fcntl.rst 6.98 KB
Newer Older
1 2
:mod:`fcntl` --- The ``fcntl`` and ``ioctl`` system calls
=========================================================
3 4 5 6 7

.. module:: fcntl
   :platform: Unix
   :synopsis: The fcntl() and ioctl() system calls.

8
.. sectionauthor:: Jaap Vermeulen
9 10

.. index::
Georg Brandl's avatar
Georg Brandl committed
11 12
   pair: UNIX; file control
   pair: UNIX; I/O control
13

14 15
----------------

16
This module performs file control and I/O control on file descriptors. It is an
17 18 19
interface to the :c:func:`fcntl` and :c:func:`ioctl` Unix routines.  For a
complete description of these calls, see :manpage:`fcntl(2)` and
:manpage:`ioctl(2)` Unix manual pages.
20 21 22

All functions in this module take a file descriptor *fd* as their first
argument.  This can be an integer file descriptor, such as returned by
23
``sys.stdin.fileno()``, or an :class:`io.IOBase` object, such as ``sys.stdin``
24 25
itself, which provides a :meth:`~io.IOBase.fileno` that returns a genuine file
descriptor.
26

27
.. versionchanged:: 3.3
28 29
   Operations in this module used to raise an :exc:`IOError` where they now
   raise an :exc:`OSError`.
30 31


32 33 34
The module defines the following functions:


35
.. function:: fcntl(fd, cmd, arg=0)
36

37
   Perform the operation *cmd* on file descriptor *fd* (file objects providing
38
   a :meth:`~io.IOBase.fileno` method are accepted as well).  The values used
39
   for *cmd* are operating system dependent, and are available as constants
40
   in the :mod:`fcntl` module, using the same names as used in the relevant C
41 42 43 44 45 46 47 48 49 50 51 52
   header files. The argument *arg* can either be an integer value, or a
   :class:`bytes` object. With an integer value, the return value of this
   function is the integer return value of the C :c:func:`fcntl` call.  When
   the argument is bytes it represents a binary structure, e.g. created by
   :func:`struct.pack`. The binary data is copied to a buffer whose address is
   passed to the C :c:func:`fcntl` call.  The return value after a successful
   call is the contents of the buffer, converted to a :class:`bytes` object.
   The length of the returned object will be the same as the length of the
   *arg* argument. This is limited to 1024 bytes. If the information returned
   in the buffer by the operating system is larger than 1024 bytes, this is
   most likely to result in a segmentation violation or a more subtle data
   corruption.
53

54
   If the :c:func:`fcntl` fails, an :exc:`OSError` is raised.
55 56


57
.. function:: ioctl(fd, request, arg=0, mutate_flag=True)
58

59 60
   This function is identical to the :func:`~fcntl.fcntl` function, except
   that the argument handling is even more complicated.
61

62 63
   The *request* parameter is limited to values that can fit in 32-bits.
   Additional constants of interest for use as the *request* argument can be
64 65
   found in the :mod:`termios` module, under the same names as used in
   the relevant C header files.
66

67 68 69
   The parameter *arg* can be one of an integer, an object supporting the
   read-only buffer interface (like :class:`bytes`) or an object supporting
   the read-write buffer interface (like :class:`bytearray`).
70

71 72
   In all but the last case, behaviour is as for the :func:`~fcntl.fcntl`
   function.
73 74 75 76 77 78

   If a mutable buffer is passed, then the behaviour is determined by the value of
   the *mutate_flag* parameter.

   If it is false, the buffer's mutability is ignored and behaviour is as for a
   read-only buffer, except that the 1024 byte limit mentioned above is avoided --
79
   so long as the buffer you pass is at least as long as what the operating system
80 81
   wants to put there, things should work.

82 83 84 85 86 87 88
   If *mutate_flag* is true (the default), then the buffer is (in effect) passed
   to the underlying :func:`ioctl` system call, the latter's return code is
   passed back to the calling Python, and the buffer's new contents reflect the
   action of the :func:`ioctl`.  This is a slight simplification, because if the
   supplied buffer is less than 1024 bytes long it is first copied into a static
   buffer 1024 bytes long which is then passed to :func:`ioctl` and copied back
   into the supplied buffer.
89

90
   If the :c:func:`ioctl` fails, an :exc:`OSError` exception is raised.
91

92 93 94 95 96 97 98 99 100 101 102 103 104 105
   An example::

      >>> import array, fcntl, struct, termios, os
      >>> os.getpgrp()
      13341
      >>> struct.unpack('h', fcntl.ioctl(0, termios.TIOCGPGRP, "  "))[0]
      13341
      >>> buf = array.array('h', [0])
      >>> fcntl.ioctl(0, termios.TIOCGPGRP, buf, 1)
      0
      >>> buf
      array('h', [13341])


106
.. function:: flock(fd, operation)
107

108
   Perform the lock operation *operation* on file descriptor *fd* (file objects providing
109
   a :meth:`~io.IOBase.fileno` method are accepted as well). See the Unix manual
110
   :manpage:`flock(2)` for details.  (On some systems, this function is emulated
111
   using :c:func:`fcntl`.)
112

113
   If the :c:func:`flock` fails, an :exc:`OSError` exception is raised.
114

115

116
.. function:: lockf(fd, cmd, len=0, start=0, whence=0)
117

118
   This is essentially a wrapper around the :func:`~fcntl.fcntl` locking calls.
119
   *fd* is the file descriptor of the file to lock or unlock, and *cmd*
120
   is one of the following values:
121 122 123 124 125

   * :const:`LOCK_UN` -- unlock
   * :const:`LOCK_SH` -- acquire a shared lock
   * :const:`LOCK_EX` -- acquire an exclusive lock

126
   When *cmd* is :const:`LOCK_SH` or :const:`LOCK_EX`, it can also be
127
   bitwise ORed with :const:`LOCK_NB` to avoid blocking on lock acquisition.
128
   If :const:`LOCK_NB` is used and the lock cannot be acquired, an
129
   :exc:`OSError` will be raised and the exception will have an *errno*
130 131 132 133 134
   attribute set to :const:`EACCES` or :const:`EAGAIN` (depending on the
   operating system; for portability, check for both values).  On at least some
   systems, :const:`LOCK_EX` can only be used if the file descriptor refers to a
   file opened for writing.

135
   *len* is the number of bytes to lock, *start* is the byte offset at
136 137
   which the lock starts, relative to *whence*, and *whence* is as with
   :func:`io.IOBase.seek`, specifically:
138

139 140 141
   * :const:`0` -- relative to the start of the file (:data:`os.SEEK_SET`)
   * :const:`1` -- relative to the current buffer position (:data:`os.SEEK_CUR`)
   * :const:`2` -- relative to the end of the file (:data:`os.SEEK_END`)
142 143

   The default for *start* is 0, which means to start at the beginning of the file.
144
   The default for *len* is 0 which means to lock to the end of the file.  The
145 146 147 148 149 150 151 152 153 154 155 156 157
   default for *whence* is also 0.

Examples (all on a SVR4 compliant system)::

   import struct, fcntl, os

   f = open(...)
   rv = fcntl.fcntl(f, fcntl.F_SETFL, os.O_NDELAY)

   lockdata = struct.pack('hhllhh', fcntl.F_WRLCK, 0, 0, 0, 0, 0)
   rv = fcntl.fcntl(f, fcntl.F_SETLKW, lockdata)

Note that in the first example the return value variable *rv* will hold an
158 159 160
integer value; in the second example it will hold a :class:`bytes` object.  The
structure lay-out for the *lockdata* variable is system dependent --- therefore
using the :func:`flock` call may be better.
161 162 163 164 165


.. seealso::

   Module :mod:`os`
166 167 168 169
      If the locking flags :data:`~os.O_SHLOCK` and :data:`~os.O_EXLOCK` are
      present in the :mod:`os` module (on BSD only), the :func:`os.open`
      function provides an alternative to the :func:`lockf` and :func:`flock`
      functions.
170