6 ASYNC_init_thread, ASYNC_cleanup_thread, ASYNC_start_job, ASYNC_pause_job,
7 ASYNC_get_current_job, ASYNC_block_pause, ASYNC_unblock_pause, ASYNC_is_capable
8 - asynchronous job management functions
12 #include <openssl/async.h>
14 int ASYNC_init_thread(size_t max_size, size_t init_size);
15 void ASYNC_cleanup_thread(void);
17 int ASYNC_start_job(ASYNC_JOB **job, ASYNC_WAIT_CTX *ctx, int *ret,
18 int (*func)(void *), void *args, size_t size);
19 int ASYNC_pause_job(void);
21 ASYNC_JOB *ASYNC_get_current_job(void);
22 ASYNC_WAIT_CTX *ASYNC_get_wait_ctx(ASYNC_JOB *job);
23 void ASYNC_block_pause(void);
24 void ASYNC_unblock_pause(void);
26 int ASYNC_is_capable(void);
30 OpenSSL implements asynchronous capabilities through an B<ASYNC_JOB>. This
31 represents code that can be started and executes until some event occurs. At
32 that point the code can be paused and control returns to user code until some
33 subsequent event indicates that the job can be resumed.
35 The creation of an B<ASYNC_JOB> is a relatively expensive operation. Therefore,
36 for efficiency reasons, jobs can be created up front and reused many times. They
37 are held in a pool until they are needed, at which point they are removed from
38 the pool, used, and then returned to the pool when the job completes. If the
39 user application is multi-threaded, then ASYNC_init_thread() may be called for
40 each thread that will initiate asynchronous jobs. Before
41 user code exits per-thread resources need to be cleaned up. This will normally
42 occur automatically (see L<OPENSSL_init_crypto(3)>) but may be explicitly
43 initiated by using ASYNC_cleanup_thread(). No asynchronous jobs must be
44 outstanding for the thread when ASYNC_cleanup_thread() is called. Failing to
45 ensure this will result in memory leaks.
47 The I<max_size> argument limits the number of B<ASYNC_JOB>s that will be held in
48 the pool. If I<max_size> is set to 0 then no upper limit is set. When an
49 B<ASYNC_JOB> is needed but there are none available in the pool already then one
50 will be automatically created, as long as the total of B<ASYNC_JOB>s managed by
51 the pool does not exceed I<max_size>. When the pool is first initialised
52 I<init_size> B<ASYNC_JOB>s will be created immediately. If ASYNC_init_thread()
53 is not called before the pool is first used then it will be called automatically
54 with a I<max_size> of 0 (no upper limit) and an I<init_size> of 0 (no
55 B<ASYNC_JOB>s created up front).
57 An asynchronous job is started by calling the ASYNC_start_job() function.
58 Initially I<*job> should be NULL. I<ctx> should point to an B<ASYNC_WAIT_CTX>
59 object created through the L<ASYNC_WAIT_CTX_new(3)> function. I<ret> should
60 point to a location where the return value of the asynchronous function should
61 be stored on completion of the job. I<func> represents the function that should
62 be started asynchronously. The data pointed to by I<args> and of size I<size>
63 will be copied and then passed as an argument to I<func> when the job starts.
64 ASYNC_start_job will return one of the following values:
70 An error occurred trying to start the job. Check the OpenSSL error queue (e.g.
71 see L<ERR_print_errors(3)>) for more details.
73 =item B<ASYNC_NO_JOBS>
75 There are no jobs currently available in the pool. This call can be retried
76 again at a later time.
80 The job was successfully started but was "paused" before it completed (see
81 ASYNC_pause_job() below). A handle to the job is placed in I<*job>. Other work
82 can be performed (if desired) and the job restarted at a later time. To restart
83 a job call ASYNC_start_job() again passing the job handle in I<*job>. The
84 I<func>, I<args> and I<size> parameters will be ignored when restarting a job.
85 When restarting a job ASYNC_start_job() B<must> be called from the same thread
86 that the job was originally started from.
90 The job completed. I<*job> will be NULL and the return value from I<func> will
95 At any one time there can be a maximum of one job actively running per thread
96 (you can have many that are paused). ASYNC_get_current_job() can be used to get
97 a pointer to the currently executing B<ASYNC_JOB>. If no job is currently
98 executing then this will return NULL.
100 If executing within the context of a job (i.e. having been called directly or
101 indirectly by the function "func" passed as an argument to ASYNC_start_job())
102 then ASYNC_pause_job() will immediately return control to the calling
103 application with B<ASYNC_PAUSE> returned from the ASYNC_start_job() call. A
104 subsequent call to ASYNC_start_job passing in the relevant B<ASYNC_JOB> in the
105 I<*job> parameter will resume execution from the ASYNC_pause_job() call. If
106 ASYNC_pause_job() is called whilst not within the context of a job then no
107 action is taken and ASYNC_pause_job() returns immediately.
109 ASYNC_get_wait_ctx() can be used to get a pointer to the B<ASYNC_WAIT_CTX>
110 for the I<job>. B<ASYNC_WAIT_CTX>s contain two different ways to notify
111 applications that a job is ready to be resumed. One is a "wait" file
112 descriptor, and the other is a "callback" mechanism.
114 The "wait" file descriptor associated with B<ASYNC_WAIT_CTX> is used for
115 applications to wait for the file descriptor to be ready for "read" using a
116 system function call such as select or poll (being ready for "read" indicates
117 that the job should be resumed). If no file descriptor is made available then
118 an application will have to periodically "poll" the job by attempting to restart
119 it to see if it is ready to continue.
121 B<ASYNC_WAIT_CTX>s also have a "callback" mechanism to notify applications. The
122 callback is set by an application, and it will be automatically called when an
123 engine completes a cryptography operation, so that the application can resume
124 the paused work flow without polling. An engine could be written to look whether
125 the callback has been set. If it has then it would use the callback mechanism
126 in preference to the file descriptor notifications. If a callback is not set
127 then the engine may use file descriptor based notifications. Please note that
128 not all engines may support the callback mechanism, so the callback may not be
129 used even if it has been set. See ASYNC_WAIT_CTX_new() for more details.
131 The ASYNC_block_pause() function will prevent the currently active job from
132 pausing. The block will remain in place until a subsequent call to
133 ASYNC_unblock_pause(). These functions can be nested, e.g. if you call
134 ASYNC_block_pause() twice then you must call ASYNC_unblock_pause() twice in
135 order to re-enable pausing. If these functions are called while there is no
136 currently active job then they have no effect. This functionality can be useful
137 to avoid deadlock scenarios. For example during the execution of an B<ASYNC_JOB>
138 an application acquires a lock. It then calls some cryptographic function which
139 invokes ASYNC_pause_job(). This returns control back to the code that created
140 the B<ASYNC_JOB>. If that code then attempts to acquire the same lock before
141 resuming the original job then a deadlock can occur. By calling
142 ASYNC_block_pause() immediately after acquiring the lock and
143 ASYNC_unblock_pause() immediately before releasing it then this situation cannot
146 Some platforms cannot support async operations. The ASYNC_is_capable() function
147 can be used to detect whether the current platform is async capable or not.
151 ASYNC_init_thread returns 1 on success or 0 otherwise.
153 ASYNC_start_job returns one of B<ASYNC_ERR>, B<ASYNC_NO_JOBS>, B<ASYNC_PAUSE> or
154 B<ASYNC_FINISH> as described above.
156 ASYNC_pause_job returns 0 if an error occurred or 1 on success. If called when
157 not within the context of an B<ASYNC_JOB> then this is counted as success so 1
160 ASYNC_get_current_job returns a pointer to the currently executing B<ASYNC_JOB>
161 or NULL if not within the context of a job.
163 ASYNC_get_wait_ctx() returns a pointer to the B<ASYNC_WAIT_CTX> for the job.
165 ASYNC_is_capable() returns 1 if the current platform is async capable or 0
170 On Windows platforms the openssl/async.h header is dependent on some
171 of the types customarily made available by including windows.h. The
172 application developer is likely to require control over when the latter
173 is included, commonly as one of the first included headers. Therefore
174 it is defined as an application developer's responsibility to include
175 windows.h prior to async.h.
179 The following example demonstrates how to use most of the core async APIs:
182 # include <windows.h>
186 #include <openssl/async.h>
187 #include <openssl/crypto.h>
191 void cleanup(ASYNC_WAIT_CTX *ctx, const void *key, OSSL_ASYNC_FD r, void *vw)
193 OSSL_ASYNC_FD *w = (OSSL_ASYNC_FD *)vw;
200 int jobfunc(void *arg)
204 int pipefds[2] = {0, 0};
208 currjob = ASYNC_get_current_job();
209 if (currjob != NULL) {
210 printf("Executing within a job\n");
212 printf("Not executing within a job - should not happen\n");
216 msg = (unsigned char *)arg;
217 printf("Passed in message is: %s\n", msg);
219 if (pipe(pipefds) != 0) {
220 printf("Failed to create pipe\n");
223 wptr = OPENSSL_malloc(sizeof(OSSL_ASYNC_FD));
225 printf("Failed to malloc\n");
229 ASYNC_WAIT_CTX_set_wait_fd(ASYNC_get_wait_ctx(currjob), &unique,
230 pipefds[0], wptr, cleanup);
233 * Normally some external event would cause this to happen at some
234 * later point - but we do it here for demo purposes, i.e.
235 * immediately signalling that the job is ready to be woken up after
236 * we return to main via ASYNC_pause_job().
238 write(pipefds[1], &buf, 1);
240 /* Return control back to main */
243 /* Clear the wake signal */
244 read(pipefds[0], &buf, 1);
246 printf ("Resumed the job after a pause\n");
253 ASYNC_JOB *job = NULL;
254 ASYNC_WAIT_CTX *ctx = NULL;
256 OSSL_ASYNC_FD waitfd;
259 unsigned char msg[13] = "Hello world!";
261 printf("Starting...\n");
263 ctx = ASYNC_WAIT_CTX_new();
265 printf("Failed to create ASYNC_WAIT_CTX\n");
270 switch (ASYNC_start_job(&job, ctx, &ret, jobfunc, msg, sizeof(msg))) {
273 printf("An error occurred\n");
276 printf("Job was paused\n");
279 printf("Job finished with return value %d\n", ret);
283 /* Wait for the job to be woken */
284 printf("Waiting for the job to be woken up\n");
286 if (!ASYNC_WAIT_CTX_get_all_fds(ctx, NULL, &numfds)
288 printf("Unexpected number of fds\n");
291 ASYNC_WAIT_CTX_get_all_fds(ctx, &waitfd, &numfds);
293 FD_SET(waitfd, &waitfdset);
294 select(waitfd + 1, &waitfdset, NULL, NULL, NULL);
298 ASYNC_WAIT_CTX_free(ctx);
299 printf("Finishing\n");
304 The expected output from executing the above example program is:
307 Executing within a job
308 Passed in message is: Hello world!
310 Waiting for the job to be woken up
311 Resumed the job after a pause
312 Job finished with return value 1
317 L<crypto(7)>, L<ERR_print_errors(3)>
321 ASYNC_init_thread, ASYNC_cleanup_thread,
322 ASYNC_start_job, ASYNC_pause_job, ASYNC_get_current_job, ASYNC_get_wait_ctx(),
323 ASYNC_block_pause(), ASYNC_unblock_pause() and ASYNC_is_capable() were first
324 added in OpenSSL 1.1.0.
328 Copyright 2015-2016 The OpenSSL Project Authors. All Rights Reserved.
330 Licensed under the Apache License 2.0 (the "License"). You may not use
331 this file except in compliance with the License. You can obtain a copy
332 in the file LICENSE in the source distribution or at
333 L<https://www.openssl.org/source/license.html>.