1VFORK(2) Linux Programmer's Manual VFORK(2)
2
6 vfork - create a child process and block parent
7
9 #include <unistd.h>
10 pid_t vfork(void);
12 Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
14 vfork():
16 Since glibc 2.12:
17 (_XOPEN_SOURCE >= 500) && ! (_POSIX_C_SOURCE >= 200809L)
18 || /* Since glibc 2.19: */ _DEFAULT_SOURCE
19 || /* Glibc <= 2.19: */ _BSD_SOURCE
20 Before glibc 2.12:
21 _BSD_SOURCE || _XOPEN_SOURCE >= 500
22
24 Standard description
25 (From POSIX.1) The vfork() function has the same effect as fork(2), ex‐
26 cept that the behavior is undefined if the process created by vfork()
27 either modifies any data other than a variable of type pid_t used to
28 store the return value from vfork(), or returns from the function in
29 which vfork() was called, or calls any other function before success‐
30 fully calling _exit(2) or one of the exec(3) family of functions.
31 Linux description
33 vfork(), just like fork(2), creates a child process of the calling
34 process. For details and return value and errors, see fork(2).
35 vfork() is a special case of clone(2). It is used to create new pro‐
37 cesses without copying the page tables of the parent process. It may
38 be useful in performance-sensitive applications where a child is cre‐
39 ated which then immediately issues an execve(2).
40 vfork() differs from fork(2) in that the calling thread is suspended
42 until the child terminates (either normally, by calling _exit(2), or
43 abnormally, after delivery of a fatal signal), or it makes a call to
44 execve(2). Until that point, the child shares all memory with its par‐
45 ent, including the stack. The child must not return from the current
46 function or call exit(3) (which would have the effect of calling exit
47 handlers established by the parent process and flushing the parent's
48 stdio(3) buffers), but may call _exit(2).
49 As with fork(2), the child process created by vfork() inherits copies
51 of various of the caller's process attributes (e.g., file descriptors,
52 signal dispositions, and current working directory); the vfork() call
53 differs only in the treatment of the virtual address space, as de‐
54 scribed above.
55 Signals sent to the parent arrive after the child releases the parent's
57 memory (i.e., after the child terminates or calls execve(2)).
58 Historic description
60 Under Linux, fork(2) is implemented using copy-on-write pages, so the
61 only penalty incurred by fork(2) is the time and memory required to du‐
62 plicate the parent's page tables, and to create a unique task structure
63 for the child. However, in the bad old days a fork(2) would require
64 making a complete copy of the caller's data space, often needlessly,
65 since usually immediately afterward an exec(3) is done. Thus, for
66 greater efficiency, BSD introduced the vfork() system call, which did
67 not fully copy the address space of the parent process, but borrowed
68 the parent's memory and thread of control until a call to execve(2) or
69 an exit occurred. The parent process was suspended while the child was
70 using its resources. The use of vfork() was tricky: for example, not
71 modifying data in the parent process depended on knowing which vari‐
72 ables were held in a register.
73
75 4.3BSD; POSIX.1-2001 (but marked OBSOLETE). POSIX.1-2008 removes the
76 specification of vfork().
77 The requirements put on vfork() by the standards are weaker than those
79 put on fork(2), so an implementation where the two are synonymous is
80 compliant. In particular, the programmer cannot rely on the parent re‐
81 maining blocked until the child either terminates or calls execve(2),
82 and cannot rely on any specific behavior with respect to shared memory.
83
85 Some consider the semantics of vfork() to be an architectural blemish,
86 and the 4.2BSD man page stated: "This system call will be eliminated
87 when proper system sharing mechanisms are implemented. Users should
88 not depend on the memory sharing semantics of vfork() as it will, in
89 that case, be made synonymous to fork(2)." However, even though modern
90 memory management hardware has decreased the performance difference be‐
91 tween fork(2) and vfork(), there are various reasons why Linux and
92 other systems have retained vfork():
93 * Some performance-critical applications require the small performance
95 advantage conferred by vfork().
96 * vfork() can be implemented on systems that lack a memory-management
98 unit (MMU), but fork(2) can't be implemented on such systems.
99 (POSIX.1-2008 removed vfork() from the standard; the POSIX rationale
100 for the posix_spawn(3) function notes that that function, which pro‐
101 vides functionality equivalent to fork(2)+exec(3), is designed to be
102 implementable on systems that lack an MMU.)
103 * On systems where memory is constrained, vfork() avoids the need to
105 temporarily commit memory (see the description of /proc/sys/vm/over‐
106 commit_memory in proc(5)) in order to execute a new program. (This
107 can be especially beneficial where a large parent process wishes to
108 execute a small helper program in a child process.) By contrast,
109 using fork(2) in this scenario requires either committing an amount
110 of memory equal to the size of the parent process (if strict over‐
111 committing is in force) or overcommitting memory with the risk that
112 a process is terminated by the out-of-memory (OOM) killer.
113 Caveats
115 The child process should take care not to modify the memory in unin‐
116 tended ways, since such changes will be seen by the parent process once
117 the child terminates or executes another program. In this regard, sig‐
118 nal handlers can be especially problematic: if a signal handler that is
119 invoked in the child of vfork() changes memory, those changes may re‐
120 sult in an inconsistent process state from the perspective of the par‐
121 ent process (e.g., memory changes would be visible in the parent, but
122 changes to the state of open file descriptors would not be visible).
123 When vfork() is called in a multithreaded process, only the calling
125 thread is suspended until the child terminates or executes a new pro‐
126 gram. This means that the child is sharing an address space with other
127 running code. This can be dangerous if another thread in the parent
128 process changes credentials (using setuid(2) or similar), since there
129 are now two processes with different privilege levels running in the
130 same address space. As an example of the dangers, suppose that a mul‐
131 tithreaded program running as root creates a child using vfork(). Af‐
132 ter the vfork(), a thread in the parent process drops the process to an
133 unprivileged user in order to run some untrusted code (e.g., perhaps
134 via plug-in opened with dlopen(3)). In this case, attacks are possible
135 where the parent process uses mmap(2) to map in code that will be exe‐
136 cuted by the privileged child process.
137 Linux notes
139 Fork handlers established using pthread_atfork(3) are not called when a
140 multithreaded program employing the NPTL threading library calls
141 vfork(). Fork handlers are called in this case in a program using the
142 LinuxThreads threading library. (See pthreads(7) for a description of
143 Linux threading libraries.)
144 A call to vfork() is equivalent to calling clone(2) with flags speci‐
146 fied as:
147 CLONE_VM | CLONE_VFORK | SIGCHLD
149 History
151 The vfork() system call appeared in 3.0BSD. In 4.4BSD it was made syn‐
152 onymous to fork(2) but NetBSD introduced it again; see
153 ⟨http://www.netbsd.org/Documentation/kernel/vfork.html⟩. In Linux, it
154 has been equivalent to fork(2) until 2.2.0-pre6 or so. Since
155 2.2.0-pre9 (on i386, somewhat later on other architectures) it is an
156 independent system call. Support was added in glibc 2.0.112.
157
159 Details of the signal handling are obscure and differ between systems.
160 The BSD man page states: "To avoid a possible deadlock situation, pro‐
161 cesses that are children in the middle of a vfork() are never sent
162 SIGTTOU or SIGTTIN signals; rather, output or ioctls are allowed and
163 input attempts result in an end-of-file indication."
164
166 clone(2), execve(2), _exit(2), fork(2), unshare(2), wait(2)
167
169 This page is part of release 5.13 of the Linux man-pages project. A
170 description of the project, information about reporting bugs, and the
171 latest version of this page, can be found at
172 https://www.kernel.org/doc/man-pages/.
173Linux 2021-03-22 VFORK(2)