 The ``idle=nomwait`` option prevents the use o !! 711 The ``idle=nomwait`` option disables the ``intel_idle`` driver and causes
639 the CPU to enter idle states. When this option !! 712 ``acpi_idle`` to be used (as long as all of the information needed by it is
640 driver will use the ``HLT`` instruction instea !! 713 there in the system's ACPI tables), but it is not allowed to use the
641 running Intel processors, this option disables !! 714 ``MWAIT`` instruction of the CPUs to ask the hardware to enter idle states.
642 and forces the use of the ``acpi_idle`` driver <<
643 case, ``acpi_idle`` driver will function only <<
644 by it is in the system's ACPI tables. <<
645 715
646 In addition to the architecture-level kernel c 716 In addition to the architecture-level kernel command line options affecting CPU
647 idle time management, there are parameters aff 717 idle time management, there are parameters affecting individual ``CPUIdle``
648 drivers that can be passed to them via the ker 718 drivers that can be passed to them via the kernel command line. Specifically,
649 the ``intel_idle.max_cstate=<n>`` and ``proces 719 the ``intel_idle.max_cstate=<n>`` and ``processor.max_cstate=<n>`` parameters,
650 where ``<n>`` is an idle state index also used 720 where ``<n>`` is an idle state index also used in the name of the given
651 state's directory in ``sysfs`` (see 721 state's directory in ``sysfs`` (see
652 `Representation of Idle States <idle-states-re 722 `Representation of Idle States <idle-states-representation_>`_), causes the
653 ``intel_idle`` and ``acpi_idle`` drivers, resp 723 ``intel_idle`` and ``acpi_idle`` drivers, respectively, to discard all of the
654 idle states deeper than idle state ``<n>``. I 724 idle states deeper than idle state ``<n>``. In that case, they will never ask
655 for any of those idle states or expose them to 725 for any of those idle states or expose them to the governor. [The behavior of
656 the two drivers is different for ``<n>`` equal 726 the two drivers is different for ``<n>`` equal to ``0``. Adding
657 ``intel_idle.max_cstate=0`` to the kernel comm 727 ``intel_idle.max_cstate=0`` to the kernel command line disables the
658 ``intel_idle`` driver and allows ``acpi_idle`` 728 ``intel_idle`` driver and allows ``acpi_idle`` to be used, whereas
659 ``processor.max_cstate=0`` is equivalent to `` 729 ``processor.max_cstate=0`` is equivalent to ``processor.max_cstate=1``.
660 Also, the ``acpi_idle`` driver is part of the 730 Also, the ``acpi_idle`` driver is part of the ``processor`` kernel module that
661 can be loaded separately and ``max_cstate=<n>` 731 can be loaded separately and ``max_cstate=<n>`` can be passed to it as a module
662 parameter when it is loaded.] 732 parameter when it is loaded.]
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.