man2/dup.2

   1 .\" This manpage is Copyright (C) 1992 Drew Eckhardt;
   2 .\" and Copyright (C) 1993 Michael Haardt, Ian Jackson.
   3 .\" and Copyright (C) 2005, 2008 Michael Kerrisk <mtk.manpages@gmail.com>
   4 .\" and Copyright (C) 2014 Michael Kerrisk <mtk.manpages@gmail.com>
   5 .\"
   6 .\" %%%LICENSE_START(VERBATIM)
   7 .\" Permission is granted to make and distribute verbatim copies of this
   8 .\" manual provided the copyright notice and this permission notice are
   9 .\" preserved on all copies.
  10 .\"
  11 .\" Permission is granted to copy and distribute modified versions of this
  12 .\" manual under the conditions for verbatim copying, provided that the
  13 .\" entire resulting derived work is distributed under the terms of a
  14 .\" permission notice identical to this one.
  15 .\"
  16 .\" Since the Linux kernel and libraries are constantly changing, this
  17 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  18 .\" responsibility for errors or omissions, or for damages resulting from
  19 .\" the use of the information contained herein.  The author(s) may not
  20 .\" have taken the same level of care in the production of this manual,
  21 .\" which is licensed free of charge, as they might when working
  22 .\" professionally.
  23 .\"
  24 .\" Formatted or processed versions of this manual, if unaccompanied by
  25 .\" the source, must acknowledge the copyright and authors of this work.
  26 .\" %%%LICENSE_END
  27 .\"
  28 .\" Modified 1993-07-21, Rik Faith <faith@cs.unc.edu>
  29 .\" Modified 1994-08-21, Michael Chastain <mec@shell.portal.com>:
  30 .\"   Fixed typoes.
  31 .\" Modified 1997-01-31, Eric S. Raymond <esr@thyrsus.com>
  32 .\" Modified 2002-09-28, aeb
  33 .\" 2009-01-12, mtk, reordered text in DESCRIPTION and added some
  34 .\"     details for dup2().
  35 .\" 2008-10-09, mtk: add description of dup3()
  36 .\"
  37 .TH DUP 2 2015-01-22 "Linux" "Linux Programmer's Manual"
  38 .SH NAME
  39 dup, dup2, dup3 \- duplicate a file descriptor
  40 .SH SYNOPSIS
  41 .nf
  42 .B #include <unistd.h>
  43 .sp
  44 .BI "int dup(int " oldfd );
  45 .BI "int dup2(int " oldfd ", int " newfd );
  46 .sp
  47 .BR "#define _GNU_SOURCE" "             /* See feature_test_macros(7) */"
  48 .BR "#include <fcntl.h>" "              /* Obtain O_* constant definitions */
  49 .B #include <unistd.h>
  50 .sp
  51 .BI "int dup3(int " oldfd ", int " newfd ", int " flags );
  52 .fi
  53 .SH DESCRIPTION
  54 The
  55 .BR dup ()
  56 system call creates a copy of the file descriptor
  57 .IR oldfd ,
  58 using the lowest-numbered unused descriptor for the new descriptor.
  59
  60 After a successful return,
  61 the old and new file descriptors may be used interchangeably.
  62 They refer to the same open file description (see
  63 .BR open (2))
  64 and thus share file offset and file status flags;
  65 for example, if the file offset is modified by using
  66 .BR lseek (2)
  67 on one of the descriptors, the offset is also changed for the other.
  68
  69 The two descriptors do not share file descriptor flags
  70 (the close-on-exec flag).
  71 The close-on-exec flag
  72 .RB ( FD_CLOEXEC ;
  73 see
  74 .BR fcntl (2))
  75 for the duplicate descriptor is off.
  76 .\"
  77 .SS dup2()
  78 The
  79 .BR dup2 ()
  80 system call performs the same task as
  81 .BR dup (),
  82 but instead of using the lowest-numbered unused file descriptor,
  83 it uses the descriptor number specified in
  84 .IR newfd .
  85 If the descriptor
  86 .IR newfd
  87 was previously open, it is silently closed before being reused.
  88
  89 The steps of closing and reusing the file descriptor
  90 .IR newfd
  91 are performed
  92 .IR atomically .
  93 This is important, because trying to implement equivalent functionality using
  94 .BR close (2)
  95 and
  96 .BR dup ()
  97 would be
  98 subject to race conditions, whereby
  99 .I newfd
 100 might be reused between the two steps.
 101 Such reuse could happen because the main program is interrupted
 102 by a signal handler that allocates a file descriptor,
 103 or because a parallel thread allocates a file descriptor.
 104
 105 Note the following points:
 106 .IP * 3
 107 If
 108 .I oldfd
 109 is not a valid file descriptor, then the call fails, and
 110 .I newfd
 111 is not closed.
 112 .IP *
 113 If
 114 .I oldfd
 115 is a valid file descriptor, and
 116 .I newfd
 117 has the same value as
 118 .IR oldfd ,
 119 then
 120 .BR dup2 ()
 121 does nothing, and returns
 122 .IR newfd .
 123 .\"
 124 .SS dup3()
 125 .BR dup3 ()
 126 is the same as
 127 .BR dup2 (),
 128 except that:
 129 .IP * 3
 130 The caller can force the close-on-exec flag to be set
 131 for the new file descriptor by specifying
 132 .BR O_CLOEXEC
 133 in
 134 .IR flags .
 135 See the description of the same flag in
 136 .BR open (2)
 137 for reasons why this may be useful.
 138 .IP *
 139 .\" Ulrich Drepper, LKML, 2008-10-09:
 140 .\"     We deliberately decided on this change.  Otherwise, what is the
 141 .\"     result of dup3(fd, fd, O_CLOEXEC)?
 142 If
 143 .IR oldfd
 144 equals
 145 .IR newfd ,
 146 then
 147 .BR dup3 ()
 148 fails with the error
 149 .BR EINVAL .
 150 .SH RETURN VALUE
 151 On success, these system calls
 152 return the new descriptor.
 153 On error, \-1 is returned, and
 154 .I errno
 155 is set appropriately.
 156 .SH ERRORS
 157 .TP
 158 .B EBADF
 159 .I oldfd
 160 isn't an open file descriptor.
 161 .TP
 162 .B EBADF
 163 .I newfd
 164 is out of the allowed range for file descriptors (see the discussion of
 165 .BR RLIMIT_NOFILE
 166 in
 167 .BR getrlimit (2)).
 168 .TP
 169 .B EBUSY
 170 (Linux only) This may be returned by
 171 .BR dup2 ()
 172 or
 173 .BR dup3 ()
 174 during a race condition with
 175 .BR open (2)
 176 and
 177 .BR dup ().
 178 .TP
 179 .B EINTR
 180 The
 181 .BR dup2 ()
 182 or
 183 .BR dup3 ()
 184 call was interrupted by a signal; see
 185 .BR signal (7).
 186 .TP
 187 .B EINVAL
 188 .RB ( dup3 ())
 189 .I flags
 190 contain an invalid value.
 191 .TP
 192 .B EINVAL
 193 .RB ( dup3 ())
 194 .\" FIXME . To confirm with Al Viro that this was intended, and its rationale
 195 .I oldfd
 196 was equal to
 197 .IR newfd .
 198 .TP
 199 .B EMFILE
 200 The process already has the maximum number of file
 201 descriptors open and tried to open a new one (see the discussion of
 202 .BR RLIMIT_NOFILE
 203 in
 204 .BR getrlimit (2)).
 205 .SH VERSIONS
 206 .BR dup3 ()
 207 was added to Linux in version 2.6.27;
 208 glibc support is available starting with
 209 version 2.9.
 210 .SH CONFORMING TO
 211 .BR dup (),
 212 .BR dup2 ():
 213 POSIX.1-2001, POSIX.1-2008, SVr4, 4.3BSD.
 214
 215 .BR dup3 ()
 216 is Linux-specific.
 217 .\" SVr4 documents additional
 218 .\" EINTR and ENOLINK error conditions.  POSIX.1 adds EINTR.
 219 .\" The EBUSY return is Linux-specific.
 220 .SH NOTES
 221 The error returned by
 222 .BR dup2 ()
 223 is different from that returned by
 224 .BR fcntl( "..., " F_DUPFD ", ..." )
 225 when
 226 .I newfd
 227 is out of range.
 228 On some systems,
 229 .BR dup2 ()
 230 also sometimes returns
 231 .B EINVAL
 232 like
 233 .BR F_DUPFD .
 234
 235 If
 236 .I newfd
 237 was open, any errors that would have been reported at
 238 .BR close (2)
 239 time are lost.
 240 If this is of concern,
 241 then\(emunless the program is single-threaded and does not allocate
 242 file descriptors in signal handlers\(emthe correct approach is
 243 .I not
 244 to close
 245 .I newfd
 246 before calling
 247 .BR dup2 (),
 248 because of the race condition described above.
 249 Instead, code something like the following could be used:
 250
 251 .nf
 252     /* Obtain a duplicate of 'newfd' that can subsequently
 253        be used to check for close() errors; an EBADF error
 254        means that 'newfd' was not open. */
 255
 256     tmpfd = dup(newfd);
 257     if (tmpfd == \-1 && errno != EBADF) {
 258         /* Handle unexpected dup() error */
 259     }
 260
 261     /* Atomically duplicate 'oldfd' on 'newfd' */
 262
 263     if (dup2(oldfd, newfd) == \-1) {
 264         /* Handle dup2() error */
 265     }
 266
 267     /* Now check for close() errors on the file originally
 268        referred to by 'newfd' */
 269
 270     if (tmpfd != \-1) {
 271         if (close(tmpfd) == \-1) {
 272             /* Handle errors from close */
 273         }
 274     }
 275 .fi
 276 .SH SEE ALSO
 277 .BR close (2),
 278 .BR fcntl (2),
 279 .BR open (2)