1POSIX_SPAWN(3P) POSIX Programmer's Manual POSIX_SPAWN(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 posix_spawn, posix_spawnp — spawn a process (ADVANCED REALTIME)
13
15 #include <spawn.h>
16 int posix_spawn(pid_t *restrict pid, const char *restrict path,
18 const posix_spawn_file_actions_t *file_actions,
19 const posix_spawnattr_t *restrict attrp,
20 char *const argv[restrict], char *const envp[restrict]);
21 int posix_spawnp(pid_t *restrict pid, const char *restrict file,
22 const posix_spawn_file_actions_t *file_actions,
23 const posix_spawnattr_t *restrict attrp,
24 char *const argv[restrict], char *const envp[restrict]);
25
27 The posix_spawn() and posix_spawnp() functions shall create a new
28 process (child process) from the specified process image. The new
29 process image shall be constructed from a regular executable file
30 called the new process image file.
31 When a C program is executed as the result of this call, it shall be
33 entered as a C-language function call as follows:
34 int main(int argc, char *argv[]);
37 where argc is the argument count and argv is an array of character
39 pointers to the arguments themselves. In addition, the following vari‐
40 able:
41 extern char **environ;
44 shall be initialized as a pointer to an array of character pointers to
46 the environment strings.
47 The argument argv is an array of character pointers to null-terminated
49 strings. The last member of this array shall be a null pointer and is
50 not counted in argc. These strings constitute the argument list avail‐
51 able to the new process image. The value in argv[0] should point to a
52 filename string that is associated with the process image being started
53 by the posix_spawn() or posix_spawnp() function.
54 The argument envp is an array of character pointers to null-terminated
56 strings. These strings constitute the environment for the new process
57 image. The environment array is terminated by a null pointer.
58 The number of bytes available for the combined argument and environment
60 lists of the child process is {ARG_MAX}. The implementation shall
61 specify in the system documentation (see the Base Definitions volume of
62 POSIX.1‐2017, Chapter 2, Conformance) whether any list overhead, such
63 as length words, null terminators, pointers, or alignment bytes, is
64 included in this total.
65 The path argument to posix_spawn() is a pathname that identifies the
67 new process image file to execute.
68 The file parameter to posix_spawnp() shall be used to construct a path‐
70 name that identifies the new process image file. If the file parameter
71 contains a <slash> character, the file parameter shall be used as the
72 pathname for the new process image file. Otherwise, the path prefix for
73 this file shall be obtained by a search of the directories passed as
74 the environment variable PATH (see the Base Definitions volume of
75 POSIX.1‐2017, Chapter 8, Environment Variables). If this environment
76 variable is not defined, the results of the search are implementation-
77 defined.
78 If file_actions is a null pointer, then file descriptors open in the
80 calling process shall remain open in the child process, except for
81 those whose close-on-exec flag FD_CLOEXEC is set (see fcntl()). For
82 those file descriptors that remain open, the child process shall not
83 inherit any file locks, but all remaining attributes of the correspond‐
84 ing open file descriptions (see fcntl()), shall remain unchanged.
85 If file_actions is not NULL, then the file descriptors open in the
87 child process shall be those open in the calling process as modified by
88 the spawn file actions object pointed to by file_actions and the
89 FD_CLOEXEC flag of each remaining open file descriptor after the spawn
90 file actions have been processed. The effective order of processing the
91 spawn file actions shall be:
92 1. The set of open file descriptors for the child process shall ini‐
94 tially be the same set as is open for the calling process. The
95 child process shall not inherit any file locks, but all remaining
96 attributes of the corresponding open file descriptions (see
97 fcntl()), shall remain unchanged.
98 2. The signal mask, signal default actions, and the effective user and
100 group IDs for the child process shall be changed as specified in
101 the attributes object referenced by attrp.
102 3. The file actions specified by the spawn file actions object shall
104 be performed in the order in which they were added to the spawn
105 file actions object.
106 4. Any file descriptor that has its FD_CLOEXEC flag set (see fcntl())
108 shall be closed.
109 If file descriptor 0, 1, or 2 would otherwise be closed in the new
111 process image created by posix_spawn() or posix_spawnp(), implementa‐
112 tions may open an unspecified file for the file descriptor in the new
113 process image. If a standard utility or a conforming application is
114 executed with file descriptor 0 not open for reading or with file
115 descriptor 1 or 2 not open for writing, the environment in which the
116 utility or application is executed shall be deemed non-conforming, and
117 consequently the utility or application might not behave as described
118 in this standard.
119 The posix_spawnattr_t spawn attributes object type is defined in
121 <spawn.h>. It shall contain at least the attributes defined below.
122 If the POSIX_SPAWN_SETPGROUP flag is set in the spawn-flags attribute
124 of the object referenced by attrp, and the spawn-pgroup attribute of
125 the same object is non-zero, then the child's process group shall be as
126 specified in the spawn-pgroup attribute of the object referenced by
127 attrp.
128 As a special case, if the POSIX_SPAWN_SETPGROUP flag is set in the
130 spawn-flags attribute of the object referenced by attrp, and the spawn-
131 pgroup attribute of the same object is set to zero, then the child
132 shall be in a new process group with a process group ID equal to its
133 process ID.
134 If the POSIX_SPAWN_SETPGROUP flag is not set in the spawn-flags
136 attribute of the object referenced by attrp, the new child process
137 shall inherit the parent's process group.
138 If the POSIX_SPAWN_SETSCHEDPARAM flag is set in the spawn-flags
140 attribute of the object referenced by attrp, but POSIX_SPAWN_SETSCHED‐
141 ULER is not set, the new process image shall initially have the sched‐
142 uling policy of the calling process with the scheduling parameters
143 specified in the spawn-schedparam attribute of the object referenced by
144 attrp.
145 If the POSIX_SPAWN_SETSCHEDULER flag is set in the spawn-flags
147 attribute of the object referenced by attrp (regardless of the setting
148 of the POSIX_SPAWN_SETSCHEDPARAM flag), the new process image shall
149 initially have the scheduling policy specified in the spawn-schedpolicy
150 attribute of the object referenced by attrp and the scheduling parame‐
151 ters specified in the spawn-schedparam attribute of the same object.
152 The POSIX_SPAWN_RESETIDS flag in the spawn-flags attribute of the
154 object referenced by attrp governs the effective user ID of the child
155 process. If this flag is not set, the child process shall inherit the
156 effective user ID of the parent process. If this flag is set, the
157 effective user ID of the child process shall be reset to the parent's
158 real user ID. In either case, if the set-user-ID mode bit of the new
159 process image file is set, the effective user ID of the child process
160 shall become that file's owner ID before the new process image begins
161 execution.
162 The POSIX_SPAWN_RESETIDS flag in the spawn-flags attribute of the
164 object referenced by attrp also governs the effective group ID of the
165 child process. If this flag is not set, the child process shall inherit
166 the effective group ID of the parent process. If this flag is set, the
167 effective group ID of the child process shall be reset to the parent's
168 real group ID. In either case, if the set-group-ID mode bit of the new
169 process image file is set, the effective group ID of the child process
170 shall become that file's group ID before the new process image begins
171 execution.
172 If the POSIX_SPAWN_SETSIGMASK flag is set in the spawn-flags attribute
174 of the object referenced by attrp, the child process shall initially
175 have the signal mask specified in the spawn-sigmask attribute of the
176 object referenced by attrp.
177 If the POSIX_SPAWN_SETSIGDEF flag is set in the spawn-flags attribute
179 of the object referenced by attrp, the signals specified in the spawn-
180 sigdefault attribute of the same object shall be set to their default
181 actions in the child process. Signals set to the default action in the
182 parent process shall be set to the default action in the child process.
183 Signals set to be caught by the calling process shall be set to the
185 default action in the child process.
186 Except for SIGCHLD, signals set to be ignored by the calling process
188 image shall be set to be ignored by the child process, unless otherwise
189 specified by the POSIX_SPAWN_SETSIGDEF flag being set in the spawn-
190 flags attribute of the object referenced by attrp and the signals being
191 indicated in the spawn-sigdefault attribute of the object referenced by
192 attrp.
193 If the SIGCHLD signal is set to be ignored by the calling process, it
195 is unspecified whether the SIGCHLD signal is set to be ignored or to
196 the default action in the child process, unless otherwise specified by
197 the POSIX_SPAWN_SETSIGDEF flag being set in the spawn_flags attribute
198 of the object referenced by attrp and the SIGCHLD signal being indi‐
199 cated in the spawn_sigdefault attribute of the object referenced by
200 attrp.
201 If the value of the attrp pointer is NULL, then the default values are
203 used.
204 All process attributes, other than those influenced by the attributes
206 set in the object referenced by attrp as specified above or by the file
207 descriptor manipulations specified in file_actions, shall appear in the
208 new process image as though fork() had been called to create a child
209 process and then a member of the exec family of functions had been
210 called by the child process to execute the new process image.
211 It is implementation-defined whether the fork handlers are run when
213 posix_spawn() or posix_spawnp() is called.
214
216 Upon successful completion, posix_spawn() and posix_spawnp() shall
217 return the process ID of the child process to the parent process, in
218 the variable pointed to by a non-NULL pid argument, and shall return
219 zero as the function return value. Otherwise, no child process shall
220 be created, the value stored into the variable pointed to by a non-NULL
221 pid is unspecified, and an error number shall be returned as the func‐
222 tion return value to indicate the error. If the pid argument is a null
223 pointer, the process ID of the child is not returned to the caller.
224
226 These functions may fail if:
227 EINVAL The value specified by file_actions or attrp is invalid.
229 If this error occurs after the calling process successfully returns
231 from the posix_spawn() or posix_spawnp() function, the child process
232 may exit with exit status 127.
233 If posix_spawn() or posix_spawnp() fail for any of the reasons that
235 would cause fork() or one of the exec family of functions to fail, an
236 error value shall be returned as described by fork() and exec, respec‐
237 tively (or, if the error occurs after the calling process successfully
238 returns, the child process shall exit with exit status 127).
239 If POSIX_SPAWN_SETPGROUP is set in the spawn-flags attribute of the
241 object referenced by attrp, and posix_spawn() or posix_spawnp() fails
242 while changing the child's process group, an error value shall be
243 returned as described by setpgid() (or, if the error occurs after the
244 calling process successfully returns, the child process shall exit with
245 exit status 127).
246 If POSIX_SPAWN_SETSCHEDPARAM is set and POSIX_SPAWN_SETSCHEDULER is not
248 set in the spawn-flags attribute of the object referenced by attrp,
249 then if posix_spawn() or posix_spawnp() fails for any of the reasons
250 that would cause sched_setparam() to fail, an error value shall be
251 returned as described by sched_setparam() (or, if the error occurs
252 after the calling process successfully returns, the child process shall
253 exit with exit status 127).
254 If POSIX_SPAWN_SETSCHEDULER is set in the spawn-flags attribute of the
256 object referenced by attrp, and if posix_spawn() or posix_spawnp()
257 fails for any of the reasons that would cause sched_setscheduler() to
258 fail, an error value shall be returned as described by sched_setsched‐
259 uler() (or, if the error occurs after the calling process successfully
260 returns, the child process shall exit with exit status 127).
261 If the file_actions argument is not NULL, and specifies any close,
263 dup2, or open actions to be performed, and if posix_spawn() or
264 posix_spawnp() fails for any of the reasons that would cause close(),
265 dup2(), or open() to fail, an error value shall be returned as
266 described by close(), dup2(), and open(), respectively (or, if the
267 error occurs after the calling process successfully returns, the child
268 process shall exit with exit status 127). An open file action may, by
269 itself, result in any of the errors described by close() or dup2(), in
270 addition to those described by open().
271 The following sections are informative.
273
275 None.
276
278 These functions are part of the Spawn option and need not be provided
279 on all implementations.
280 See also the APPLICATION USAGE section for exec.
282
284 The posix_spawn() function and its close relation posix_spawnp() have
285 been introduced to overcome the following perceived difficulties with
286 fork(): the fork() function is difficult or impossible to implement
287 without swapping or dynamic address translation.
288 * Swapping is generally too slow for a realtime environment.
290 * Dynamic address translation is not available everywhere that POSIX
292 might be useful.
293 * Processes are too useful to simply option out of POSIX whenever it
295 must run without address translation or other MMU services.
296 Thus, POSIX needs process creation and file execution primitives that
298 can be efficiently implemented without address translation or other MMU
299 services.
300 The posix_spawn() function is implementable as a library routine, but
302 both posix_spawn() and posix_spawnp() are designed as kernel opera‐
303 tions. Also, although they may be an efficient replacement for many
304 fork()/exec pairs, their goal is to provide useful process creation
305 primitives for systems that have difficulty with fork(), not to provide
306 drop-in replacements for fork()/exec.
307 This view of the role of posix_spawn() and posix_spawnp() influenced
309 the design of their API. It does not attempt to provide the full func‐
310 tionality of fork()/exec in which arbitrary user-specified operations
311 of any sort are permitted between the creation of the child process and
312 the execution of the new process image; any attempt to reach that level
313 would need to provide a programming language as parameters. Instead,
314 posix_spawn() and posix_spawnp() are process creation primitives like
315 the Start_Process and Start_Process_Search Ada language bindings pack‐
316 age POSIX_Process_Primitives and also like those in many operating sys‐
317 tems that are not UNIX systems, but with some POSIX-specific additions.
318 To achieve its coverage goals, posix_spawn() and posix_spawnp() have
320 control of six types of inheritance: file descriptors, process group
321 ID, user and group ID, signal mask, scheduling, and whether each signal
322 ignored in the parent will remain ignored in the child, or be reset to
323 its default action in the child.
324 Control of file descriptors is required to allow an independently writ‐
326 ten child process image to access data streams opened by and even gen‐
327 erated or read by the parent process without being specifically coded
328 to know which parent files and file descriptors are to be used. Con‐
329 trol of the process group ID is required to control how the job control
330 of the child process relates to that of the parent.
331 Control of the signal mask and signal defaulting is sufficient to sup‐
333 port the implementation of system(). Although support for system() is
334 not explicitly one of the goals for posix_spawn() and posix_spawnp(),
335 it is covered under the ``at least 50%'' coverage goal.
336 The intention is that the normal file descriptor inheritance across
338 fork(), the subsequent effect of the specified spawn file actions, and
339 the normal file descriptor inheritance across one of the exec family of
340 functions should fully specify open file inheritance. The implementa‐
341 tion need make no decisions regarding the set of open file descriptors
342 when the child process image begins execution, those decisions having
343 already been made by the caller and expressed as the set of open file
344 descriptors and their FD_CLOEXEC flags at the time of the call and the
345 spawn file actions object specified in the call. We have been assured
346 that in cases where the POSIX Start_Process Ada primitives have been
347 implemented in a library, this method of controlling file descriptor
348 inheritance may be implemented very easily.
349 We can identify several problems with posix_spawn() and posix_spawnp(),
351 but there does not appear to be a solution that introduces fewer prob‐
352 lems. Environment modification for child process attributes not speci‐
353 fiable via the attrp or file_actions arguments must be done in the par‐
354 ent process, and since the parent generally wants to save its context,
355 it is more costly than similar functionality with fork()/exec. It is
356 also complicated to modify the environment of a multi-threaded process
357 temporarily, since all threads must agree when it is safe for the envi‐
358 ronment to be changed. However, this cost is only borne by those invo‐
359 cations of posix_spawn() and posix_spawnp() that use the additional
360 functionality. Since extensive modifications are not the usual case,
361 and are particularly unlikely in time-critical code, keeping much of
362 the environment control out of posix_spawn() and posix_spawnp() is
363 appropriate design.
364 The posix_spawn() and posix_spawnp() functions do not have all the
366 power of fork()/exec. This is to be expected. The fork() function is a
367 wonderfully powerful operation. We do not expect to duplicate its func‐
368 tionality in a simple, fast function with no special hardware require‐
369 ments. It is worth noting that posix_spawn() and posix_spawnp() are
370 very similar to the process creation operations on many operating sys‐
371 tems that are not UNIX systems.
372 Requirements
374 The requirements for posix_spawn() and posix_spawnp() are:
375 * They must be implementable without an MMU or unusual hardware.
377 * They must be compatible with existing POSIX standards.
379 Additional goals are:
381 * They should be efficiently implementable.
383 * They should be able to replace at least 50% of typical executions
385 of fork().
386 * A system with posix_spawn() and posix_spawnp() and without fork()
388 should be useful, at least for realtime applications.
389 * A system with fork() and the exec family should be able to imple‐
391 ment posix_spawn() and posix_spawnp() as library routines.
392 Two-Syntax
394 POSIX exec has several calling sequences with approximately the same
395 functionality. These appear to be required for compatibility with
396 existing practice. Since the existing practice for the posix_spawn*()
397 functions is otherwise substantially unlike POSIX, we feel that sim‐
398 plicity outweighs compatibility. There are, therefore, only two names
399 for the posix_spawn*() functions.
400 The parameter list does not differ between posix_spawn() and
402 posix_spawnp(); posix_spawnp() interprets the second parameter more
403 elaborately than posix_spawn().
404 Compatibility with POSIX.5 (Ada)
406 The Start_Process and Start_Process_Search procedures from the
407 POSIX_Process_Primitives package from the Ada language binding to
408 POSIX.1 encapsulate fork() and exec functionality in a manner similar
409 to that of posix_spawn() and posix_spawnp(). Originally, in keeping
410 with our simplicity goal, the standard developers had limited the capa‐
411 bilities of posix_spawn() and posix_spawnp() to a subset of the capa‐
412 bilities of Start_Process and Start_Process_Search; certain non-default
413 capabilities were not supported. However, based on suggestions by the
414 ballot group to improve file descriptor mapping or drop it, and on the
415 advice of an Ada Language Bindings working group member, the standard
416 developers decided that posix_spawn() and posix_spawnp() should be suf‐
417 ficiently powerful to implement Start_Process and Start_Process_Search.
418 The rationale is that if the Ada language binding to such a primitive
419 had already been approved as an IEEE standard, there can be little jus‐
420 tification for not approving the functionally-equivalent parts of a C
421 binding. The only three capabilities provided by posix_spawn() and
422 posix_spawnp() that are not provided by Start_Process and
423 Start_Process_Search are optionally specifying the child's process
424 group ID, the set of signals to be reset to default signal handling in
425 the child process, and the child's scheduling policy and parameters.
426 For the Ada language binding for Start_Process to be implemented with
428 posix_spawn(), that binding would need to explicitly pass an empty sig‐
429 nal mask and the parent's environment to posix_spawn() whenever the
430 caller of Start_Process allowed these arguments to default, since
431 posix_spawn() does not provide such defaults. The ability of
432 Start_Process to mask user-specified signals during its execution is
433 functionally unique to the Ada language binding and must be dealt with
434 in the binding separately from the call to posix_spawn().
435 Process Group
437 The process group inheritance field can be used to join the child
438 process with an existing process group. By assigning a value of zero to
439 the spawn-pgroup attribute of the object referenced by attrp, the
440 setpgid() mechanism will place the child process in a new process
441 group.
442 Threads
444 Without the posix_spawn() and posix_spawnp() functions, systems without
445 address translation can still use threads to give an abstraction of
446 concurrency. In many cases, thread creation suffices, but it is not
447 always a good substitute. The posix_spawn() and posix_spawnp() func‐
448 tions are considerably ``heavier'' than thread creation. Processes have
449 several important attributes that threads do not. Even without address
450 translation, a process may have base-and-bound memory protection. Each
451 process has a process environment including security attributes and
452 file capabilities, and powerful scheduling attributes. Processes
453 abstract the behavior of non-uniform-memory-architecture multi-proces‐
454 sors better than threads, and they are more convenient to use for
455 activities that are not closely linked.
456 The posix_spawn() and posix_spawnp() functions may not bring support
458 for multiple processes to every configuration. Process creation is not
459 the only piece of operating system support required to support multiple
460 processes. The total cost of support for multiple processes may be
461 quite high in some circumstances. Existing practice shows that support
462 for multiple processes is uncommon and threads are common among ``tiny
463 kernels''. There should, therefore, probably continue to be AEPs for
464 operating systems with only one process.
465 Asynchronous Error Notification
467 A library implementation of posix_spawn() or posix_spawnp() may not be
468 able to detect all possible errors before it forks the child process.
469 POSIX.1‐2008 provides for an error indication returned from a child
470 process which could not successfully complete the spawn operation via a
471 special exit status which may be detected using the status value
472 returned by wait(), waitid(), and waitpid().
473 The stat_val interface and the macros used to interpret it are not well
475 suited to the purpose of returning API errors, but they are the only
476 path available to a library implementation. Thus, an implementation may
477 cause the child process to exit with exit status 127 for any error
478 detected during the spawn process after the posix_spawn() or
479 posix_spawnp() function has successfully returned.
480 The standard developers had proposed using two additional macros to
482 interpret stat_val. The first, WIFSPAWNFAIL, would have detected a
483 status that indicated that the child exited because of an error
484 detected during the posix_spawn() or posix_spawnp() operations rather
485 than during actual execution of the child process image; the second,
486 WSPAWNERRNO, would have extracted the error value if WIFSPAWNFAIL indi‐
487 cated a failure. Unfortunately, the ballot group strongly opposed this
488 because it would make a library implementation of posix_spawn() or
489 posix_spawnp() dependent on kernel modifications to waitpid() to be
490 able to embed special information in stat_val to indicate a spawn fail‐
491 ure.
492 The 8 bits of child process exit status that are guaranteed by
494 POSIX.1‐2008 to be accessible to the waiting parent process are insuf‐
495 ficient to disambiguate a spawn error from any other kind of error that
496 may be returned by an arbitrary process image. No other bits of the
497 exit status are required to be visible in stat_val, so these macros
498 could not be strictly implemented at the library level. Reserving an
499 exit status of 127 for such spawn errors is consistent with the use of
500 this value by system() and popen() to signal failures in these opera‐
501 tions that occur after the function has returned but before a shell is
502 able to execute. The exit status of 127 does not uniquely identify this
503 class of error, nor does it provide any detailed information on the
504 nature of the failure. Note that a kernel implementation of
505 posix_spawn() or posix_spawnp() is permitted (and encouraged) to return
506 any possible error as the function value, thus providing more detailed
507 failure information to the parent process.
508 Thus, no special macros are available to isolate asynchronous
510 posix_spawn() or posix_spawnp() errors. Instead, errors detected by the
511 posix_spawn() or posix_spawnp() operations in the context of the child
512 process before the new process image executes are reported by setting
513 the child's exit status to 127. The calling process may use the WIFEX‐
514 ITED and WEXITSTATUS macros on the stat_val stored by the wait() or
515 waitpid() functions to detect spawn failures to the extent that other
516 status values with which the child process image may exit (before the
517 parent can conclusively determine that the child process image has
518 begun execution) are distinct from exit status 127.
519
521 None.
522
524 alarm(), chmod(), close(), dup(), exec, exit(), fcntl(), fork(),
525 fstatat(), kill(), open(), posix_spawn_file_actions_addclose(),
526 posix_spawn_file_actions_adddup2(), posix_spawn_file_actions_destroy(),
527 posix_spawnattr_destroy(), posix_spawnattr_getsigdefault(), posix_spaw‐
528 nattr_getflags(), posix_spawnattr_getpgroup(), posix_spaw‐
529 nattr_getschedparam(), posix_spawnattr_getschedpolicy(), posix_spaw‐
530 nattr_getsigmask(), sched_setparam(), sched_setscheduler(), setpgid(),
531 setuid(), times(), wait(), waitid()
532 The Base Definitions volume of POSIX.1‐2017, Chapter 8, Environment
534 Variables, <spawn.h>
535
537 Portions of this text are reprinted and reproduced in electronic form
538 from IEEE Std 1003.1-2017, Standard for Information Technology -- Por‐
539 table Operating System Interface (POSIX), The Open Group Base Specifi‐
540 cations Issue 7, 2018 Edition, Copyright (C) 2018 by the Institute of
541 Electrical and Electronics Engineers, Inc and The Open Group. In the
542 event of any discrepancy between this version and the original IEEE and
543 The Open Group Standard, the original IEEE and The Open Group Standard
544 is the referee document. The original Standard can be obtained online
545 at http://www.opengroup.org/unix/online.html .
546 Any typographical or formatting errors that appear in this page are
548 most likely to have been introduced during the conversion of the source
549 files to man page format. To report such errors, see https://www.ker‐
550 nel.org/doc/man-pages/reporting_bugs.html .
551IEEE/The Open Group 2017 POSIX_SPAWN(3P)