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