1# SPDX-License-Identifier: GPL-2.0-only 2# 3# RCU-related configuration options 4# 5 6menu "RCU Subsystem" 7 8config TREE_RCU 9 bool 10 default y if SMP 11 # Dynticks-idle tracking 12 select CONTEXT_TRACKING_IDLE 13 help 14 This option selects the RCU implementation that is 15 designed for very large SMP system with hundreds or 16 thousands of CPUs. It also scales down nicely to 17 smaller systems. 18 19config PREEMPT_RCU 20 bool 21 default y if (PREEMPT || PREEMPT_RT || PREEMPT_DYNAMIC) 22 select TREE_RCU 23 help 24 This option selects the RCU implementation that is 25 designed for very large SMP systems with hundreds or 26 thousands of CPUs, but for which real-time response 27 is also required. It also scales down nicely to 28 smaller systems. 29 30 Select this option if you are unsure. 31 32config TINY_RCU 33 bool 34 default y if !PREEMPT_RCU && !SMP 35 help 36 This option selects the RCU implementation that is 37 designed for UP systems from which real-time response 38 is not required. This option greatly reduces the 39 memory footprint of RCU. 40 41config RCU_EXPERT 42 bool "Make expert-level adjustments to RCU configuration" 43 default n 44 help 45 This option needs to be enabled if you wish to make 46 expert-level adjustments to RCU configuration. By default, 47 no such adjustments can be made, which has the often-beneficial 48 side-effect of preventing "make oldconfig" from asking you all 49 sorts of detailed questions about how you would like numerous 50 obscure RCU options to be set up. 51 52 Say Y if you need to make expert-level adjustments to RCU. 53 54 Say N if you are unsure. 55 56config TINY_SRCU 57 bool 58 default y if TINY_RCU 59 help 60 This option selects the single-CPU non-preemptible version of SRCU. 61 62config TREE_SRCU 63 bool 64 default y if !TINY_RCU 65 help 66 This option selects the full-fledged version of SRCU. 67 68config FORCE_NEED_SRCU_NMI_SAFE 69 bool "Force selection of NEED_SRCU_NMI_SAFE" 70 depends on !TINY_SRCU 71 depends on RCU_EXPERT 72 depends on ARCH_HAS_NMI_SAFE_THIS_CPU_OPS 73 select NEED_SRCU_NMI_SAFE 74 default n 75 help 76 This option forces selection of the NEED_SRCU_NMI_SAFE 77 Kconfig option, allowing testing of srcu_read_lock_nmisafe() 78 and srcu_read_unlock_nmisafe() on architectures (like x86) 79 that select the ARCH_HAS_NMI_SAFE_THIS_CPU_OPS Kconfig option. 80 81config NEED_SRCU_NMI_SAFE 82 def_bool HAVE_NMI && !ARCH_HAS_NMI_SAFE_THIS_CPU_OPS && !TINY_SRCU 83 84config TASKS_RCU_GENERIC 85 def_bool TASKS_RCU || TASKS_RUDE_RCU || TASKS_TRACE_RCU 86 help 87 This option enables generic infrastructure code supporting 88 task-based RCU implementations. Not for manual selection. 89 90config FORCE_TASKS_RCU 91 bool "Force selection of TASKS_RCU" 92 depends on RCU_EXPERT 93 select TASKS_RCU 94 default n 95 help 96 This option force-enables a task-based RCU implementation 97 that uses only voluntary context switch (not preemption!), 98 idle, and user-mode execution as quiescent states. Not for 99 manual selection in most cases. 100 101config NEED_TASKS_RCU 102 bool 103 default n 104 105config TASKS_RCU 106 bool 107 default NEED_TASKS_RCU && PREEMPTION 108 select IRQ_WORK 109 110config FORCE_TASKS_RUDE_RCU 111 bool "Force selection of Tasks Rude RCU" 112 depends on RCU_EXPERT 113 select TASKS_RUDE_RCU 114 default n 115 help 116 This option force-enables a task-based RCU implementation 117 that uses only context switch (including preemption) and 118 user-mode execution as quiescent states. It forces IPIs and 119 context switches on all online CPUs, including idle ones, 120 so use with caution. Not for manual selection in most cases. 121 122config TASKS_RUDE_RCU 123 bool 124 default n 125 select IRQ_WORK 126 127config FORCE_TASKS_TRACE_RCU 128 bool "Force selection of Tasks Trace RCU" 129 depends on RCU_EXPERT 130 select TASKS_TRACE_RCU 131 default n 132 help 133 This option enables a task-based RCU implementation that uses 134 explicit rcu_read_lock_trace() read-side markers, and allows 135 these readers to appear in the idle loop as well as on the 136 CPU hotplug code paths. It can force IPIs on online CPUs, 137 including idle ones, so use with caution. Not for manual 138 selection in most cases. 139 140config TASKS_TRACE_RCU 141 bool 142 default n 143 select IRQ_WORK 144 145config RCU_STALL_COMMON 146 def_bool TREE_RCU 147 help 148 This option enables RCU CPU stall code that is common between 149 the TINY and TREE variants of RCU. The purpose is to allow 150 the tiny variants to disable RCU CPU stall warnings, while 151 making these warnings mandatory for the tree variants. 152 153config RCU_NEED_SEGCBLIST 154 def_bool ( TREE_RCU || TREE_SRCU || TASKS_RCU_GENERIC ) 155 156config RCU_FANOUT 157 int "Tree-based hierarchical RCU fanout value" 158 range 2 64 if 64BIT 159 range 2 32 if !64BIT 160 depends on TREE_RCU && RCU_EXPERT 161 default 64 if 64BIT 162 default 32 if !64BIT 163 help 164 This option controls the fanout of hierarchical implementations 165 of RCU, allowing RCU to work efficiently on machines with 166 large numbers of CPUs. This value must be at least the fourth 167 root of NR_CPUS, which allows NR_CPUS to be insanely large. 168 The default value of RCU_FANOUT should be used for production 169 systems, but if you are stress-testing the RCU implementation 170 itself, small RCU_FANOUT values allow you to test large-system 171 code paths on small(er) systems. 172 173 Select a specific number if testing RCU itself. 174 Take the default if unsure. 175 176config RCU_FANOUT_LEAF 177 int "Tree-based hierarchical RCU leaf-level fanout value" 178 range 2 64 if 64BIT && !RCU_STRICT_GRACE_PERIOD 179 range 2 32 if !64BIT && !RCU_STRICT_GRACE_PERIOD 180 range 2 3 if RCU_STRICT_GRACE_PERIOD 181 depends on TREE_RCU && RCU_EXPERT 182 default 16 if !RCU_STRICT_GRACE_PERIOD 183 default 2 if RCU_STRICT_GRACE_PERIOD 184 help 185 This option controls the leaf-level fanout of hierarchical 186 implementations of RCU, and allows trading off cache misses 187 against lock contention. Systems that synchronize their 188 scheduling-clock interrupts for energy-efficiency reasons will 189 want the default because the smaller leaf-level fanout keeps 190 lock contention levels acceptably low. Very large systems 191 (hundreds or thousands of CPUs) will instead want to set this 192 value to the maximum value possible in order to reduce the 193 number of cache misses incurred during RCU's grace-period 194 initialization. These systems tend to run CPU-bound, and thus 195 are not helped by synchronized interrupts, and thus tend to 196 skew them, which reduces lock contention enough that large 197 leaf-level fanouts work well. That said, setting leaf-level 198 fanout to a large number will likely cause problematic 199 lock contention on the leaf-level rcu_node structures unless 200 you boot with the skew_tick kernel parameter. 201 202 Select a specific number if testing RCU itself. 203 204 Select the maximum permissible value for large systems, but 205 please understand that you may also need to set the skew_tick 206 kernel boot parameter to avoid contention on the rcu_node 207 structure's locks. 208 209 Take the default if unsure. 210 211config RCU_BOOST 212 bool "Enable RCU priority boosting" 213 depends on (RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT) || PREEMPT_RT 214 default y if PREEMPT_RT 215 help 216 This option boosts the priority of preempted RCU readers that 217 block the current preemptible RCU grace period for too long. 218 This option also prevents heavy loads from blocking RCU 219 callback invocation. 220 221 Say Y here if you are working with real-time apps or heavy loads 222 Say N here if you are unsure. 223 224config RCU_BOOST_DELAY 225 int "Milliseconds to delay boosting after RCU grace-period start" 226 range 0 3000 227 depends on RCU_BOOST 228 default 500 229 help 230 This option specifies the time to wait after the beginning of 231 a given grace period before priority-boosting preempted RCU 232 readers blocking that grace period. Note that any RCU reader 233 blocking an expedited RCU grace period is boosted immediately. 234 235 Accept the default if unsure. 236 237config RCU_EXP_KTHREAD 238 bool "Perform RCU expedited work in a real-time kthread" 239 depends on RCU_BOOST && RCU_EXPERT 240 default !PREEMPT_RT && NR_CPUS <= 32 241 help 242 Use this option to further reduce the latencies of expedited 243 grace periods at the expense of being more disruptive. 244 245 This option is disabled by default on PREEMPT_RT=y kernels which 246 disable expedited grace periods after boot by unconditionally 247 setting rcupdate.rcu_normal_after_boot=1. 248 249 Accept the default if unsure. 250 251config RCU_NOCB_CPU 252 bool "Offload RCU callback processing from boot-selected CPUs" 253 depends on TREE_RCU 254 depends on RCU_EXPERT || NO_HZ_FULL 255 default n 256 help 257 Use this option to reduce OS jitter for aggressive HPC or 258 real-time workloads. It can also be used to offload RCU 259 callback invocation to energy-efficient CPUs in battery-powered 260 asymmetric multiprocessors. The price of this reduced jitter 261 is that the overhead of call_rcu() increases and that some 262 workloads will incur significant increases in context-switch 263 rates. 264 265 This option offloads callback invocation from the set of 266 CPUs specified at boot time by the rcu_nocbs parameter. 267 For each such CPU, a kthread ("rcuox/N") will be created to 268 invoke callbacks, where the "N" is the CPU being offloaded, 269 and where the "x" is "p" for RCU-preempt (PREEMPTION kernels) 270 and "s" for RCU-sched (!PREEMPTION kernels). This option 271 also creates another kthread for each sqrt(nr_cpu_ids) CPUs 272 ("rcuog/N", where N is the first CPU in that group to come 273 online), which handles grace periods for its group. Nothing 274 prevents these kthreads from running on the specified CPUs, 275 but (1) the kthreads may be preempted between each callback, 276 and (2) affinity or cgroups can be used to force the kthreads 277 to run on whatever set of CPUs is desired. 278 279 The sqrt(nr_cpu_ids) grouping may be overridden using the 280 rcutree.rcu_nocb_gp_stride kernel boot parameter. This can 281 be especially helpful for smaller numbers of CPUs, where 282 sqrt(nr_cpu_ids) can be a bit of a blunt instrument. 283 284 Say Y here if you need reduced OS jitter, despite added overhead. 285 Say N here if you are unsure. 286 287config RCU_NOCB_CPU_DEFAULT_ALL 288 bool "Offload RCU callback processing from all CPUs by default" 289 depends on RCU_NOCB_CPU 290 default n 291 help 292 Use this option to offload callback processing from all CPUs 293 by default, in the absence of the rcu_nocbs or nohz_full boot 294 parameter. This also avoids the need to use any boot parameters 295 to achieve the effect of offloading all CPUs on boot. 296 297 Say Y here if you want offload all CPUs by default on boot. 298 Say N here if you are unsure. 299 300config RCU_NOCB_CPU_CB_BOOST 301 bool "Offload RCU callback from real-time kthread" 302 depends on RCU_NOCB_CPU && RCU_BOOST 303 default y if PREEMPT_RT 304 help 305 Use this option to invoke offloaded callbacks as SCHED_FIFO 306 to avoid starvation by heavy SCHED_OTHER background load. 307 Of course, running as SCHED_FIFO during callback floods will 308 cause the rcuo[ps] kthreads to monopolize the CPU for hundreds 309 of milliseconds or more. Therefore, when enabling this option, 310 it is your responsibility to ensure that latency-sensitive 311 tasks either run with higher priority or run on some other CPU. 312 313 Say Y here if you want to set RT priority for offloading kthreads. 314 Say N here if you are building a !PREEMPT_RT kernel and are unsure. 315 316config TASKS_TRACE_RCU_READ_MB 317 bool "Tasks Trace RCU readers use memory barriers in user and idle" 318 depends on RCU_EXPERT && TASKS_TRACE_RCU 319 default PREEMPT_RT || NR_CPUS < 8 320 help 321 Use this option to further reduce the number of IPIs sent 322 to CPUs executing in userspace or idle during tasks trace 323 RCU grace periods. Given that a reasonable setting of 324 the rcupdate.rcu_task_ipi_delay kernel boot parameter 325 eliminates such IPIs for many workloads, proper setting 326 of this Kconfig option is important mostly for aggressive 327 real-time installations and for battery-powered devices, 328 hence the default chosen above. 329 330 Say Y here if you hate IPIs. 331 Say N here if you hate read-side memory barriers. 332 Take the default if you are unsure. 333 334config RCU_LAZY 335 bool "RCU callback lazy invocation functionality" 336 depends on RCU_NOCB_CPU 337 default n 338 help 339 To save power, batch RCU callbacks and delay starting the 340 corresponding grace period for multiple seconds. The grace 341 period will be started after this delay, in case of memory 342 pressure, or if the corresponding CPU's callback list grows 343 too large. 344 345 These delays happen only on rcu_nocbs CPUs, that is, CPUs 346 whose callbacks have been offloaded. 347 348 Use the rcutree.enable_rcu_lazy=0 kernel-boot parameter to 349 globally disable these delays. 350 351config RCU_LAZY_DEFAULT_OFF 352 bool "Turn RCU lazy invocation off by default" 353 depends on RCU_LAZY 354 default n 355 help 356 Build the kernel with CONFIG_RCU_LAZY=y, but cause the kernel 357 to boot with these energy-efficiency delays disabled. Use the 358 rcutree.enable_rcu_lazy=0 kernel-boot parameter to override 359 the this option at boot time, thus re-enabling these delays. 360 361config RCU_DOUBLE_CHECK_CB_TIME 362 bool "RCU callback-batch backup time check" 363 depends on RCU_EXPERT 364 default n 365 help 366 Use this option to provide more precise enforcement of the 367 rcutree.rcu_resched_ns module parameter in situations where 368 a single RCU callback might run for hundreds of microseconds, 369 thus defeating the 32-callback batching used to amortize the 370 cost of the fine-grained but expensive local_clock() function. 371 372 This option rounds rcutree.rcu_resched_ns up to the next 373 jiffy, and overrides the 32-callback batching if this limit 374 is exceeded. 375 376 Say Y here if you need tighter callback-limit enforcement. 377 Say N here if you are unsure. 378 379endmenu # "RCU Subsystem" 380