1vfork(2) System Calls Manual vfork(2)
2
6 vfork - create a child process and block parent
7
9 Standard C library (libc, -lc)
10
12 #include <unistd.h>
13 pid_t vfork(void);
15 Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
17 vfork():
19 Since glibc 2.12:
20 (_XOPEN_SOURCE >= 500) && ! (_POSIX_C_SOURCE >= 200809L)
21 || /* Since glibc 2.19: */ _DEFAULT_SOURCE
22 || /* glibc <= 2.19: */ _BSD_SOURCE
23 Before glibc 2.12:
24 _BSD_SOURCE || _XOPEN_SOURCE >= 500
25
27 Standard description
28 (From POSIX.1) The vfork() function has the same effect as fork(2), ex‐
29 cept that the behavior is undefined if the process created by vfork()
30 either modifies any data other than a variable of type pid_t used to
31 store the return value from vfork(), or returns from the function in
32 which vfork() was called, or calls any other function before success‐
33 fully calling _exit(2) or one of the exec(3) family of functions.
34 Linux description
36 vfork(), just like fork(2), creates a child process of the calling
37 process. For details and return value and errors, see fork(2).
38 vfork() is a special case of clone(2). It is used to create new pro‐
40 cesses without copying the page tables of the parent process. It may
41 be useful in performance-sensitive applications where a child is cre‐
42 ated which then immediately issues an execve(2).
43 vfork() differs from fork(2) in that the calling thread is suspended
45 until the child terminates (either normally, by calling _exit(2), or
46 abnormally, after delivery of a fatal signal), or it makes a call to
47 execve(2). Until that point, the child shares all memory with its par‐
48 ent, including the stack. The child must not return from the current
49 function or call exit(3) (which would have the effect of calling exit
50 handlers established by the parent process and flushing the parent's
51 stdio(3) buffers), but may call _exit(2).
52 As with fork(2), the child process created by vfork() inherits copies
54 of various of the caller's process attributes (e.g., file descriptors,
55 signal dispositions, and current working directory); the vfork() call
56 differs only in the treatment of the virtual address space, as de‐
57 scribed above.
58 Signals sent to the parent arrive after the child releases the parent's
60 memory (i.e., after the child terminates or calls execve(2)).
61 Historic description
63 Under Linux, fork(2) is implemented using copy-on-write pages, so the
64 only penalty incurred by fork(2) is the time and memory required to du‐
65 plicate the parent's page tables, and to create a unique task structure
66 for the child. However, in the bad old days a fork(2) would require
67 making a complete copy of the caller's data space, often needlessly,
68 since usually immediately afterward an exec(3) is done. Thus, for
69 greater efficiency, BSD introduced the vfork() system call, which did
70 not fully copy the address space of the parent process, but borrowed
71 the parent's memory and thread of control until a call to execve(2) or
72 an exit occurred. The parent process was suspended while the child was
73 using its resources. The use of vfork() was tricky: for example, not
74 modifying data in the parent process depended on knowing which vari‐
75 ables were held in a register.
76
78 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 Some consider the semantics of vfork() to be an architectural blemish,
85 and the 4.2BSD man page stated: “This system call will be eliminated
86 when proper system sharing mechanisms are implemented. Users should
87 not depend on the memory sharing semantics of vfork as it will, in that
88 case, be made synonymous to fork.” However, even though modern memory
89 management hardware has decreased the performance difference between
90 fork(2) and vfork(), there are various reasons why Linux and other sys‐
91 tems have retained vfork():
92 • Some performance-critical applications require the small performance
94 advantage conferred by vfork().
95 • vfork() can be implemented on systems that lack a memory-management
97 unit (MMU), but fork(2) can't be implemented on such systems.
98 (POSIX.1-2008 removed vfork() from the standard; the POSIX rationale
99 for the posix_spawn(3) function notes that that function, which pro‐
100 vides functionality equivalent to fork(2)+exec(3), is designed to be
101 implementable on systems that lack an MMU.)
102 • On systems where memory is constrained, vfork() avoids the need to
104 temporarily commit memory (see the description of /proc/sys/vm/over‐
105 commit_memory in proc(5)) in order to execute a new program. (This
106 can be especially beneficial where a large parent process wishes to
107 execute a small helper program in a child process.) By contrast,
108 using fork(2) in this scenario requires either committing an amount
109 of memory equal to the size of the parent process (if strict over‐
110 committing is in force) or overcommitting memory with the risk that
111 a process is terminated by the out-of-memory (OOM) killer.
112 Linux notes
114 Fork handlers established using pthread_atfork(3) are not called when a
115 multithreaded program employing the NPTL threading library calls
116 vfork(). Fork handlers are called in this case in a program using the
117 LinuxThreads threading library. (See pthreads(7) for a description of
118 Linux threading libraries.)
119 A call to vfork() is equivalent to calling clone(2) with flags speci‐
121 fied as:
122 CLONE_VM | CLONE_VFORK | SIGCHLD
124
126 None.
127
129 4.3BSD; POSIX.1-2001 (but marked OBSOLETE). POSIX.1-2008 removes the
130 specification of vfork().
131 The vfork() system call appeared in 3.0BSD. In 4.4BSD it was made syn‐
133 onymous to fork(2) but NetBSD introduced it again; see
134 ⟨http://www.netbsd.org/Documentation/kernel/vfork.html⟩. In Linux, it
135 has been equivalent to fork(2) until Linux 2.2.0-pre6 or so. Since
136 Linux 2.2.0-pre9 (on i386, somewhat later on other architectures) it is
137 an independent system call. Support was added in glibc 2.0.112.
138
140 The child process should take care not to modify the memory in unin‐
141 tended ways, since such changes will be seen by the parent process once
142 the child terminates or executes another program. In this regard, sig‐
143 nal handlers can be especially problematic: if a signal handler that is
144 invoked in the child of vfork() changes memory, those changes may re‐
145 sult in an inconsistent process state from the perspective of the par‐
146 ent process (e.g., memory changes would be visible in the parent, but
147 changes to the state of open file descriptors would not be visible).
148 When vfork() is called in a multithreaded process, only the calling
150 thread is suspended until the child terminates or executes a new pro‐
151 gram. This means that the child is sharing an address space with other
152 running code. This can be dangerous if another thread in the parent
153 process changes credentials (using setuid(2) or similar), since there
154 are now two processes with different privilege levels running in the
155 same address space. As an example of the dangers, suppose that a mul‐
156 tithreaded program running as root creates a child using vfork(). Af‐
157 ter the vfork(), a thread in the parent process drops the process to an
158 unprivileged user in order to run some untrusted code (e.g., perhaps
159 via plug-in opened with dlopen(3)). In this case, attacks are possible
160 where the parent process uses mmap(2) to map in code that will be exe‐
161 cuted by the privileged child process.
162
164 Details of the signal handling are obscure and differ between systems.
165 The BSD man page states: "To avoid a possible deadlock situation, pro‐
166 cesses that are children in the middle of a vfork() are never sent
167 SIGTTOU or SIGTTIN signals; rather, output or ioctls are allowed and
168 input attempts result in an end-of-file indication."
169
171 clone(2), execve(2), _exit(2), fork(2), unshare(2), wait(2)
172Linux man-pages 6.05 2023-07-28 vfork(2)