1PTHREAD_CLEANUP_POP(3P) POSIX Programmer's Manual PTHREAD_CLEANUP_POP(3P)
2
3
4
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_cleanup_pop, pthread_cleanup_push — establish cancellation han‐
13 dlers
14
16 #include <pthread.h>
17
18 void pthread_cleanup_pop(int execute);
19 void pthread_cleanup_push(void (*routine)(void*), void *arg);
20
22 The pthread_cleanup_pop() function shall remove the routine at the top
23 of the calling thread's cancellation cleanup stack and optionally
24 invoke it (if execute is non-zero).
25
26 The pthread_cleanup_push() function shall push the specified cancella‐
27 tion cleanup handler routine onto the calling thread's cancellation
28 cleanup stack. The cancellation cleanup handler shall be popped from
29 the cancellation cleanup stack and invoked with the argument arg when:
30
31 * The thread exits (that is, calls pthread_exit()).
32
33 * The thread acts upon a cancellation request.
34
35 * The thread calls pthread_cleanup_pop() with a non-zero execute
36 argument.
37
38 It is unspecified whether pthread_cleanup_push() and
39 pthread_cleanup_pop() are macros or functions. If a macro definition is
40 suppressed in order to access an actual function, or a program defines
41 an external identifier with any of these names, the behavior is unde‐
42 fined. The application shall ensure that they appear as statements,
43 and in pairs within the same lexical scope (that is, the
44 pthread_cleanup_push() macro may be thought to expand to a token list
45 whose first token is '{' with pthread_cleanup_pop() expanding to a
46 token list whose last token is the corresponding '}').
47
48 The effect of calling longjmp() or siglongjmp() is undefined if there
49 have been any calls to pthread_cleanup_push() or pthread_cleanup_pop()
50 made without the matching call since the jump buffer was filled. The
51 effect of calling longjmp() or siglongjmp() from inside a cancellation
52 cleanup handler is also undefined unless the jump buffer was also
53 filled in the cancellation cleanup handler.
54
55 The effect of the use of return, break, continue, and goto to prema‐
56 turely leave a code block described by a pair of pthread_cleanup_push()
57 and pthread_cleanup_pop() functions calls is undefined.
58
60 The pthread_cleanup_push() and pthread_cleanup_pop() functions shall
61 not return a value.
62
64 No errors are defined.
65
66 These functions shall not return an error code of [EINTR].
67
68 The following sections are informative.
69
71 The following is an example using thread primitives to implement a can‐
72 celable, writers-priority read-write lock:
73
74
75 typedef struct {
76 pthread_mutex_t lock;
77 pthread_cond_t rcond,
78 wcond;
79 int lock_count; /* < 0 .. Held by writer. */
80 /* > 0 .. Held by lock_count readers. */
81 /* = 0 .. Held by nobody. */
82 int waiting_writers; /* Count of waiting writers. */
83 } rwlock;
84
85 void
86 waiting_reader_cleanup(void *arg)
87 {
88 rwlock *l;
89
90 l = (rwlock *) arg;
91 pthread_mutex_unlock(&l->lock);
92 }
93
94 void
95 lock_for_read(rwlock *l)
96 {
97 pthread_mutex_lock(&l->lock);
98 pthread_cleanup_push(waiting_reader_cleanup, l);
99 while ((l->lock_count < 0) || (l->waiting_writers != 0))
100 pthread_cond_wait(&l->rcond, &l->lock);
101 l->lock_count++;
102 /*
103 * Note the pthread_cleanup_pop executes
104 * waiting_reader_cleanup.
105 */
106 pthread_cleanup_pop(1);
107 }
108
109 void
110 release_read_lock(rwlock *l)
111 {
112 pthread_mutex_lock(&l->lock);
113 if (--l->lock_count == 0)
114 pthread_cond_signal(&l->wcond);
115 pthread_mutex_unlock(&l->lock);
116 }
117
118 void
119 waiting_writer_cleanup(void *arg)
120 {
121 rwlock *l;
122
123 l = (rwlock *) arg;
124 if ((--l->waiting_writers == 0) && (l->lock_count >= 0)) {
125 /*
126 * This only happens if we have been canceled. If the
127 * lock is not held by a writer, there may be readers who
128 * were blocked because waiting_writers was positive; they
129 * can now be unblocked.
130 */
131 pthread_cond_broadcast(&l->rcond);
132 }
133 pthread_mutex_unlock(&l->lock);
134 }
135
136 void
137 lock_for_write(rwlock *l)
138 {
139 pthread_mutex_lock(&l->lock);
140 l->waiting_writers++;
141 pthread_cleanup_push(waiting_writer_cleanup, l);
142 while (l->lock_count != 0)
143 pthread_cond_wait(&l->wcond, &l->lock);
144 l->lock_count = -1;
145 /*
146 * Note the pthread_cleanup_pop executes
147 * waiting_writer_cleanup.
148 */
149 pthread_cleanup_pop(1);
150 }
151
152 void
153 release_write_lock(rwlock *l)
154 {
155 pthread_mutex_lock(&l->lock);
156 l->lock_count = 0;
157 if (l->waiting_writers == 0)
158 pthread_cond_broadcast(&l->rcond);
159 else
160 pthread_cond_signal(&l->wcond);
161 pthread_mutex_unlock(&l->lock);
162 }
163
164 /*
165 * This function is called to initialize the read/write lock.
166 */
167 void
168 initialize_rwlock(rwlock *l)
169 {
170 pthread_mutex_init(&l->lock, pthread_mutexattr_default);
171 pthread_cond_init(&l->wcond, pthread_condattr_default);
172 pthread_cond_init(&l->rcond, pthread_condattr_default);
173 l->lock_count = 0;
174 l->waiting_writers = 0;
175 }
176
177 reader_thread()
178 {
179 lock_for_read(&lock);
180 pthread_cleanup_push(release_read_lock, &lock);
181 /*
182 * Thread has read lock.
183 */
184 pthread_cleanup_pop(1);
185 }
186
187 writer_thread()
188 {
189 lock_for_write(&lock);
190 pthread_cleanup_push(release_write_lock, &lock);
191 /*
192 * Thread has write lock.
193 */
194 pthread_cleanup_pop(1);
195 }
196
198 The two routines that push and pop cancellation cleanup handlers,
199 pthread_cleanup_push() and pthread_cleanup_pop(), can be thought of as
200 left and right-parentheses. They always need to be matched.
201
203 The restriction that the two routines that push and pop cancellation
204 cleanup handlers, pthread_cleanup_push() and pthread_cleanup_pop(),
205 have to appear in the same lexical scope allows for efficient macro or
206 compiler implementations and efficient storage management. A sample
207 implementation of these routines as macros might look like this:
208
209
210 #define pthread_cleanup_push(rtn,arg) { \
211 struct _pthread_handler_rec __cleanup_handler, **__head; \
212 __cleanup_handler.rtn = rtn; \
213 __cleanup_handler.arg = arg; \
214 (void) pthread_getspecific(_pthread_handler_key, &__head); \
215 __cleanup_handler.next = *__head; \
216 *__head = &__cleanup_handler;
217
218 #define pthread_cleanup_pop(ex) \
219 *__head = __cleanup_handler.next; \
220 if (ex) (*__cleanup_handler.rtn)(__cleanup_handler.arg); \
221 }
222
223 A more ambitious implementation of these routines might do even better
224 by allowing the compiler to note that the cancellation cleanup handler
225 is a constant and can be expanded inline.
226
227 This volume of POSIX.1‐2017 currently leaves unspecified the effect of
228 calling longjmp() from a signal handler executing in a POSIX System
229 Interfaces function. If an implementation wants to allow this and give
230 the programmer reasonable behavior, the longjmp() function has to call
231 all cancellation cleanup handlers that have been pushed but not popped
232 since the time setjmp() was called.
233
234 Consider a multi-threaded function called by a thread that uses sig‐
235 nals. If a signal were delivered to a signal handler during the opera‐
236 tion of qsort() and that handler were to call longjmp() (which, in
237 turn, did not call the cancellation cleanup handlers) the helper
238 threads created by the qsort() function would not be canceled. Instead,
239 they would continue to execute and write into the argument array even
240 though the array might have been popped off the stack.
241
242 Note that the specified cleanup handling mechanism is especially tied
243 to the C language and, while the requirement for a uniform mechanism
244 for expressing cleanup is language-independent, the mechanism used in
245 other languages may be quite different. In addition, this mechanism is
246 really only necessary due to the lack of a real exception mechanism in
247 the C language, which would be the ideal solution.
248
249 There is no notion of a cancellation cleanup-safe function. If an
250 application has no cancellation points in its signal handlers, blocks
251 any signal whose handler may have cancellation points while calling
252 async-unsafe functions, or disables cancellation while calling async-
253 unsafe functions, all functions may be safely called from cancellation
254 cleanup routines.
255
257 None.
258
260 pthread_cancel(), pthread_setcancelstate()
261
262 The Base Definitions volume of POSIX.1‐2017, <pthread.h>
263
265 Portions of this text are reprinted and reproduced in electronic form
266 from IEEE Std 1003.1-2017, Standard for Information Technology -- Por‐
267 table Operating System Interface (POSIX), The Open Group Base Specifi‐
268 cations Issue 7, 2018 Edition, Copyright (C) 2018 by the Institute of
269 Electrical and Electronics Engineers, Inc and The Open Group. In the
270 event of any discrepancy between this version and the original IEEE and
271 The Open Group Standard, the original IEEE and The Open Group Standard
272 is the referee document. The original Standard can be obtained online
273 at http://www.opengroup.org/unix/online.html .
274
275 Any typographical or formatting errors that appear in this page are
276 most likely to have been introduced during the conversion of the source
277 files to man page format. To report such errors, see https://www.ker‐
278 nel.org/doc/man-pages/reporting_bugs.html .
279
280
281
282IEEE/The Open Group 2017 PTHREAD_CLEANUP_POP(3P)