man2/read.2

   1 .\" This manpage is Copyright (C) 1992 Drew Eckhardt;
   2 .\"             and Copyright (C) 1993 Michael Haardt, Ian Jackson.
   3 .\"
   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.
   8 .\"
   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.
  13 .\"
  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  15 .\" manual page may be incorrect or out-of-date.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
  20 .\" professionally.
  21 .\"
  22 .\" Formatted or processed versions of this manual, if unaccompanied by
  23 .\" the source, must acknowledge the copyright and authors of this work.
  24 .\" %%%LICENSE_END
  25 .\"
  26 .\" Modified Sat Jul 24 00:06:00 1993 by Rik Faith <faith@cs.unc.edu>
  27 .\" Modified Wed Jan 17 16:02:32 1996 by Michael Haardt
  28 .\"   <michael@cantor.informatik.rwth-aachen.de>
  29 .\" Modified Thu Apr 11 19:26:35 1996 by Andries Brouwer <aeb@cwi.nl>
  30 .\" Modified Sun Jul 21 18:59:33 1996 by Andries Brouwer <aeb@cwi.nl>
  31 .\" Modified Fri Jan 31 16:47:33 1997 by Eric S. Raymond <esr@thyrsus.com>
  32 .\" Modified Sat Jul 12 20:45:39 1997 by Michael Haardt
  33 .\"   <michael@cantor.informatik.rwth-aachen.de>
  34 .\"
  35 .TH READ 2 2018-02-02 "Linux" "Linux Programmer's Manual"
  36 .SH NAME
  37 read \- read from a file descriptor
  38 .SH SYNOPSIS
  39 .nf
  40 .B #include <unistd.h>
  41 .PP
  42 .BI "ssize_t read(int " fd ", void *" buf ", size_t " count );
  43 .fi
  44 .SH DESCRIPTION
  45 .BR read ()
  46 attempts to read up to
  47 .I count
  48 bytes from file descriptor
  49 .I fd
  50 into the buffer starting at
  51 .IR buf .
  52 .PP
  53 On files that support seeking,
  54 the read operation commences at the file offset,
  55 and the file offset is incremented by the number of bytes read.
  56 If the file offset is at or past the end of file,
  57 no bytes are read, and
  58 .BR read ()
  59 returns zero.
  60 .PP
  61 If
  62 .I count
  63 is zero,
  64 .BR read ()
  65 .I may
  66 detect the errors described below.
  67 In the absence of any errors,
  68 or if
  69 .BR read ()
  70 does not check for errors, a
  71 .BR read ()
  72 with a
  73 .I count
  74 of 0 returns zero and has no other effects.
  75 .PP
  76 According to POSIX.1, if
  77 .I count
  78 is greater than
  79 .BR SSIZE_MAX ,
  80 the result is implementation-defined;
  81 see NOTES for the upper limit on Linux.
  82 .SH RETURN VALUE
  83 On success, the number of bytes read is returned (zero indicates end of
  84 file), and the file position is advanced by this number.
  85 It is not an error if this number is smaller than the number of bytes
  86 requested; this may happen for example because fewer bytes are actually
  87 available right now (maybe because we were close to end-of-file, or
  88 because we are reading from a pipe, or from a terminal), or because
  89 .BR read ()
  90 was interrupted by a signal.
  91 See also NOTES.
  92 .PP
  93 On error, \-1 is returned, and
  94 .I errno
  95 is set appropriately.
  96 In this case, it is left unspecified whether
  97 the file position (if any) changes.
  98 .SH ERRORS
  99 .TP
 100 .B EAGAIN
 101 The file descriptor
 102 .I fd
 103 refers to a file other than a socket and has been marked nonblocking
 104 .RB ( O_NONBLOCK ),
 105 and the read would block.
 106 See
 107 .BR open (2)
 108 for further details on the
 109 .BR O_NONBLOCK
 110 flag.
 111 .TP
 112 .BR EAGAIN " or " EWOULDBLOCK
 113 .\" Actually EAGAIN on Linux
 114 The file descriptor
 115 .I fd
 116 refers to a socket and has been marked nonblocking
 117 .RB ( O_NONBLOCK ),
 118 and the read would block.
 119 POSIX.1-2001 allows either error to be returned for this case,
 120 and does not require these constants to have the same value,
 121 so a portable application should check for both possibilities.
 122 .TP
 123 .B EBADF
 124 .I fd
 125 is not a valid file descriptor or is not open for reading.
 126 .TP
 127 .B EFAULT
 128 .I buf
 129 is outside your accessible address space.
 130 .TP
 131 .B EINTR
 132 The call was interrupted by a signal before any data was read; see
 133 .BR signal (7).
 134 .TP
 135 .B EINVAL
 136 .I fd
 137 is attached to an object which is unsuitable for reading;
 138 or the file was opened with the
 139 .B O_DIRECT
 140 flag, and either the address specified in
 141 .IR buf ,
 142 the value specified in
 143 .IR count ,
 144 or the file offset is not suitably aligned.
 145 .TP
 146 .B EINVAL
 147 .I fd
 148 was created via a call to
 149 .BR timerfd_create (2)
 150 and the wrong size buffer was given to
 151 .BR read ();
 152 see
 153 .BR timerfd_create (2)
 154 for further information.
 155 .TP
 156 .B EIO
 157 I/O error.
 158 This will happen for example when the process is in a
 159 background process group, tries to read from its controlling terminal,
 160 and either it is ignoring or blocking
 161 .B SIGTTIN
 162 or its process group
 163 is orphaned.
 164 It may also occur when there is a low-level I/O error
 165 while reading from a disk or tape.
 166 A further possible cause of
 167 .B EIO
 168 on networked filesystems is when an advisory lock had been taken
 169 out on the file descriptor and this lock has been lost.
 170 See the
 171 .I "Lost locks"
 172 section of
 173 .BR fcntl (2)
 174 for further details.
 175 .TP
 176 .B EISDIR
 177 .I fd
 178 refers to a directory.
 179 .PP
 180 Other errors may occur, depending on the object connected to
 181 .IR fd .
 182 .SH CONFORMING TO
 183 SVr4, 4.3BSD, POSIX.1-2001.
 184 .SH NOTES
 185 The types
 186 .I size_t
 187 and
 188 .I ssize_t
 189 are, respectively,
 190 unsigned and signed integer data types specified by POSIX.1.
 191 .PP
 192 On Linux,
 193 .BR read ()
 194 (and similar system calls) will transfer at most
 195 0x7ffff000 (2,147,479,552) bytes,
 196 returning the number of bytes actually transferred.
 197 .\" commit e28cc71572da38a5a12c1cfe4d7032017adccf69
 198 (This is true on both 32-bit and 64-bit systems.)
 199 .PP
 200 On NFS filesystems, reading small amounts of data will update the
 201 timestamp only the first time, subsequent calls may not do so.
 202 This is caused
 203 by client side attribute caching, because most if not all NFS clients
 204 leave
 205 .I st_atime
 206 (last file access time)
 207 updates to the server, and client side reads satisfied from the
 208 client's cache will not cause
 209 .I st_atime
 210 updates on the server as there are no
 211 server-side reads.
 212 UNIX semantics can be obtained by disabling client-side attribute caching,
 213 but in most situations this will substantially
 214 increase server load and decrease performance.
 215 .SH BUGS
 216 According to POSIX.1-2008/SUSv4 Section XSI 2.9.7
 217 ("Thread Interactions with Regular File Operations"):
 218 .PP
 219 .RS 4
 220 All of the following functions shall be atomic with respect to
 221 each other in the effects specified in POSIX.1-2008 when they
 222 operate on regular files or symbolic links: ...
 223 .RE
 224 .PP
 225 Among the APIs subsequently listed are
 226 .BR read ()
 227 and
 228 .BR readv (2).
 229 And among the effects that should be atomic across threads (and processes)
 230 are updates of the file offset.
 231 However, on Linux before version 3.14,
 232 this was not the case: if two processes that share
 233 an open file description (see
 234 .BR open (2))
 235 perform a
 236 .BR read ()
 237 (or
 238 .BR readv (2))
 239 at the same time, then the I/O operations were not atomic
 240 with respect updating the file offset,
 241 with the result that the reads in the two processes
 242 might (incorrectly) overlap in the blocks of data that they obtained.
 243 This problem was fixed in Linux 3.14.
 244 .\" http://thread.gmane.org/gmane.linux.kernel/1649458
 245 .\"    From: Michael Kerrisk (man-pages <mtk.manpages <at> gmail.com>
 246 .\"    Subject: Update of file offset on write() etc. is non-atomic with I/O
 247 .\"    Date: 2014-02-17 15:41:37 GMT
 248 .\"    Newsgroups: gmane.linux.kernel, gmane.linux.file-systems
 249 .\" commit 9c225f2655e36a470c4f58dbbc99244c5fc7f2d4
 250 .\"    Author: Linus Torvalds <torvalds@linux-foundation.org>
 251 .\"    Date:   Mon Mar 3 09:36:58 2014 -0800
 252 .\"
 253 .\"        vfs: atomic f_pos accesses as per POSIX
 254 .SH SEE ALSO
 255 .BR close (2),
 256 .BR fcntl (2),
 257 .BR ioctl (2),
 258 .BR lseek (2),
 259 .BR open (2),
 260 .BR pread (2),
 261 .BR readdir (2),
 262 .BR readlink (2),
 263 .BR readv (2),
 264 .BR select (2),
 265 .BR write (2),
 266 .BR fread (3)