1UNSHARE(2) Linux Programmer's Manual UNSHARE(2)
2
6 unshare - disassociate parts of the process execution context
7
9 #define _GNU_SOURCE
10 #include <sched.h>
11 int unshare(int flags);
13
15 unshare() allows a process (or thread) to disassociate parts of its
16 execution context that are currently being shared with other processes
17 (or threads). Part of the execution context, such as the mount names‐
18 pace, is shared implicitly when a new process is created using fork(2)
19 or vfork(2), while other parts, such as virtual memory, may be shared
20 by explicit request when creating a process or thread using clone(2).
21 The main use of unshare() is to allow a process to control its shared
23 execution context without creating a new process.
24 The flags argument is a bit mask that specifies which parts of the exe‐
26 cution context should be unshared. This argument is specified by ORing
27 together zero or more of the following constants:
28 CLONE_FILES
30 Reverse the effect of the clone(2) CLONE_FILES flag. Unshare
31 the file descriptor table, so that the calling process no longer
32 shares its file descriptors with any other process.
33 CLONE_FS
35 Reverse the effect of the clone(2) CLONE_FS flag. Unshare
36 filesystem attributes, so that the calling process no longer
37 shares its root directory (chroot(2)), current directory
38 (chdir(2)), or umask (umask(2)) attributes with any other
39 process.
40 CLONE_NEWCGROUP (since Linux 4.6)
42 This flag has the same effect as the clone(2) CLONE_NEWCGROUP
43 flag. Unshare the cgroup namespace. Use of CLONE_NEWCGROUP
44 requires the CAP_SYS_ADMIN capability.
45 CLONE_NEWIPC (since Linux 2.6.19)
47 This flag has the same effect as the clone(2) CLONE_NEWIPC flag.
48 Unshare the IPC namespace, so that the calling process has a
49 private copy of the IPC namespace which is not shared with any
50 other process. Specifying this flag automatically implies
51 CLONE_SYSVSEM as well. Use of CLONE_NEWIPC requires the
52 CAP_SYS_ADMIN capability.
53 CLONE_NEWNET (since Linux 2.6.24)
55 This flag has the same effect as the clone(2) CLONE_NEWNET flag.
56 Unshare the network namespace, so that the calling process is
57 moved into a new network namespace which is not shared with any
58 previously existing process. Use of CLONE_NEWNET requires the
59 CAP_SYS_ADMIN capability.
60 CLONE_NEWNS
62 This flag has the same effect as the clone(2) CLONE_NEWNS flag.
63 Unshare the mount namespace, so that the calling process has a
64 private copy of its namespace which is not shared with any other
65 process. Specifying this flag automatically implies CLONE_FS as
66 well. Use of CLONE_NEWNS requires the CAP_SYS_ADMIN capability.
67 For further information, see mount_namespaces(7).
68 CLONE_NEWPID (since Linux 3.8)
70 This flag has the same effect as the clone(2) CLONE_NEWPID flag.
71 Unshare the PID namespace, so that the calling process has a new
72 PID namespace for its children which is not shared with any pre‐
73 viously existing process. The calling process is not moved into
74 the new namespace. The first child created by the calling
75 process will have the process ID 1 and will assume the role of
76 init(1) in the new namespace. CLONE_NEWPID automatically
77 implies CLONE_THREAD as well. Use of CLONE_NEWPID requires the
78 CAP_SYS_ADMIN capability. For further information, see
79 pid_namespaces(7).
80 CLONE_NEWTIME (since Linux 5.6)
82 Unshare the time namespace, so that the calling process has a
83 new time namespace for its children which is not shared with any
84 previously existing process. The calling process is not moved
85 into the new namespace. Use of CLONE_NEWTIME requires the
86 CAP_SYS_ADMIN capability. For further information, see
87 time_namespaces(7).
88 CLONE_NEWUSER (since Linux 3.8)
90 This flag has the same effect as the clone(2) CLONE_NEWUSER
91 flag. Unshare the user namespace, so that the calling process
92 is moved into a new user namespace which is not shared with any
93 previously existing process. As with the child process created
94 by clone(2) with the CLONE_NEWUSER flag, the caller obtains a
95 full set of capabilities in the new namespace.
96 CLONE_NEWUSER requires that the calling process is not threaded;
98 specifying CLONE_NEWUSER automatically implies CLONE_THREAD.
99 Since Linux 3.9, CLONE_NEWUSER also automatically implies
100 CLONE_FS. CLONE_NEWUSER requires that the user ID and group ID
101 of the calling process are mapped to user IDs and group IDs in
102 the user namespace of the calling process at the time of the
103 call.
104 For further information on user namespaces, see user_names‐
106 paces(7).
107 CLONE_NEWUTS (since Linux 2.6.19)
109 This flag has the same effect as the clone(2) CLONE_NEWUTS flag.
110 Unshare the UTS IPC namespace, so that the calling process has a
111 private copy of the UTS namespace which is not shared with any
112 other process. Use of CLONE_NEWUTS requires the CAP_SYS_ADMIN
113 capability.
114 CLONE_SYSVSEM (since Linux 2.6.26)
116 This flag reverses the effect of the clone(2) CLONE_SYSVSEM
117 flag. Unshare System V semaphore adjustment (semadj) values, so
118 that the calling process has a new empty semadj list that is not
119 shared with any other process. If this is the last process that
120 has a reference to the process's current semadj list, then the
121 adjustments in that list are applied to the corresponding sema‐
122 phores, as described in semop(2).
123 In addition, CLONE_THREAD, CLONE_SIGHAND, and CLONE_VM can be specified
125 in flags if the caller is single threaded (i.e., it is not sharing its
126 address space with another process or thread). In this case, these
127 flags have no effect. (Note also that specifying CLONE_THREAD automat‐
128 ically implies CLONE_VM, and specifying CLONE_VM automatically implies
129 CLONE_SIGHAND.) If the process is multithreaded, then the use of these
130 flags results in an error.
131 If flags is specified as zero, then unshare() is a no-op; no changes
133 are made to the calling process's execution context.
134
136 On success, zero returned. On failure, -1 is returned and errno is set
137 to indicate the error.
138
140 EINVAL An invalid bit was specified in flags.
141 EINVAL CLONE_THREAD, CLONE_SIGHAND, or CLONE_VM was specified in flags,
143 and the caller is multithreaded.
144 EINVAL CLONE_NEWIPC was specified in flags, but the kernel was not con‐
146 figured with the CONFIG_SYSVIPC and CONFIG_IPC_NS options.
147 EINVAL CLONE_NEWNET was specified in flags, but the kernel was not con‐
149 figured with the CONFIG_NET_NS option.
150 EINVAL CLONE_NEWPID was specified in flags, but the kernel was not con‐
152 figured with the CONFIG_PID_NS option.
153 EINVAL CLONE_NEWUSER was specified in flags, but the kernel was not
155 configured with the CONFIG_USER_NS option.
156 EINVAL CLONE_NEWUTS was specified in flags, but the kernel was not con‐
158 figured with the CONFIG_UTS_NS option.
159 EINVAL CLONE_NEWPID was specified in flags, but the process has previ‐
161 ously called unshare() with the CLONE_NEWPID flag.
162 ENOMEM Cannot allocate sufficient memory to copy parts of caller's con‐
164 text that need to be unshared.
165 ENOSPC (since Linux 3.7)
167 CLONE_NEWPID was specified in flags, but the limit on the nest‐
168 ing depth of PID namespaces would have been exceeded; see
169 pid_namespaces(7).
170 ENOSPC (since Linux 4.9; beforehand EUSERS)
172 CLONE_NEWUSER was specified in flags, and the call would cause
173 the limit on the number of nested user namespaces to be
174 exceeded. See user_namespaces(7).
175 From Linux 3.11 to Linux 4.8, the error diagnosed in this case
177 was EUSERS.
178 ENOSPC (since Linux 4.9)
180 One of the values in flags specified the creation of a new user
181 namespace, but doing so would have caused the limit defined by
182 the corresponding file in /proc/sys/user to be exceeded. For
183 further details, see namespaces(7).
184 EPERM The calling process did not have the required privileges for
186 this operation.
187 EPERM CLONE_NEWUSER was specified in flags, but either the effective
189 user ID or the effective group ID of the caller does not have a
190 mapping in the parent namespace (see user_namespaces(7)).
191 EPERM (since Linux 3.9)
193 CLONE_NEWUSER was specified in flags and the caller is in a
194 chroot environment (i.e., the caller's root directory does not
195 match the root directory of the mount namespace in which it
196 resides).
197 EUSERS (from Linux 3.11 to Linux 4.8)
199 CLONE_NEWUSER was specified in flags, and the limit on the num‐
200 ber of nested user namespaces would be exceeded. See the dis‐
201 cussion of the ENOSPC error above.
202
204 The unshare() system call was added to Linux in kernel 2.6.16.
205
207 The unshare() system call is Linux-specific.
208
210 Not all of the process attributes that can be shared when a new process
211 is created using clone(2) can be unshared using unshare(). In particu‐
212 lar, as at kernel 3.8, unshare() does not implement flags that reverse
213 the effects of CLONE_SIGHAND, CLONE_THREAD, or CLONE_VM. Such func‐
214 tionality may be added in the future, if required.
215
217 The program below provides a simple implementation of the unshare(1)
218 command, which unshares one or more namespaces and executes the command
219 supplied in its command-line arguments. Here's an example of the use
220 of this program, running a shell in a new mount namespace, and verify‐
221 ing that the original shell and the new shell are in separate mount
222 namespaces:
223 $ readlink /proc/$$/ns/mnt
225 mnt:[4026531840]
226 $ sudo ./unshare -m /bin/bash
227 # readlink /proc/$$/ns/mnt
228 mnt:[4026532325]
229 The differing output of the two readlink(1) commands shows that the two
231 shells are in different mount namespaces.
232 Program source
234 /* unshare.c
236 A simple implementation of the unshare(1) command: unshare
238 namespaces and execute a command.
239 */
240 #define _GNU_SOURCE
241 #include <sched.h>
242 #include <unistd.h>
243 #include <stdlib.h>
244 #include <stdio.h>
245 /* A simple error-handling function: print an error message based
247 on the value in 'errno' and terminate the calling process */
248 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
250 } while (0)
251 static void
253 usage(char *pname)
254 {
255 fprintf(stderr, "Usage: %s [options] program [arg...]\n", pname);
256 fprintf(stderr, "Options can be:\n");
257 fprintf(stderr, " -C unshare cgroup namespace\n");
258 fprintf(stderr, " -i unshare IPC namespace\n");
259 fprintf(stderr, " -m unshare mount namespace\n");
260 fprintf(stderr, " -n unshare network namespace\n");
261 fprintf(stderr, " -p unshare PID namespace\n");
262 fprintf(stderr, " -t unshare time namespace\n");
263 fprintf(stderr, " -u unshare UTS namespace\n");
264 fprintf(stderr, " -U unshare user namespace\n");
265 exit(EXIT_FAILURE);
266 }
267 int
269 main(int argc, char *argv[])
270 {
271 int flags, opt;
272 flags = 0;
274 while ((opt = getopt(argc, argv, "CimnptuU")) != -1) {
276 switch (opt) {
277 case 'C': flags |= CLONE_NEWCGROUP; break;
278 case 'i': flags |= CLONE_NEWIPC; break;
279 case 'm': flags |= CLONE_NEWNS; break;
280 case 'n': flags |= CLONE_NEWNET; break;
281 case 'p': flags |= CLONE_NEWPID; break;
282 case 't': flags |= CLONE_NEWTIME; break;
283 case 'u': flags |= CLONE_NEWUTS; break;
284 case 'U': flags |= CLONE_NEWUSER; break;
285 default: usage(argv[0]);
286 }
287 }
288 if (optind >= argc)
290 usage(argv[0]);
291 if (unshare(flags) == -1)
293 errExit("unshare");
294 execvp(argv[optind], &argv[optind]);
296 errExit("execvp");
297 }
298
300 unshare(1), clone(2), fork(2), kcmp(2), setns(2), vfork(2), names‐
301 paces(7)
302 Documentation/userspace-api/unshare.rst in the Linux kernel source tree
304 (or Documentation/unshare.txt before Linux 4.12)
305
307 This page is part of release 5.07 of the Linux man-pages project. A
308 description of the project, information about reporting bugs, and the
309 latest version of this page, can be found at
310 https://www.kernel.org/doc/man-pages/.
311Linux 2020-04-11 UNSHARE(2)