virtiofsd: Use --thread-pool-size=0 to mean no thread pool
[qemu/kevin.git] / util / rcu.c
blob13ac0f75cb2a7de3b61f13174fb545b8fd08e314
1 /*
2 * urcu-mb.c
4 * Userspace RCU library with explicit memory barriers
6 * Copyright (c) 2009 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
7 * Copyright (c) 2009 Paul E. McKenney, IBM Corporation.
8 * Copyright 2015 Red Hat, Inc.
10 * Ported to QEMU by Paolo Bonzini <pbonzini@redhat.com>
12 * This library is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU Lesser General Public
14 * License as published by the Free Software Foundation; either
15 * version 2.1 of the License, or (at your option) any later version.
17 * This library is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 * Lesser General Public License for more details.
22 * You should have received a copy of the GNU Lesser General Public
23 * License along with this library; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
26 * IBM's contributions to this file may be relicensed under LGPLv2 or later.
29 #include "qemu/osdep.h"
30 #include "qemu/rcu.h"
31 #include "qemu/atomic.h"
32 #include "qemu/thread.h"
33 #include "qemu/main-loop.h"
34 #include "qemu/lockable.h"
35 #if defined(CONFIG_MALLOC_TRIM)
36 #include <malloc.h>
37 #endif
40 * Global grace period counter. Bit 0 is always one in rcu_gp_ctr.
41 * Bits 1 and above are defined in synchronize_rcu.
43 #define RCU_GP_LOCKED (1UL << 0)
44 #define RCU_GP_CTR (1UL << 1)
46 unsigned long rcu_gp_ctr = RCU_GP_LOCKED;
48 QemuEvent rcu_gp_event;
49 static QemuMutex rcu_registry_lock;
50 static QemuMutex rcu_sync_lock;
53 * Check whether a quiescent state was crossed between the beginning of
54 * update_counter_and_wait and now.
56 static inline int rcu_gp_ongoing(unsigned long *ctr)
58 unsigned long v;
60 v = qatomic_read(ctr);
61 return v && (v != rcu_gp_ctr);
64 /* Written to only by each individual reader. Read by both the reader and the
65 * writers.
67 __thread struct rcu_reader_data rcu_reader;
69 /* Protected by rcu_registry_lock. */
70 typedef QLIST_HEAD(, rcu_reader_data) ThreadList;
71 static ThreadList registry = QLIST_HEAD_INITIALIZER(registry);
73 /* Wait for previous parity/grace period to be empty of readers. */
74 static void wait_for_readers(void)
76 ThreadList qsreaders = QLIST_HEAD_INITIALIZER(qsreaders);
77 struct rcu_reader_data *index, *tmp;
79 for (;;) {
80 /* We want to be notified of changes made to rcu_gp_ongoing
81 * while we walk the list.
83 qemu_event_reset(&rcu_gp_event);
85 /* Instead of using qatomic_mb_set for index->waiting, and
86 * qatomic_mb_read for index->ctr, memory barriers are placed
87 * manually since writes to different threads are independent.
88 * qemu_event_reset has acquire semantics, so no memory barrier
89 * is needed here.
91 QLIST_FOREACH(index, &registry, node) {
92 qatomic_set(&index->waiting, true);
95 /* Here, order the stores to index->waiting before the loads of
96 * index->ctr. Pairs with smp_mb_placeholder() in rcu_read_unlock(),
97 * ensuring that the loads of index->ctr are sequentially consistent.
99 smp_mb_global();
101 QLIST_FOREACH_SAFE(index, &registry, node, tmp) {
102 if (!rcu_gp_ongoing(&index->ctr)) {
103 QLIST_REMOVE(index, node);
104 QLIST_INSERT_HEAD(&qsreaders, index, node);
106 /* No need for mb_set here, worst of all we
107 * get some extra futex wakeups.
109 qatomic_set(&index->waiting, false);
113 if (QLIST_EMPTY(&registry)) {
114 break;
117 /* Wait for one thread to report a quiescent state and try again.
118 * Release rcu_registry_lock, so rcu_(un)register_thread() doesn't
119 * wait too much time.
121 * rcu_register_thread() may add nodes to &registry; it will not
122 * wake up synchronize_rcu, but that is okay because at least another
123 * thread must exit its RCU read-side critical section before
124 * synchronize_rcu is done. The next iteration of the loop will
125 * move the new thread's rcu_reader from &registry to &qsreaders,
126 * because rcu_gp_ongoing() will return false.
128 * rcu_unregister_thread() may remove nodes from &qsreaders instead
129 * of &registry if it runs during qemu_event_wait. That's okay;
130 * the node then will not be added back to &registry by QLIST_SWAP
131 * below. The invariant is that the node is part of one list when
132 * rcu_registry_lock is released.
134 qemu_mutex_unlock(&rcu_registry_lock);
135 qemu_event_wait(&rcu_gp_event);
136 qemu_mutex_lock(&rcu_registry_lock);
139 /* put back the reader list in the registry */
140 QLIST_SWAP(&registry, &qsreaders, node);
143 void synchronize_rcu(void)
145 QEMU_LOCK_GUARD(&rcu_sync_lock);
147 /* Write RCU-protected pointers before reading p_rcu_reader->ctr.
148 * Pairs with smp_mb_placeholder() in rcu_read_lock().
150 smp_mb_global();
152 QEMU_LOCK_GUARD(&rcu_registry_lock);
153 if (!QLIST_EMPTY(&registry)) {
154 /* In either case, the qatomic_mb_set below blocks stores that free
155 * old RCU-protected pointers.
157 if (sizeof(rcu_gp_ctr) < 8) {
158 /* For architectures with 32-bit longs, a two-subphases algorithm
159 * ensures we do not encounter overflow bugs.
161 * Switch parity: 0 -> 1, 1 -> 0.
163 qatomic_mb_set(&rcu_gp_ctr, rcu_gp_ctr ^ RCU_GP_CTR);
164 wait_for_readers();
165 qatomic_mb_set(&rcu_gp_ctr, rcu_gp_ctr ^ RCU_GP_CTR);
166 } else {
167 /* Increment current grace period. */
168 qatomic_mb_set(&rcu_gp_ctr, rcu_gp_ctr + RCU_GP_CTR);
171 wait_for_readers();
176 #define RCU_CALL_MIN_SIZE 30
178 /* Multi-producer, single-consumer queue based on urcu/static/wfqueue.h
179 * from liburcu. Note that head is only used by the consumer.
181 static struct rcu_head dummy;
182 static struct rcu_head *head = &dummy, **tail = &dummy.next;
183 static int rcu_call_count;
184 static QemuEvent rcu_call_ready_event;
186 static void enqueue(struct rcu_head *node)
188 struct rcu_head **old_tail;
190 node->next = NULL;
191 old_tail = qatomic_xchg(&tail, &node->next);
192 qatomic_mb_set(old_tail, node);
195 static struct rcu_head *try_dequeue(void)
197 struct rcu_head *node, *next;
199 retry:
200 /* Test for an empty list, which we do not expect. Note that for
201 * the consumer head and tail are always consistent. The head
202 * is consistent because only the consumer reads/writes it.
203 * The tail, because it is the first step in the enqueuing.
204 * It is only the next pointers that might be inconsistent.
206 if (head == &dummy && qatomic_mb_read(&tail) == &dummy.next) {
207 abort();
210 /* If the head node has NULL in its next pointer, the value is
211 * wrong and we need to wait until its enqueuer finishes the update.
213 node = head;
214 next = qatomic_mb_read(&head->next);
215 if (!next) {
216 return NULL;
219 /* Since we are the sole consumer, and we excluded the empty case
220 * above, the queue will always have at least two nodes: the
221 * dummy node, and the one being removed. So we do not need to update
222 * the tail pointer.
224 head = next;
226 /* If we dequeued the dummy node, add it back at the end and retry. */
227 if (node == &dummy) {
228 enqueue(node);
229 goto retry;
232 return node;
235 static void *call_rcu_thread(void *opaque)
237 struct rcu_head *node;
239 rcu_register_thread();
241 for (;;) {
242 int tries = 0;
243 int n = qatomic_read(&rcu_call_count);
245 /* Heuristically wait for a decent number of callbacks to pile up.
246 * Fetch rcu_call_count now, we only must process elements that were
247 * added before synchronize_rcu() starts.
249 while (n == 0 || (n < RCU_CALL_MIN_SIZE && ++tries <= 5)) {
250 g_usleep(10000);
251 if (n == 0) {
252 qemu_event_reset(&rcu_call_ready_event);
253 n = qatomic_read(&rcu_call_count);
254 if (n == 0) {
255 #if defined(CONFIG_MALLOC_TRIM)
256 malloc_trim(4 * 1024 * 1024);
257 #endif
258 qemu_event_wait(&rcu_call_ready_event);
261 n = qatomic_read(&rcu_call_count);
264 qatomic_sub(&rcu_call_count, n);
265 synchronize_rcu();
266 qemu_mutex_lock_iothread();
267 while (n > 0) {
268 node = try_dequeue();
269 while (!node) {
270 qemu_mutex_unlock_iothread();
271 qemu_event_reset(&rcu_call_ready_event);
272 node = try_dequeue();
273 if (!node) {
274 qemu_event_wait(&rcu_call_ready_event);
275 node = try_dequeue();
277 qemu_mutex_lock_iothread();
280 n--;
281 node->func(node);
283 qemu_mutex_unlock_iothread();
285 abort();
288 void call_rcu1(struct rcu_head *node, void (*func)(struct rcu_head *node))
290 node->func = func;
291 enqueue(node);
292 qatomic_inc(&rcu_call_count);
293 qemu_event_set(&rcu_call_ready_event);
297 struct rcu_drain {
298 struct rcu_head rcu;
299 QemuEvent drain_complete_event;
302 static void drain_rcu_callback(struct rcu_head *node)
304 struct rcu_drain *event = (struct rcu_drain *)node;
305 qemu_event_set(&event->drain_complete_event);
309 * This function ensures that all pending RCU callbacks
310 * on the current thread are done executing
312 * drops big qemu lock during the wait to allow RCU thread
313 * to process the callbacks
317 void drain_call_rcu(void)
319 struct rcu_drain rcu_drain;
320 bool locked = qemu_mutex_iothread_locked();
322 memset(&rcu_drain, 0, sizeof(struct rcu_drain));
323 qemu_event_init(&rcu_drain.drain_complete_event, false);
325 if (locked) {
326 qemu_mutex_unlock_iothread();
331 * RCU callbacks are invoked in the same order as in which they
332 * are registered, thus we can be sure that when 'drain_rcu_callback'
333 * is called, all RCU callbacks that were registered on this thread
334 * prior to calling this function are completed.
336 * Note that since we have only one global queue of the RCU callbacks,
337 * we also end up waiting for most of RCU callbacks that were registered
338 * on the other threads, but this is a side effect that shoudn't be
339 * assumed.
342 call_rcu1(&rcu_drain.rcu, drain_rcu_callback);
343 qemu_event_wait(&rcu_drain.drain_complete_event);
345 if (locked) {
346 qemu_mutex_lock_iothread();
351 void rcu_register_thread(void)
353 assert(rcu_reader.ctr == 0);
354 qemu_mutex_lock(&rcu_registry_lock);
355 QLIST_INSERT_HEAD(&registry, &rcu_reader, node);
356 qemu_mutex_unlock(&rcu_registry_lock);
359 void rcu_unregister_thread(void)
361 qemu_mutex_lock(&rcu_registry_lock);
362 QLIST_REMOVE(&rcu_reader, node);
363 qemu_mutex_unlock(&rcu_registry_lock);
366 static void rcu_init_complete(void)
368 QemuThread thread;
370 qemu_mutex_init(&rcu_registry_lock);
371 qemu_mutex_init(&rcu_sync_lock);
372 qemu_event_init(&rcu_gp_event, true);
374 qemu_event_init(&rcu_call_ready_event, false);
376 /* The caller is assumed to have iothread lock, so the call_rcu thread
377 * must have been quiescent even after forking, just recreate it.
379 qemu_thread_create(&thread, "call_rcu", call_rcu_thread,
380 NULL, QEMU_THREAD_DETACHED);
382 rcu_register_thread();
385 static int atfork_depth = 1;
387 void rcu_enable_atfork(void)
389 atfork_depth++;
392 void rcu_disable_atfork(void)
394 atfork_depth--;
397 #ifdef CONFIG_POSIX
398 static void rcu_init_lock(void)
400 if (atfork_depth < 1) {
401 return;
404 qemu_mutex_lock(&rcu_sync_lock);
405 qemu_mutex_lock(&rcu_registry_lock);
408 static void rcu_init_unlock(void)
410 if (atfork_depth < 1) {
411 return;
414 qemu_mutex_unlock(&rcu_registry_lock);
415 qemu_mutex_unlock(&rcu_sync_lock);
418 static void rcu_init_child(void)
420 if (atfork_depth < 1) {
421 return;
424 memset(&registry, 0, sizeof(registry));
425 rcu_init_complete();
427 #endif
429 static void __attribute__((__constructor__)) rcu_init(void)
431 smp_mb_global_init();
432 #ifdef CONFIG_POSIX
433 pthread_atfork(rcu_init_lock, rcu_init_unlock, rcu_init_child);
434 #endif
435 rcu_init_complete();