1EXECVE(2) Linux Programmer's Manual EXECVE(2)
2
3
4
6 execve - execute program
7
9 #include <unistd.h>
10
11 int execve(const char *filename, char *const argv[],
12 char *const envp[]);
13
15 execve() executes the program pointed to by filename. filename must be
16 either a binary executable, or a script starting with a line of the
17 form:
18
19 #! interpreter [optional-arg]
20
21 For details of the latter case, see "Interpreter scripts" below.
22
23 argv is an array of argument strings passed to the new program. By
24 convention, the first of these strings should contain the filename
25 associated with the file being executed. envp is an array of strings,
26 conventionally of the form key=value, which are passed as environment
27 to the new program. Both argv and envp must be terminated by a NULL
28 pointer. The argument vector and environment can be accessed by the
29 called program's main function, when it is defined as:
30
31 int main(int argc, char *argv[], char *envp[])
32
33 execve() does not return on success, and the text, data, bss, and stack
34 of the calling process are overwritten by that of the program loaded.
35
36 If the current program is being ptraced, a SIGTRAP is sent to it after
37 a successful execve().
38
39 If the set-user-ID bit is set on the program file pointed to by file‐
40 name, and the underlying file system is not mounted nosuid (the
41 MS_NOSUID flag for mount(2)), and the calling process is not being
42 ptraced, then the effective user ID of the calling process is changed
43 to that of the owner of the program file. Similarly, when the set-
44 group-ID bit of the program file is set the effective group ID of the
45 calling process is set to the group of the program file.
46
47 The effective user ID of the process is copied to the saved set-user-
48 ID; similarly, the effective group ID is copied to the saved set-group-
49 ID. This copying takes place after any effective ID changes that occur
50 because of the set-user-ID and set-group-ID permission bits.
51
52 If the executable is an a.out dynamically linked binary executable con‐
53 taining shared-library stubs, the Linux dynamic linker ld.so(8) is
54 called at the start of execution to bring needed shared libraries into
55 memory and link the executable with them.
56
57 If the executable is a dynamically linked ELF executable, the inter‐
58 preter named in the PT_INTERP segment is used to load the needed shared
59 libraries. This interpreter is typically /lib/ld-linux.so.2 for bina‐
60 ries linked with glibc 2. (For binaries linked with the old Linux
61 libc5, the interpreter was typically /lib/ld-linux.so.1.)
62
63 All process attributes are preserved during an execve(), except the
64 following:
65
66 * The dispositions of any signals that are being caught are reset to
67 the default (signal(7)).
68
69 * Any alternate signal stack is not preserved (sigaltstack(2)).
70
71 * Memory mappings are not preserved (mmap(2)).
72
73 * Attached System V shared memory segments are detached (shmat(2)).
74
75 * POSIX shared memory regions are unmapped (shm_open(3)).
76
77 * Open POSIX message queue descriptors are closed (mq_overview(7)).
78
79 * Any open POSIX named semaphores are closed (sem_overview(7)).
80
81 * POSIX timers are not preserved (timer_create(2)).
82
83 * Any open directory streams are closed (opendir(3)).
84
85 * Memory locks are not preserved (mlock(2), mlockall(2)).
86
87 * Exit handlers are not preserved (atexit(3), on_exit(3)).
88
89 * The floating-point environment is reset to the default (see
90 fenv(3)).
91
92 The process attributes in the preceding list are all specified in
93 POSIX.1-2001. The following Linux-specific process attributes are also
94 not preserved during an execve():
95
96 * The prctl(2) PR_SET_DUMPABLE flag is set, unless a set-user-ID or
97 set-group ID program is being executed, in which case it is cleared.
98
99 * The prctl(2) PR_SET_KEEPCAPS flag is cleared.
100
101 * (Since Linux 2.4.36 / 2.6.23) If a set-user-ID or set-group-ID pro‐
102 gram is being executed, then the parent death signal set by prctl(2)
103 PR_SET_PDEATHSIG flag is cleared.
104
105 * The process name, as set by prctl(2) PR_SET_NAME (and displayed by
106 ps -o comm), is reset to the name of the new executable file.
107
108 * The SECBIT_KEEP_CAPS securebits flag is cleared. See capabili‐
109 ties(7).
110
111 * The termination signal is reset to SIGCHLD (see clone(2)).
112
113 Note the following further points:
114
115 * All threads other than the calling thread are destroyed during an
116 execve(). Mutexes, condition variables, and other pthreads objects
117 are not preserved.
118
119 * The equivalent of setlocale(LC_ALL, "C") is executed at program
120 start-up.
121
122 * POSIX.1-2001 specifies that the dispositions of any signals that are
123 ignored or set to the default are left unchanged. POSIX.1-2001
124 specifies one exception: if SIGCHLD is being ignored, then an imple‐
125 mentation may leave the disposition unchanged or reset it to the
126 default; Linux does the former.
127
128 * Any outstanding asynchronous I/O operations are canceled
129 (aio_read(3), aio_write(3)).
130
131 * For the handling of capabilities during execve(), see capabili‐
132 ties(7).
133
134 * By default, file descriptors remain open across an execve(). File
135 descriptors that are marked close-on-exec are closed; see the
136 description of FD_CLOEXEC in fcntl(2). (If a file descriptor is
137 closed, this will cause the release of all record locks obtained on
138 the underlying file by this process. See fcntl(2) for details.)
139 POSIX.1-2001 says that if file descriptors 0, 1, and 2 would other‐
140 wise be closed after a successful execve(), and the process would
141 gain privilege because the set-user_ID or set-group_ID permission
142 bit was set on the executed file, then the system may open an
143 unspecified file for each of these file descriptors. As a general
144 principle, no portable program, whether privileged or not, can
145 assume that these three file descriptors will remain closed across
146 an execve().
147
148 Interpreter scripts
149 An interpreter script is a text file that has execute permission
150 enabled and whose first line is of the form:
151
152 #! interpreter [optional-arg]
153
154 The interpreter must be a valid pathname for an executable which is not
155 itself a script. If the filename argument of execve() specifies an
156 interpreter script, then interpreter will be invoked with the following
157 arguments:
158
159 interpreter [optional-arg] filename arg...
160
161 where arg... is the series of words pointed to by the argv argument of
162 execve().
163
164 For portable use, optional-arg should either be absent, or be specified
165 as a single word (i.e., it should not contain white space); see NOTES
166 below.
167
168 Limits on size of arguments and environment
169 Most UNIX implementations impose some limit on the total size of the
170 command-line argument (argv) and environment (envp) strings that may be
171 passed to a new program. POSIX.1 allows an implementation to advertise
172 this limit using the ARG_MAX constant (either defined in <limits.h> or
173 available at run time using the call sysconf(_SC_ARG_MAX)).
174
175 On Linux prior to kernel 2.6.23, the memory used to store the environ‐
176 ment and argument strings was limited to 32 pages (defined by the ker‐
177 nel constant MAX_ARG_PAGES). On architectures with a 4-kB page size,
178 this yields a maximum size of 128 kB.
179
180 On kernel 2.6.23 and later, most architectures support a size limit
181 derived from the soft RLIMIT_STACK resource limit (see getrlimit(2))
182 that is in force at the time of the execve() call. (Architectures with
183 no memory management unit are excepted: they maintain the limit that
184 was in effect before kernel 2.6.23.) This change allows programs to
185 have a much larger argument and/or environment list. For these archi‐
186 tectures, the total size is limited to 1/4 of the allowed stack size.
187 (Imposing the 1/4-limit ensures that the new program always has some
188 stack space.) Since Linux 2.6.25, the kernel places a floor of 32
189 pages on this size limit, so that, even when RLIMIT_STACK is set very
190 low, applications are guaranteed to have at least as much argument and
191 environment space as was provided by Linux 2.6.23 and earlier. (This
192 guarantee was not provided in Linux 2.6.23 and 2.6.24.) Additionally,
193 the limit per string is 32 pages (the kernel constant MAX_ARG_STRLEN),
194 and the maximum number of strings is 0x7FFFFFFF.
195
197 On success, execve() does not return, on error -1 is returned, and
198 errno is set appropriately.
199
201 E2BIG The total number of bytes in the environment (envp) and argument
202 list (argv) is too large.
203
204 EACCES Search permission is denied on a component of the path prefix of
205 filename or the name of a script interpreter. (See also
206 path_resolution(7).)
207
208 EACCES The file or a script interpreter is not a regular file.
209
210 EACCES Execute permission is denied for the file or a script or ELF
211 interpreter.
212
213 EACCES The file system is mounted noexec.
214
215 EFAULT filename points outside your accessible address space.
216
217 EINVAL An ELF executable had more than one PT_INTERP segment (i.e.,
218 tried to name more than one interpreter).
219
220 EIO An I/O error occurred.
221
222 EISDIR An ELF interpreter was a directory.
223
224 ELIBBAD
225 An ELF interpreter was not in a recognized format.
226
227 ELOOP Too many symbolic links were encountered in resolving filename
228 or the name of a script or ELF interpreter.
229
230 EMFILE The process has the maximum number of files open.
231
232 ENAMETOOLONG
233 filename is too long.
234
235 ENFILE The system limit on the total number of open files has been
236 reached.
237
238 ENOENT The file filename or a script or ELF interpreter does not exist,
239 or a shared library needed for file or interpreter cannot be
240 found.
241
242 ENOEXEC
243 An executable is not in a recognized format, is for the wrong
244 architecture, or has some other format error that means it can‐
245 not be executed.
246
247 ENOMEM Insufficient kernel memory was available.
248
249 ENOTDIR
250 A component of the path prefix of filename or a script or ELF
251 interpreter is not a directory.
252
253 EPERM The file system is mounted nosuid, the user is not the supe‐
254 ruser, and the file has the set-user-ID or set-group-ID bit set.
255
256 EPERM The process is being traced, the user is not the superuser and
257 the file has the set-user-ID or set-group-ID bit set.
258
259 ETXTBSY
260 Executable was open for writing by one or more processes.
261
263 SVr4, 4.3BSD, POSIX.1-2001. POSIX.1-2001 does not document the #!
264 behavior but is otherwise compatible.
265
267 Set-user-ID and set-group-ID processes can not be ptrace(2)d.
268
269 Linux ignores the set-user-ID and set-group-ID bits on scripts.
270
271 The result of mounting a file system nosuid varies across Linux kernel
272 versions: some will refuse execution of set-user-ID and set-group-ID
273 executables when this would give the user powers she did not have
274 already (and return EPERM), some will just ignore the set-user-ID and
275 set-group-ID bits and exec() successfully.
276
277 A maximum line length of 127 characters is allowed for the first line
278 in a #! executable shell script.
279
280 The semantics of the optional-arg argument of an interpreter script
281 vary across implementations. On Linux, the entire string following the
282 interpreter name is passed as a single argument to the interpreter, and
283 this string can include white space. However, behavior differs on some
284 other systems. Some systems use the first white space to terminate
285 optional-arg. On some systems, an interpreter script can have multiple
286 arguments, and white spaces in optional-arg are used to delimit the
287 arguments.
288
289 On Linux, either argv or envp can be specified as NULL, which has the
290 same effect as specifying these arguments as a pointer to a list con‐
291 taining a single NULL pointer. Do not take advantage of this misfea‐
292 ture! It is nonstandard and nonportable: on most other UNIX systems
293 doing this will result in an error (EFAULT).
294
295 POSIX.1-2001 says that values returned by sysconf(3) should be invari‐
296 ant over the lifetime of a process. However, since Linux 2.6.23, if
297 the RLIMIT_STACK resource limit changes, then the value reported by
298 _SC_ARG_MAX will also change, to reflect the fact that the limit on
299 space for holding command-line arguments and environment variables has
300 changed.
301
302 Historical
303 With UNIX V6 the argument list of an exec() call was ended by 0, while
304 the argument list of main was ended by -1. Thus, this argument list
305 was not directly usable in a further exec() call. Since UNIX V7 both
306 are NULL.
307
309 The following program is designed to be execed by the second program
310 below. It just echoes its command-line one per line.
311
312 /* myecho.c */
313
314 #include <stdio.h>
315 #include <stdlib.h>
316
317 int
318 main(int argc, char *argv[])
319 {
320 int j;
321
322 for (j = 0; j < argc; j++)
323 printf("argv[%d]: %s\n", j, argv[j]);
324
325 exit(EXIT_SUCCESS);
326 }
327
328 This program can be used to exec the program named in its command-line
329 argument:
330
331 /* execve.c */
332
333 #include <stdio.h>
334 #include <stdlib.h>
335 #include <unistd.h>
336
337 int
338 main(int argc, char *argv[])
339 {
340 char *newargv[] = { NULL, "hello", "world", NULL };
341 char *newenviron[] = { NULL };
342
343 if (argc != 2) {
344 fprintf(stderr, "Usage: %s <file-to-exec>\n", argv[0]);
345 exit(EXIT_FAILURE);
346 }
347
348 newargv[0] = argv[1];
349
350 execve(argv[1], newargv, newenviron);
351 perror("execve"); /* execve() only returns on error */
352 exit(EXIT_FAILURE);
353 }
354
355 We can use the second program to exec the first as follows:
356
357 $ cc myecho.c -o myecho
358 $ cc execve.c -o execve
359 $ ./execve ./myecho
360 argv[0]: ./myecho
361 argv[1]: hello
362 argv[2]: world
363
364 We can also use these programs to demonstrate the use of a script
365 interpreter. To do this we create a script whose "interpreter" is our
366 myecho program:
367
368 $ cat > script.sh
369 #! ./myecho script-arg
370 ^D
371 $ chmod +x script.sh
372
373 We can then use our program to exec the script:
374
375 $ ./execve ./script.sh
376 argv[0]: ./myecho
377 argv[1]: script-arg
378 argv[2]: ./script.sh
379 argv[3]: hello
380 argv[4]: world
381
383 chmod(2), fork(2), ptrace(2), execl(3), fexecve(3), getopt(3), creden‐
384 tials(7), environ(7), path_resolution(7), ld.so(8)
385
387 This page is part of release 3.53 of the Linux man-pages project. A
388 description of the project, and information about reporting bugs, can
389 be found at http://www.kernel.org/doc/man-pages/.
390
391
392
393Linux 2013-07-04 EXECVE(2)