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