1PTHREAD_COND_TIMEDWAIT(3P) POSIX Programmer's ManualPTHREAD_COND_TIMEDWAIT(3P)
2
6 This manual page is part of the POSIX Programmer's Manual. The Linux
7 implementation of this interface may differ (consult the corresponding
8 Linux manual page for details of Linux behavior), or the interface may
9 not be implemented on Linux.
10
12 pthread_cond_timedwait, pthread_cond_wait - wait on a condition
13
15 #include <pthread.h>
16 int pthread_cond_timedwait(pthread_cond_t *restrict cond,
18 pthread_mutex_t *restrict mutex,
19 const struct timespec *restrict abstime);
20 int pthread_cond_wait(pthread_cond_t *restrict cond,
21 pthread_mutex_t *restrict mutex);
22
25 The pthread_cond_timedwait() and pthread_cond_wait() functions shall
26 block on a condition variable. They shall be called with mutex locked
27 by the calling thread or undefined behavior results.
28 These functions atomically release mutex and cause the calling thread
30 to block on the condition variable cond; atomically here means "atomi‐
31 cally with respect to access by another thread to the mutex and then
32 the condition variable". That is, if another thread is able to acquire
33 the mutex after the about-to-block thread has released it, then a sub‐
34 sequent call to pthread_cond_broadcast() or pthread_cond_signal() in
35 that thread shall behave as if it were issued after the about-to-block
36 thread has blocked.
37 Upon successful return, the mutex shall have been locked and shall be
39 owned by the calling thread.
40 When using condition variables there is always a Boolean predicate
42 involving shared variables associated with each condition wait that is
43 true if the thread should proceed. Spurious wakeups from the
44 pthread_cond_timedwait() or pthread_cond_wait() functions may occur.
45 Since the return from pthread_cond_timedwait() or pthread_cond_wait()
46 does not imply anything about the value of this predicate, the predi‐
47 cate should be re-evaluated upon such return.
48 The effect of using more than one mutex for concurrent
50 pthread_cond_timedwait() or pthread_cond_wait() operations on the same
51 condition variable is undefined; that is, a condition variable becomes
52 bound to a unique mutex when a thread waits on the condition variable,
53 and this (dynamic) binding shall end when the wait returns.
54 A condition wait (whether timed or not) is a cancellation point. When
56 the cancelability enable state of a thread is set to PTHREAD_CAN‐
57 CEL_DEFERRED, a side effect of acting upon a cancellation request while
58 in a condition wait is that the mutex is (in effect) re-acquired before
59 calling the first cancellation cleanup handler. The effect is as if the
60 thread were unblocked, allowed to execute up to the point of returning
61 from the call to pthread_cond_timedwait() or pthread_cond_wait(), but
62 at that point notices the cancellation request and instead of returning
63 to the caller of pthread_cond_timedwait() or pthread_cond_wait(),
64 starts the thread cancellation activities, which includes calling can‐
65 cellation cleanup handlers.
66 A thread that has been unblocked because it has been canceled while
68 blocked in a call to pthread_cond_timedwait() or pthread_cond_wait()
69 shall not consume any condition signal that may be directed concur‐
70 rently at the condition variable if there are other threads blocked on
71 the condition variable.
72 The pthread_cond_timedwait() function shall be equivalent to
74 pthread_cond_wait(), except that an error is returned if the absolute
75 time specified by abstime passes (that is, system time equals or
76 exceeds abstime) before the condition cond is signaled or broadcasted,
77 or if the absolute time specified by abstime has already been passed at
78 the time of the call.
79 If the Clock Selection option is supported, the condition variable
81 shall have a clock attribute which specifies the clock that shall be
82 used to measure the time specified by the abstime argument. When such
83 timeouts occur, pthread_cond_timedwait() shall nonetheless release and
84 re-acquire the mutex referenced by mutex. The pthread_cond_timedwait()
85 function is also a cancellation point.
86 If a signal is delivered to a thread waiting for a condition variable,
88 upon return from the signal handler the thread resumes waiting for the
89 condition variable as if it was not interrupted, or it shall return
90 zero due to spurious wakeup.
91
93 Except in the case of [ETIMEDOUT], all these error checks shall act as
94 if they were performed immediately at the beginning of processing for
95 the function and shall cause an error return, in effect, prior to modi‐
96 fying the state of the mutex specified by mutex or the condition vari‐
97 able specified by cond.
98 Upon successful completion, a value of zero shall be returned; other‐
100 wise, an error number shall be returned to indicate the error.
101
103 The pthread_cond_timedwait() function shall fail if:
104 ETIMEDOUT
106 The time specified by abstime to pthread_cond_timedwait() has
107 passed.
108 The pthread_cond_timedwait() and pthread_cond_wait() functions may fail
111 if:
112 EINVAL The value specified by cond, mutex, or abstime is invalid.
114 EINVAL Different mutexes were supplied for concurrent
116 pthread_cond_timedwait() or pthread_cond_wait() operations on
117 the same condition variable.
118 EPERM The mutex was not owned by the current thread at the time of the
120 call.
121 These functions shall not return an error code of [EINTR].
124 The following sections are informative.
126
128 None.
129
131 None.
132
134 Condition Wait Semantics
135 It is important to note that when pthread_cond_wait() and
136 pthread_cond_timedwait() return without error, the associated predicate
137 may still be false. Similarly, when pthread_cond_timedwait() returns
138 with the timeout error, the associated predicate may be true due to an
139 unavoidable race between the expiration of the timeout and the predi‐
140 cate state change.
141 Some implementations, particularly on a multi-processor, may sometimes
143 cause multiple threads to wake up when the condition variable is sig‐
144 naled simultaneously on different processors.
145 In general, whenever a condition wait returns, the thread has to re-
147 evaluate the predicate associated with the condition wait to determine
148 whether it can safely proceed, should wait again, or should declare a
149 timeout. A return from the wait does not imply that the associated
150 predicate is either true or false.
151 It is thus recommended that a condition wait be enclosed in the equiva‐
153 lent of a "while loop" that checks the predicate.
154 Timed Wait Semantics
156 An absolute time measure was chosen for specifying the timeout parame‐
157 ter for two reasons. First, a relative time measure can be easily
158 implemented on top of a function that specifies absolute time, but
159 there is a race condition associated with specifying an absolute time‐
160 out on top of a function that specifies relative timeouts. For exam‐
161 ple, assume that clock_gettime() returns the current time and cond_rel‐
162 ative_timed_wait() uses relative timeouts:
163 clock_gettime(CLOCK_REALTIME, &now)
166 reltime = sleep_til_this_absolute_time -now;
167 cond_relative_timed_wait(c, m, &reltime);
168 If the thread is preempted between the first statement and the last
170 statement, the thread blocks for too long. Blocking, however, is irrel‐
171 evant if an absolute timeout is used. An absolute timeout also need not
172 be recomputed if it is used multiple times in a loop, such as that
173 enclosing a condition wait.
174 For cases when the system clock is advanced discontinuously by an oper‐
176 ator, it is expected that implementations process any timed wait expir‐
177 ing at an intervening time as if that time had actually occurred.
178 Cancellation and Condition Wait
180 A condition wait, whether timed or not, is a cancellation point. That
181 is, the functions pthread_cond_wait() or pthread_cond_timedwait() are
182 points where a pending (or concurrent) cancellation request is noticed.
183 The reason for this is that an indefinite wait is possible at these
184 points-whatever event is being waited for, even if the program is
185 totally correct, might never occur; for example, some input data being
186 awaited might never be sent. By making condition wait a cancellation
187 point, the thread can be canceled and perform its cancellation cleanup
188 handler even though it may be stuck in some indefinite wait.
189 A side effect of acting on a cancellation request while a thread is
191 blocked on a condition variable is to re-acquire the mutex before call‐
192 ing any of the cancellation cleanup handlers. This is done in order to
193 ensure that the cancellation cleanup handler is executed in the same
194 state as the critical code that lies both before and after the call to
195 the condition wait function. This rule is also required when interfac‐
196 ing to POSIX threads from languages, such as Ada or C++, which may
197 choose to map cancellation onto a language exception; this rule ensures
198 that each exception handler guarding a critical section can always
199 safely depend upon the fact that the associated mutex has already been
200 locked regardless of exactly where within the critical section the
201 exception was raised. Without this rule, there would not be a uniform
202 rule that exception handlers could follow regarding the lock, and so
203 coding would become very cumbersome.
204 Therefore, since some statement has to be made regarding the state of
206 the lock when a cancellation is delivered during a wait, a definition
207 has been chosen that makes application coding most convenient and error
208 free.
209 When acting on a cancellation request while a thread is blocked on a
211 condition variable, the implementation is required to ensure that the
212 thread does not consume any condition signals directed at that condi‐
213 tion variable if there are any other threads waiting on that condition
214 variable. This rule is specified in order to avoid deadlock conditions
215 that could occur if these two independent requests (one acting on a
216 thread and the other acting on the condition variable) were not pro‐
217 cessed independently.
218 Performance of Mutexes and Condition Variables
220 Mutexes are expected to be locked only for a few instructions. This
221 practice is almost automatically enforced by the desire of programmers
222 to avoid long serial regions of execution (which would reduce total
223 effective parallelism).
224 When using mutexes and condition variables, one tries to ensure that
226 the usual case is to lock the mutex, access shared data, and unlock the
227 mutex. Waiting on a condition variable should be a relatively rare sit‐
228 uation. For example, when implementing a read-write lock, code that
229 acquires a read-lock typically needs only to increment the count of
230 readers (under mutual-exclusion) and return. The calling thread would
231 actually wait on the condition variable only when there is already an
232 active writer. So the efficiency of a synchronization operation is
233 bounded by the cost of mutex lock/unlock and not by condition wait.
234 Note that in the usual case there is no context switch.
235 This is not to say that the efficiency of condition waiting is unimpor‐
237 tant. Since there needs to be at least one context switch per Ada ren‐
238 dezvous, the efficiency of waiting on a condition variable is impor‐
239 tant. The cost of waiting on a condition variable should be little more
240 than the minimal cost for a context switch plus the time to unlock and
241 lock the mutex.
242 Features of Mutexes and Condition Variables
244 It had been suggested that the mutex acquisition and release be decou‐
245 pled from condition wait. This was rejected because it is the combined
246 nature of the operation that, in fact, facilitates realtime implementa‐
247 tions. Those implementations can atomically move a high-priority thread
248 between the condition variable and the mutex in a manner that is trans‐
249 parent to the caller. This can prevent extra context switches and pro‐
250 vide more deterministic acquisition of a mutex when the waiting thread
251 is signaled. Thus, fairness and priority issues can be dealt with
252 directly by the scheduling discipline. Furthermore, the current condi‐
253 tion wait operation matches existing practice.
254 Scheduling Behavior of Mutexes and Condition Variables
256 Synchronization primitives that attempt to interfere with scheduling
257 policy by specifying an ordering rule are considered undesirable.
258 Threads waiting on mutexes and condition variables are selected to pro‐
259 ceed in an order dependent upon the scheduling policy rather than in
260 some fixed order (for example, FIFO or priority). Thus, the scheduling
261 policy determines which thread(s) are awakened and allowed to proceed.
262 Timed Condition Wait
264 The pthread_cond_timedwait() function allows an application to give up
265 waiting for a particular condition after a given amount of time. An
266 example of its use follows:
267 (void) pthread_mutex_lock(&t.mn);
270 t.waiters++;
271 clock_gettime(CLOCK_REALTIME, &ts);
272 ts.tv_sec += 5;
273 rc = 0;
274 while (! mypredicate(&t) && rc == 0)
275 rc = pthread_cond_timedwait(&t.cond, &t.mn, &ts);
276 t.waiters--;
277 if (rc == 0) setmystate(&t);
278 (void) pthread_mutex_unlock(&t.mn);
279 By making the timeout parameter absolute, it does not need to be recom‐
281 puted each time the program checks its blocking predicate. If the
282 timeout was relative, it would have to be recomputed before each call.
283 This would be especially difficult since such code would need to take
284 into account the possibility of extra wakeups that result from extra
285 broadcasts or signals on the condition variable that occur before
286 either the predicate is true or the timeout is due.
287
289 None.
290
292 pthread_cond_signal(), pthread_cond_broadcast(), the Base Definitions
293 volume of IEEE Std 1003.1-2001, <pthread.h>
294
296 Portions of this text are reprinted and reproduced in electronic form
297 from IEEE Std 1003.1, 2003 Edition, Standard for Information Technology
298 -- Portable Operating System Interface (POSIX), The Open Group Base
299 Specifications Issue 6, Copyright (C) 2001-2003 by the Institute of
300 Electrical and Electronics Engineers, Inc and The Open Group. In the
301 event of any discrepancy between this version and the original IEEE and
302 The Open Group Standard, the original IEEE and The Open Group Standard
303 is the referee document. The original Standard can be obtained online
304 at http://www.opengroup.org/unix/online.html .
305IEEE/The Open Group 2003 PTHREAD_COND_TIMEDWAIT(3P)