1PTHREAD_COND_TIMEDWAIT(P) POSIX Programmer's Manual PTHREAD_COND_TIMEDWAIT(P)
2
6 pthread_cond_timedwait, pthread_cond_wait - wait on a condition
7
9 #include <pthread.h>
10 int pthread_cond_timedwait(pthread_cond_t *restrict cond,
12 pthread_mutex_t *restrict mutex,
13 const struct timespec *restrict abstime);
14 int pthread_cond_wait(pthread_cond_t *restrict cond,
15 pthread_mutex_t *restrict mutex);
16
19 The pthread_cond_timedwait() and pthread_cond_wait() functions shall
20 block on a condition variable. They shall be called with mutex locked
21 by the calling thread or undefined behavior results.
22 These functions atomically release mutex and cause the calling thread
24 to block on the condition variable cond; atomically here means "atomi‐
25 cally with respect to access by another thread to the mutex and then
26 the condition variable". That is, if another thread is able to acquire
27 the mutex after the about-to-block thread has released it, then a sub‐
28 sequent call to pthread_cond_broadcast() or pthread_cond_signal() in
29 that thread shall behave as if it were issued after the about-to-block
30 thread has blocked.
31 Upon successful return, the mutex shall have been locked and shall be
33 owned by the calling thread.
34 When using condition variables there is always a Boolean predicate
36 involving shared variables associated with each condition wait that is
37 true if the thread should proceed. Spurious wakeups from the
38 pthread_cond_timedwait() or pthread_cond_wait() functions may occur.
39 Since the return from pthread_cond_timedwait() or pthread_cond_wait()
40 does not imply anything about the value of this predicate, the predi‐
41 cate should be re-evaluated upon such return.
42 The effect of using more than one mutex for concurrent
44 pthread_cond_timedwait() or pthread_cond_wait() operations on the same
45 condition variable is undefined; that is, a condition variable becomes
46 bound to a unique mutex when a thread waits on the condition variable,
47 and this (dynamic) binding shall end when the wait returns.
48 A condition wait (whether timed or not) is a cancellation point. When
50 the cancelability enable state of a thread is set to PTHREAD_CAN‐
51 CEL_DEFERRED, a side effect of acting upon a cancellation request while
52 in a condition wait is that the mutex is (in effect) re-acquired before
53 calling the first cancellation cleanup handler. The effect is as if the
54 thread were unblocked, allowed to execute up to the point of returning
55 from the call to pthread_cond_timedwait() or pthread_cond_wait(), but
56 at that point notices the cancellation request and instead of returning
57 to the caller of pthread_cond_timedwait() or pthread_cond_wait(),
58 starts the thread cancellation activities, which includes calling can‐
59 cellation cleanup handlers.
60 A thread that has been unblocked because it has been canceled while
62 blocked in a call to pthread_cond_timedwait() or pthread_cond_wait()
63 shall not consume any condition signal that may be directed concur‐
64 rently at the condition variable if there are other threads blocked on
65 the condition variable.
66 The pthread_cond_timedwait() function shall be equivalent to
68 pthread_cond_wait(), except that an error is returned if the absolute
69 time specified by abstime passes (that is, system time equals or
70 exceeds abstime) before the condition cond is signaled or broadcasted,
71 or if the absolute time specified by abstime has already been passed at
72 the time of the call.
73 If the Clock Selection option is supported, the condition variable
75 shall have a clock attribute which specifies the clock that shall be
76 used to measure the time specified by the abstime argument. When such
77 timeouts occur, pthread_cond_timedwait() shall nonetheless release and
78 re-acquire the mutex referenced by mutex. The pthread_cond_timedwait()
79 function is also a cancellation point.
80 If a signal is delivered to a thread waiting for a condition variable,
82 upon return from the signal handler the thread resumes waiting for the
83 condition variable as if it was not interrupted, or it shall return
84 zero due to spurious wakeup.
85
87 Except in the case of [ETIMEDOUT], all these error checks shall act as
88 if they were performed immediately at the beginning of processing for
89 the function and shall cause an error return, in effect, prior to modi‐
90 fying the state of the mutex specified by mutex or the condition vari‐
91 able specified by cond.
92 Upon successful completion, a value of zero shall be returned; other‐
94 wise, an error number shall be returned to indicate the error.
95
97 The pthread_cond_timedwait() function shall fail if:
98 ETIMEDOUT
100 The time specified by abstime to pthread_cond_timedwait() has
101 passed.
102 The pthread_cond_timedwait() and pthread_cond_wait() functions may fail
105 if:
106 EINVAL The value specified by cond, mutex, or abstime is invalid.
108 EINVAL Different mutexes were supplied for concurrent
110 pthread_cond_timedwait() or pthread_cond_wait() operations on
111 the same condition variable.
112 EPERM The mutex was not owned by the current thread at the time of the
114 call.
115 These functions shall not return an error code of [EINTR].
118 The following sections are informative.
120
122 None.
123
125 None.
126
128 Condition Wait Semantics
129 It is important to note that when pthread_cond_wait() and
130 pthread_cond_timedwait() return without error, the associated predicate
131 may still be false. Similarly, when pthread_cond_timedwait() returns
132 with the timeout error, the associated predicate may be true due to an
133 unavoidable race between the expiration of the timeout and the predi‐
134 cate state change.
135 Some implementations, particularly on a multi-processor, may sometimes
137 cause multiple threads to wake up when the condition variable is sig‐
138 naled simultaneously on different processors.
139 In general, whenever a condition wait returns, the thread has to re-
141 evaluate the predicate associated with the condition wait to determine
142 whether it can safely proceed, should wait again, or should declare a
143 timeout. A return from the wait does not imply that the associated
144 predicate is either true or false.
145 It is thus recommended that a condition wait be enclosed in the equiva‐
147 lent of a "while loop" that checks the predicate.
148 Timed Wait Semantics
150 An absolute time measure was chosen for specifying the timeout parame‐
151 ter for two reasons. First, a relative time measure can be easily
152 implemented on top of a function that specifies absolute time, but
153 there is a race condition associated with specifying an absolute time‐
154 out on top of a function that specifies relative timeouts. For exam‐
155 ple, assume that clock_gettime() returns the current time and cond_rel‐
156 ative_timed_wait() uses relative timeouts:
157 clock_gettime(CLOCK_REALTIME, &now)
160 reltime = sleep_til_this_absolute_time -now;
161 cond_relative_timed_wait(c, m, &reltime);
162 If the thread is preempted between the first statement and the last
164 statement, the thread blocks for too long. Blocking, however, is irrel‐
165 evant if an absolute timeout is used. An absolute timeout also need not
166 be recomputed if it is used multiple times in a loop, such as that
167 enclosing a condition wait.
168 For cases when the system clock is advanced discontinuously by an oper‐
170 ator, it is expected that implementations process any timed wait expir‐
171 ing at an intervening time as if that time had actually occurred.
172 Cancellation and Condition Wait
174 A condition wait, whether timed or not, is a cancellation point. That
175 is, the functions pthread_cond_wait() or pthread_cond_timedwait() are
176 points where a pending (or concurrent) cancellation request is noticed.
177 The reason for this is that an indefinite wait is possible at these
178 points-whatever event is being waited for, even if the program is
179 totally correct, might never occur; for example, some input data being
180 awaited might never be sent. By making condition wait a cancellation
181 point, the thread can be canceled and perform its cancellation cleanup
182 handler even though it may be stuck in some indefinite wait.
183 A side effect of acting on a cancellation request while a thread is
185 blocked on a condition variable is to re-acquire the mutex before call‐
186 ing any of the cancellation cleanup handlers. This is done in order to
187 ensure that the cancellation cleanup handler is executed in the same
188 state as the critical code that lies both before and after the call to
189 the condition wait function. This rule is also required when interfac‐
190 ing to POSIX threads from languages, such as Ada or C++, which may
191 choose to map cancellation onto a language exception; this rule ensures
192 that each exception handler guarding a critical section can always
193 safely depend upon the fact that the associated mutex has already been
194 locked regardless of exactly where within the critical section the
195 exception was raised. Without this rule, there would not be a uniform
196 rule that exception handlers could follow regarding the lock, and so
197 coding would become very cumbersome.
198 Therefore, since some statement has to be made regarding the state of
200 the lock when a cancellation is delivered during a wait, a definition
201 has been chosen that makes application coding most convenient and error
202 free.
203 When acting on a cancellation request while a thread is blocked on a
205 condition variable, the implementation is required to ensure that the
206 thread does not consume any condition signals directed at that condi‐
207 tion variable if there are any other threads waiting on that condition
208 variable. This rule is specified in order to avoid deadlock conditions
209 that could occur if these two independent requests (one acting on a
210 thread and the other acting on the condition variable) were not pro‐
211 cessed independently.
212 Performance of Mutexes and Condition Variables
214 Mutexes are expected to be locked only for a few instructions. This
215 practice is almost automatically enforced by the desire of programmers
216 to avoid long serial regions of execution (which would reduce total
217 effective parallelism).
218 When using mutexes and condition variables, one tries to ensure that
220 the usual case is to lock the mutex, access shared data, and unlock the
221 mutex. Waiting on a condition variable should be a relatively rare sit‐
222 uation. For example, when implementing a read-write lock, code that
223 acquires a read-lock typically needs only to increment the count of
224 readers (under mutual-exclusion) and return. The calling thread would
225 actually wait on the condition variable only when there is already an
226 active writer. So the efficiency of a synchronization operation is
227 bounded by the cost of mutex lock/unlock and not by condition wait.
228 Note that in the usual case there is no context switch.
229 This is not to say that the efficiency of condition waiting is unimpor‐
231 tant. Since there needs to be at least one context switch per Ada ren‐
232 dezvous, the efficiency of waiting on a condition variable is impor‐
233 tant. The cost of waiting on a condition variable should be little more
234 than the minimal cost for a context switch plus the time to unlock and
235 lock the mutex.
236 Features of Mutexes and Condition Variables
238 It had been suggested that the mutex acquisition and release be decou‐
239 pled from condition wait. This was rejected because it is the combined
240 nature of the operation that, in fact, facilitates realtime implementa‐
241 tions. Those implementations can atomically move a high-priority thread
242 between the condition variable and the mutex in a manner that is trans‐
243 parent to the caller. This can prevent extra context switches and pro‐
244 vide more deterministic acquisition of a mutex when the waiting thread
245 is signaled. Thus, fairness and priority issues can be dealt with
246 directly by the scheduling discipline. Furthermore, the current condi‐
247 tion wait operation matches existing practice.
248 Scheduling Behavior of Mutexes and Condition Variables
250 Synchronization primitives that attempt to interfere with scheduling
251 policy by specifying an ordering rule are considered undesirable.
252 Threads waiting on mutexes and condition variables are selected to pro‐
253 ceed in an order dependent upon the scheduling policy rather than in
254 some fixed order (for example, FIFO or priority). Thus, the scheduling
255 policy determines which thread(s) are awakened and allowed to proceed.
256 Timed Condition Wait
258 The pthread_cond_timedwait() function allows an application to give up
259 waiting for a particular condition after a given amount of time. An
260 example of its use follows:
261 (void) pthread_mutex_lock(&t.mn);
264 t.waiters++;
265 clock_gettime(CLOCK_REALTIME, &ts);
266 ts.tv_sec += 5;
267 rc = 0;
268 while (! mypredicate(&t) && rc == 0)
269 rc = pthread_cond_timedwait(&t.cond, &t.mn, &ts);
270 t.waiters--;
271 if (rc == 0) setmystate(&t);
272 (void) pthread_mutex_unlock(&t.mn);
273 By making the timeout parameter absolute, it does not need to be recom‐
275 puted each time the program checks its blocking predicate. If the
276 timeout was relative, it would have to be recomputed before each call.
277 This would be especially difficult since such code would need to take
278 into account the possibility of extra wakeups that result from extra
279 broadcasts or signals on the condition variable that occur before
280 either the predicate is true or the timeout is due.
281
283 None.
284
286 pthread_cond_signal() , pthread_cond_broadcast() , the Base Definitions
287 volume of IEEE Std 1003.1-2001, <pthread.h>
288
290 Portions of this text are reprinted and reproduced in electronic form
291 from IEEE Std 1003.1, 2003 Edition, Standard for Information Technology
292 -- Portable Operating System Interface (POSIX), The Open Group Base
293 Specifications Issue 6, Copyright (C) 2001-2003 by the Institute of
294 Electrical and Electronics Engineers, Inc and The Open Group. In the
295 event of any discrepancy between this version and the original IEEE and
296 The Open Group Standard, the original IEEE and The Open Group Standard
297 is the referee document. The original Standard can be obtained online
298 at http://www.opengroup.org/unix/online.html .
299IEEE/The Open Group 2003 PTHREAD_COND_TIMEDWAIT(P)