1# 2# RCU-related configuration options 3# 4 5menu "RCU Subsystem" 6 7config TREE_RCU 8 bool 9 default y if !PREEMPT && SMP 10 help 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 14 smaller systems. 15 16config PREEMPT_RCU 17 bool 18 default y if PREEMPT 19 help 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 24 smaller systems. 25 26 Select this option if you are unsure. 27 28config TINY_RCU 29 bool 30 default y if !PREEMPT && !SMP 31 help 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. 36 37config RCU_EXPERT 38 bool "Make expert-level adjustments to RCU configuration" 39 default n 40 help 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. 47 48 Say Y if you need to make expert-level adjustments to RCU. 49 50 Say N if you are unsure. 51 52config 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 59config TINY_SRCU 60 bool 61 default y if SRCU && TINY_RCU 62 help 63 This option selects the single-CPU non-preemptible version of SRCU. 64 65config TREE_SRCU 66 bool 67 default y if SRCU && !TINY_RCU 68 help 69 This option selects the full-fledged version of SRCU. 70 71config TASKS_RCU 72 def_bool PREEMPT 73 select SRCU 74 help 75 This option enables a task-based RCU implementation that uses 76 only voluntary context switch (not preemption!), idle, and 77 user-mode execution as quiescent states. 78 79config RCU_STALL_COMMON 80 def_bool ( TREE_RCU || PREEMPT_RCU ) 81 help 82 This option enables RCU CPU stall code that is common between 83 the TINY and TREE variants of RCU. The purpose is to allow 84 the tiny variants to disable RCU CPU stall warnings, while 85 making these warnings mandatory for the tree variants. 86 87config RCU_NEED_SEGCBLIST 88 def_bool ( TREE_RCU || PREEMPT_RCU || TREE_SRCU ) 89 90config RCU_FANOUT 91 int "Tree-based hierarchical RCU fanout value" 92 range 2 64 if 64BIT 93 range 2 32 if !64BIT 94 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT 95 default 64 if 64BIT 96 default 32 if !64BIT 97 help 98 This option controls the fanout of hierarchical implementations 99 of RCU, allowing RCU to work efficiently on machines with 100 large numbers of CPUs. This value must be at least the fourth 101 root of NR_CPUS, which allows NR_CPUS to be insanely large. 102 The default value of RCU_FANOUT should be used for production 103 systems, but if you are stress-testing the RCU implementation 104 itself, small RCU_FANOUT values allow you to test large-system 105 code paths on small(er) systems. 106 107 Select a specific number if testing RCU itself. 108 Take the default if unsure. 109 110config RCU_FANOUT_LEAF 111 int "Tree-based hierarchical RCU leaf-level fanout value" 112 range 2 64 if 64BIT 113 range 2 32 if !64BIT 114 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT 115 default 16 116 help 117 This option controls the leaf-level fanout of hierarchical 118 implementations of RCU, and allows trading off cache misses 119 against lock contention. Systems that synchronize their 120 scheduling-clock interrupts for energy-efficiency reasons will 121 want the default because the smaller leaf-level fanout keeps 122 lock contention levels acceptably low. Very large systems 123 (hundreds or thousands of CPUs) will instead want to set this 124 value to the maximum value possible in order to reduce the 125 number of cache misses incurred during RCU's grace-period 126 initialization. These systems tend to run CPU-bound, and thus 127 are not helped by synchronized interrupts, and thus tend to 128 skew them, which reduces lock contention enough that large 129 leaf-level fanouts work well. That said, setting leaf-level 130 fanout to a large number will likely cause problematic 131 lock contention on the leaf-level rcu_node structures unless 132 you boot with the skew_tick kernel parameter. 133 134 Select a specific number if testing RCU itself. 135 136 Select the maximum permissible value for large systems, but 137 please understand that you may also need to set the skew_tick 138 kernel boot parameter to avoid contention on the rcu_node 139 structure's locks. 140 141 Take the default if unsure. 142 143config 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 161config RCU_BOOST 162 bool "Enable RCU priority boosting" 163 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT 164 default n 165 help 166 This option boosts the priority of preempted RCU readers that 167 block the current preemptible RCU grace period for too long. 168 This option also prevents heavy loads from blocking RCU 169 callback invocation. 170 171 Say Y here if you are working with real-time apps or heavy loads 172 Say N here if you are unsure. 173 174config RCU_BOOST_DELAY 175 int "Milliseconds to delay boosting after RCU grace-period start" 176 range 0 3000 177 depends on RCU_BOOST 178 default 500 179 help 180 This option specifies the time to wait after the beginning of 181 a given grace period before priority-boosting preempted RCU 182 readers blocking that grace period. Note that any RCU reader 183 blocking an expedited RCU grace period is boosted immediately. 184 185 Accept the default if unsure. 186 187config RCU_NOCB_CPU 188 bool "Offload RCU callback processing from boot-selected CPUs" 189 depends on TREE_RCU || PREEMPT_RCU 190 depends on RCU_EXPERT || NO_HZ_FULL 191 default n 192 help 193 Use this option to reduce OS jitter for aggressive HPC or 194 real-time workloads. It can also be used to offload RCU 195 callback invocation to energy-efficient CPUs in battery-powered 196 asymmetric multiprocessors. 197 198 This option offloads callback invocation from the set of CPUs 199 specified at boot time by the rcu_nocbs parameter. For each 200 such CPU, a kthread ("rcuox/N") will be created to invoke 201 callbacks, where the "N" is the CPU being offloaded, and where 202 the "p" for RCU-preempt (PREEMPT kernels) and "s" for RCU-sched 203 (!PREEMPT kernels). Nothing prevents this kthread from running 204 on the specified CPUs, but (1) the kthreads may be preempted 205 between each callback, and (2) affinity or cgroups can be used 206 to force the kthreads to run on whatever set of CPUs is desired. 207 208 Say Y here if you want to help to debug reduced OS jitter. 209 Say N here if you are unsure. 210 211endmenu # "RCU Subsystem" 212