1 # SPDX-License-Identifier: GPL-2.0-only 1 # SPDX-License-Identifier: GPL-2.0-only
2 # 2 #
3 # RCU-related configuration options 3 # RCU-related configuration options
4 # 4 #
5 5
6 menu "RCU Subsystem" 6 menu "RCU Subsystem"
7 7
8 config TREE_RCU 8 config TREE_RCU
9 bool 9 bool
10 default y if SMP 10 default y if SMP
11 # Dynticks-idle tracking 11 # Dynticks-idle tracking
12 select CONTEXT_TRACKING_IDLE 12 select CONTEXT_TRACKING_IDLE
13 help 13 help
14 This option selects the RCU implemen 14 This option selects the RCU implementation that is
15 designed for very large SMP system w 15 designed for very large SMP system with hundreds or
16 thousands of CPUs. It also scales d 16 thousands of CPUs. It also scales down nicely to
17 smaller systems. 17 smaller systems.
18 18
19 config PREEMPT_RCU 19 config PREEMPT_RCU
20 bool 20 bool
21 default y if PREEMPTION 21 default y if PREEMPTION
22 select TREE_RCU 22 select TREE_RCU
23 help 23 help
24 This option selects the RCU implemen 24 This option selects the RCU implementation that is
25 designed for very large SMP systems 25 designed for very large SMP systems with hundreds or
26 thousands of CPUs, but for which rea 26 thousands of CPUs, but for which real-time response
27 is also required. It also scales do 27 is also required. It also scales down nicely to
28 smaller systems. 28 smaller systems.
29 29
30 Select this option if you are unsure 30 Select this option if you are unsure.
31 31
32 config TINY_RCU 32 config TINY_RCU
33 bool 33 bool
34 default y if !PREEMPT_RCU && !SMP !! 34 default y if !PREEMPTION && !SMP
35 help 35 help
36 This option selects the RCU implemen 36 This option selects the RCU implementation that is
37 designed for UP systems from which r 37 designed for UP systems from which real-time response
38 is not required. This option greatl 38 is not required. This option greatly reduces the
39 memory footprint of RCU. 39 memory footprint of RCU.
40 40
41 config RCU_EXPERT 41 config RCU_EXPERT
42 bool "Make expert-level adjustments to 42 bool "Make expert-level adjustments to RCU configuration"
43 default n 43 default n
44 help 44 help
45 This option needs to be enabled if y 45 This option needs to be enabled if you wish to make
46 expert-level adjustments to RCU conf 46 expert-level adjustments to RCU configuration. By default,
47 no such adjustments can be made, whi 47 no such adjustments can be made, which has the often-beneficial
48 side-effect of preventing "make oldc 48 side-effect of preventing "make oldconfig" from asking you all
49 sorts of detailed questions about ho 49 sorts of detailed questions about how you would like numerous
50 obscure RCU options to be set up. 50 obscure RCU options to be set up.
51 51
52 Say Y if you need to make expert-lev 52 Say Y if you need to make expert-level adjustments to RCU.
53 53
54 Say N if you are unsure. 54 Say N if you are unsure.
55 55
>> 56 config SRCU
>> 57 def_bool y
>> 58
56 config TINY_SRCU 59 config TINY_SRCU
57 bool 60 bool
58 default y if TINY_RCU 61 default y if TINY_RCU
59 help 62 help
60 This option selects the single-CPU n 63 This option selects the single-CPU non-preemptible version of SRCU.
61 64
62 config TREE_SRCU 65 config TREE_SRCU
63 bool 66 bool
64 default y if !TINY_RCU 67 default y if !TINY_RCU
65 help 68 help
66 This option selects the full-fledged 69 This option selects the full-fledged version of SRCU.
67 70
68 config NEED_SRCU_NMI_SAFE 71 config NEED_SRCU_NMI_SAFE
69 def_bool HAVE_NMI && !ARCH_HAS_NMI_SAF 72 def_bool HAVE_NMI && !ARCH_HAS_NMI_SAFE_THIS_CPU_OPS && !TINY_SRCU
70 73
71 config TASKS_RCU_GENERIC 74 config TASKS_RCU_GENERIC
72 def_bool TASKS_RCU || TASKS_RUDE_RCU | 75 def_bool TASKS_RCU || TASKS_RUDE_RCU || TASKS_TRACE_RCU
73 help 76 help
74 This option enables generic infrastr 77 This option enables generic infrastructure code supporting
75 task-based RCU implementations. Not 78 task-based RCU implementations. Not for manual selection.
76 79
77 config FORCE_TASKS_RCU 80 config FORCE_TASKS_RCU
78 bool "Force selection of TASKS_RCU" 81 bool "Force selection of TASKS_RCU"
79 depends on RCU_EXPERT 82 depends on RCU_EXPERT
80 select TASKS_RCU 83 select TASKS_RCU
81 default n 84 default n
82 help 85 help
83 This option force-enables a task-bas 86 This option force-enables a task-based RCU implementation
84 that uses only voluntary context swi 87 that uses only voluntary context switch (not preemption!),
85 idle, and user-mode execution as qui 88 idle, and user-mode execution as quiescent states. Not for
86 manual selection in most cases. 89 manual selection in most cases.
87 90
88 config NEED_TASKS_RCU <<
89 bool <<
90 default n <<
91 <<
92 config TASKS_RCU 91 config TASKS_RCU
93 bool 92 bool
94 default NEED_TASKS_RCU && (PREEMPTION !! 93 default n
95 select IRQ_WORK 94 select IRQ_WORK
96 95
97 config FORCE_TASKS_RUDE_RCU 96 config FORCE_TASKS_RUDE_RCU
98 bool "Force selection of Tasks Rude RC 97 bool "Force selection of Tasks Rude RCU"
99 depends on RCU_EXPERT 98 depends on RCU_EXPERT
100 select TASKS_RUDE_RCU 99 select TASKS_RUDE_RCU
101 default n 100 default n
102 help 101 help
103 This option force-enables a task-bas 102 This option force-enables a task-based RCU implementation
104 that uses only context switch (inclu 103 that uses only context switch (including preemption) and
105 user-mode execution as quiescent sta 104 user-mode execution as quiescent states. It forces IPIs and
106 context switches on all online CPUs, 105 context switches on all online CPUs, including idle ones,
107 so use with caution. Not for manual 106 so use with caution. Not for manual selection in most cases.
108 107
109 config TASKS_RUDE_RCU 108 config TASKS_RUDE_RCU
110 bool 109 bool
111 default n 110 default n
112 select IRQ_WORK 111 select IRQ_WORK
113 112
114 config FORCE_TASKS_TRACE_RCU 113 config FORCE_TASKS_TRACE_RCU
115 bool "Force selection of Tasks Trace R 114 bool "Force selection of Tasks Trace RCU"
116 depends on RCU_EXPERT 115 depends on RCU_EXPERT
117 select TASKS_TRACE_RCU 116 select TASKS_TRACE_RCU
118 default n 117 default n
119 help 118 help
120 This option enables a task-based RCU 119 This option enables a task-based RCU implementation that uses
121 explicit rcu_read_lock_trace() read- 120 explicit rcu_read_lock_trace() read-side markers, and allows
122 these readers to appear in the idle 121 these readers to appear in the idle loop as well as on the
123 CPU hotplug code paths. It can forc 122 CPU hotplug code paths. It can force IPIs on online CPUs,
124 including idle ones, so use with cau 123 including idle ones, so use with caution. Not for manual
125 selection in most cases. 124 selection in most cases.
126 125
127 config TASKS_TRACE_RCU 126 config TASKS_TRACE_RCU
128 bool 127 bool
129 default n 128 default n
130 select IRQ_WORK 129 select IRQ_WORK
131 130
132 config RCU_STALL_COMMON 131 config RCU_STALL_COMMON
133 def_bool TREE_RCU 132 def_bool TREE_RCU
134 help 133 help
135 This option enables RCU CPU stall co 134 This option enables RCU CPU stall code that is common between
136 the TINY and TREE variants of RCU. 135 the TINY and TREE variants of RCU. The purpose is to allow
137 the tiny variants to disable RCU CPU 136 the tiny variants to disable RCU CPU stall warnings, while
138 making these warnings mandatory for 137 making these warnings mandatory for the tree variants.
139 138
140 config RCU_NEED_SEGCBLIST 139 config RCU_NEED_SEGCBLIST
141 def_bool ( TREE_RCU || TREE_SRCU || TA 140 def_bool ( TREE_RCU || TREE_SRCU || TASKS_RCU_GENERIC )
142 141
143 config RCU_FANOUT 142 config RCU_FANOUT
144 int "Tree-based hierarchical RCU fanou 143 int "Tree-based hierarchical RCU fanout value"
145 range 2 64 if 64BIT 144 range 2 64 if 64BIT
146 range 2 32 if !64BIT 145 range 2 32 if !64BIT
147 depends on TREE_RCU && RCU_EXPERT 146 depends on TREE_RCU && RCU_EXPERT
148 default 64 if 64BIT 147 default 64 if 64BIT
149 default 32 if !64BIT 148 default 32 if !64BIT
150 help 149 help
151 This option controls the fanout of h 150 This option controls the fanout of hierarchical implementations
152 of RCU, allowing RCU to work efficie 151 of RCU, allowing RCU to work efficiently on machines with
153 large numbers of CPUs. This value m 152 large numbers of CPUs. This value must be at least the fourth
154 root of NR_CPUS, which allows NR_CPU 153 root of NR_CPUS, which allows NR_CPUS to be insanely large.
155 The default value of RCU_FANOUT shou 154 The default value of RCU_FANOUT should be used for production
156 systems, but if you are stress-testi 155 systems, but if you are stress-testing the RCU implementation
157 itself, small RCU_FANOUT values allo 156 itself, small RCU_FANOUT values allow you to test large-system
158 code paths on small(er) systems. 157 code paths on small(er) systems.
159 158
160 Select a specific number if testing 159 Select a specific number if testing RCU itself.
161 Take the default if unsure. 160 Take the default if unsure.
162 161
163 config RCU_FANOUT_LEAF 162 config RCU_FANOUT_LEAF
164 int "Tree-based hierarchical RCU leaf- 163 int "Tree-based hierarchical RCU leaf-level fanout value"
165 range 2 64 if 64BIT && !RCU_STRICT_GRA 164 range 2 64 if 64BIT && !RCU_STRICT_GRACE_PERIOD
166 range 2 32 if !64BIT && !RCU_STRICT_GR 165 range 2 32 if !64BIT && !RCU_STRICT_GRACE_PERIOD
167 range 2 3 if RCU_STRICT_GRACE_PERIOD 166 range 2 3 if RCU_STRICT_GRACE_PERIOD
168 depends on TREE_RCU && RCU_EXPERT 167 depends on TREE_RCU && RCU_EXPERT
169 default 16 if !RCU_STRICT_GRACE_PERIOD 168 default 16 if !RCU_STRICT_GRACE_PERIOD
170 default 2 if RCU_STRICT_GRACE_PERIOD 169 default 2 if RCU_STRICT_GRACE_PERIOD
171 help 170 help
172 This option controls the leaf-level 171 This option controls the leaf-level fanout of hierarchical
173 implementations of RCU, and allows t 172 implementations of RCU, and allows trading off cache misses
174 against lock contention. Systems th 173 against lock contention. Systems that synchronize their
175 scheduling-clock interrupts for ener 174 scheduling-clock interrupts for energy-efficiency reasons will
176 want the default because the smaller 175 want the default because the smaller leaf-level fanout keeps
177 lock contention levels acceptably lo 176 lock contention levels acceptably low. Very large systems
178 (hundreds or thousands of CPUs) will 177 (hundreds or thousands of CPUs) will instead want to set this
179 value to the maximum value possible 178 value to the maximum value possible in order to reduce the
180 number of cache misses incurred duri 179 number of cache misses incurred during RCU's grace-period
181 initialization. These systems tend 180 initialization. These systems tend to run CPU-bound, and thus
182 are not helped by synchronized inter 181 are not helped by synchronized interrupts, and thus tend to
183 skew them, which reduces lock conten 182 skew them, which reduces lock contention enough that large
184 leaf-level fanouts work well. That 183 leaf-level fanouts work well. That said, setting leaf-level
185 fanout to a large number will likely 184 fanout to a large number will likely cause problematic
186 lock contention on the leaf-level rc 185 lock contention on the leaf-level rcu_node structures unless
187 you boot with the skew_tick kernel p 186 you boot with the skew_tick kernel parameter.
188 187
189 Select a specific number if testing 188 Select a specific number if testing RCU itself.
190 189
191 Select the maximum permissible value 190 Select the maximum permissible value for large systems, but
192 please understand that you may also 191 please understand that you may also need to set the skew_tick
193 kernel boot parameter to avoid conte 192 kernel boot parameter to avoid contention on the rcu_node
194 structure's locks. 193 structure's locks.
195 194
196 Take the default if unsure. 195 Take the default if unsure.
197 196
198 config RCU_BOOST 197 config RCU_BOOST
199 bool "Enable RCU priority boosting" 198 bool "Enable RCU priority boosting"
200 depends on (RT_MUTEXES && PREEMPT_RCU 199 depends on (RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT) || PREEMPT_RT
201 default y if PREEMPT_RT 200 default y if PREEMPT_RT
202 help 201 help
203 This option boosts the priority of p 202 This option boosts the priority of preempted RCU readers that
204 block the current preemptible RCU gr 203 block the current preemptible RCU grace period for too long.
205 This option also prevents heavy load 204 This option also prevents heavy loads from blocking RCU
206 callback invocation. 205 callback invocation.
207 206
208 Say Y here if you are working with r 207 Say Y here if you are working with real-time apps or heavy loads
209 Say N here if you are unsure. 208 Say N here if you are unsure.
210 209
211 config RCU_BOOST_DELAY 210 config RCU_BOOST_DELAY
212 int "Milliseconds to delay boosting af 211 int "Milliseconds to delay boosting after RCU grace-period start"
213 range 0 3000 212 range 0 3000
214 depends on RCU_BOOST 213 depends on RCU_BOOST
215 default 500 214 default 500
216 help 215 help
217 This option specifies the time to wa 216 This option specifies the time to wait after the beginning of
218 a given grace period before priority 217 a given grace period before priority-boosting preempted RCU
219 readers blocking that grace period. 218 readers blocking that grace period. Note that any RCU reader
220 blocking an expedited RCU grace peri 219 blocking an expedited RCU grace period is boosted immediately.
221 220
222 Accept the default if unsure. 221 Accept the default if unsure.
223 222
224 config RCU_EXP_KTHREAD 223 config RCU_EXP_KTHREAD
225 bool "Perform RCU expedited work in a 224 bool "Perform RCU expedited work in a real-time kthread"
226 depends on RCU_BOOST && RCU_EXPERT 225 depends on RCU_BOOST && RCU_EXPERT
227 default !PREEMPT_RT && NR_CPUS <= 32 226 default !PREEMPT_RT && NR_CPUS <= 32
228 help 227 help
229 Use this option to further reduce th 228 Use this option to further reduce the latencies of expedited
230 grace periods at the expense of bein 229 grace periods at the expense of being more disruptive.
231 230
232 This option is disabled by default o 231 This option is disabled by default on PREEMPT_RT=y kernels which
233 disable expedited grace periods afte 232 disable expedited grace periods after boot by unconditionally
234 setting rcupdate.rcu_normal_after_bo 233 setting rcupdate.rcu_normal_after_boot=1.
235 234
236 Accept the default if unsure. 235 Accept the default if unsure.
237 236
238 config RCU_NOCB_CPU 237 config RCU_NOCB_CPU
239 bool "Offload RCU callback processing 238 bool "Offload RCU callback processing from boot-selected CPUs"
240 depends on TREE_RCU 239 depends on TREE_RCU
241 depends on RCU_EXPERT || NO_HZ_FULL 240 depends on RCU_EXPERT || NO_HZ_FULL
242 default n 241 default n
243 help 242 help
244 Use this option to reduce OS jitter 243 Use this option to reduce OS jitter for aggressive HPC or
245 real-time workloads. It can also be 244 real-time workloads. It can also be used to offload RCU
246 callback invocation to energy-effici 245 callback invocation to energy-efficient CPUs in battery-powered
247 asymmetric multiprocessors. The pri 246 asymmetric multiprocessors. The price of this reduced jitter
248 is that the overhead of call_rcu() i 247 is that the overhead of call_rcu() increases and that some
249 workloads will incur significant inc 248 workloads will incur significant increases in context-switch
250 rates. 249 rates.
251 250
252 This option offloads callback invoca 251 This option offloads callback invocation from the set of CPUs
253 specified at boot time by the rcu_no 252 specified at boot time by the rcu_nocbs parameter. For each
254 such CPU, a kthread ("rcuox/N") will 253 such CPU, a kthread ("rcuox/N") will be created to invoke
255 callbacks, where the "N" is the CPU 254 callbacks, where the "N" is the CPU being offloaded, and where
256 the "x" is "p" for RCU-preempt (PREE 255 the "x" is "p" for RCU-preempt (PREEMPTION kernels) and "s" for
257 RCU-sched (!PREEMPTION kernels). No 256 RCU-sched (!PREEMPTION kernels). Nothing prevents this kthread
258 from running on the specified CPUs, 257 from running on the specified CPUs, but (1) the kthreads may be
259 preempted between each callback, and 258 preempted between each callback, and (2) affinity or cgroups can
260 be used to force the kthreads to run 259 be used to force the kthreads to run on whatever set of CPUs is
261 desired. 260 desired.
262 261
263 Say Y here if you need reduced OS ji 262 Say Y here if you need reduced OS jitter, despite added overhead.
264 Say N here if you are unsure. 263 Say N here if you are unsure.
265 264
266 config RCU_NOCB_CPU_DEFAULT_ALL 265 config RCU_NOCB_CPU_DEFAULT_ALL
267 bool "Offload RCU callback processing 266 bool "Offload RCU callback processing from all CPUs by default"
268 depends on RCU_NOCB_CPU 267 depends on RCU_NOCB_CPU
269 default n 268 default n
270 help 269 help
271 Use this option to offload callback 270 Use this option to offload callback processing from all CPUs
272 by default, in the absence of the rc 271 by default, in the absence of the rcu_nocbs or nohz_full boot
273 parameter. This also avoids the need 272 parameter. This also avoids the need to use any boot parameters
274 to achieve the effect of offloading 273 to achieve the effect of offloading all CPUs on boot.
275 274
276 Say Y here if you want offload all C 275 Say Y here if you want offload all CPUs by default on boot.
277 Say N here if you are unsure. 276 Say N here if you are unsure.
278 277
279 config RCU_NOCB_CPU_CB_BOOST 278 config RCU_NOCB_CPU_CB_BOOST
280 bool "Offload RCU callback from real-t 279 bool "Offload RCU callback from real-time kthread"
281 depends on RCU_NOCB_CPU && RCU_BOOST 280 depends on RCU_NOCB_CPU && RCU_BOOST
282 default y if PREEMPT_RT 281 default y if PREEMPT_RT
283 help 282 help
284 Use this option to invoke offloaded 283 Use this option to invoke offloaded callbacks as SCHED_FIFO
285 to avoid starvation by heavy SCHED_O 284 to avoid starvation by heavy SCHED_OTHER background load.
286 Of course, running as SCHED_FIFO dur 285 Of course, running as SCHED_FIFO during callback floods will
287 cause the rcuo[ps] kthreads to monop 286 cause the rcuo[ps] kthreads to monopolize the CPU for hundreds
288 of milliseconds or more. Therefore, 287 of milliseconds or more. Therefore, when enabling this option,
289 it is your responsibility to ensure 288 it is your responsibility to ensure that latency-sensitive
290 tasks either run with higher priorit 289 tasks either run with higher priority or run on some other CPU.
291 290
292 Say Y here if you want to set RT pri 291 Say Y here if you want to set RT priority for offloading kthreads.
293 Say N here if you are building a !PR 292 Say N here if you are building a !PREEMPT_RT kernel and are unsure.
294 293
295 config TASKS_TRACE_RCU_READ_MB 294 config TASKS_TRACE_RCU_READ_MB
296 bool "Tasks Trace RCU readers use memo 295 bool "Tasks Trace RCU readers use memory barriers in user and idle"
297 depends on RCU_EXPERT && TASKS_TRACE_R 296 depends on RCU_EXPERT && TASKS_TRACE_RCU
298 default PREEMPT_RT || NR_CPUS < 8 297 default PREEMPT_RT || NR_CPUS < 8
299 help 298 help
300 Use this option to further reduce th 299 Use this option to further reduce the number of IPIs sent
301 to CPUs executing in userspace or id 300 to CPUs executing in userspace or idle during tasks trace
302 RCU grace periods. Given that a rea 301 RCU grace periods. Given that a reasonable setting of
303 the rcupdate.rcu_task_ipi_delay kern 302 the rcupdate.rcu_task_ipi_delay kernel boot parameter
304 eliminates such IPIs for many worklo 303 eliminates such IPIs for many workloads, proper setting
305 of this Kconfig option is important 304 of this Kconfig option is important mostly for aggressive
306 real-time installations and for batt 305 real-time installations and for battery-powered devices,
307 hence the default chosen above. 306 hence the default chosen above.
308 307
309 Say Y here if you hate IPIs. 308 Say Y here if you hate IPIs.
310 Say N here if you hate read-side mem 309 Say N here if you hate read-side memory barriers.
311 Take the default if you are unsure. 310 Take the default if you are unsure.
312 311
313 config RCU_LAZY 312 config RCU_LAZY
314 bool "RCU callback lazy invocation fun 313 bool "RCU callback lazy invocation functionality"
315 depends on RCU_NOCB_CPU 314 depends on RCU_NOCB_CPU
316 default n 315 default n
317 help 316 help
318 To save power, batch RCU callbacks a 317 To save power, batch RCU callbacks and flush after delay, memory
319 pressure, or callback list growing t 318 pressure, or callback list growing too big.
320 <<
321 Requires rcu_nocbs=all to be set. <<
322 <<
323 Use rcutree.enable_rcu_lazy=0 to tur <<
324 <<
325 config RCU_LAZY_DEFAULT_OFF <<
326 bool "Turn RCU lazy invocation off by <<
327 depends on RCU_LAZY <<
328 default n <<
329 help <<
330 Allows building the kernel with CONF <<
331 off. Boot time param rcutree.enable_ <<
332 it back on. <<
333 <<
334 config RCU_DOUBLE_CHECK_CB_TIME <<
335 bool "RCU callback-batch backup time c <<
336 depends on RCU_EXPERT <<
337 default n <<
338 help <<
339 Use this option to provide more prec <<
340 rcutree.rcu_resched_ns module parame <<
341 a single RCU callback might run for <<
342 thus defeating the 32-callback batch <<
343 cost of the fine-grained but expensi <<
344 <<
345 This option rounds rcutree.rcu_resch <<
346 jiffy, and overrides the 32-callback <<
347 is exceeded. <<
348 <<
349 Say Y here if you need tighter callb <<
350 Say N here if you are unsure. <<
351 319
352 endmenu # "RCU Subsystem" 320 endmenu # "RCU Subsystem"
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.