man2/vfork.2

   1 .\" Copyright (c) 1999 Andries Brouwer (aeb@cwi.nl), 1 Nov 1999
   2 .\"
   3 .\" Permission is granted to make and distribute verbatim copies of this
   4 .\" manual provided the copyright notice and this permission notice are
   5 .\" preserved on all copies.
   6 .\"
   7 .\" Permission is granted to copy and distribute modified versions of this
   8 .\" manual under the conditions for verbatim copying, provided that the
   9 .\" entire resulting derived work is distributed under the terms of a
  10 .\" permission notice identical to this one.
  11 .\"
  12 .\" Since the Linux kernel and libraries are constantly changing, this
  13 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  14 .\" responsibility for errors or omissions, or for damages resulting from
  15 .\" the use of the information contained herein.  The author(s) may not
  16 .\" have taken the same level of care in the production of this manual,
  17 .\" which is licensed free of charge, as they might when working
  18 .\" professionally.
  19 .\"
  20 .\" Formatted or processed versions of this manual, if unaccompanied by
  21 .\" the source, must acknowledge the copyright and authors of this work.
  22 .\"
  23 .\" 1999-11-10: Merged text taken from the page contributed by
  24 .\" Reed H. Petty (rhp@draper.net)
  25 .\"
  26 .TH VFORK 2 2007-07-26 "Linux" "Linux Programmer's Manual"
  27 .SH NAME
  28 vfork \- create a child process and block parent
  29 .SH SYNOPSIS
  30 .B #include <sys/types.h>
  31 .br
  32 .B #include <unistd.h>
  33 .sp
  34 .B pid_t vfork(void);
  35 .sp
  36 .in -4n
  37 Feature Test Macro Requirements for glibc (see
  38 .BR feature_test_macros (7)):
  39 .in
  40 .sp
  41 .BR vfork ():
  42 _BSD_SOURCE || _XOPEN_SOURCE\ >=\ 500
  43 .SH DESCRIPTION
  44 .SS "Standard Description"
  45 (From SUSv2 / POSIX draft.)
  46 The
  47 .BR vfork ()
  48 function has the same effect as
  49 .BR fork (2),
  50 except that the behavior is undefined if the process created by
  51 .BR vfork ()
  52 either modifies any data other than a variable of type
  53 .I pid_t
  54 used to store the return value from
  55 .BR vfork (),
  56 or returns from the function in which
  57 .BR vfork ()
  58 was called, or calls any other function before successfully calling
  59 .BR _exit (2)
  60 or one of the
  61 .BR exec (3)
  62 family of functions.
  63 .SS "Linux Description"
  64 .BR vfork (),
  65 just like
  66 .BR fork (2),
  67 creates a child process of the calling process.
  68 For details and return value and errors, see
  69 .BR fork (2).
  70 .PP
  71 .BR vfork ()
  72 is a special case of
  73 .BR clone (2).
  74 It is used to create new processes without copying the page tables of
  75 the parent process.
  76 It may be useful in performance sensitive applications
  77 where a child will be created which then immediately issues an
  78 .BR execve (2) .
  79 .PP
  80 .BR vfork ()
  81 differs from
  82 .BR fork (2)
  83 in that the parent is suspended until the child makes a call to
  84 .BR execve (2)
  85 or
  86 .BR _exit (2).
  87 The child shares all memory with its parent, including the stack, until
  88 .BR execve (2)
  89 is issued by the child.
  90 The child must not return from the current function or call
  91 .BR exit (3),
  92 but may call
  93 .BR _exit (2).
  94 .PP
  95 Signal handlers are inherited, but not shared.
  96 Signals to the parent
  97 arrive after the child releases the parent's memory.
  98 .SS "Historic Description"
  99 Under Linux,
 100 .BR fork (2)
 101 is implemented using copy-on-write pages, so the only penalty incurred by
 102 .BR fork (2)
 103 is the time and memory required to duplicate the parent's page tables,
 104 and to create a unique task structure for the child.
 105 However, in the bad old days a
 106 .BR fork (2)
 107 would require making a complete copy of the caller's data space,
 108 often needlessly, since usually immediately afterwards an
 109 .BR exec (3)
 110 is done.
 111 Thus, for greater efficiency, BSD introduced the
 112 .BR vfork ()
 113 system call, that did not fully copy the address space of
 114 the parent process, but borrowed the parent's memory and thread
 115 of control until a call to
 116 .BR execve (2)
 117 or an exit occurred.
 118 The parent process was suspended while the
 119 child was using its resources.
 120 The use of
 121 .BR vfork ()
 122 was tricky: for example, not modifying data
 123 in the parent process depended on knowing which variables are
 124 held in a register.
 125 .SH "CONFORMING TO"
 126 4.3BSD, POSIX.1-2001.
 127 .\" FIXME . Mar 07: in the draft of the next POSIX revision, the spec for
 128 .\" vfork() has been removed.  See if this remains so in final version.
 129 The requirements put on
 130 .BR vfork ()
 131 by the standards are weaker than those put on
 132 .BR fork (2),
 133 so an implementation where the two are synonymous is compliant.
 134 In particular, the programmer cannot
 135 rely on the parent remaining blocked until a call of
 136 .BR execve (2)
 137 or
 138 .BR _exit (2)
 139 and cannot rely on any specific behavior wth respect to shared memory.
 140 .\" In AIXv3.1 vfork is equivalent to fork.
 141 .SH NOTES
 142 .SS Linux Notes
 143 Fork handlers established using
 144 .BR pthread_atfork (3)
 145 are not called when a multithreaded program employing
 146 the NPTL threading library calls
 147 .BR vfork ().
 148 Fork handlers are called in this case in a program using the
 149 LinuxThreads threading library.
 150 (See
 151 .BR pthreads (7)
 152 for a description of Linux threading libraries.)
 153 .SS History
 154 The
 155 .BR vfork ()
 156 system call appeared in 3.0BSD.
 157 .\" In the release notes for 4.2BSD Sam Leffler wrote: `vfork: Is still
 158 .\" present, but definitely on its way out'.
 159 In 4.4BSD it was made synonymous to
 160 .BR fork (2)
 161 but NetBSD introduced it again,
 162 cf. http://www.netbsd.org/Documentation/kernel/vfork.html .
 163 In Linux, it has been equivalent to
 164 .BR fork (2)
 165 until 2.2.0-pre6 or so.
 166 Since 2.2.0-pre9 (on i386, somewhat later on
 167 other architectures) it is an independent system call.
 168 Support was added in glibc 2.0.112.
 169 .SH BUGS
 170 It is rather unfortunate that Linux revived this specter from the past.
 171 The BSD man page states:
 172 "This system call will be eliminated when proper system sharing mechanisms
 173 are implemented.
 174 Users should not depend on the memory sharing semantics of
 175 .BR vfork ()
 176 as it will, in that case, be made synonymous to
 177 .BR fork (2).\c
 178 "
 179
 180 Details of the signal handling are obscure and differ between systems.
 181 The BSD man page states:
 182 "To avoid a possible deadlock situation, processes that are children
 183 in the middle of a
 184 .BR vfork ()
 185 are never sent
 186 .B SIGTTOU
 187 or
 188 .B SIGTTIN
 189 signals; rather, output or
 190 .IR ioctl s
 191 are allowed and input attempts result in an end-of-file indication."
 192 .\"
 193 .\" As far as I can tell, the following is not true in 2.6.19:
 194 .\" Currently (Linux 2.3.25),
 195 .\" .BR strace (1)
 196 .\" cannot follow
 197 .\" .BR vfork ()
 198 .\" and requires a kernel patch.
 199 .SH "SEE ALSO"
 200 .BR clone (2),
 201 .BR execve (2),
 202 .BR fork (2),
 203 .BR unshare (2),
 204 .BR wait (2)