1WRITE(3P) POSIX Programmer's Manual WRITE(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 pwrite, write — write on a file
13
15 #include <unistd.h>
16 ssize_t pwrite(int fildes, const void *buf, size_t nbyte,
18 off_t offset);
19 ssize_t write(int fildes, const void *buf, size_t nbyte);
20
22 The write() function shall attempt to write nbyte bytes from the buffer
23 pointed to by buf to the file associated with the open file descriptor,
24 fildes.
25 Before any action described below is taken, and if nbyte is zero and
27 the file is a regular file, the write() function may detect and return
28 errors as described below. In the absence of errors, or if error detec‐
29 tion is not performed, the write() function shall return zero and have
30 no other results. If nbyte is zero and the file is not a regular file,
31 the results are unspecified.
32 On a regular file or other file capable of seeking, the actual writing
34 of data shall proceed from the position in the file indicated by the
35 file offset associated with fildes. Before successful return from
36 write(), the file offset shall be incremented by the number of bytes
37 actually written. On a regular file, if the position of the last byte
38 written is greater than or equal to the length of the file, the length
39 of the file shall be set to this position plus one.
40 On a file not capable of seeking, writing shall always take place
42 starting at the current position. The value of a file offset associated
43 with such a device is undefined.
44 If the O_APPEND flag of the file status flags is set, the file offset
46 shall be set to the end of the file prior to each write and no inter‐
47 vening file modification operation shall occur between changing the
48 file offset and the write operation.
49 If a write() requests that more bytes be written than there is room for
51 (for example, the file size limit of the process or the physical end of
52 a medium), only as many bytes as there is room for shall be written.
53 For example, suppose there is space for 20 bytes more in a file before
54 reaching a limit. A write of 512 bytes will return 20. The next write
55 of a non-zero number of bytes would give a failure return (except as
56 noted below).
57 If the request would cause the file size to exceed the soft file size
59 limit for the process and there is no room for any bytes to be written,
60 the request shall fail and the implementation shall generate the
61 SIGXFSZ signal for the thread.
62 If write() is interrupted by a signal before it writes any data, it
64 shall return -1 with errno set to [EINTR].
65 If write() is interrupted by a signal after it successfully writes some
67 data, it shall return the number of bytes written.
68 If the value of nbyte is greater than {SSIZE_MAX}, the result is imple‐
70 mentation-defined.
71 After a write() to a regular file has successfully returned:
73 * Any successful read() from each byte position in the file that was
75 modified by that write shall return the data specified by the
76 write() for that position until such byte positions are again modi‐
77 fied.
78 * Any subsequent successful write() to the same byte position in the
80 file shall overwrite that file data.
81 Write requests to a pipe or FIFO shall be handled in the same way as a
83 regular file with the following exceptions:
84 * There is no file offset associated with a pipe, hence each write
86 request shall append to the end of the pipe.
87 * Write requests of {PIPE_BUF} bytes or less shall not be interleaved
89 with data from other processes doing writes on the same pipe.
90 Writes of greater than {PIPE_BUF} bytes may have data interleaved,
91 on arbitrary boundaries, with writes by other processes, whether or
92 not the O_NONBLOCK flag of the file status flags is set.
93 * If the O_NONBLOCK flag is clear, a write request may cause the
95 thread to block, but on normal completion it shall return nbyte.
96 * If the O_NONBLOCK flag is set, write() requests shall be handled
98 differently, in the following ways:
99 -- The write() function shall not block the thread.
101 -- A write request for {PIPE_BUF} or fewer bytes shall have the
103 following effect: if there is sufficient space available in the
104 pipe, write() shall transfer all the data and return the number
105 of bytes requested. Otherwise, write() shall transfer no data
106 and return -1 with errno set to [EAGAIN].
107 -- A write request for more than {PIPE_BUF} bytes shall cause one
109 of the following:
110 -- When at least one byte can be written, transfer what it can
112 and return the number of bytes written. When all data pre‐
113 viously written to the pipe is read, it shall transfer at
114 least {PIPE_BUF} bytes.
115 -- When no data can be written, transfer no data, and return
117 -1 with errno set to [EAGAIN].
118 When attempting to write to a file descriptor (other than a pipe or
120 FIFO) that supports non-blocking writes and cannot accept the data
121 immediately:
122 * If the O_NONBLOCK flag is clear, write() shall block the calling
124 thread until the data can be accepted.
125 * If the O_NONBLOCK flag is set, write() shall not block the thread.
127 If some data can be written without blocking the thread, write()
128 shall write what it can and return the number of bytes written.
129 Otherwise, it shall return -1 and set errno to [EAGAIN].
130 Upon successful completion, where nbyte is greater than 0, write()
132 shall mark for update the last data modification and last file status
133 change timestamps of the file, and if the file is a regular file, the
134 S_ISUID and S_ISGID bits of the file mode may be cleared.
135 For regular files, no data transfer shall occur past the offset maximum
137 established in the open file description associated with fildes.
138 If fildes refers to a socket, write() shall be equivalent to send()
140 with no flags set.
141 If the O_DSYNC bit has been set, write I/O operations on the file
143 descriptor shall complete as defined by synchronized I/O data integrity
144 completion.
145 If the O_SYNC bit has been set, write I/O operations on the file
147 descriptor shall complete as defined by synchronized I/O file integrity
148 completion.
149 If fildes refers to a shared memory object, the result of the write()
151 function is unspecified.
152 If fildes refers to a typed memory object, the result of the write()
154 function is unspecified.
155 If fildes refers to a STREAM, the operation of write() shall be deter‐
157 mined by the values of the minimum and maximum nbyte range (packet
158 size) accepted by the STREAM. These values are determined by the top‐
159 most STREAM module. If nbyte falls within the packet size range, nbyte
160 bytes shall be written. If nbyte does not fall within the range and the
161 minimum packet size value is 0, write() shall break the buffer into
162 maximum packet size segments prior to sending the data downstream (the
163 last segment may contain less than the maximum packet size). If nbyte
164 does not fall within the range and the minimum value is non-zero,
165 write() shall fail with errno set to [ERANGE]. Writing a zero-length
166 buffer (nbyte is 0) to a STREAMS device sends 0 bytes with 0 returned.
167 However, writing a zero-length buffer to a STREAMS-based pipe or FIFO
168 sends no message and 0 is returned. The process may issue I_SWROPT
169 ioctl() to enable zero-length messages to be sent across the pipe or
170 FIFO.
171 When writing to a STREAM, data messages are created with a priority
173 band of 0. When writing to a STREAM that is not a pipe or FIFO:
174 * If O_NONBLOCK is clear, and the STREAM cannot accept data (the
176 STREAM write queue is full due to internal flow control condi‐
177 tions), write() shall block until data can be accepted.
178 * If O_NONBLOCK is set and the STREAM cannot accept data, write()
180 shall return -1 and set errno to [EAGAIN].
181 * If O_NONBLOCK is set and part of the buffer has been written while
183 a condition in which the STREAM cannot accept additional data
184 occurs, write() shall terminate and return the number of bytes
185 written.
186 In addition, write() shall fail if the STREAM head has processed an
188 asynchronous error before the call. In this case, the value of errno
189 does not reflect the result of write(), but reflects the prior error.
190 The pwrite() function shall be equivalent to write(), except that it
192 writes into a given position and does not change the file offset
193 (regardless of whether O_APPEND is set). The first three arguments to
194 pwrite() are the same as write() with the addition of a fourth argument
195 offset for the desired position inside the file. An attempt to perform
196 a pwrite() on a file that is incapable of seeking shall result in an
197 error.
198
200 Upon successful completion, these functions shall return the number of
201 bytes actually written to the file associated with fildes. This number
202 shall never be greater than nbyte. Otherwise, -1 shall be returned and
203 errno set to indicate the error.
204
206 These functions shall fail if:
207 EAGAIN The file is neither a pipe, nor a FIFO, nor a socket, the O_NON‐
209 BLOCK flag is set for the file descriptor, and the thread would
210 be delayed in the write() operation.
211 EBADF The fildes argument is not a valid file descriptor open for
213 writing.
214 EFBIG An attempt was made to write a file that exceeds the implementa‐
216 tion-defined maximum file size or the file size limit of the
217 process, and there was no room for any bytes to be written.
218 EFBIG The file is a regular file, nbyte is greater than 0, and the
220 starting position is greater than or equal to the offset maximum
221 established in the open file description associated with fildes.
222 EINTR The write operation was terminated due to the receipt of a sig‐
224 nal, and no data was transferred.
225 EIO The process is a member of a background process group attempting
227 to write to its controlling terminal, TOSTOP is set, the calling
228 thread is not blocking SIGTTOU, the process is not ignoring
229 SIGTTOU, and the process group of the process is orphaned. This
230 error may also be returned under implementation-defined condi‐
231 tions.
232 ENOSPC There was no free space remaining on the device containing the
234 file.
235 ERANGE The transfer request size was outside the range supported by the
237 STREAMS file associated with fildes.
238 The pwrite() function shall fail if:
240 EINVAL The file is a regular file or block special file, and the offset
242 argument is negative. The file offset shall remain unchanged.
243 ESPIPE The file is incapable of seeking.
245 The write() function shall fail if:
247 EAGAIN The file is a pipe or FIFO, the O_NONBLOCK flag is set for the
249 file descriptor, and the thread would be delayed in the write
250 operation.
251 EAGAIN or EWOULDBLOCK
253 The file is a socket, the O_NONBLOCK flag is set for the file
254 descriptor, and the thread would be delayed in the write opera‐
255 tion.
256 ECONNRESET
258 A write was attempted on a socket that is not connected.
259 EPIPE An attempt is made to write to a pipe or FIFO that is not open
261 for reading by any process, or that only has one end open. A
262 SIGPIPE signal shall also be sent to the thread.
263 EPIPE A write was attempted on a socket that is shut down for writing,
265 or is no longer connected. In the latter case, if the socket is
266 of type SOCK_STREAM, a SIGPIPE signal shall also be sent to the
267 thread.
268 These functions may fail if:
270 EINVAL The STREAM or multiplexer referenced by fildes is linked
272 (directly or indirectly) downstream from a multiplexer.
273 EIO A physical I/O error has occurred.
275 ENOBUFS
277 Insufficient resources were available in the system to perform
278 the operation.
279 ENXIO A request was made of a nonexistent device, or the request was
281 outside the capabilities of the device.
282 ENXIO A hangup occurred on the STREAM being written to.
284 A write to a STREAMS file may fail if an error message has been
286 received at the STREAM head. In this case, errno is set to the value
287 included in the error message.
288 The write() function may fail if:
290 EACCES A write was attempted on a socket and the calling process does
292 not have appropriate privileges.
293 ENETDOWN
295 A write was attempted on a socket and the local network inter‐
296 face used to reach the destination is down.
297 ENETUNREACH
299 A write was attempted on a socket and no route to the network is
300 present.
301 The following sections are informative.
303
305 Writing from a Buffer
306 The following example writes data from the buffer pointed to by buf to
307 the file associated with the file descriptor fd.
308 #include <sys/types.h>
311 #include <string.h>
312 ...
313 char buf[20];
314 size_t nbytes;
315 ssize_t bytes_written;
316 int fd;
317 ...
318 strcpy(buf, "This is a test\n");
319 nbytes = strlen(buf);
320 bytes_written = write(fd, buf, nbytes);
322 ...
323
325 None.
326
328 See also the RATIONALE section in read().
329 An attempt to write to a pipe or FIFO has several major characteris‐
331 tics:
332 * Atomic/non-atomic: A write is atomic if the whole amount written in
334 one operation is not interleaved with data from any other process.
335 This is useful when there are multiple writers sending data to a
336 single reader. Applications need to know how large a write request
337 can be expected to be performed atomically. This maximum is called
338 {PIPE_BUF}. This volume of POSIX.1‐2017 does not say whether write
339 requests for more than {PIPE_BUF} bytes are atomic, but requires
340 that writes of {PIPE_BUF} or fewer bytes shall be atomic.
341 * Blocking/immediate: Blocking is only possible with O_NONBLOCK
343 clear. If there is enough space for all the data requested to be
344 written immediately, the implementation should do so. Otherwise,
345 the calling thread may block; that is, pause until enough space is
346 available for writing. The effective size of a pipe or FIFO (the
347 maximum amount that can be written in one operation without block‐
348 ing) may vary dynamically, depending on the implementation, so it
349 is not possible to specify a fixed value for it.
350 * Complete/partial/deferred: A write request:
352 int fildes;
355 size_t nbyte;
356 ssize_t ret;
357 char *buf;
358 ret = write(fildes, buf, nbyte);
360 may return:
362 Complete ret=nbyte
364 Partial ret<nbyte
366 This shall never happen if nbyte≤{PIPE_BUF}. If it does
368 happen (with nbyte>{PIPE_BUF}), this volume of
369 POSIX.1‐2017 does not guarantee atomicity, even if
370 ret≤{PIPE_BUF}, because atomicity is guaranteed according
371 to the amount requested, not the amount written.
372 Deferred: ret=-1, errno=[EAGAIN]
374 This error indicates that a later request may succeed. It
376 does not indicate that it shall succeed, even if
377 nbyte≤{PIPE_BUF}, because if no process reads from the
378 pipe or FIFO, the write never succeeds. An application
379 could usefully count the number of times [EAGAIN] is
380 caused by a particular value of nbyte>{PIPE_BUF} and per‐
381 haps do later writes with a smaller value, on the assump‐
382 tion that the effective size of the pipe may have
383 decreased.
384 Partial and deferred writes are only possible with O_NONBLOCK set.
386 The relations of these properties are shown in the following tables:
388 ┌───────────────────────────────────────────────────────────────────────┐
390 │ Write to a Pipe or FIFO with O_NONBLOCK clear │
391 ├─────────────────────┬─────────────────────────────────────────────────┤
392 │Immediately Writable:│ None Some nbyte │
393 ├─────────────────────┼─────────────────────────────────────────────────┤
394 │nbyte≤{PIPE_BUF} │Atomic blocking Atomic blocking Atomic immediate │
395 │ │nbyte nbyte nbyte │
396 ├─────────────────────┼─────────────────────────────────────────────────┤
397 │nbyte>{PIPE_BUF} │Blocking nbyte Blocking nbyte Blocking nbyte │
398 └─────────────────────┴─────────────────────────────────────────────────┘
399 If the O_NONBLOCK flag is clear, a write request shall block if the
400 amount writable immediately is less than that requested. If the flag is
401 set (by fcntl()), a write request shall never block.
402 ┌───────────────────────────────────────────────────────────────┐
404 │ Write to a Pipe or FIFO with O_NONBLOCK set │
405 ├─────────────────────┬─────────────────────────────────────────┤
406 │Immediately Writable:│ None Some nbyte │
407 ├─────────────────────┼─────────────────────────────────────────┤
408 │nbyte≤{PIPE_BUF} │-1, [EAGAIN] -1, [EAGAIN] Atomic nbyte │
409 ├─────────────────────┼─────────────────────────────────────────┤
410 │nbyte>{PIPE_BUF} │-1, [EAGAIN] <nbyte or -1, ≤nbyte or -1, │
411 │ │ [EAGAIN] [EAGAIN] │
412 └─────────────────────┴─────────────────────────────────────────┘
413 There is no exception regarding partial writes when O_NONBLOCK is set.
414 With the exception of writing to an empty pipe, this volume of
415 POSIX.1‐2017 does not specify exactly when a partial write is performed
416 since that would require specifying internal details of the implementa‐
417 tion. Every application should be prepared to handle partial writes
418 when O_NONBLOCK is set and the requested amount is greater than
419 {PIPE_BUF}, just as every application should be prepared to handle par‐
420 tial writes on other kinds of file descriptors.
421 The intent of forcing writing at least one byte if any can be written
423 is to assure that each write makes progress if there is any room in the
424 pipe. If the pipe is empty, {PIPE_BUF} bytes must be written; if not,
425 at least some progress must have been made.
426 Where this volume of POSIX.1‐2017 requires -1 to be returned and errno
428 set to [EAGAIN], most historical implementations return zero (with the
429 O_NDELAY flag set, which is the historical predecessor of O_NONBLOCK,
430 but is not itself in this volume of POSIX.1‐2017). The error indica‐
431 tions in this volume of POSIX.1‐2017 were chosen so that an application
432 can distinguish these cases from end-of-file. While write() cannot
433 receive an indication of end-of-file, read() can, and the two functions
434 have similar return values. Also, some existing systems (for example,
435 Eighth Edition) permit a write of zero bytes to mean that the reader
436 should get an end-of-file indication; for those systems, a return value
437 of zero from write() indicates a successful write of an end-of-file
438 indication.
439 Implementations are allowed, but not required, to perform error check‐
441 ing for write() requests of zero bytes.
442 The concept of a {PIPE_MAX} limit (indicating the maximum number of
444 bytes that can be written to a pipe in a single operation) was consid‐
445 ered, but rejected, because this concept would unnecessarily limit
446 application writing.
447 See also the discussion of O_NONBLOCK in read().
449 Writes can be serialized with respect to other reads and writes. If a
451 read() of file data can be proven (by any means) to occur after a
452 write() of the data, it must reflect that write(), even if the calls
453 are made by different processes. A similar requirement applies to mul‐
454 tiple write operations to the same file position. This is needed to
455 guarantee the propagation of data from write() calls to subsequent
456 read() calls. This requirement is particularly significant for net‐
457 worked file systems, where some caching schemes violate these seman‐
458 tics.
459 Note that this is specified in terms of read() and write(). The XSI
461 extensions readv() and writev() also obey these semantics. A new
462 ``high-performance'' write analog that did not follow these serializa‐
463 tion requirements would also be permitted by this wording. This volume
464 of POSIX.1‐2017 is also silent about any effects of application-level
465 caching (such as that done by stdio).
466 This volume of POSIX.1‐2017 does not specify the value of the file off‐
468 set after an error is returned; there are too many cases. For program‐
469 ming errors, such as [EBADF], the concept is meaningless since no file
470 is involved. For errors that are detected immediately, such as
471 [EAGAIN], clearly the pointer should not change. After an interrupt or
472 hardware error, however, an updated value would be very useful and is
473 the behavior of many implementations.
474 This volume of POSIX.1‐2017 does not specify the behavior of concurrent
476 writes to a regular file from multiple threads, except that each write
477 is atomic (see Section 2.9.7, Thread Interactions with Regular File
478 Operations). Applications should use some form of concurrency control.
479 This volume of POSIX.1‐2017 intentionally does not specify any pwrite()
481 errors related to pipes, FIFOs, and sockets other than [ESPIPE].
482
484 None.
485
487 chmod(), creat(), dup(), fcntl(), getrlimit(), lseek(), open(), pipe(),
488 read(), ulimit(), writev()
489 The Base Definitions volume of POSIX.1‐2017, <limits.h>, <stropts.h>,
491 <sys_uio.h>, <unistd.h>
492
494 Portions of this text are reprinted and reproduced in electronic form
495 from IEEE Std 1003.1-2017, Standard for Information Technology -- Por‐
496 table Operating System Interface (POSIX), The Open Group Base Specifi‐
497 cations Issue 7, 2018 Edition, Copyright (C) 2018 by the Institute of
498 Electrical and Electronics Engineers, Inc and The Open Group. In the
499 event of any discrepancy between this version and the original IEEE and
500 The Open Group Standard, the original IEEE and The Open Group Standard
501 is the referee document. The original Standard can be obtained online
502 at http://www.opengroup.org/unix/online.html .
503 Any typographical or formatting errors that appear in this page are
505 most likely to have been introduced during the conversion of the source
506 files to man page format. To report such errors, see https://www.ker‐
507 nel.org/doc/man-pages/reporting_bugs.html .
508IEEE/The Open Group 2017 WRITE(3P)