1 # SPDX-License-Identifier: GPL-2.0-only
3 # RCU-related configuration options
12 This option selects the RCU implementation that is
13 designed for very large SMP system with hundreds or
14 thousands of CPUs. It also scales down nicely to
19 default y if PREEMPTION
22 This option selects the RCU implementation that is
23 designed for very large SMP systems with hundreds or
24 thousands of CPUs, but for which real-time response
25 is also required. It also scales down nicely to
28 Select this option if you are unsure.
32 default y if !PREEMPTION && !SMP
34 This option selects the RCU implementation that is
35 designed for UP systems from which real-time response
36 is not required. This option greatly reduces the
37 memory footprint of RCU.
40 bool "Make expert-level adjustments to RCU configuration"
43 This option needs to be enabled if you wish to make
44 expert-level adjustments to RCU configuration. By default,
45 no such adjustments can be made, which has the often-beneficial
46 side-effect of preventing "make oldconfig" from asking you all
47 sorts of detailed questions about how you would like numerous
48 obscure RCU options to be set up.
50 Say Y if you need to make expert-level adjustments to RCU.
52 Say N if you are unsure.
57 This option selects the sleepable version of RCU. This version
58 permits arbitrary sleeping or blocking within RCU read-side critical
63 default y if SRCU && TINY_RCU
65 This option selects the single-CPU non-preemptible version of SRCU.
69 default y if SRCU && !TINY_RCU
71 This option selects the full-fledged version of SRCU.
73 config TASKS_RCU_GENERIC
74 def_bool TASKS_RCU || TASKS_RUDE_RCU || TASKS_TRACE_RCU
77 This option enables generic infrastructure code supporting
78 task-based RCU implementations. Not for manual selection.
83 This option enables a task-based RCU implementation that uses
84 only voluntary context switch (not preemption!), idle, and
85 user-mode execution as quiescent states. Not for manual selection.
90 This option enables a task-based RCU implementation that uses
91 only context switch (including preemption) and user-mode
92 execution as quiescent states. It forces IPIs and context
93 switches on all online CPUs, including idle ones, so use
96 config TASKS_TRACE_RCU
99 This option enables a task-based RCU implementation that uses
100 explicit rcu_read_lock_trace() read-side markers, and allows
101 these readers to appear in the idle loop as well as on the CPU
102 hotplug code paths. It can force IPIs on online CPUs, including
103 idle ones, so use with caution.
105 config RCU_STALL_COMMON
108 This option enables RCU CPU stall code that is common between
109 the TINY and TREE variants of RCU. The purpose is to allow
110 the tiny variants to disable RCU CPU stall warnings, while
111 making these warnings mandatory for the tree variants.
113 config RCU_NEED_SEGCBLIST
114 def_bool ( TREE_RCU || TREE_SRCU )
117 int "Tree-based hierarchical RCU fanout value"
120 depends on TREE_RCU && RCU_EXPERT
124 This option controls the fanout of hierarchical implementations
125 of RCU, allowing RCU to work efficiently on machines with
126 large numbers of CPUs. This value must be at least the fourth
127 root of NR_CPUS, which allows NR_CPUS to be insanely large.
128 The default value of RCU_FANOUT should be used for production
129 systems, but if you are stress-testing the RCU implementation
130 itself, small RCU_FANOUT values allow you to test large-system
131 code paths on small(er) systems.
133 Select a specific number if testing RCU itself.
134 Take the default if unsure.
136 config RCU_FANOUT_LEAF
137 int "Tree-based hierarchical RCU leaf-level fanout value"
138 range 2 64 if 64BIT && !RCU_STRICT_GRACE_PERIOD
139 range 2 32 if !64BIT && !RCU_STRICT_GRACE_PERIOD
140 range 2 3 if RCU_STRICT_GRACE_PERIOD
141 depends on TREE_RCU && RCU_EXPERT
142 default 16 if !RCU_STRICT_GRACE_PERIOD
143 default 2 if RCU_STRICT_GRACE_PERIOD
145 This option controls the leaf-level fanout of hierarchical
146 implementations of RCU, and allows trading off cache misses
147 against lock contention. Systems that synchronize their
148 scheduling-clock interrupts for energy-efficiency reasons will
149 want the default because the smaller leaf-level fanout keeps
150 lock contention levels acceptably low. Very large systems
151 (hundreds or thousands of CPUs) will instead want to set this
152 value to the maximum value possible in order to reduce the
153 number of cache misses incurred during RCU's grace-period
154 initialization. These systems tend to run CPU-bound, and thus
155 are not helped by synchronized interrupts, and thus tend to
156 skew them, which reduces lock contention enough that large
157 leaf-level fanouts work well. That said, setting leaf-level
158 fanout to a large number will likely cause problematic
159 lock contention on the leaf-level rcu_node structures unless
160 you boot with the skew_tick kernel parameter.
162 Select a specific number if testing RCU itself.
164 Select the maximum permissible value for large systems, but
165 please understand that you may also need to set the skew_tick
166 kernel boot parameter to avoid contention on the rcu_node
169 Take the default if unsure.
171 config RCU_FAST_NO_HZ
172 bool "Accelerate last non-dyntick-idle CPU's grace periods"
173 depends on NO_HZ_COMMON && SMP && RCU_EXPERT
176 This option permits CPUs to enter dynticks-idle state even if
177 they have RCU callbacks queued, and prevents RCU from waking
178 these CPUs up more than roughly once every four jiffies (by
179 default, you can adjust this using the rcutree.rcu_idle_gp_delay
180 parameter), thus improving energy efficiency. On the other
181 hand, this option increases the duration of RCU grace periods,
182 for example, slowing down synchronize_rcu().
184 Say Y if energy efficiency is critically important, and you
185 don't care about increased grace-period durations.
187 Say N if you are unsure.
190 bool "Enable RCU priority boosting"
191 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT
194 This option boosts the priority of preempted RCU readers that
195 block the current preemptible RCU grace period for too long.
196 This option also prevents heavy loads from blocking RCU
199 Say Y here if you are working with real-time apps or heavy loads
200 Say N here if you are unsure.
202 config RCU_BOOST_DELAY
203 int "Milliseconds to delay boosting after RCU grace-period start"
208 This option specifies the time to wait after the beginning of
209 a given grace period before priority-boosting preempted RCU
210 readers blocking that grace period. Note that any RCU reader
211 blocking an expedited RCU grace period is boosted immediately.
213 Accept the default if unsure.
216 bool "Offload RCU callback processing from boot-selected CPUs"
218 depends on RCU_EXPERT || NO_HZ_FULL
221 Use this option to reduce OS jitter for aggressive HPC or
222 real-time workloads. It can also be used to offload RCU
223 callback invocation to energy-efficient CPUs in battery-powered
224 asymmetric multiprocessors. The price of this reduced jitter
225 is that the overhead of call_rcu() increases and that some
226 workloads will incur significant increases in context-switch
229 This option offloads callback invocation from the set of CPUs
230 specified at boot time by the rcu_nocbs parameter. For each
231 such CPU, a kthread ("rcuox/N") will be created to invoke
232 callbacks, where the "N" is the CPU being offloaded, and where
233 the "x" is "p" for RCU-preempt (PREEMPTION kernels) and "s" for
234 RCU-sched (!PREEMPTION kernels). Nothing prevents this kthread
235 from running on the specified CPUs, but (1) the kthreads may be
236 preempted between each callback, and (2) affinity or cgroups can
237 be used to force the kthreads to run on whatever set of CPUs is
240 Say Y here if you need reduced OS jitter, despite added overhead.
241 Say N here if you are unsure.
243 config TASKS_TRACE_RCU_READ_MB
244 bool "Tasks Trace RCU readers use memory barriers in user and idle"
245 depends on RCU_EXPERT
246 default PREEMPT_RT || NR_CPUS < 8
248 Use this option to further reduce the number of IPIs sent
249 to CPUs executing in userspace or idle during tasks trace
250 RCU grace periods. Given that a reasonable setting of
251 the rcupdate.rcu_task_ipi_delay kernel boot parameter
252 eliminates such IPIs for many workloads, proper setting
253 of this Kconfig option is important mostly for aggressive
254 real-time installations and for battery-powered devices,
255 hence the default chosen above.
257 Say Y here if you hate IPIs.
258 Say N here if you hate read-side memory barriers.
259 Take the default if you are unsure.
261 endmenu # "RCU Subsystem"