1SYSTEM(3) Linux Programmer's Manual SYSTEM(3)
2
6 system - execute a shell command
7
9 #include <stdlib.h>
10 int system(const char *command);
12
14 The system() library function uses fork(2) to create a child process
15 that executes the shell command specified in command using execl(3) as
16 follows:
17 execl("/bin/sh", "sh", "-c", command, (char *) NULL);
19 system() returns after the command has been completed.
21 During execution of the command, SIGCHLD will be blocked, and SIGINT
23 and SIGQUIT will be ignored, in the process that calls system().
24 (These signals will be handled according to their defaults inside the
25 child process that executes command.)
26 If command is NULL, then system() returns a status indicating whether a
28 shell is available on the system.
29
31 The return value of system() is one of the following:
32 * If command is NULL, then a nonzero value if a shell is available, or
34 0 if no shell is available.
35 * If a child process could not be created, or its status could not be
37 retrieved, the return value is -1 and errno is set to indicate the
38 error.
39 * If a shell could not be executed in the child process, then the re‐
41 turn value is as though the child shell terminated by calling
42 _exit(2) with the status 127.
43 * If all system calls succeed, then the return value is the termina‐
45 tion status of the child shell used to execute command. (The termi‐
46 nation status of a shell is the termination status of the last com‐
47 mand it executes.)
48 In the last two cases, the return value is a "wait status" that can be
50 examined using the macros described in waitpid(2). (i.e., WIFEXITED(),
51 WEXITSTATUS(), and so on).
52 system() does not affect the wait status of any other children.
54
56 system() can fail with any of the same errors as fork(2).
57
59 For an explanation of the terms used in this section, see at‐
60 tributes(7).
61 ┌────────────────────────────────────────────┬───────────────┬─────────┐
63 │Interface │ Attribute │ Value │
64 ├────────────────────────────────────────────┼───────────────┼─────────┤
65 │system() │ Thread safety │ MT-Safe │
66 └────────────────────────────────────────────┴───────────────┴─────────┘
67
69 POSIX.1-2001, POSIX.1-2008, C89, C99.
70
72 system() provides simplicity and convenience: it handles all of the de‐
73 tails of calling fork(2), execl(3), and waitpid(2), as well as the nec‐
74 essary manipulations of signals; in addition, the shell performs the
75 usual substitutions and I/O redirections for command. The main cost of
76 system() is inefficiency: additional system calls are required to cre‐
77 ate the process that runs the shell and to execute the shell.
78 If the _XOPEN_SOURCE feature test macro is defined (before including
80 any header files), then the macros described in waitpid(2) (WEXITSTA‐
81 TUS(), etc.) are made available when including <stdlib.h>.
82 As mentioned, system() ignores SIGINT and SIGQUIT. This may make pro‐
84 grams that call it from a loop uninterruptible, unless they take care
85 themselves to check the exit status of the child. For example:
86 while (something) {
88 int ret = system("foo");
89 if (WIFSIGNALED(ret) &&
91 (WTERMSIG(ret) == SIGINT || WTERMSIG(ret) == SIGQUIT))
92 break;
93 }
94 According to POSIX.1, it is unspecified whether handlers registered us‐
96 ing pthread_atfork(3) are called during the execution of system(). In
97 the glibc implementation, such handlers are not called.
98 In versions of glibc before 2.1.3, the check for the availability of
100 /bin/sh was not actually performed if command was NULL; instead it was
101 always assumed to be available, and system() always returned 1 in this
102 case. Since glibc 2.1.3, this check is performed because, even though
103 POSIX.1-2001 requires a conforming implementation to provide a shell,
104 that shell may not be available or executable if the calling program
105 has previously called chroot(2) (which is not specified by
106 POSIX.1-2001).
107 It is possible for the shell command to terminate with a status of 127,
109 which yields a system() return value that is indistinguishable from the
110 case where a shell could not be executed in the child process.
111 Caveats
113 Do not use system() from a privileged program (a set-user-ID or set-
114 group-ID program, or a program with capabilities) because strange val‐
115 ues for some environment variables might be used to subvert system in‐
116 tegrity. For example, PATH could be manipulated so that an arbitrary
117 program is executed with privilege. Use the exec(3) family of func‐
118 tions instead, but not execlp(3) or execvp(3) (which also use the PATH
119 environment variable to search for an executable).
120 system() will not, in fact, work properly from programs with set-user-
122 ID or set-group-ID privileges on systems on which /bin/sh is bash ver‐
123 sion 2: as a security measure, bash 2 drops privileges on startup.
124 (Debian uses a different shell, dash(1), which does not do this when
125 invoked as sh.)
126 Any user input that is employed as part of command should be carefully
128 sanitized, to ensure that unexpected shell commands or command options
129 are not executed. Such risks are especially grave when using system()
130 from a privileged program.
131
133 If the command name starts with a hyphen, sh(1) interprets the command
134 name as an option, and the behavior is undefined. (See the -c option
135 to sh(1).) To work around this problem, prepend the command with a
136 space as in the following call:
137 system(" -unfortunate-command-name");
139
141 sh(1), execve(2), fork(2), sigaction(2), sigprocmask(2), wait(2),
142 exec(3), signal(7)
143
145 This page is part of release 5.12 of the Linux man-pages project. A
146 description of the project, information about reporting bugs, and the
147 latest version of this page, can be found at
148 https://www.kernel.org/doc/man-pages/.
149 2021-03-22 SYSTEM(3)