1SETNS(2) Linux Programmer's Manual SETNS(2)
2
6 setns - reassociate thread with a namespace
7
9 #define _GNU_SOURCE /* See feature_test_macros(7) */
10 #include <sched.h>
11 int setns(int fd, int nstype);
13
15 The setns() system call allows the calling thread to move into differ‐
16 ent namespaces. The fd argument is one of the following:
17 • a file descriptor referring to one of the magic links in a
19 /proc/[pid]/ns/ directory (or a bind mount to such a link);
20 • a PID file descriptor (see pidfd_open(2)).
22 The nstype argument is interpreted differently in each case.
24 fd refers to a /proc/[pid]/ns/ link
26 If fd refers to a /proc/[pid]/ns/ link, then setns() reassociates the
27 calling thread with the namespace associated with that link, subject to
28 any constraints imposed by the nstype argument. In this usage, each
29 call to setns() changes just one of the caller's namespace memberships.
30 The nstype argument specifies which type of namespace the calling
32 thread may be reassociated with. This argument can have one of the
33 following values:
34 0 Allow any type of namespace to be joined.
36 CLONE_NEWCGROUP (since Linux 4.6)
38 fd must refer to a cgroup namespace.
39 CLONE_NEWIPC (since Linux 3.0)
41 fd must refer to an IPC namespace.
42 CLONE_NEWNET (since Linux 3.0)
44 fd must refer to a network namespace.
45 CLONE_NEWNS (since Linux 3.8)
47 fd must refer to a mount namespace.
48 CLONE_NEWPID (since Linux 3.8)
50 fd must refer to a descendant PID namespace.
51 CLONE_NEWTIME (since Linux 5.8)
53 fd must refer to a time namespace.
54 CLONE_NEWUSER (since Linux 3.8)
56 fd must refer to a user namespace.
57 CLONE_NEWUTS (since Linux 3.0)
59 fd must refer to a UTS namespace.
60 Specifying nstype as 0 suffices if the caller knows (or does not care)
62 what type of namespace is referred to by fd. Specifying a nonzero
63 value for nstype is useful if the caller does not know what type of
64 namespace is referred to by fd and wants to ensure that the namespace
65 is of a particular type. (The caller might not know the type of the
66 namespace referred to by fd if the file descriptor was opened by an‐
67 other process and, for example, passed to the caller via a UNIX domain
68 socket.)
69 fd is a PID file descriptor
71 Since Linux 5.8, fd may refer to a PID file descriptor obtained from
72 pidfd_open(2) or clone(2). In this usage, setns() atomically moves the
73 calling thread into one or more of the same namespaces as the thread
74 referred to by fd.
75 The nstype argument is a bit mask specified by ORing together one or
77 more of the CLONE_NEW* namespace constants listed above. The caller is
78 moved into each of the target thread's namespaces that is specified in
79 nstype; the caller's memberships in the remaining namespaces are left
80 unchanged.
81 For example, the following code would move the caller into the same
83 user, network, and UTS namespaces as PID 1234, but would leave the
84 caller's other namespace memberships unchanged:
85 int fd = pidfd_open(1234, 0);
87 setns(fd, CLONE_NEWUSER | CLONE_NEWNET | CLONE_NEWUTS);
88 Details for specific namespace types
90 Note the following details and restrictions when reassociating with
91 specific namespace types:
92 User namespaces
94 A process reassociating itself with a user namespace must have
95 the CAP_SYS_ADMIN capability in the target user namespace.
96 (This necessarily implies that it is only possible to join a de‐
97 scendant user namespace.) Upon successfully joining a user
98 namespace, a process is granted all capabilities in that name‐
99 space, regardless of its user and group IDs.
100 A multithreaded process may not change user namespace with
102 setns().
103 It is not permitted to use setns() to reenter the caller's cur‐
105 rent user namespace. This prevents a caller that has dropped
106 capabilities from regaining those capabilities via a call to
107 setns().
108 For security reasons, a process can't join a new user namespace
110 if it is sharing filesystem-related attributes (the attributes
111 whose sharing is controlled by the clone(2) CLONE_FS flag) with
112 another process.
113 For further details on user namespaces, see user_namespaces(7).
115 Mount namespaces
117 Changing the mount namespace requires that the caller possess
118 both CAP_SYS_CHROOT and CAP_SYS_ADMIN capabilities in its own
119 user namespace and CAP_SYS_ADMIN in the user namespace that owns
120 the target mount namespace.
121 A process can't join a new mount namespace if it is sharing
123 filesystem-related attributes (the attributes whose sharing is
124 controlled by the clone(2) CLONE_FS flag) with another process.
125 See user_namespaces(7) for details on the interaction of user
127 namespaces and mount namespaces.
128 PID namespaces
130 In order to reassociate itself with a new PID namespace, the
131 caller must have the CAP_SYS_ADMIN capability both in its own
132 user namespace and in the user namespace that owns the target
133 PID namespace.
134 Reassociating the PID namespace has somewhat different from
136 other namespace types. Reassociating the calling thread with a
137 PID namespace changes only the PID namespace that subsequently
138 created child processes of the caller will be placed in; it does
139 not change the PID namespace of the caller itself.
140 Reassociating with a PID namespace is allowed only if the target
142 PID namespace is a descendant (child, grandchild, etc.) of, or
143 is the same as, the current PID namespace of the caller.
144 For further details on PID namespaces, see pid_namespaces(7).
146 Cgroup namespaces
148 In order to reassociate itself with a new cgroup namespace, the
149 caller must have the CAP_SYS_ADMIN capability both in its own
150 user namespace and in the user namespace that owns the target
151 cgroup namespace.
152 Using setns() to change the caller's cgroup namespace does not
154 change the caller's cgroup memberships.
155 Network, IPC, time, and UTS namespaces
157 In order to reassociate itself with a new network, IPC, time, or
158 UTS namespace, the caller must have the CAP_SYS_ADMIN capability
159 both in its own user namespace and in the user namespace that
160 owns the target namespace.
161
163 On success, setns() returns 0. On failure, -1 is returned and errno is
164 set to indicate the error.
165
167 EBADF fd is not a valid file descriptor.
168 EINVAL fd refers to a namespace whose type does not match that speci‐
170 fied in nstype.
171 EINVAL There is problem with reassociating the thread with the speci‐
173 fied namespace.
174 EINVAL The caller tried to join an ancestor (parent, grandparent, and
176 so on) PID namespace.
177 EINVAL The caller attempted to join the user namespace in which it is
179 already a member.
180 EINVAL The caller shares filesystem (CLONE_FS) state (in particular,
182 the root directory) with other processes and tried to join a new
183 user namespace.
184 EINVAL The caller is multithreaded and tried to join a new user name‐
186 space.
187 EINVAL fd is a PID file descriptor and nstype is invalid (e.g., it is
189 0).
190 ENOMEM Cannot allocate sufficient memory to change the specified name‐
192 space.
193 EPERM The calling thread did not have the required capability for this
195 operation.
196 ESRCH fd is a PID file descriptor but the process it refers to no
198 longer exists (i.e., it has terminated and been waited on).
199
201 The setns() system call first appeared in Linux in kernel 3.0; library
202 support was added to glibc in version 2.14.
203
205 The setns() system call is Linux-specific.
206
208 For further information on the /proc/[pid]/ns/ magic links, see name‐
209 spaces(7).
210 Not all of the attributes that can be shared when a new thread is cre‐
212 ated using clone(2) can be changed using setns().
213
215 The program below takes two or more arguments. The first argument
216 specifies the pathname of a namespace file in an existing
217 /proc/[pid]/ns/ directory. The remaining arguments specify a command
218 and its arguments. The program opens the namespace file, joins that
219 namespace using setns(), and executes the specified command inside that
220 namespace.
221 The following shell session demonstrates the use of this program (com‐
223 piled as a binary named ns_exec) in conjunction with the CLONE_NEWUTS
224 example program in the clone(2) man page (complied as a binary named
225 newuts).
226 We begin by executing the example program in clone(2) in the back‐
228 ground. That program creates a child in a separate UTS namespace. The
229 child changes the hostname in its namespace, and then both processes
230 display the hostnames in their UTS namespaces, so that we can see that
231 they are different.
232 $ su # Need privilege for namespace operations
234 Password:
235 # ./newuts bizarro &
236 [1] 3549
237 clone() returned 3550
238 uts.nodename in child: bizarro
239 uts.nodename in parent: antero
240 # uname -n # Verify hostname in the shell
241 antero
242 We then run the program shown below, using it to execute a shell. In‐
244 side that shell, we verify that the hostname is the one set by the
245 child created by the first program:
246 # ./ns_exec /proc/3550/ns/uts /bin/bash
248 # uname -n # Executed in shell started by ns_exec
249 bizarro
250 Program source
252 #define _GNU_SOURCE
253 #include <fcntl.h>
254 #include <sched.h>
255 #include <unistd.h>
256 #include <stdlib.h>
257 #include <stdio.h>
258 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
260 } while (0)
261 int
263 main(int argc, char *argv[])
264 {
265 int fd;
266 if (argc < 3) {
268 fprintf(stderr, "%s /proc/PID/ns/FILE cmd args...\n", argv[0]);
269 exit(EXIT_FAILURE);
270 }
271 /* Get file descriptor for namespace; the file descriptor is opened
273 with O_CLOEXEC so as to ensure that it is not inherited by the
274 program that is later executed. */
275 fd = open(argv[1], O_RDONLY | O_CLOEXEC);
277 if (fd == -1)
278 errExit("open");
279 if (setns(fd, 0) == -1) /* Join that namespace */
281 errExit("setns");
282 execvp(argv[2], &argv[2]); /* Execute a command in namespace */
284 errExit("execvp");
285 }
286
288 nsenter(1), clone(2), fork(2), unshare(2), vfork(2), namespaces(7),
289 unix(7)
290
292 This page is part of release 5.13 of the Linux man-pages project. A
293 description of the project, information about reporting bugs, and the
294 latest version of this page, can be found at
295 https://www.kernel.org/doc/man-pages/.
296Linux 2020-08-13 SETNS(2)