1 .\" This manpage is copyright (C) 1992 Drew Eckhardt,
2 .\" copyright (C) 1995 Michael Shields.
4 .\" %%%LICENSE_START(VERBATIM)
5 .\" Permission is granted to make and distribute verbatim copies of this
6 .\" manual provided the copyright notice and this permission notice are
7 .\" preserved on all copies.
9 .\" Permission is granted to copy and distribute modified versions of this
10 .\" manual under the conditions for verbatim copying, provided that the
11 .\" entire resulting derived work is distributed under the terms of a
12 .\" permission notice identical to this one.
14 .\" Since the Linux kernel and libraries are constantly changing, this
15 .\" manual page may be incorrect or out-of-date. The author(s) assume no
16 .\" responsibility for errors or omissions, or for damages resulting from
17 .\" the use of the information contained herein. The author(s) may not
18 .\" have taken the same level of care in the production of this manual,
19 .\" which is licensed free of charge, as they might when working
22 .\" Formatted or processed versions of this manual, if unaccompanied by
23 .\" the source, must acknowledge the copyright and authors of this work.
26 .\" Modified 1993-07-24 by Rik Faith <faith@cs.unc.edu>
27 .\" Modified 1995-05-18 by Jim Van Zandt <jrv@vanzandt.mv.com>
28 .\" Sun Feb 11 14:07:00 MET 1996 Martin Schulze <joey@linux.de>
29 .\" * layout slightly modified
31 .\" Modified Mon Oct 21 23:05:29 EDT 1996 by Eric S. Raymond <esr@thyrsus.com>
32 .\" Modified Thu Feb 24 01:41:09 CET 2000 by aeb
33 .\" Modified Thu Feb 9 22:32:09 CET 2001 by bert hubert <ahu@ds9a.nl>, aeb
34 .\" Modified Mon Nov 11 14:35:00 PST 2002 by Ben Woodard <ben@zork.net>
35 .\" 2005-03-11, mtk, modified pselect() text (it is now a system
38 .TH SELECT 2 2017-03-13 "Linux" "Linux Programmer's Manual"
40 select, pselect, FD_CLR, FD_ISSET, FD_SET, FD_ZERO \-
41 synchronous I/O multiplexing
44 /* According to POSIX.1-2001, POSIX.1-2008 */
46 .B #include <sys/select.h>
48 /* According to earlier standards */
50 .B #include <sys/time.h>
52 .B #include <sys/types.h>
54 .B #include <unistd.h>
56 .BI "int select(int " nfds ", fd_set *" readfds ", fd_set *" writefds ,
57 .BI " fd_set *" exceptfds ", struct timeval *" timeout );
59 .BI "void FD_CLR(int " fd ", fd_set *" set );
61 .BI "int FD_ISSET(int " fd ", fd_set *" set );
63 .BI "void FD_SET(int " fd ", fd_set *" set );
65 .BI "void FD_ZERO(fd_set *" set );
67 .B #include <sys/select.h>
69 .BI "int pselect(int " nfds ", fd_set *" readfds ", fd_set *" writefds ,
70 .BI " fd_set *" exceptfds ", const struct timespec *" timeout ,
71 .BI " const sigset_t *" sigmask );
75 Feature Test Macro Requirements for glibc (see
76 .BR feature_test_macros (7)):
80 _POSIX_C_SOURCE\ >=\ 200112L
85 allow a program to monitor multiple file descriptors,
86 waiting until one or more of the file descriptors become "ready"
87 for some class of I/O operation (e.g., input possible).
88 A file descriptor is considered ready if it is possible to
89 perform a corresponding I/O operation (e.g.,
91 without blocking, or a sufficiently small
95 can monitor only file descriptors numbers that are less than
98 does not have this limitation.
105 is identical, other than these three differences:
109 uses a timeout that is a
111 (with seconds and microseconds), while
115 (with seconds and nanoseconds).
121 argument to indicate how much time was left.
123 does not change this argument.
129 argument, and behaves as
134 Three independent sets of file descriptors are watched.
137 will be watched to see if characters become
138 available for reading (more precisely, to see if a read will not
139 block; in particular, a file descriptor is also ready on end-of-file),
142 will be watched to see if space is available for write (though a large
143 write may still block), and those in
145 will be watched for exceptions.
146 On exit, the sets are modified in place
147 to indicate which file descriptors actually changed status.
148 Each of the three file descriptor sets may be specified as NULL
149 if no file descriptors are to be watched for the corresponding class
152 Four macros are provided to manipulate the sets.
158 respectively add and remove a given file descriptor from a set.
160 tests to see if a file descriptor is part of the set;
166 should be set to the highest-numbered file descriptor in any
167 of the three sets, plus 1.
168 The indicated file descriptors in each set are checked, up to this limit
173 argument specifies the interval that
175 should block waiting for a file descriptor to become ready.
176 The call will block until either:
178 a file descriptor becomes ready;
180 the call is interrupted by a signal handler; or
186 interval will be rounded up to the system clock granularity,
187 and kernel scheduling delays mean that the blocking interval
188 may overrun by a small amount.
189 If both fields of the
191 structure are zero, then
194 (This is useful for polling.)
197 is NULL (no timeout),
199 can block indefinitely.
202 is a pointer to a signal mask (see
203 .BR sigprocmask (2));
204 if it is not NULL, then
206 first replaces the current signal mask by the one pointed to by
208 then does the "select" function, and then restores the original
211 Other than the difference in the precision of the
213 argument, the following
218 ready = pselect(nfds, &readfds, &writefds, &exceptfds,
224 executing the following calls:
229 pthread_sigmask(SIG_SETMASK, &sigmask, &origmask);
230 ready = select(nfds, &readfds, &writefds, &exceptfds, timeout);
231 pthread_sigmask(SIG_SETMASK, &origmask, NULL);
236 is needed is that if one wants to wait for either a signal
237 or for a file descriptor to become ready, then
238 an atomic test is needed to prevent race conditions.
239 (Suppose the signal handler sets a global flag and
241 Then a test of this global flag followed by a call of
243 could hang indefinitely if the signal arrived just after the test
244 but just before the call.
247 allows one to first block signals, handle the signals that have come in,
254 The time structures involved are defined in
261 long tv_sec; /* seconds */
262 long tv_usec; /* microseconds */
272 long tv_sec; /* seconds */
273 long tv_nsec; /* nanoseconds */
278 (However, see below on the POSIX.1 versions.)
282 with all three sets empty,
286 as a fairly portable way to sleep with subsecond precision.
292 to reflect the amount of time not slept; most other implementations
294 (POSIX.1 permits either behavior.)
295 This causes problems both when Linux code which reads
297 is ported to other operating systems, and when code is ported to Linux
298 that reuses a \fIstruct timeval\fP for multiple
300 in a loop without reinitializing it.
303 to be undefined after
306 .\" .PP - it is rumored that:
307 .\" On BSD, when a timeout occurs, the file descriptor bits are not changed.
308 .\" - it is certainly true that:
309 .\" Linux follows SUSv2 and sets the bit masks to zero upon a timeout.
315 return the number of file descriptors contained in the three returned
316 descriptor sets (that is, the total number of bits that are set in
320 which may be zero if the timeout expires before anything interesting happens.
321 On error, \-1 is returned, and
323 is set to indicate the error;
324 the file descriptor sets are unmodified,
331 An invalid file descriptor was given in one of the sets.
332 (Perhaps a file descriptor that was already closed,
333 or one on which an error has occurred.)
337 A signal was caught; see
342 is negative or exceeds the
348 The value contained within
353 Unable to allocate memory for internal tables.
356 was added to Linux in kernel 2.6.16.
359 was emulated in glibc (but see BUGS).
362 conforms to POSIX.1-2001, POSIX.1-2008, and
365 first appeared in 4.2BSD).
366 Generally portable to/from
367 non-BSD systems supporting clones of the BSD socket layer (including
369 However, note that the System\ V variant typically
370 sets the timeout variable before exit, but the BSD variant does not.
373 is defined in POSIX.1g, and in
374 POSIX.1-2001 and POSIX.1-2008.
378 is a fixed size buffer.
385 that is negative or is equal to or larger than
388 in undefined behavior.
389 Moreover, POSIX requires
391 to be a valid file descriptor.
393 On some other UNIX systems,
394 .\" Darwin, according to a report by Jeremy Sequoia, relayed by Josh Triplett
396 can fail with the error
398 if the system fails to allocate kernel-internal resources, rather than
401 POSIX specifies this error for
405 Portable programs may wish to check for
407 and loop, just as with
412 reliable (and more portable) signal trapping can be achieved
413 using the self-pipe trick.
415 a signal handler writes a byte to a pipe whose other end
419 (To avoid possibly blocking when writing to a pipe that may be full
420 or reading from a pipe that may be empty,
421 nonblocking I/O is used when reading from and writing to the pipe.)
423 Concerning the types involved, the classical situation is that
426 structure are typed as
428 (as shown above), and the structure is defined in
430 The POSIX.1 situation is
435 time_t tv_sec; /* seconds */
436 suseconds_t tv_usec; /* microseconds */
441 where the structure is defined in
450 Concerning prototypes, the classical situation is that one should
455 The POSIX.1 situation is that one should include
464 gives the wrong prototype for
466 Under glibc 2.1 to 2.2.1, it gives
471 Since glibc 2.2.2, the requirements are as shown in the SYNOPSIS.
473 .SS Correspondence between select() and poll() notifications
474 Within the Linux kernel source,
476 we find the following definitions which show the correspondence
477 between the readable, writable, and exceptional condition notifications of
479 and the event notifications provided by
486 #define POLLIN_SET (POLLRDNORM | POLLRDBAND | POLLIN | POLLHUP |
488 /* Ready for reading */
489 #define POLLOUT_SET (POLLWRBAND | POLLWRNORM | POLLOUT | POLLERR)
490 /* Ready for writing */
491 #define POLLEX_SET (POLLPRI)
492 /* Exceptional condition */
496 .SS Multithreaded applications
497 If a file descriptor being monitored by
499 is closed in another thread, the result is unspecified.
500 On some UNIX systems,
502 unblocks and returns, with an indication that the file descriptor is ready
503 (a subsequent I/O operation will likely fail with an error,
504 unless another the file descriptor reopened between the time
506 returned and the I/O operations was performed).
507 On Linux (and some other systems),
508 closing the file descriptor in another thread has no effect on
510 In summary, any application that relies on a particular behavior
511 in this scenario must be considered buggy.
513 .SS C library/kernel differences
514 The Linux kernel allows file descriptor sets of arbitrary size,
515 determining the length of the sets to be checked from the value of
517 However, in the glibc implementation, the
519 type is fixed in size.
524 interface described in this page is implemented by glibc.
525 The underlying Linux system call is named
527 This system call has somewhat different behavior from the glibc
532 system call modifies its
535 However, the glibc wrapper function hides this behavior
536 by using a local variable for the timeout argument that
537 is passed to the system call.
540 function does not modify its
543 this is the behavior required by POSIX.1-2001.
545 The final argument of the
549 pointer, but is instead a structure of the form:
554 const kernel_sigset_t *ss; /* Pointer to signal set */
555 size_t ss_len; /* Size (in bytes) of object
556 pointed to by 'ss' */
561 This allows the system call to obtain both
562 a pointer to the signal set and its size,
563 while allowing for the fact that most architectures
564 support a maximum of 6 arguments to a system call.
567 for a discussion of the difference between the kernel and libc
568 notion of the signal set.
570 POSIX allows an implementation to define an upper limit,
571 advertised via the constant
573 on the range of file descriptors that can be specified
574 in a file descriptor set.
575 The Linux kernel imposes no fixed limit, but the glibc implementation makes
577 a fixed-size type, with
579 defined as 1024, and the
581 macros operating according to that limit.
582 To monitor file descriptors greater than 1023, use
588 should check all specified file descriptors in the three file descriptor sets,
591 However, the current implementation ignores any file descriptor in
592 these sets that is greater than the maximum file descriptor number
593 that the process currently has open.
594 According to POSIX, any such file descriptor that is specified in one
595 of the sets should result in the error
598 Glibc 2.0 provided a version of
604 Starting with version 2.1, glibc provided an emulation of
606 that was implemented using
610 This implementation remained vulnerable to the very race condition that
612 was designed to prevent.
613 Modern versions of glibc use the (race-free)
615 system call on kernels where it is provided.
619 may report a socket file descriptor as "ready for reading", while
620 nevertheless a subsequent read blocks.
621 This could for example
622 happen when data has arrived but upon examination has wrong
623 checksum and is discarded.
624 There may be other circumstances
625 in which a file descriptor is spuriously reported as ready.
626 .\" Stevens discusses a case where accept can block after select
627 .\" returns successfully because of an intervening RST from the client.
628 Thus it may be safer to use
630 on sockets that should not block.
631 .\" Maybe the kernel should have returned EIO in such a situation?
637 if the call is interrupted by a signal handler (i.e., the
640 This is not permitted by POSIX.1.
643 system call has the same behavior,
644 but the glibc wrapper hides this behavior by internally copying the
646 to a local variable and passing that variable to the system call.
651 #include <sys/time.h>
652 #include <sys/types.h>
662 /* Watch stdin (fd 0) to see when it has input. */
667 /* Wait up to five seconds. */
672 retval = select(1, &rfds, NULL, NULL, &tv);
673 /* Don't rely on the value of tv now! */
678 printf("Data is available now.\\n");
679 /* FD_ISSET(0, &rfds) will be true. */
681 printf("No data within five seconds.\\n");
692 .BR restart_syscall (2),
699 For a tutorial with discussion and examples, see