2 # RCU-related configuration options
9 default y if !PREEMPT && SMP
11 This option selects the RCU implementation that is
12 designed for very large SMP system with hundreds or
13 thousands of CPUs. It also scales down nicely to
20 This option selects the RCU implementation that is
21 designed for very large SMP systems with hundreds or
22 thousands of CPUs, but for which real-time response
23 is also required. It also scales down nicely to
26 Select this option if you are unsure.
30 default y if !PREEMPT && !SMP
32 This option selects the RCU implementation that is
33 designed for UP systems from which real-time response
34 is not required. This option greatly reduces the
35 memory footprint of RCU.
38 bool "Make expert-level adjustments to RCU configuration"
41 This option needs to be enabled if you wish to make
42 expert-level adjustments to RCU configuration. By default,
43 no such adjustments can be made, which has the often-beneficial
44 side-effect of preventing "make oldconfig" from asking you all
45 sorts of detailed questions about how you would like numerous
46 obscure RCU options to be set up.
48 Say Y if you need to make expert-level adjustments to RCU.
50 Say N if you are unsure.
55 This option selects the sleepable version of RCU. This version
56 permits arbitrary sleeping or blocking within RCU read-side critical
61 default y if SRCU && TINY_RCU
63 This option selects the single-CPU non-preemptible version of SRCU.
67 default y if SRCU && !TINY_RCU
69 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 )
91 config CONTEXT_TRACKING
94 config CONTEXT_TRACKING_FORCE
95 bool "Force context tracking"
96 depends on CONTEXT_TRACKING
97 default y if !NO_HZ_FULL
99 The major pre-requirement for full dynticks to work is to
100 support the context tracking subsystem. But there are also
101 other dependencies to provide in order to make the full
104 This option stands for testing when an arch implements the
105 context tracking backend but doesn't yet fullfill all the
106 requirements to make the full dynticks feature working.
107 Without the full dynticks, there is no way to test the support
108 for context tracking and the subsystems that rely on it: RCU
109 userspace extended quiescent state and tickless cputime
110 accounting. This option copes with the absence of the full
111 dynticks subsystem by forcing the context tracking on all
114 Say Y only if you're working on the development of an
115 architecture backend for the context tracking.
117 Say N otherwise, this option brings an overhead that you
118 don't want in production.
122 int "Tree-based hierarchical RCU fanout value"
125 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
129 This option controls the fanout of hierarchical implementations
130 of RCU, allowing RCU to work efficiently on machines with
131 large numbers of CPUs. This value must be at least the fourth
132 root of NR_CPUS, which allows NR_CPUS to be insanely large.
133 The default value of RCU_FANOUT should be used for production
134 systems, but if you are stress-testing the RCU implementation
135 itself, small RCU_FANOUT values allow you to test large-system
136 code paths on small(er) systems.
138 Select a specific number if testing RCU itself.
139 Take the default if unsure.
141 config RCU_FANOUT_LEAF
142 int "Tree-based hierarchical RCU leaf-level fanout value"
145 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
148 This option controls the leaf-level fanout of hierarchical
149 implementations of RCU, and allows trading off cache misses
150 against lock contention. Systems that synchronize their
151 scheduling-clock interrupts for energy-efficiency reasons will
152 want the default because the smaller leaf-level fanout keeps
153 lock contention levels acceptably low. Very large systems
154 (hundreds or thousands of CPUs) will instead want to set this
155 value to the maximum value possible in order to reduce the
156 number of cache misses incurred during RCU's grace-period
157 initialization. These systems tend to run CPU-bound, and thus
158 are not helped by synchronized interrupts, and thus tend to
159 skew them, which reduces lock contention enough that large
160 leaf-level fanouts work well. That said, setting leaf-level
161 fanout to a large number will likely cause problematic
162 lock contention on the leaf-level rcu_node structures unless
163 you boot with the skew_tick kernel parameter.
165 Select a specific number if testing RCU itself.
167 Select the maximum permissible value for large systems, but
168 please understand that you may also need to set the skew_tick
169 kernel boot parameter to avoid contention on the rcu_node
172 Take the default if unsure.
174 config RCU_FAST_NO_HZ
175 bool "Accelerate last non-dyntick-idle CPU's grace periods"
176 depends on NO_HZ_COMMON && SMP && RCU_EXPERT
179 This option permits CPUs to enter dynticks-idle state even if
180 they have RCU callbacks queued, and prevents RCU from waking
181 these CPUs up more than roughly once every four jiffies (by
182 default, you can adjust this using the rcutree.rcu_idle_gp_delay
183 parameter), thus improving energy efficiency. On the other
184 hand, this option increases the duration of RCU grace periods,
185 for example, slowing down synchronize_rcu().
187 Say Y if energy efficiency is critically important, and you
188 don't care about increased grace-period durations.
190 Say N if you are unsure.
193 bool "Enable RCU priority boosting"
194 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT
197 This option boosts the priority of preempted RCU readers that
198 block the current preemptible RCU grace period for too long.
199 This option also prevents heavy loads from blocking RCU
200 callback invocation for all flavors of RCU.
202 Say Y here if you are working with real-time apps or heavy loads
203 Say N here if you are unsure.
205 config RCU_BOOST_DELAY
206 int "Milliseconds to delay boosting after RCU grace-period start"
211 This option specifies the time to wait after the beginning of
212 a given grace period before priority-boosting preempted RCU
213 readers blocking that grace period. Note that any RCU reader
214 blocking an expedited RCU grace period is boosted immediately.
216 Accept the default if unsure.
219 bool "Offload RCU callback processing from boot-selected CPUs"
220 depends on TREE_RCU || PREEMPT_RCU
221 depends on RCU_EXPERT || NO_HZ_FULL
224 Use this option to reduce OS jitter for aggressive HPC or
225 real-time workloads. It can also be used to offload RCU
226 callback invocation to energy-efficient CPUs in battery-powered
227 asymmetric multiprocessors.
229 This option offloads callback invocation from the set of
230 CPUs specified at boot time by the rcu_nocbs parameter.
231 For each such CPU, a kthread ("rcuox/N") will be created to
232 invoke callbacks, where the "N" is the CPU being offloaded,
233 and where the "x" is "b" for RCU-bh, "p" for RCU-preempt, and
234 "s" for RCU-sched. Nothing prevents this kthread from running
235 on the specified CPUs, but (1) the kthreads may be preempted
236 between each callback, and (2) affinity or cgroups can be used
237 to force the kthreads to run on whatever set of CPUs is desired.
239 Say Y here if you want to help to debug reduced OS jitter.
240 Say N here if you are unsure.
242 endmenu # "RCU Subsystem"