1 # SPDX-License-Identifier: GPL-2.0-only
3 # RCU-related configuration options
10 default y if !PREEMPT && SMP
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
21 This option selects the RCU implementation that is
22 designed for very large SMP systems with hundreds or
23 thousands of CPUs, but for which real-time response
24 is also required. It also scales down nicely to
27 Select this option if you are unsure.
31 default y if !PREEMPT && !SMP
33 This option selects the RCU implementation that is
34 designed for UP systems from which real-time response
35 is not required. This option greatly reduces the
36 memory footprint of RCU.
39 bool "Make expert-level adjustments to RCU configuration"
42 This option needs to be enabled if you wish to make
43 expert-level adjustments to RCU configuration. By default,
44 no such adjustments can be made, which has the often-beneficial
45 side-effect of preventing "make oldconfig" from asking you all
46 sorts of detailed questions about how you would like numerous
47 obscure RCU options to be set up.
49 Say Y if you need to make expert-level adjustments to RCU.
51 Say N if you are unsure.
56 This option selects the sleepable version of RCU. This version
57 permits arbitrary sleeping or blocking within RCU read-side critical
62 default y if SRCU && TINY_RCU
64 This option selects the single-CPU non-preemptible version of SRCU.
68 default y if SRCU && !TINY_RCU
70 This option selects the full-fledged version of SRCU.
76 This option enables a task-based RCU implementation that uses
77 only voluntary context switch (not preemption!), idle, and
78 user-mode execution as quiescent states.
80 config RCU_STALL_COMMON
81 def_bool ( TREE_RCU || PREEMPT_RCU )
83 This option enables RCU CPU stall code that is common between
84 the TINY and TREE variants of RCU. The purpose is to allow
85 the tiny variants to disable RCU CPU stall warnings, while
86 making these warnings mandatory for the tree variants.
88 config RCU_NEED_SEGCBLIST
89 def_bool ( TREE_RCU || PREEMPT_RCU || TREE_SRCU )
92 int "Tree-based hierarchical RCU fanout value"
95 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
99 This option controls the fanout of hierarchical implementations
100 of RCU, allowing RCU to work efficiently on machines with
101 large numbers of CPUs. This value must be at least the fourth
102 root of NR_CPUS, which allows NR_CPUS to be insanely large.
103 The default value of RCU_FANOUT should be used for production
104 systems, but if you are stress-testing the RCU implementation
105 itself, small RCU_FANOUT values allow you to test large-system
106 code paths on small(er) systems.
108 Select a specific number if testing RCU itself.
109 Take the default if unsure.
111 config RCU_FANOUT_LEAF
112 int "Tree-based hierarchical RCU leaf-level fanout value"
115 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
118 This option controls the leaf-level fanout of hierarchical
119 implementations of RCU, and allows trading off cache misses
120 against lock contention. Systems that synchronize their
121 scheduling-clock interrupts for energy-efficiency reasons will
122 want the default because the smaller leaf-level fanout keeps
123 lock contention levels acceptably low. Very large systems
124 (hundreds or thousands of CPUs) will instead want to set this
125 value to the maximum value possible in order to reduce the
126 number of cache misses incurred during RCU's grace-period
127 initialization. These systems tend to run CPU-bound, and thus
128 are not helped by synchronized interrupts, and thus tend to
129 skew them, which reduces lock contention enough that large
130 leaf-level fanouts work well. That said, setting leaf-level
131 fanout to a large number will likely cause problematic
132 lock contention on the leaf-level rcu_node structures unless
133 you boot with the skew_tick kernel parameter.
135 Select a specific number if testing RCU itself.
137 Select the maximum permissible value for large systems, but
138 please understand that you may also need to set the skew_tick
139 kernel boot parameter to avoid contention on the rcu_node
142 Take the default if unsure.
144 config RCU_FAST_NO_HZ
145 bool "Accelerate last non-dyntick-idle CPU's grace periods"
146 depends on NO_HZ_COMMON && SMP && RCU_EXPERT
149 This option permits CPUs to enter dynticks-idle state even if
150 they have RCU callbacks queued, and prevents RCU from waking
151 these CPUs up more than roughly once every four jiffies (by
152 default, you can adjust this using the rcutree.rcu_idle_gp_delay
153 parameter), thus improving energy efficiency. On the other
154 hand, this option increases the duration of RCU grace periods,
155 for example, slowing down synchronize_rcu().
157 Say Y if energy efficiency is critically important, and you
158 don't care about increased grace-period durations.
160 Say N if you are unsure.
163 bool "Enable RCU priority boosting"
164 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT
167 This option boosts the priority of preempted RCU readers that
168 block the current preemptible RCU grace period for too long.
169 This option also prevents heavy loads from blocking RCU
172 Say Y here if you are working with real-time apps or heavy loads
173 Say N here if you are unsure.
175 config RCU_BOOST_DELAY
176 int "Milliseconds to delay boosting after RCU grace-period start"
181 This option specifies the time to wait after the beginning of
182 a given grace period before priority-boosting preempted RCU
183 readers blocking that grace period. Note that any RCU reader
184 blocking an expedited RCU grace period is boosted immediately.
186 Accept the default if unsure.
189 bool "Offload RCU callback processing from boot-selected CPUs"
190 depends on TREE_RCU || PREEMPT_RCU
191 depends on RCU_EXPERT || NO_HZ_FULL
194 Use this option to reduce OS jitter for aggressive HPC or
195 real-time workloads. It can also be used to offload RCU
196 callback invocation to energy-efficient CPUs in battery-powered
197 asymmetric multiprocessors.
199 This option offloads callback invocation from the set of CPUs
200 specified at boot time by the rcu_nocbs parameter. For each
201 such CPU, a kthread ("rcuox/N") will be created to invoke
202 callbacks, where the "N" is the CPU being offloaded, and where
203 the "p" for RCU-preempt (PREEMPT kernels) and "s" for RCU-sched
204 (!PREEMPT kernels). Nothing prevents this kthread from running
205 on the specified CPUs, but (1) the kthreads may be preempted
206 between each callback, and (2) affinity or cgroups can be used
207 to force the kthreads to run on whatever set of CPUs is desired.
209 Say Y here if you want to help to debug reduced OS jitter.
210 Say N here if you are unsure.
212 endmenu # "RCU Subsystem"