blob: e0d6d99b8f9bb0dc17f647dbd8256aad1282dd2d [file] [log] [blame]
Paul E. McKenneydd81eca2005-09-10 00:26:24 -07001What is RCU?
2
3RCU is a synchronization mechanism that was added to the Linux kernel
4during the 2.5 development effort that is optimized for read-mostly
5situations. Although RCU is actually quite simple once you understand it,
6getting there can sometimes be a challenge. Part of the problem is that
7most of the past descriptions of RCU have been written with the mistaken
8assumption that there is "one true way" to describe RCU. Instead,
9the experience has been that different people must take different paths
10to arrive at an understanding of RCU. This document provides several
11different paths, as follows:
12
131. RCU OVERVIEW
142. WHAT IS RCU'S CORE API?
153. WHAT ARE SOME EXAMPLE USES OF CORE RCU API?
164. WHAT IF MY UPDATING THREAD CANNOT BLOCK?
175. WHAT ARE SOME SIMPLE IMPLEMENTATIONS OF RCU?
186. ANALOGY WITH READER-WRITER LOCKING
197. FULL LIST OF RCU APIs
208. ANSWERS TO QUICK QUIZZES
21
22People who prefer starting with a conceptual overview should focus on
23Section 1, though most readers will profit by reading this section at
24some point. People who prefer to start with an API that they can then
25experiment with should focus on Section 2. People who prefer to start
26with example uses should focus on Sections 3 and 4. People who need to
27understand the RCU implementation should focus on Section 5, then dive
28into the kernel source code. People who reason best by analogy should
29focus on Section 6. Section 7 serves as an index to the docbook API
30documentation, and Section 8 is the traditional answer key.
31
32So, start with the section that makes the most sense to you and your
33preferred method of learning. If you need to know everything about
34everything, feel free to read the whole thing -- but if you are really
35that type of person, you have perused the source code and will therefore
36never need this document anyway. ;-)
37
38
391. RCU OVERVIEW
40
41The basic idea behind RCU is to split updates into "removal" and
42"reclamation" phases. The removal phase removes references to data items
43within a data structure (possibly by replacing them with references to
44new versions of these data items), and can run concurrently with readers.
45The reason that it is safe to run the removal phase concurrently with
46readers is the semantics of modern CPUs guarantee that readers will see
47either the old or the new version of the data structure rather than a
48partially updated reference. The reclamation phase does the work of reclaiming
49(e.g., freeing) the data items removed from the data structure during the
50removal phase. Because reclaiming data items can disrupt any readers
51concurrently referencing those data items, the reclamation phase must
52not start until readers no longer hold references to those data items.
53
54Splitting the update into removal and reclamation phases permits the
55updater to perform the removal phase immediately, and to defer the
56reclamation phase until all readers active during the removal phase have
57completed, either by blocking until they finish or by registering a
58callback that is invoked after they finish. Only readers that are active
59during the removal phase need be considered, because any reader starting
60after the removal phase will be unable to gain a reference to the removed
61data items, and therefore cannot be disrupted by the reclamation phase.
62
63So the typical RCU update sequence goes something like the following:
64
65a. Remove pointers to a data structure, so that subsequent
66 readers cannot gain a reference to it.
67
68b. Wait for all previous readers to complete their RCU read-side
69 critical sections.
70
71c. At this point, there cannot be any readers who hold references
72 to the data structure, so it now may safely be reclaimed
73 (e.g., kfree()d).
74
75Step (b) above is the key idea underlying RCU's deferred destruction.
76The ability to wait until all readers are done allows RCU readers to
77use much lighter-weight synchronization, in some cases, absolutely no
78synchronization at all. In contrast, in more conventional lock-based
79schemes, readers must use heavy-weight synchronization in order to
80prevent an updater from deleting the data structure out from under them.
81This is because lock-based updaters typically update data items in place,
82and must therefore exclude readers. In contrast, RCU-based updaters
83typically take advantage of the fact that writes to single aligned
84pointers are atomic on modern CPUs, allowing atomic insertion, removal,
85and replacement of data items in a linked structure without disrupting
86readers. Concurrent RCU readers can then continue accessing the old
87versions, and can dispense with the atomic operations, memory barriers,
88and communications cache misses that are so expensive on present-day
89SMP computer systems, even in absence of lock contention.
90
91In the three-step procedure shown above, the updater is performing both
92the removal and the reclamation step, but it is often helpful for an
93entirely different thread to do the reclamation, as is in fact the case
94in the Linux kernel's directory-entry cache (dcache). Even if the same
95thread performs both the update step (step (a) above) and the reclamation
96step (step (c) above), it is often helpful to think of them separately.
97For example, RCU readers and updaters need not communicate at all,
98but RCU provides implicit low-overhead communication between readers
99and reclaimers, namely, in step (b) above.
100
101So how the heck can a reclaimer tell when a reader is done, given
102that readers are not doing any sort of synchronization operations???
103Read on to learn about how RCU's API makes this easy.
104
105
1062. WHAT IS RCU'S CORE API?
107
108The core RCU API is quite small:
109
110a. rcu_read_lock()
111b. rcu_read_unlock()
112c. synchronize_rcu() / call_rcu()
113d. rcu_assign_pointer()
114e. rcu_dereference()
115
116There are many other members of the RCU API, but the rest can be
117expressed in terms of these five, though most implementations instead
118express synchronize_rcu() in terms of the call_rcu() callback API.
119
120The five core RCU APIs are described below, the other 18 will be enumerated
121later. See the kernel docbook documentation for more info, or look directly
122at the function header comments.
123
124rcu_read_lock()
125
126 void rcu_read_lock(void);
127
128 Used by a reader to inform the reclaimer that the reader is
129 entering an RCU read-side critical section. It is illegal
130 to block while in an RCU read-side critical section, though
131 kernels built with CONFIG_PREEMPT_RCU can preempt RCU read-side
132 critical sections. Any RCU-protected data structure accessed
133 during an RCU read-side critical section is guaranteed to remain
134 unreclaimed for the full duration of that critical section.
135 Reference counts may be used in conjunction with RCU to maintain
136 longer-term references to data structures.
137
138rcu_read_unlock()
139
140 void rcu_read_unlock(void);
141
142 Used by a reader to inform the reclaimer that the reader is
143 exiting an RCU read-side critical section. Note that RCU
144 read-side critical sections may be nested and/or overlapping.
145
146synchronize_rcu()
147
148 void synchronize_rcu(void);
149
150 Marks the end of updater code and the beginning of reclaimer
151 code. It does this by blocking until all pre-existing RCU
152 read-side critical sections on all CPUs have completed.
153 Note that synchronize_rcu() will -not- necessarily wait for
154 any subsequent RCU read-side critical sections to complete.
155 For example, consider the following sequence of events:
156
157 CPU 0 CPU 1 CPU 2
158 ----------------- ------------------------- ---------------
159 1. rcu_read_lock()
160 2. enters synchronize_rcu()
161 3. rcu_read_lock()
162 4. rcu_read_unlock()
163 5. exits synchronize_rcu()
164 6. rcu_read_unlock()
165
166 To reiterate, synchronize_rcu() waits only for ongoing RCU
167 read-side critical sections to complete, not necessarily for
168 any that begin after synchronize_rcu() is invoked.
169
170 Of course, synchronize_rcu() does not necessarily return
171 -immediately- after the last pre-existing RCU read-side critical
172 section completes. For one thing, there might well be scheduling
173 delays. For another thing, many RCU implementations process
174 requests in batches in order to improve efficiencies, which can
175 further delay synchronize_rcu().
176
177 Since synchronize_rcu() is the API that must figure out when
178 readers are done, its implementation is key to RCU. For RCU
179 to be useful in all but the most read-intensive situations,
180 synchronize_rcu()'s overhead must also be quite small.
181
182 The call_rcu() API is a callback form of synchronize_rcu(),
183 and is described in more detail in a later section. Instead of
184 blocking, it registers a function and argument which are invoked
185 after all ongoing RCU read-side critical sections have completed.
186 This callback variant is particularly useful in situations where
Paul E. McKenney165d6c72006-06-25 05:48:44 -0700187 it is illegal to block or where update-side performance is
188 critically important.
189
190 However, the call_rcu() API should not be used lightly, as use
191 of the synchronize_rcu() API generally results in simpler code.
192 In addition, the synchronize_rcu() API has the nice property
193 of automatically limiting update rate should grace periods
194 be delayed. This property results in system resilience in face
195 of denial-of-service attacks. Code using call_rcu() should limit
196 update rate in order to gain this same sort of resilience. See
197 checklist.txt for some approaches to limiting the update rate.
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700198
199rcu_assign_pointer()
200
201 typeof(p) rcu_assign_pointer(p, typeof(p) v);
202
203 Yes, rcu_assign_pointer() -is- implemented as a macro, though it
204 would be cool to be able to declare a function in this manner.
205 (Compiler experts will no doubt disagree.)
206
207 The updater uses this function to assign a new value to an
208 RCU-protected pointer, in order to safely communicate the change
209 in value from the updater to the reader. This function returns
210 the new value, and also executes any memory-barrier instructions
211 required for a given CPU architecture.
212
Paul E. McKenneyd19720a2006-02-01 03:06:42 -0800213 Perhaps just as important, it serves to document (1) which
214 pointers are protected by RCU and (2) the point at which a
215 given structure becomes accessible to other CPUs. That said,
216 rcu_assign_pointer() is most frequently used indirectly, via
217 the _rcu list-manipulation primitives such as list_add_rcu().
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700218
219rcu_dereference()
220
221 typeof(p) rcu_dereference(p);
222
223 Like rcu_assign_pointer(), rcu_dereference() must be implemented
224 as a macro.
225
226 The reader uses rcu_dereference() to fetch an RCU-protected
227 pointer, which returns a value that may then be safely
228 dereferenced. Note that rcu_deference() does not actually
229 dereference the pointer, instead, it protects the pointer for
230 later dereferencing. It also executes any needed memory-barrier
231 instructions for a given CPU architecture. Currently, only Alpha
232 needs memory barriers within rcu_dereference() -- on other CPUs,
233 it compiles to nothing, not even a compiler directive.
234
235 Common coding practice uses rcu_dereference() to copy an
236 RCU-protected pointer to a local variable, then dereferences
237 this local variable, for example as follows:
238
239 p = rcu_dereference(head.next);
240 return p->data;
241
242 However, in this case, one could just as easily combine these
243 into one statement:
244
245 return rcu_dereference(head.next)->data;
246
247 If you are going to be fetching multiple fields from the
248 RCU-protected structure, using the local variable is of
249 course preferred. Repeated rcu_dereference() calls look
250 ugly and incur unnecessary overhead on Alpha CPUs.
251
252 Note that the value returned by rcu_dereference() is valid
253 only within the enclosing RCU read-side critical section.
254 For example, the following is -not- legal:
255
256 rcu_read_lock();
257 p = rcu_dereference(head.next);
258 rcu_read_unlock();
259 x = p->address;
260 rcu_read_lock();
261 y = p->data;
262 rcu_read_unlock();
263
264 Holding a reference from one RCU read-side critical section
265 to another is just as illegal as holding a reference from
266 one lock-based critical section to another! Similarly,
267 using a reference outside of the critical section in which
268 it was acquired is just as illegal as doing so with normal
269 locking.
270
271 As with rcu_assign_pointer(), an important function of
Paul E. McKenneyd19720a2006-02-01 03:06:42 -0800272 rcu_dereference() is to document which pointers are protected by
273 RCU, in particular, flagging a pointer that is subject to changing
274 at any time, including immediately after the rcu_dereference().
275 And, again like rcu_assign_pointer(), rcu_dereference() is
276 typically used indirectly, via the _rcu list-manipulation
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700277 primitives, such as list_for_each_entry_rcu().
278
279The following diagram shows how each API communicates among the
280reader, updater, and reclaimer.
281
282
283 rcu_assign_pointer()
284 +--------+
285 +---------------------->| reader |---------+
286 | +--------+ |
287 | | |
288 | | | Protect:
289 | | | rcu_read_lock()
290 | | | rcu_read_unlock()
291 | rcu_dereference() | |
292 +---------+ | |
293 | updater |<---------------------+ |
294 +---------+ V
295 | +-----------+
296 +----------------------------------->| reclaimer |
297 +-----------+
298 Defer:
299 synchronize_rcu() & call_rcu()
300
301
302The RCU infrastructure observes the time sequence of rcu_read_lock(),
303rcu_read_unlock(), synchronize_rcu(), and call_rcu() invocations in
304order to determine when (1) synchronize_rcu() invocations may return
305to their callers and (2) call_rcu() callbacks may be invoked. Efficient
306implementations of the RCU infrastructure make heavy use of batching in
307order to amortize their overhead over many uses of the corresponding APIs.
308
309There are no fewer than three RCU mechanisms in the Linux kernel; the
310diagram above shows the first one, which is by far the most commonly used.
311The rcu_dereference() and rcu_assign_pointer() primitives are used for
312all three mechanisms, but different defer and protect primitives are
313used as follows:
314
315 Defer Protect
316
317a. synchronize_rcu() rcu_read_lock() / rcu_read_unlock()
318 call_rcu()
319
320b. call_rcu_bh() rcu_read_lock_bh() / rcu_read_unlock_bh()
321
322c. synchronize_sched() preempt_disable() / preempt_enable()
323 local_irq_save() / local_irq_restore()
324 hardirq enter / hardirq exit
325 NMI enter / NMI exit
326
327These three mechanisms are used as follows:
328
329a. RCU applied to normal data structures.
330
331b. RCU applied to networking data structures that may be subjected
332 to remote denial-of-service attacks.
333
334c. RCU applied to scheduler and interrupt/NMI-handler tasks.
335
336Again, most uses will be of (a). The (b) and (c) cases are important
337for specialized uses, but are relatively uncommon.
338
339
3403. WHAT ARE SOME EXAMPLE USES OF CORE RCU API?
341
342This section shows a simple use of the core RCU API to protect a
Paul E. McKenneyd19720a2006-02-01 03:06:42 -0800343global pointer to a dynamically allocated structure. More-typical
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700344uses of RCU may be found in listRCU.txt, arrayRCU.txt, and NMI-RCU.txt.
345
346 struct foo {
347 int a;
348 char b;
349 long c;
350 };
351 DEFINE_SPINLOCK(foo_mutex);
352
353 struct foo *gbl_foo;
354
355 /*
356 * Create a new struct foo that is the same as the one currently
357 * pointed to by gbl_foo, except that field "a" is replaced
358 * with "new_a". Points gbl_foo to the new structure, and
359 * frees up the old structure after a grace period.
360 *
361 * Uses rcu_assign_pointer() to ensure that concurrent readers
362 * see the initialized version of the new structure.
363 *
364 * Uses synchronize_rcu() to ensure that any readers that might
365 * have references to the old structure complete before freeing
366 * the old structure.
367 */
368 void foo_update_a(int new_a)
369 {
370 struct foo *new_fp;
371 struct foo *old_fp;
372
Baruch Evende0dfcd2006-03-24 18:25:25 +0100373 new_fp = kmalloc(sizeof(*new_fp), GFP_KERNEL);
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700374 spin_lock(&foo_mutex);
375 old_fp = gbl_foo;
376 *new_fp = *old_fp;
377 new_fp->a = new_a;
378 rcu_assign_pointer(gbl_foo, new_fp);
379 spin_unlock(&foo_mutex);
380 synchronize_rcu();
381 kfree(old_fp);
382 }
383
384 /*
385 * Return the value of field "a" of the current gbl_foo
386 * structure. Use rcu_read_lock() and rcu_read_unlock()
387 * to ensure that the structure does not get deleted out
388 * from under us, and use rcu_dereference() to ensure that
389 * we see the initialized version of the structure (important
390 * for DEC Alpha and for people reading the code).
391 */
392 int foo_get_a(void)
393 {
394 int retval;
395
396 rcu_read_lock();
397 retval = rcu_dereference(gbl_foo)->a;
398 rcu_read_unlock();
399 return retval;
400 }
401
402So, to sum up:
403
404o Use rcu_read_lock() and rcu_read_unlock() to guard RCU
405 read-side critical sections.
406
407o Within an RCU read-side critical section, use rcu_dereference()
408 to dereference RCU-protected pointers.
409
410o Use some solid scheme (such as locks or semaphores) to
411 keep concurrent updates from interfering with each other.
412
413o Use rcu_assign_pointer() to update an RCU-protected pointer.
414 This primitive protects concurrent readers from the updater,
415 -not- concurrent updates from each other! You therefore still
416 need to use locking (or something similar) to keep concurrent
417 rcu_assign_pointer() primitives from interfering with each other.
418
419o Use synchronize_rcu() -after- removing a data element from an
420 RCU-protected data structure, but -before- reclaiming/freeing
421 the data element, in order to wait for the completion of all
422 RCU read-side critical sections that might be referencing that
423 data item.
424
425See checklist.txt for additional rules to follow when using RCU.
Paul E. McKenneyd19720a2006-02-01 03:06:42 -0800426And again, more-typical uses of RCU may be found in listRCU.txt,
427arrayRCU.txt, and NMI-RCU.txt.
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700428
429
4304. WHAT IF MY UPDATING THREAD CANNOT BLOCK?
431
432In the example above, foo_update_a() blocks until a grace period elapses.
433This is quite simple, but in some cases one cannot afford to wait so
434long -- there might be other high-priority work to be done.
435
436In such cases, one uses call_rcu() rather than synchronize_rcu().
437The call_rcu() API is as follows:
438
439 void call_rcu(struct rcu_head * head,
440 void (*func)(struct rcu_head *head));
441
442This function invokes func(head) after a grace period has elapsed.
443This invocation might happen from either softirq or process context,
444so the function is not permitted to block. The foo struct needs to
445have an rcu_head structure added, perhaps as follows:
446
447 struct foo {
448 int a;
449 char b;
450 long c;
451 struct rcu_head rcu;
452 };
453
454The foo_update_a() function might then be written as follows:
455
456 /*
457 * Create a new struct foo that is the same as the one currently
458 * pointed to by gbl_foo, except that field "a" is replaced
459 * with "new_a". Points gbl_foo to the new structure, and
460 * frees up the old structure after a grace period.
461 *
462 * Uses rcu_assign_pointer() to ensure that concurrent readers
463 * see the initialized version of the new structure.
464 *
465 * Uses call_rcu() to ensure that any readers that might have
466 * references to the old structure complete before freeing the
467 * old structure.
468 */
469 void foo_update_a(int new_a)
470 {
471 struct foo *new_fp;
472 struct foo *old_fp;
473
Baruch Evende0dfcd2006-03-24 18:25:25 +0100474 new_fp = kmalloc(sizeof(*new_fp), GFP_KERNEL);
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700475 spin_lock(&foo_mutex);
476 old_fp = gbl_foo;
477 *new_fp = *old_fp;
478 new_fp->a = new_a;
479 rcu_assign_pointer(gbl_foo, new_fp);
480 spin_unlock(&foo_mutex);
481 call_rcu(&old_fp->rcu, foo_reclaim);
482 }
483
484The foo_reclaim() function might appear as follows:
485
486 void foo_reclaim(struct rcu_head *rp)
487 {
488 struct foo *fp = container_of(rp, struct foo, rcu);
489
490 kfree(fp);
491 }
492
493The container_of() primitive is a macro that, given a pointer into a
494struct, the type of the struct, and the pointed-to field within the
495struct, returns a pointer to the beginning of the struct.
496
497The use of call_rcu() permits the caller of foo_update_a() to
498immediately regain control, without needing to worry further about the
499old version of the newly updated element. It also clearly shows the
500RCU distinction between updater, namely foo_update_a(), and reclaimer,
501namely foo_reclaim().
502
503The summary of advice is the same as for the previous section, except
504that we are now using call_rcu() rather than synchronize_rcu():
505
506o Use call_rcu() -after- removing a data element from an
507 RCU-protected data structure in order to register a callback
508 function that will be invoked after the completion of all RCU
509 read-side critical sections that might be referencing that
510 data item.
511
512Again, see checklist.txt for additional rules governing the use of RCU.
513
514
5155. WHAT ARE SOME SIMPLE IMPLEMENTATIONS OF RCU?
516
517One of the nice things about RCU is that it has extremely simple "toy"
518implementations that are a good first step towards understanding the
519production-quality implementations in the Linux kernel. This section
520presents two such "toy" implementations of RCU, one that is implemented
521in terms of familiar locking primitives, and another that more closely
522resembles "classic" RCU. Both are way too simple for real-world use,
523lacking both functionality and performance. However, they are useful
524in getting a feel for how RCU works. See kernel/rcupdate.c for a
525production-quality implementation, and see:
526
527 http://www.rdrop.com/users/paulmck/RCU
528
529for papers describing the Linux kernel RCU implementation. The OLS'01
530and OLS'02 papers are a good introduction, and the dissertation provides
Paul E. McKenneyd19720a2006-02-01 03:06:42 -0800531more details on the current implementation as of early 2004.
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700532
533
5345A. "TOY" IMPLEMENTATION #1: LOCKING
535
536This section presents a "toy" RCU implementation that is based on
537familiar locking primitives. Its overhead makes it a non-starter for
538real-life use, as does its lack of scalability. It is also unsuitable
539for realtime use, since it allows scheduling latency to "bleed" from
540one read-side critical section to another.
541
542However, it is probably the easiest implementation to relate to, so is
543a good starting point.
544
545It is extremely simple:
546
547 static DEFINE_RWLOCK(rcu_gp_mutex);
548
549 void rcu_read_lock(void)
550 {
551 read_lock(&rcu_gp_mutex);
552 }
553
554 void rcu_read_unlock(void)
555 {
556 read_unlock(&rcu_gp_mutex);
557 }
558
559 void synchronize_rcu(void)
560 {
561 write_lock(&rcu_gp_mutex);
562 write_unlock(&rcu_gp_mutex);
563 }
564
565[You can ignore rcu_assign_pointer() and rcu_dereference() without
566missing much. But here they are anyway. And whatever you do, don't
567forget about them when submitting patches making use of RCU!]
568
569 #define rcu_assign_pointer(p, v) ({ \
570 smp_wmb(); \
571 (p) = (v); \
572 })
573
574 #define rcu_dereference(p) ({ \
575 typeof(p) _________p1 = p; \
576 smp_read_barrier_depends(); \
577 (_________p1); \
578 })
579
580
581The rcu_read_lock() and rcu_read_unlock() primitive read-acquire
582and release a global reader-writer lock. The synchronize_rcu()
583primitive write-acquires this same lock, then immediately releases
584it. This means that once synchronize_rcu() exits, all RCU read-side
Matt LaPlante53cb4722006-10-03 22:55:17 +0200585critical sections that were in progress before synchronize_rcu() was
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700586called are guaranteed to have completed -- there is no way that
587synchronize_rcu() would have been able to write-acquire the lock
588otherwise.
589
590It is possible to nest rcu_read_lock(), since reader-writer locks may
591be recursively acquired. Note also that rcu_read_lock() is immune
592from deadlock (an important property of RCU). The reason for this is
593that the only thing that can block rcu_read_lock() is a synchronize_rcu().
594But synchronize_rcu() does not acquire any locks while holding rcu_gp_mutex,
595so there can be no deadlock cycle.
596
597Quick Quiz #1: Why is this argument naive? How could a deadlock
598 occur when using this algorithm in a real-world Linux
599 kernel? How could this deadlock be avoided?
600
601
6025B. "TOY" EXAMPLE #2: CLASSIC RCU
603
604This section presents a "toy" RCU implementation that is based on
605"classic RCU". It is also short on performance (but only for updates) and
606on features such as hotplug CPU and the ability to run in CONFIG_PREEMPT
607kernels. The definitions of rcu_dereference() and rcu_assign_pointer()
608are the same as those shown in the preceding section, so they are omitted.
609
610 void rcu_read_lock(void) { }
611
612 void rcu_read_unlock(void) { }
613
614 void synchronize_rcu(void)
615 {
616 int cpu;
617
KAMEZAWA Hiroyuki3c30a752006-03-28 01:56:39 -0800618 for_each_possible_cpu(cpu)
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700619 run_on(cpu);
620 }
621
622Note that rcu_read_lock() and rcu_read_unlock() do absolutely nothing.
623This is the great strength of classic RCU in a non-preemptive kernel:
624read-side overhead is precisely zero, at least on non-Alpha CPUs.
625And there is absolutely no way that rcu_read_lock() can possibly
626participate in a deadlock cycle!
627
628The implementation of synchronize_rcu() simply schedules itself on each
629CPU in turn. The run_on() primitive can be implemented straightforwardly
630in terms of the sched_setaffinity() primitive. Of course, a somewhat less
631"toy" implementation would restore the affinity upon completion rather
632than just leaving all tasks running on the last CPU, but when I said
633"toy", I meant -toy-!
634
635So how the heck is this supposed to work???
636
637Remember that it is illegal to block while in an RCU read-side critical
638section. Therefore, if a given CPU executes a context switch, we know
639that it must have completed all preceding RCU read-side critical sections.
640Once -all- CPUs have executed a context switch, then -all- preceding
641RCU read-side critical sections will have completed.
642
643So, suppose that we remove a data item from its structure and then invoke
644synchronize_rcu(). Once synchronize_rcu() returns, we are guaranteed
645that there are no RCU read-side critical sections holding a reference
646to that data item, so we can safely reclaim it.
647
648Quick Quiz #2: Give an example where Classic RCU's read-side
649 overhead is -negative-.
650
651Quick Quiz #3: If it is illegal to block in an RCU read-side
652 critical section, what the heck do you do in
653 PREEMPT_RT, where normal spinlocks can block???
654
655
6566. ANALOGY WITH READER-WRITER LOCKING
657
658Although RCU can be used in many different ways, a very common use of
659RCU is analogous to reader-writer locking. The following unified
660diff shows how closely related RCU and reader-writer locking can be.
661
662 @@ -13,15 +14,15 @@
663 struct list_head *lp;
664 struct el *p;
665
666 - read_lock();
667 - list_for_each_entry(p, head, lp) {
668 + rcu_read_lock();
669 + list_for_each_entry_rcu(p, head, lp) {
670 if (p->key == key) {
671 *result = p->data;
672 - read_unlock();
673 + rcu_read_unlock();
674 return 1;
675 }
676 }
677 - read_unlock();
678 + rcu_read_unlock();
679 return 0;
680 }
681
682 @@ -29,15 +30,16 @@
683 {
684 struct el *p;
685
686 - write_lock(&listmutex);
687 + spin_lock(&listmutex);
688 list_for_each_entry(p, head, lp) {
689 if (p->key == key) {
Urs Thuermann82a854e2006-07-10 04:44:06 -0700690 - list_del(&p->list);
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700691 - write_unlock(&listmutex);
Urs Thuermann82a854e2006-07-10 04:44:06 -0700692 + list_del_rcu(&p->list);
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700693 + spin_unlock(&listmutex);
694 + synchronize_rcu();
695 kfree(p);
696 return 1;
697 }
698 }
699 - write_unlock(&listmutex);
700 + spin_unlock(&listmutex);
701 return 0;
702 }
703
704Or, for those who prefer a side-by-side listing:
705
706 1 struct el { 1 struct el {
707 2 struct list_head list; 2 struct list_head list;
708 3 long key; 3 long key;
709 4 spinlock_t mutex; 4 spinlock_t mutex;
710 5 int data; 5 int data;
711 6 /* Other data fields */ 6 /* Other data fields */
712 7 }; 7 };
713 8 spinlock_t listmutex; 8 spinlock_t listmutex;
714 9 struct el head; 9 struct el head;
715
716 1 int search(long key, int *result) 1 int search(long key, int *result)
717 2 { 2 {
718 3 struct list_head *lp; 3 struct list_head *lp;
719 4 struct el *p; 4 struct el *p;
720 5 5
721 6 read_lock(); 6 rcu_read_lock();
722 7 list_for_each_entry(p, head, lp) { 7 list_for_each_entry_rcu(p, head, lp) {
723 8 if (p->key == key) { 8 if (p->key == key) {
724 9 *result = p->data; 9 *result = p->data;
72510 read_unlock(); 10 rcu_read_unlock();
72611 return 1; 11 return 1;
72712 } 12 }
72813 } 13 }
72914 read_unlock(); 14 rcu_read_unlock();
73015 return 0; 15 return 0;
73116 } 16 }
732
733 1 int delete(long key) 1 int delete(long key)
734 2 { 2 {
735 3 struct el *p; 3 struct el *p;
736 4 4
737 5 write_lock(&listmutex); 5 spin_lock(&listmutex);
738 6 list_for_each_entry(p, head, lp) { 6 list_for_each_entry(p, head, lp) {
739 7 if (p->key == key) { 7 if (p->key == key) {
Urs Thuermann82a854e2006-07-10 04:44:06 -0700740 8 list_del(&p->list); 8 list_del_rcu(&p->list);
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700741 9 write_unlock(&listmutex); 9 spin_unlock(&listmutex);
742 10 synchronize_rcu();
74310 kfree(p); 11 kfree(p);
74411 return 1; 12 return 1;
74512 } 13 }
74613 } 14 }
74714 write_unlock(&listmutex); 15 spin_unlock(&listmutex);
74815 return 0; 16 return 0;
74916 } 17 }
750
751Either way, the differences are quite small. Read-side locking moves
752to rcu_read_lock() and rcu_read_unlock, update-side locking moves from
Paolo Ornati670e9f32006-10-03 22:57:56 +0200753a reader-writer lock to a simple spinlock, and a synchronize_rcu()
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700754precedes the kfree().
755
756However, there is one potential catch: the read-side and update-side
757critical sections can now run concurrently. In many cases, this will
758not be a problem, but it is necessary to check carefully regardless.
759For example, if multiple independent list updates must be seen as
760a single atomic update, converting to RCU will require special care.
761
762Also, the presence of synchronize_rcu() means that the RCU version of
763delete() can now block. If this is a problem, there is a callback-based
764mechanism that never blocks, namely call_rcu(), that can be used in
765place of synchronize_rcu().
766
767
7687. FULL LIST OF RCU APIs
769
770The RCU APIs are documented in docbook-format header comments in the
771Linux-kernel source code, but it helps to have a full list of the
772APIs, since there does not appear to be a way to categorize them
773in docbook. Here is the list, by category.
774
775Markers for RCU read-side critical sections:
776
777 rcu_read_lock
778 rcu_read_unlock
779 rcu_read_lock_bh
780 rcu_read_unlock_bh
Paul E. McKenney621934e2006-10-04 02:17:02 -0700781 srcu_read_lock
782 srcu_read_unlock
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700783
784RCU pointer/list traversal:
785
786 rcu_dereference
787 list_for_each_rcu (to be deprecated in favor of
788 list_for_each_entry_rcu)
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700789 list_for_each_entry_rcu
790 list_for_each_continue_rcu (to be deprecated in favor of new
791 list_for_each_entry_continue_rcu)
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700792 hlist_for_each_entry_rcu
793
794RCU pointer update:
795
796 rcu_assign_pointer
797 list_add_rcu
798 list_add_tail_rcu
799 list_del_rcu
800 list_replace_rcu
801 hlist_del_rcu
802 hlist_add_head_rcu
803
804RCU grace period:
805
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700806 synchronize_net
807 synchronize_sched
808 synchronize_rcu
Paul E. McKenney621934e2006-10-04 02:17:02 -0700809 synchronize_srcu
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700810 call_rcu
811 call_rcu_bh
812
813See the comment headers in the source code (or the docbook generated
814from them) for more information.
815
816
8178. ANSWERS TO QUICK QUIZZES
818
819Quick Quiz #1: Why is this argument naive? How could a deadlock
820 occur when using this algorithm in a real-world Linux
821 kernel? [Referring to the lock-based "toy" RCU
822 algorithm.]
823
824Answer: Consider the following sequence of events:
825
826 1. CPU 0 acquires some unrelated lock, call it
Paul E. McKenneyd19720a2006-02-01 03:06:42 -0800827 "problematic_lock", disabling irq via
828 spin_lock_irqsave().
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700829
830 2. CPU 1 enters synchronize_rcu(), write-acquiring
831 rcu_gp_mutex.
832
833 3. CPU 0 enters rcu_read_lock(), but must wait
834 because CPU 1 holds rcu_gp_mutex.
835
836 4. CPU 1 is interrupted, and the irq handler
837 attempts to acquire problematic_lock.
838
839 The system is now deadlocked.
840
841 One way to avoid this deadlock is to use an approach like
842 that of CONFIG_PREEMPT_RT, where all normal spinlocks
843 become blocking locks, and all irq handlers execute in
844 the context of special tasks. In this case, in step 4
845 above, the irq handler would block, allowing CPU 1 to
846 release rcu_gp_mutex, avoiding the deadlock.
847
848 Even in the absence of deadlock, this RCU implementation
849 allows latency to "bleed" from readers to other
850 readers through synchronize_rcu(). To see this,
851 consider task A in an RCU read-side critical section
852 (thus read-holding rcu_gp_mutex), task B blocked
853 attempting to write-acquire rcu_gp_mutex, and
854 task C blocked in rcu_read_lock() attempting to
855 read_acquire rcu_gp_mutex. Task A's RCU read-side
856 latency is holding up task C, albeit indirectly via
857 task B.
858
859 Realtime RCU implementations therefore use a counter-based
860 approach where tasks in RCU read-side critical sections
861 cannot be blocked by tasks executing synchronize_rcu().
862
863Quick Quiz #2: Give an example where Classic RCU's read-side
864 overhead is -negative-.
865
866Answer: Imagine a single-CPU system with a non-CONFIG_PREEMPT
867 kernel where a routing table is used by process-context
868 code, but can be updated by irq-context code (for example,
869 by an "ICMP REDIRECT" packet). The usual way of handling
870 this would be to have the process-context code disable
871 interrupts while searching the routing table. Use of
872 RCU allows such interrupt-disabling to be dispensed with.
873 Thus, without RCU, you pay the cost of disabling interrupts,
874 and with RCU you don't.
875
876 One can argue that the overhead of RCU in this
877 case is negative with respect to the single-CPU
878 interrupt-disabling approach. Others might argue that
879 the overhead of RCU is merely zero, and that replacing
880 the positive overhead of the interrupt-disabling scheme
881 with the zero-overhead RCU scheme does not constitute
882 negative overhead.
883
884 In real life, of course, things are more complex. But
885 even the theoretical possibility of negative overhead for
886 a synchronization primitive is a bit unexpected. ;-)
887
888Quick Quiz #3: If it is illegal to block in an RCU read-side
889 critical section, what the heck do you do in
890 PREEMPT_RT, where normal spinlocks can block???
891
892Answer: Just as PREEMPT_RT permits preemption of spinlock
893 critical sections, it permits preemption of RCU
894 read-side critical sections. It also permits
895 spinlocks blocking while in RCU read-side critical
896 sections.
897
898 Why the apparent inconsistency? Because it is it
899 possible to use priority boosting to keep the RCU
900 grace periods short if need be (for example, if running
901 short of memory). In contrast, if blocking waiting
902 for (say) network reception, there is no way to know
903 what should be boosted. Especially given that the
904 process we need to boost might well be a human being
905 who just went out for a pizza or something. And although
906 a computer-operated cattle prod might arouse serious
907 interest, it might also provoke serious objections.
908 Besides, how does the computer know what pizza parlor
909 the human being went to???
910
911
912ACKNOWLEDGEMENTS
913
914My thanks to the people who helped make this human-readable, including
Paul E. McKenneyd19720a2006-02-01 03:06:42 -0800915Jon Walpole, Josh Triplett, Serge Hallyn, Suzanne Wood, and Alan Stern.
Paul E. McKenneydd81eca2005-09-10 00:26:24 -0700916
917
918For more information, see http://www.rdrop.com/users/paulmck/RCU.