6 The basic idea behind RCU (read-copy update) is to split destructive
7 operations into two parts, one that prevents anyone from seeing the data
8 item being destroyed, and one that actually carries out the destruction.
9 A "grace period" must elapse between the two parts, and this grace period
10 must be long enough that any readers accessing the item being deleted have
11 since dropped their references. For example, an RCU-protected deletion
12 from a linked list would first remove the item from the list, wait for
13 a grace period to elapse, then free the element. See the
14 :ref:`Documentation/RCU/listRCU.rst <list_rcu_doc>` for more information on
15 using RCU with linked lists.
17 Frequently Asked Questions
18 --------------------------
20 - Why would anyone want to use RCU?
22 The advantage of RCU's two-part approach is that RCU readers need
23 not acquire any locks, perform any atomic instructions, write to
24 shared memory, or (on CPUs other than Alpha) execute any memory
25 barriers. The fact that these operations are quite expensive
26 on modern CPUs is what gives RCU its performance advantages
27 in read-mostly situations. The fact that RCU readers need not
28 acquire locks can also greatly simplify deadlock-avoidance code.
30 - How can the updater tell when a grace period has completed
31 if the RCU readers give no indication when they are done?
33 Just as with spinlocks, RCU readers are not permitted to
34 block, switch to user-mode execution, or enter the idle loop.
35 Therefore, as soon as a CPU is seen passing through any of these
36 three states, we know that that CPU has exited any previous RCU
37 read-side critical sections. So, if we remove an item from a
38 linked list, and then wait until all CPUs have switched context,
39 executed in user mode, or executed in the idle loop, we can
40 safely free up that item.
42 Preemptible variants of RCU (CONFIG_PREEMPT_RCU) get the
43 same effect, but require that the readers manipulate CPU-local
44 counters. These counters allow limited types of blocking within
45 RCU read-side critical sections. SRCU also uses CPU-local
46 counters, and permits general blocking within RCU read-side
47 critical sections. These variants of RCU detect grace periods
48 by sampling these counters.
50 - If I am running on a uniprocessor kernel, which can only do one
51 thing at a time, why should I wait for a grace period?
53 See :ref:`Documentation/RCU/UP.rst <up_doc>` for more information.
55 - How can I see where RCU is currently used in the Linux kernel?
57 Search for "rcu_read_lock", "rcu_read_unlock", "call_rcu",
58 "rcu_read_lock_bh", "rcu_read_unlock_bh", "srcu_read_lock",
59 "srcu_read_unlock", "synchronize_rcu", "synchronize_net",
60 "synchronize_srcu", and the other RCU primitives. Or grab one
61 of the cscope databases from:
63 (http://www.rdrop.com/users/paulmck/RCU/linuxusage/rculocktab.html).
65 - What guidelines should I follow when writing code that uses RCU?
67 See the checklist.txt file in this directory.
71 "RCU" stands for "read-copy update".
72 :ref:`Documentation/RCU/listRCU.rst <list_rcu_doc>` has more information on where
73 this name came from, search for "read-copy update" to find it.
75 - I hear that RCU is patented? What is with that?
77 Yes, it is. There are several known patents related to RCU,
78 search for the string "Patent" in Documentation/RCU/RTFP.txt to find them.
79 Of these, one was allowed to lapse by the assignee, and the
80 others have been contributed to the Linux kernel under GPL.
81 There are now also LGPL implementations of user-level RCU
82 available (https://liburcu.org/).
84 - I hear that RCU needs work in order to support realtime kernels?
86 Realtime-friendly RCU can be enabled via the CONFIG_PREEMPT_RCU
87 kernel configuration parameter.
89 - Where can I find more information on RCU?
91 See the Documentation/RCU/RTFP.txt file.
92 Or point your browser at (http://www.rdrop.com/users/paulmck/RCU/).