xref: /openbmc/qemu/docs/devel/rcu.txt (revision e6e03dcf)
1Using RCU (Read-Copy-Update) for synchronization
2================================================
3
4Read-copy update (RCU) is a synchronization mechanism that is used to
5protect read-mostly data structures.  RCU is very efficient and scalable
6on the read side (it is wait-free), and thus can make the read paths
7extremely fast.
8
9RCU supports concurrency between a single writer and multiple readers,
10thus it is not used alone.  Typically, the write-side will use a lock to
11serialize multiple updates, but other approaches are possible (e.g.,
12restricting updates to a single task).  In QEMU, when a lock is used,
13this will often be the "iothread mutex", also known as the "big QEMU
14lock" (BQL).  Also, restricting updates to a single task is done in
15QEMU using the "bottom half" API.
16
17RCU is fundamentally a "wait-to-finish" mechanism.  The read side marks
18sections of code with "critical sections", and the update side will wait
19for the execution of all *currently running* critical sections before
20proceeding, or before asynchronously executing a callback.
21
22The key point here is that only the currently running critical sections
23are waited for; critical sections that are started _after_ the beginning
24of the wait do not extend the wait, despite running concurrently with
25the updater.  This is the reason why RCU is more scalable than,
26for example, reader-writer locks.  It is so much more scalable that
27the system will have a single instance of the RCU mechanism; a single
28mechanism can be used for an arbitrary number of "things", without
29having to worry about things such as contention or deadlocks.
30
31How is this possible?  The basic idea is to split updates in two phases,
32"removal" and "reclamation".  During removal, we ensure that subsequent
33readers will not be able to get a reference to the old data.  After
34removal has completed, a critical section will not be able to access
35the old data.  Therefore, critical sections that begin after removal
36do not matter; as soon as all previous critical sections have finished,
37there cannot be any readers who hold references to the data structure,
38and these can now be safely reclaimed (e.g., freed or unref'ed).
39
40Here is a picture:
41
42        thread 1                  thread 2                  thread 3
43    -------------------    ------------------------    -------------------
44    enter RCU crit.sec.
45           |                finish removal phase
46           |                begin wait
47           |                      |                    enter RCU crit.sec.
48    exit RCU crit.sec             |                           |
49                            complete wait                     |
50                            begin reclamation phase           |
51                                                       exit RCU crit.sec.
52
53
54Note how thread 3 is still executing its critical section when thread 2
55starts reclaiming data.  This is possible, because the old version of the
56data structure was not accessible at the time thread 3 began executing
57that critical section.
58
59
60RCU API
61=======
62
63The core RCU API is small:
64
65     void rcu_read_lock(void);
66
67        Used by a reader to inform the reclaimer that the reader is
68        entering an RCU read-side critical section.
69
70     void rcu_read_unlock(void);
71
72        Used by a reader to inform the reclaimer that the reader is
73        exiting an RCU read-side critical section.  Note that RCU
74        read-side critical sections may be nested and/or overlapping.
75
76     void synchronize_rcu(void);
77
78        Blocks until all pre-existing RCU read-side critical sections
79        on all threads have completed.  This marks the end of the removal
80        phase and the beginning of reclamation phase.
81
82        Note that it would be valid for another update to come while
83        synchronize_rcu is running.  Because of this, it is better that
84        the updater releases any locks it may hold before calling
85        synchronize_rcu.  If this is not possible (for example, because
86        the updater is protected by the BQL), you can use call_rcu.
87
88     void call_rcu1(struct rcu_head * head,
89                    void (*func)(struct rcu_head *head));
90
91        This function invokes func(head) after all pre-existing RCU
92        read-side critical sections on all threads have completed.  This
93        marks the end of the removal phase, with func taking care
94        asynchronously of the reclamation phase.
95
96        The foo struct needs to have an rcu_head structure added,
97        perhaps as follows:
98
99            struct foo {
100                struct rcu_head rcu;
101                int a;
102                char b;
103                long c;
104            };
105
106        so that the reclaimer function can fetch the struct foo address
107        and free it:
108
109            call_rcu1(&foo.rcu, foo_reclaim);
110
111            void foo_reclaim(struct rcu_head *rp)
112            {
113                struct foo *fp = container_of(rp, struct foo, rcu);
114                g_free(fp);
115            }
116
117        For the common case where the rcu_head member is the first of the
118        struct, you can use the following macro.
119
120     void call_rcu(T *p,
121                   void (*func)(T *p),
122                   field-name);
123     void g_free_rcu(T *p,
124                     field-name);
125
126        call_rcu1 is typically used through these macro, in the common case
127        where the "struct rcu_head" is the first field in the struct.  If
128        the callback function is g_free, in particular, g_free_rcu can be
129        used.  In the above case, one could have written simply:
130
131            g_free_rcu(&foo, rcu);
132
133     typeof(*p) atomic_rcu_read(p);
134
135        atomic_rcu_read() is similar to atomic_mb_read(), but it makes
136        some assumptions on the code that calls it.  This allows a more
137        optimized implementation.
138
139        atomic_rcu_read assumes that whenever a single RCU critical
140        section reads multiple shared data, these reads are either
141        data-dependent or need no ordering.  This is almost always the
142        case when using RCU, because read-side critical sections typically
143        navigate one or more pointers (the pointers that are changed on
144        every update) until reaching a data structure of interest,
145        and then read from there.
146
147        RCU read-side critical sections must use atomic_rcu_read() to
148        read data, unless concurrent writes are prevented by another
149        synchronization mechanism.
150
151        Furthermore, RCU read-side critical sections should traverse the
152        data structure in a single direction, opposite to the direction
153        in which the updater initializes it.
154
155     void atomic_rcu_set(p, typeof(*p) v);
156
157        atomic_rcu_set() is also similar to atomic_mb_set(), and it also
158        makes assumptions on the code that calls it in order to allow a more
159        optimized implementation.
160
161        In particular, atomic_rcu_set() suffices for synchronization
162        with readers, if the updater never mutates a field within a
163        data item that is already accessible to readers.  This is the
164        case when initializing a new copy of the RCU-protected data
165        structure; just ensure that initialization of *p is carried out
166        before atomic_rcu_set() makes the data item visible to readers.
167        If this rule is observed, writes will happen in the opposite
168        order as reads in the RCU read-side critical sections (or if
169        there is just one update), and there will be no need for other
170        synchronization mechanism to coordinate the accesses.
171
172The following APIs must be used before RCU is used in a thread:
173
174     void rcu_register_thread(void);
175
176        Mark a thread as taking part in the RCU mechanism.  Such a thread
177        will have to report quiescent points regularly, either manually
178        or through the QemuCond/QemuSemaphore/QemuEvent APIs.
179
180     void rcu_unregister_thread(void);
181
182        Mark a thread as not taking part anymore in the RCU mechanism.
183        It is not a problem if such a thread reports quiescent points,
184        either manually or by using the QemuCond/QemuSemaphore/QemuEvent
185        APIs.
186
187Note that these APIs are relatively heavyweight, and should _not_ be
188nested.
189
190Convenience macros
191==================
192
193Two macros are provided that automatically release the read lock at the
194end of the scope.
195
196      RCU_READ_LOCK_GUARD()
197
198         Takes the lock and will release it at the end of the block it's
199         used in.
200
201      WITH_RCU_READ_LOCK_GUARD()  { code }
202
203         Is used at the head of a block to protect the code within the block.
204
205Note that 'goto'ing out of the guarded block will also drop the lock.
206
207DIFFERENCES WITH LINUX
208======================
209
210- Waiting on a mutex is possible, though discouraged, within an RCU critical
211  section.  This is because spinlocks are rarely (if ever) used in userspace
212  programming; not allowing this would prevent upgrading an RCU read-side
213  critical section to become an updater.
214
215- atomic_rcu_read and atomic_rcu_set replace rcu_dereference and
216  rcu_assign_pointer.  They take a _pointer_ to the variable being accessed.
217
218- call_rcu is a macro that has an extra argument (the name of the first
219  field in the struct, which must be a struct rcu_head), and expects the
220  type of the callback's argument to be the type of the first argument.
221  call_rcu1 is the same as Linux's call_rcu.
222
223
224RCU PATTERNS
225============
226
227Many patterns using read-writer locks translate directly to RCU, with
228the advantages of higher scalability and deadlock immunity.
229
230In general, RCU can be used whenever it is possible to create a new
231"version" of a data structure every time the updater runs.  This may
232sound like a very strict restriction, however:
233
234- the updater does not mean "everything that writes to a data structure",
235  but rather "everything that involves a reclamation step".  See the
236  array example below
237
238- in some cases, creating a new version of a data structure may actually
239  be very cheap.  For example, modifying the "next" pointer of a singly
240  linked list is effectively creating a new version of the list.
241
242Here are some frequently-used RCU idioms that are worth noting.
243
244
245RCU list processing
246-------------------
247
248TBD (not yet used in QEMU)
249
250
251RCU reference counting
252----------------------
253
254Because grace periods are not allowed to complete while there is an RCU
255read-side critical section in progress, the RCU read-side primitives
256may be used as a restricted reference-counting mechanism.  For example,
257consider the following code fragment:
258
259    rcu_read_lock();
260    p = atomic_rcu_read(&foo);
261    /* do something with p. */
262    rcu_read_unlock();
263
264The RCU read-side critical section ensures that the value of "p" remains
265valid until after the rcu_read_unlock().  In some sense, it is acquiring
266a reference to p that is later released when the critical section ends.
267The write side looks simply like this (with appropriate locking):
268
269    qemu_mutex_lock(&foo_mutex);
270    old = foo;
271    atomic_rcu_set(&foo, new);
272    qemu_mutex_unlock(&foo_mutex);
273    synchronize_rcu();
274    free(old);
275
276If the processing cannot be done purely within the critical section, it
277is possible to combine this idiom with a "real" reference count:
278
279    rcu_read_lock();
280    p = atomic_rcu_read(&foo);
281    foo_ref(p);
282    rcu_read_unlock();
283    /* do something with p. */
284    foo_unref(p);
285
286The write side can be like this:
287
288    qemu_mutex_lock(&foo_mutex);
289    old = foo;
290    atomic_rcu_set(&foo, new);
291    qemu_mutex_unlock(&foo_mutex);
292    synchronize_rcu();
293    foo_unref(old);
294
295or with call_rcu:
296
297    qemu_mutex_lock(&foo_mutex);
298    old = foo;
299    atomic_rcu_set(&foo, new);
300    qemu_mutex_unlock(&foo_mutex);
301    call_rcu(foo_unref, old, rcu);
302
303In both cases, the write side only performs removal.  Reclamation
304happens when the last reference to a "foo" object is dropped.
305Using synchronize_rcu() is undesirably expensive, because the
306last reference may be dropped on the read side.  Hence you can
307use call_rcu() instead:
308
309     foo_unref(struct foo *p) {
310        if (atomic_fetch_dec(&p->refcount) == 1) {
311            call_rcu(foo_destroy, p, rcu);
312        }
313    }
314
315
316Note that the same idioms would be possible with reader/writer
317locks:
318
319    read_lock(&foo_rwlock);         write_mutex_lock(&foo_rwlock);
320    p = foo;                        p = foo;
321    /* do something with p. */      foo = new;
322    read_unlock(&foo_rwlock);       free(p);
323                                    write_mutex_unlock(&foo_rwlock);
324                                    free(p);
325
326    ------------------------------------------------------------------
327
328    read_lock(&foo_rwlock);         write_mutex_lock(&foo_rwlock);
329    p = foo;                        old = foo;
330    foo_ref(p);                     foo = new;
331    read_unlock(&foo_rwlock);       foo_unref(old);
332    /* do something with p. */      write_mutex_unlock(&foo_rwlock);
333    read_lock(&foo_rwlock);
334    foo_unref(p);
335    read_unlock(&foo_rwlock);
336
337foo_unref could use a mechanism such as bottom halves to move deallocation
338out of the write-side critical section.
339
340
341RCU resizable arrays
342--------------------
343
344Resizable arrays can be used with RCU.  The expensive RCU synchronization
345(or call_rcu) only needs to take place when the array is resized.
346The two items to take care of are:
347
348- ensuring that the old version of the array is available between removal
349  and reclamation;
350
351- avoiding mismatches in the read side between the array data and the
352  array size.
353
354The first problem is avoided simply by not using realloc.  Instead,
355each resize will allocate a new array and copy the old data into it.
356The second problem would arise if the size and the data pointers were
357two members of a larger struct:
358
359    struct mystuff {
360        ...
361        int data_size;
362        int data_alloc;
363        T   *data;
364        ...
365    };
366
367Instead, we store the size of the array with the array itself:
368
369    struct arr {
370        int size;
371        int alloc;
372        T   data[];
373    };
374    struct arr *global_array;
375
376    read side:
377        rcu_read_lock();
378        struct arr *array = atomic_rcu_read(&global_array);
379        x = i < array->size ? array->data[i] : -1;
380        rcu_read_unlock();
381        return x;
382
383    write side (running under a lock):
384        if (global_array->size == global_array->alloc) {
385            /* Creating a new version.  */
386            new_array = g_malloc(sizeof(struct arr) +
387                                 global_array->alloc * 2 * sizeof(T));
388            new_array->size = global_array->size;
389            new_array->alloc = global_array->alloc * 2;
390            memcpy(new_array->data, global_array->data,
391                   global_array->alloc * sizeof(T));
392
393            /* Removal phase.  */
394            old_array = global_array;
395            atomic_rcu_set(&new_array->data, new_array);
396            synchronize_rcu();
397
398            /* Reclamation phase.  */
399            free(old_array);
400        }
401
402
403SOURCES
404=======
405
406* Documentation/RCU/ from the Linux kernel
407