man2/write.2

   1 .\" This manpage is Copyright (C) 1992 Drew Eckhardt;
   2 .\"             and Copyright (C) 1993 Michael Haardt, Ian Jackson.
   3 .\" and Copyright (C) 2007 Michael Kerrisk <mtk.manpages@gmail.com>
   4 .\"
   5 .\" %%%LICENSE_START(VERBATIM)
   6 .\" Permission is granted to make and distribute verbatim copies of this
   7 .\" manual provided the copyright notice and this permission notice are
   8 .\" preserved on all copies.
   9 .\"
  10 .\" Permission is granted to copy and distribute modified versions of this
  11 .\" manual under the conditions for verbatim copying, provided that the
  12 .\" entire resulting derived work is distributed under the terms of a
  13 .\" permission notice identical to this one.
  14 .\"
  15 .\" Since the Linux kernel and libraries are constantly changing, this
  16 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  17 .\" responsibility for errors or omissions, or for damages resulting from
  18 .\" the use of the information contained herein.  The author(s) may not
  19 .\" have taken the same level of care in the production of this manual,
  20 .\" which is licensed free of charge, as they might when working
  21 .\" professionally.
  22 .\"
  23 .\" Formatted or processed versions of this manual, if unaccompanied by
  24 .\" the source, must acknowledge the copyright and authors of this work.
  25 .\" %%%LICENSE_END
  26 .\"
  27 .\" Modified Sat Jul 24 13:35:59 1993 by Rik Faith <faith@cs.unc.edu>
  28 .\" Modified Sun Nov 28 17:19:01 1993 by Rik Faith <faith@cs.unc.edu>
  29 .\" Modified Sat Jan 13 12:58:08 1996 by Michael Haardt
  30 .\"   <michael@cantor.informatik.rwth-aachen.de>
  31 .\" Modified Sun Jul 21 18:59:33 1996 by Andries Brouwer <aeb@cwi.nl>
  32 .\" 2001-12-13 added remark by Zack Weinberg
  33 .\" 2007-06-18 mtk:
  34 .\"     Added details about seekable files and file offset.
  35 .\"     Noted that write() may write less than 'count' bytes, and
  36 .\"     gave some examples of why this might occur.
  37 .\"     Noted what happens if write() is interrupted by a signal.
  38 .\"
  39 .TH WRITE 2 2018-02-02 "Linux" "Linux Programmer's Manual"
  40 .SH NAME
  41 write \- write to a file descriptor
  42 .SH SYNOPSIS
  43 .B #include <unistd.h>
  44 .PP
  45 .BI "ssize_t write(int " fd ", const void *" buf ", size_t " count );
  46 .SH DESCRIPTION
  47 .BR write ()
  48 writes up to
  49 .I count
  50 bytes from the buffer starting at
  51 .I buf
  52 to the file referred to by the file descriptor
  53 .IR fd .
  54 .PP
  55 The number of bytes written may be less than
  56 .I count
  57 if, for example,
  58 there is insufficient space on the underlying physical medium, or the
  59 .B RLIMIT_FSIZE
  60 resource limit is encountered (see
  61 .BR setrlimit (2)),
  62 or the call was interrupted by a signal
  63 handler after having written less than
  64 .I count
  65 bytes.
  66 (See also
  67 .BR pipe (7).)
  68 .PP
  69 For a seekable file (i.e., one to which
  70 .BR lseek (2)
  71 may be applied, for example, a regular file)
  72 writing takes place at the file offset,
  73 and the file offset is incremented by
  74 the number of bytes actually written.
  75 If the file was
  76 .BR open (2)ed
  77 with
  78 .BR O_APPEND ,
  79 the file offset is first set to the end of the file before writing.
  80 The adjustment of the file offset and the write operation
  81 are performed as an atomic step.
  82 .PP
  83 POSIX requires that a
  84 .BR read (2)
  85 that can be proved to occur after a
  86 .BR write ()
  87 has returned will return the new data.
  88 Note that not all filesystems are POSIX conforming.
  89 .PP
  90 According to POSIX.1, if
  91 .I count
  92 is greater than
  93 .BR SSIZE_MAX ,
  94 the result is implementation-defined;
  95 see NOTES for the upper limit on Linux.
  96 .SH RETURN VALUE
  97 On success, the number of bytes written is returned.
  98 On error, \-1 is returned, and \fIerrno\fP is set
  99 to indicate the cause of the error.
 100 .PP
 101 Note that a successful
 102 .BR write ()
 103 may transfer fewer than
 104 .I count
 105 bytes.
 106 Such partial writes can occur for various reasons;
 107 for example, because there was insufficient space on the disk device
 108 to write all of the requested bytes, or because a blocked
 109 .BR write ()
 110 to a socket, pipe, or similar was interrupted by a signal handler
 111 after it had transferred some, but before it had transferred all
 112 of the requested bytes.
 113 In the event of a partial write, the caller can make another
 114 .BR write ()
 115 call to transfer the remaining bytes.
 116 The subsequent call will either transfer further bytes or
 117 may result in an error (e.g., if the disk is now full).
 118 .PP
 119 If \fIcount\fP is zero and
 120 .I fd
 121 refers to a regular file, then
 122 .BR write ()
 123 may return a failure status if one of the errors below is detected.
 124 If no errors are detected, or error detection is not performed,
 125 0 will be returned without causing any other effect.
 126 If
 127 \fIcount\fP is zero and
 128 .I fd
 129 refers to a file other than a regular file,
 130 the results are not specified.
 131 .SH ERRORS
 132 .TP
 133 .B EAGAIN
 134 The file descriptor
 135 .I fd
 136 refers to a file other than a socket and has been marked nonblocking
 137 .RB ( O_NONBLOCK ),
 138 and the write would block.
 139 See
 140 .BR open (2)
 141 for further details on the
 142 .BR O_NONBLOCK
 143 flag.
 144 .TP
 145 .BR EAGAIN " or " EWOULDBLOCK
 146 .\" Actually EAGAIN on Linux
 147 The file descriptor
 148 .I fd
 149 refers to a socket and has been marked nonblocking
 150 .RB ( O_NONBLOCK ),
 151 and the write would block.
 152 POSIX.1-2001 allows either error to be returned for this case,
 153 and does not require these constants to have the same value,
 154 so a portable application should check for both possibilities.
 155 .TP
 156 .B EBADF
 157 .I fd
 158 is not a valid file descriptor or is not open for writing.
 159 .TP
 160 .B EDESTADDRREQ
 161 .I fd
 162 refers to a datagram socket for which a peer address has not been set using
 163 .BR connect (2).
 164 .TP
 165 .B EDQUOT
 166 The user's quota of disk blocks on the filesystem containing the file
 167 referred to by
 168 .I fd
 169 has been exhausted.
 170 .TP
 171 .B EFAULT
 172 .I buf
 173 is outside your accessible address space.
 174 .TP
 175 .B EFBIG
 176 An attempt was made to write a file that exceeds the implementation-defined
 177 maximum file size or the process's file size limit,
 178 or to write at a position past the maximum allowed offset.
 179 .TP
 180 .B EINTR
 181 The call was interrupted by a signal before any data was written; see
 182 .BR signal (7).
 183 .TP
 184 .B EINVAL
 185 .I fd
 186 is attached to an object which is unsuitable for writing;
 187 or the file was opened with the
 188 .B O_DIRECT
 189 flag, and either the address specified in
 190 .IR buf ,
 191 the value specified in
 192 .IR count ,
 193 or the file offset is not suitably aligned.
 194 .TP
 195 .B EIO
 196 A low-level I/O error occurred while modifying the inode.
 197 This error may relate to the write-back of data written by an earlier
 198 .BR write (2),
 199 which may have been issued to a different file descriptor on
 200 the same file.
 201 Since Linux 4.13, errors from write-back come
 202 with a promise that they
 203 .I may
 204 be reported by subsequent.
 205 .BR write (2)
 206 requests, and
 207 .I will
 208 be reported by a subsequent
 209 .BR fsync (2)
 210 (whether or not they were also reported by
 211 .BR write (2)).
 212 .\" commit 088737f44bbf6378745f5b57b035e57ee3dc4750
 213 An alternate cause of
 214 .B EIO
 215 on networked filesystems is when an advisory lock had been taken out
 216 on the file descriptor and this lock has been lost.
 217 See the
 218 .I "Lost locks"
 219 section of
 220 .BR fcntl (2)
 221 for further details.
 222 .TP
 223 .B ENOSPC
 224 The device containing the file referred to by
 225 .I fd
 226 has no room for the data.
 227 .TP
 228 .B EPERM
 229 The operation was prevented by a file seal; see
 230 .BR fcntl (2).
 231 .TP
 232 .B EPIPE
 233 .I fd
 234 is connected to a pipe or socket whose reading end is closed.
 235 When this happens the writing process will also receive a
 236 .B SIGPIPE
 237 signal.
 238 (Thus, the write return value is seen only if the program
 239 catches, blocks or ignores this signal.)
 240 .PP
 241 Other errors may occur, depending on the object connected to
 242 .IR fd .
 243 .SH CONFORMING TO
 244 SVr4, 4.3BSD, POSIX.1-2001.
 245 .\" SVr4 documents additional error
 246 .\" conditions EDEADLK, ENOLCK, ENOLNK, ENOSR, ENXIO, or ERANGE.
 247 .PP
 248 Under SVr4 a write may be interrupted and return
 249 .B EINTR
 250 at any point,
 251 not just before any data is written.
 252 .SH NOTES
 253 The types
 254 .I size_t
 255 and
 256 .I ssize_t
 257 are, respectively,
 258 unsigned and signed integer data types specified by POSIX.1.
 259 .PP
 260 A successful return from
 261 .BR write ()
 262 does not make any guarantee that data has been committed to disk.
 263 On some filesystems, including NFS, it does not even guarantee
 264 that space has successfully been reserved for the data.
 265 In this case,
 266 some errors might be delayed until a future
 267 .BR write (2),
 268 .BR fsync (2),
 269 or even
 270 .BR close (2).
 271 The only way to be sure is to call
 272 .BR fsync (2)
 273 after you are done writing all your data.
 274 .PP
 275 If a
 276 .BR write ()
 277 is interrupted by a signal handler before any bytes are written,
 278 then the call fails with the error
 279 .BR EINTR ;
 280 if it is interrupted after at least one byte has been written,
 281 the call succeeds, and returns the number of bytes written.
 282 .PP
 283 On Linux,
 284 .BR write ()
 285 (and similar system calls) will transfer at most
 286 0x7ffff000 (2,147,479,552) bytes,
 287 returning the number of bytes actually transferred.
 288 .\" commit e28cc71572da38a5a12c1cfe4d7032017adccf69
 289 (This is true on both 32-bit and 64-bit systems.)
 290 .PP
 291 An error return value while performing
 292 .BR write ()
 293 using direct I/O does not mean the
 294 entire write has failed. Partial data may be written
 295 and the data at the file offset on which the
 296 .BR write ()
 297 was attempted should be considered inconsistent.
 298 .SH BUGS
 299 According to POSIX.1-2008/SUSv4 Section XSI 2.9.7
 300 ("Thread Interactions with Regular File Operations"):
 301 .PP
 302 .RS 4
 303 All of the following functions shall be atomic with respect to
 304 each other in the effects specified in POSIX.1-2008 when they
 305 operate on regular files or symbolic links: ...
 306 .RE
 307 .PP
 308 Among the APIs subsequently listed are
 309 .BR write ()
 310 and
 311 .BR writev (2).
 312 And among the effects that should be atomic across threads (and processes)
 313 are updates of the file offset.
 314 However, on Linux before version 3.14,
 315 this was not the case: if two processes that share
 316 an open file description (see
 317 .BR open (2))
 318 perform a
 319 .BR write ()
 320 (or
 321 .BR writev (2))
 322 at the same time, then the I/O operations were not atomic
 323 with respect updating the file offset,
 324 with the result that the blocks of data output by the two processes
 325 might (incorrectly) overlap.
 326 This problem was fixed in Linux 3.14.
 327 .\" http://thread.gmane.org/gmane.linux.kernel/1649458
 328 .\"    From: Michael Kerrisk (man-pages <mtk.manpages <at> gmail.com>
 329 .\"    Subject: Update of file offset on write() etc. is non-atomic with I/O
 330 .\"    Date: 2014-02-17 15:41:37 GMT
 331 .\"    Newsgroups: gmane.linux.kernel, gmane.linux.file-systems
 332 .\" commit 9c225f2655e36a470c4f58dbbc99244c5fc7f2d4
 333 .\"    Author: Linus Torvalds <torvalds@linux-foundation.org>
 334 .\"    Date:   Mon Mar 3 09:36:58 2014 -0800
 335 .\"
 336 .\"        vfs: atomic f_pos accesses as per POSIX
 337 .SH SEE ALSO
 338 .BR close (2),
 339 .BR fcntl (2),
 340 .BR fsync (2),
 341 .BR ioctl (2),
 342 .BR lseek (2),
 343 .BR open (2),
 344 .BR pwrite (2),
 345 .BR read (2),
 346 .BR select (2),
 347 .BR writev (2),
 348 .BR fwrite (3)