man2/unshare.2

   1 .\" Copyright (C) 2006, Janak Desai <janak@us.ibm.com>
   2 .\" and Copyright (C) 2006, 2012 Michael Kerrisk <mtk.manpages@gmail.com>
   3 .\"
   4 .\" %%%LICENSE_START(GPL_NOVERSION_ONELINE)
   5 .\" Licensed under the GPL
   6 .\" %%%LICENSE_END
   7 .\"
   8 .\" Patch Justification:
   9 .\" unshare system call is needed to implement, using PAM,
  10 .\" per-security_context and/or per-user namespace to provide
  11 .\" polyinstantiated directories. Using unshare and bind mounts, a
  12 .\" PAM module can create private namespace with appropriate
  13 .\" directories(based on user's security context) bind mounted on
  14 .\" public directories such as /tmp, thus providing an instance of
  15 .\" /tmp that is based on user's security context. Without the
  16 .\" unshare system call, namespace separation can only be achieved
  17 .\" by clone, which would require porting and maintaining all commands
  18 .\" such as login, and su, that establish a user session.
  19 .\"
  20 .TH UNSHARE 2 2013-04-17 "Linux" "Linux Programmer's Manual"
  21 .SH NAME
  22 unshare \- disassociate parts of the process execution context
  23 .SH SYNOPSIS
  24 .nf
  25 .B #include <sched.h>
  26 .sp
  27 .BI "int unshare(int " flags );
  28 .fi
  29 .sp
  30 .in -4n
  31 Feature Test Macro Requirements for glibc (see
  32 .BR feature_test_macros (7)):
  33 .in
  34 .sp
  35 .BR unshare ():
  36 .ad l
  37 .RS 4
  38 .PD 0
  39 .TP 4
  40 Since glibc 2.14:
  41 _GNU_SOURCE
  42 .TP 4
  43 .\" See http://sources.redhat.com/bugzilla/show_bug.cgi?id=4749
  44 Before glibc 2.14:
  45 _BSD_SOURCE || _SVID_SOURCE
  46     /* _GNU_SOURCE also suffices */
  47 .PD
  48 .RE
  49 .ad b
  50 .SH DESCRIPTION
  51 .BR unshare ()
  52 allows a process to disassociate parts of its execution
  53 context that are currently being shared with other processes.
  54 Part of the execution context, such as the mount namespace, is shared
  55 implicitly when a new process is created using
  56 .BR fork (2)
  57 or
  58 .BR vfork (2),
  59 while other parts, such as virtual memory, may be
  60 shared by explicit request when creating a process using
  61 .BR clone (2).
  62
  63 The main use of
  64 .BR unshare ()
  65 is to allow a process to control its
  66 shared execution context without creating a new process.
  67
  68 The
  69 .I flags
  70 argument is a bit mask that specifies which parts of
  71 the execution context should be unshared.
  72 This argument is specified by ORing together zero or more
  73 of the following constants:
  74 .TP
  75 .B CLONE_FILES
  76 Reverse the effect of the
  77 .BR clone (2)
  78 .B CLONE_FILES
  79 flag.
  80 Unshare the file descriptor table, so that the calling process
  81 no longer shares its file descriptors with any other process.
  82 .TP
  83 .B CLONE_FS
  84 Reverse the effect of the
  85 .BR clone (2)
  86 .B CLONE_FS
  87 flag.
  88 Unshare filesystem attributes, so that the calling process
  89 no longer shares its root directory
  90 .RB ( chroot (2)),
  91 current directory
  92 .RB ( chdir (2)),
  93 or umask
  94 .RB ( umask (2))
  95 attributes with any other process.
  96 .TP
  97 .BR CLONE_NEWIPC " (since Linux 2.6.19)
  98 This flag has the same effect as the
  99 .BR clone (2)
 100 .B CLONE_NEWIPC
 101 flag.
 102 Unshare the System\ V IPC namespace,
 103 so that the calling process has a private copy of the
 104 System\ V IPC namespace which is not shared with any other process.
 105 Specifying this flag automatically implies
 106 .BR CLONE_SYSVSEM
 107 as well.
 108 Use of
 109 .BR CLONE_NEWIPC
 110 requires the
 111 .BR CAP_SYS_ADMIN
 112 capability.
 113 .TP
 114 .BR CLONE_NEWNET " (since Linux 2.6.24)
 115 This flag has the same effect as the
 116 .BR clone (2)
 117 .B CLONE_NEWNET
 118 flag.
 119 Unshare the network namespace,
 120 so that the calling process is moved into a
 121 new network namespace which is not shared
 122 with any previously existing process.
 123 Use of
 124 .BR CLONE_NEWNET
 125 requires the
 126 .BR CAP_SYS_ADMIN
 127 capability.
 128 .TP
 129 .B CLONE_NEWNS
 130 .\" These flag name are inconsistent:
 131 .\" CLONE_NEWNS does the same thing in clone(), but CLONE_VM,
 132 .\" CLONE_FS, and CLONE_FILES reverse the action of the clone()
 133 .\" flags of the same name.
 134 This flag has the same effect as the
 135 .BR clone (2)
 136 .B CLONE_NEWNS
 137 flag.
 138 Unshare the mount namespace,
 139 so that the calling process has a private copy of
 140 its namespace which is not shared with any other process.
 141 Specifying this flag automatically implies
 142 .B CLONE_FS
 143 as well.
 144 Use of
 145 .BR CLONE_NEWNS
 146 requires the
 147 .BR CAP_SYS_ADMIN
 148 capability.
 149 .TP
 150 .BR CLONE_NEWPID " (since Linux 3.8)"
 151 This flag has the same effect as the
 152 .BR clone (2)
 153 .B CLONE_NEWPID
 154 flag.
 155 Unshare the PID namespace,
 156 so that the calling process has a new PID namespace for its children
 157 which is not shared with any previously existing process.
 158 The calling process is
 159 .I not
 160 moved into the new namespace.
 161 The first child created by the calling process will have
 162 the process ID 1 and will assume the role of
 163 .BR init (1)
 164 in the new namespace.
 165 Use of
 166 .BR CLONE_NEWPID
 167 requires the
 168 .BR CAP_SYS_ADMIN
 169 capability.
 170 For further information, see
 171 .BR pid_namespaces (7).
 172 .TP
 173 .BR CLONE_NEWUSER " (since Linux 3.8)"
 174 This flag has the same effect as the
 175 .BR clone (2)
 176 .B CLONE_NEWUSER
 177 flag.
 178 Unshare the user namespace,
 179 so that the calling process is moved into a new user namespace
 180 which is not shared with any previously existing process.
 181 like the child process created by
 182 .BR clone (2)
 183 with the
 184 .B CLONE_NEWUSER
 185 flag, the caller obtains a full set of capabilities in the new namespace.
 186 .BR CLONE_NEWUSER
 187 requires that the calling process is not threaded and the user ID and group ID
 188 of the calling process are mapped user IDs and group IDs in the
 189 user namespace of the calling process at the time of the call.
 190 .TP
 191 .BR CLONE_NEWUTS " (since Linux 2.6.19)"
 192 This flag has the same effect as the
 193 .BR clone (2)
 194 .B CLONE_NEWUTS
 195 flag.
 196 Unshare the UTS IPC namespace,
 197 so that the calling process has a private copy of the
 198 UTS namespace which is not shared with any other process.
 199 Use of
 200 .BR CLONE_NEWUTS
 201 requires the
 202 .BR CAP_SYS_ADMIN
 203 capability.
 204 .TP
 205 .BR CLONE_SYSVSEM " (since Linux 2.6.26)
 206 .\" commit 9edff4ab1f8d82675277a04e359d0ed8bf14a7b7
 207 This flag reverses the effect of the
 208 .BR clone (2)
 209 .B CLONE_SYSVSEM
 210 flag.
 211 Unshare System\ V semaphore undo values,
 212 so that the calling process has a private copy
 213 which is not shared with any other process.
 214 Use of
 215 .BR CLONE_SYSVSEM
 216 requires the
 217 .BR CAP_SYS_ADMIN
 218 capability.
 219 .\" As at 2.6.16, the following forced implications also apply,
 220 .\" although the relevant flags are not yet implemented.
 221 .\" If CLONE_THREAD is set force CLONE_VM.
 222 .\" If CLONE_VM is set, force CLONE_SIGHAND.
 223 .\" CLONE_NEWNS If CLONE_SIGHAND is set and signals are also being shared
 224 .\" (i.e., current->signal->count > 1), force CLONE_THREAD.
 225 .\"
 226 .\" FIXME . CLONE_VM is not (yet, as at 2.6.16) implemented.
 227 .\" .TP
 228 .\" .B CLONE_VM
 229 .\" Reverse the effect of the
 230 .\" .BR clone (2)
 231 .\" .B CLONE_VM
 232 .\" flag.
 233 .\" .RB ( CLONE_VM
 234 .\" is also implicitly set by
 235 .\" .BR vfork (2),
 236 .\" and can be reversed using this
 237 .\" .BR unshare ()
 238 .\" flag.)
 239 .\" Unshare virtual memory, so that the calling process no
 240 .\" longer shares its virtual address space with any other process.
 241 .PP
 242 In addition,
 243 .BR CLONE_THREAD ,
 244 .BR CLONE_SIGHAND ,
 245 and
 246 .BR CLONE_VM
 247 can be specified in
 248 .I flags
 249 if the caller is single threaded (i.e., it is not sharing
 250 its address space with another process or thread).
 251 In this case, these flags have no effect.
 252 If the process is multithreaded, then
 253 the use of these flags results in an error.
 254 .\" See kernel/fork.c::check_unshare_flags()
 255 .PP
 256 If
 257 .I flags
 258 is specified as zero, then
 259 .BR unshare ()
 260 is a no-op;
 261 no changes are made to the calling process's execution context.
 262 .SH RETURN VALUE
 263 On success, zero returned.
 264 On failure, \-1 is returned and
 265 .I errno
 266 is set to indicate the error.
 267 .SH ERRORS
 268 .TP
 269 .B EINVAL
 270 An invalid bit was specified in
 271 .IR flags .
 272 .TP
 273 .B EINVAL
 274 .BR CLONE_THREAD ,
 275 .BR CLONE_SIGHAND ,
 276 or
 277 .BR CLONE_VM
 278 was specified in
 279 .IR flags ,
 280 and the caller is multithreaded.
 281 .TP
 282 .B ENOMEM
 283 Cannot allocate sufficient memory to copy parts of caller's
 284 context that need to be unshared.
 285 .TP
 286 .B EPERM
 287 The calling process did not have the required privileges for this operation.
 288 .TP
 289 .B EPERM
 290 .BR CLONE_NEWUSER
 291 was specified in
 292 .IR flags ,
 293 but either the effective user ID or the effective group ID of the caller
 294 does not have a mapping in the parent namespace (see
 295 .BR user_namespaces (7)).
 296 .SH VERSIONS
 297 The
 298 .BR unshare ()
 299 system call was added to Linux in kernel 2.6.16.
 300 .SH CONFORMING TO
 301 The
 302 .BR unshare ()
 303 system call is Linux-specific.
 304 .SH NOTES
 305 Not all of the process attributes that can be shared when
 306 a new process is created using
 307 .BR clone (2)
 308 can be unshared using
 309 .BR unshare ().
 310 In particular, as at kernel 3.8,
 311 .\" FIXME all of the following needs to be reviewed for the current kernel
 312 .BR unshare ()
 313 does not implement flags that reverse the effects of
 314 .BR CLONE_SIGHAND ,
 315 .\" However, we can do unshare(CLONE_SIGHAND) if CLONE_SIGHAND
 316 .\" was not specified when doing clone(); i.e., unsharing
 317 .\" signal handlers is permitted if we are not actually
 318 .\" sharing signal handlers.   mtk
 319 .BR CLONE_THREAD ,
 320 or
 321 .BR CLONE_VM .
 322 .\" However, we can do unshare(CLONE_VM) if CLONE_VM
 323 .\" was not specified when doing clone(); i.e., unsharing
 324 .\" virtual memory is permitted if we are not actually
 325 .\" sharing virtual memory.   mtk
 326 Such functionality may be added in the future, if required.
 327 .\"
 328 .\"9) Future Work
 329 .\"--------------
 330 .\"The current implementation of unshare does not allow unsharing of
 331 .\"signals and signal handlers. Signals are complex to begin with and
 332 .\"to unshare signals and/or signal handlers of a currently running
 333 .\"process is even more complex. If in the future there is a specific
 334 .\"need to allow unsharing of signals and/or signal handlers, it can
 335 .\"be incrementally added to unshare without affecting legacy
 336 .\"applications using unshare.
 337 .\"
 338 .SH EXAMPLE
 339 The program below provides a simple implementation of the
 340 .BR unshare (1)
 341 command, which unshares one or more namespaces and executes the
 342 command supplied in its command line arguments.
 343 Here's an example of the use of this program,
 344 running a shell in a new mount namespace,
 345 and verifying that the original shell and the
 346 new shell are in separate mount namespaces:
 347 .in +4n
 348 .nf
 349
 350 $ \fBreadlink /proc/$$/ns/mnt\fP
 351 mnt:[4026531840]
 352 $ \fBsudo ./unshare -m /bin/bash\fP
 353 [sudo] password for cecilia:
 354 # \fBreadlink /proc/$$/ns/mnt\fP
 355 mnt:[4026532325]
 356 .fi
 357 .in
 358
 359 The differing output of the two
 360 .BR readlink (1)
 361 commands shows that the two shells are in different mount namespaces.
 362 .SS Program source
 363 \&
 364 .nf
 365 /* unshare.c
 366
 367    A simple implementation of the unshare(1) command: unshare
 368    namespaces and execute a command.
 369 */
 370 #define _GNU_SOURCE
 371 #include <sched.h>
 372 #include <unistd.h>
 373 #include <stdlib.h>
 374 #include <stdio.h>
 375
 376 /* A simple error\-handling function: print an error message based
 377    on the value in \(aqerrno\(aq and terminate the calling process */
 378
 379 #define errExit(msg)    do { perror(msg); exit(EXIT_FAILURE); \\
 380                         } while (0)
 381
 382 static void
 383 usage(char *pname)
 384 {
 385     fprintf(stderr, "Usage: %s [options] program [arg...]\\n", pname);
 386     fprintf(stderr, "Options can be:\\n");
 387     fprintf(stderr, "    \-i   unshare IPC namespace\\n");
 388     fprintf(stderr, "    \-m   unshare mount namespace\\n");
 389     fprintf(stderr, "    \-n   unshare network namespace\\n");
 390     fprintf(stderr, "    \-p   unshare PID namespace\\n");
 391     fprintf(stderr, "    \-u   unshare UTS namespace\\n");
 392     fprintf(stderr, "    \-U   unshare user namespace\\n");
 393     exit(EXIT_FAILURE);
 394 }
 395
 396 int
 397 main(int argc, char *argv[])
 398 {
 399     int flags, opt;
 400
 401     flags = 0;
 402
 403     while ((opt = getopt(argc, argv, "imnpuU")) != \-1) {
 404         switch (opt) {
 405         case \(aqi\(aq: flags |= CLONE_NEWIPC;        break;
 406         case \(aqm\(aq: flags |= CLONE_NEWNS;         break;
 407         case \(aqn\(aq: flags |= CLONE_NEWNET;        break;
 408         case \(aqp\(aq: flags |= CLONE_NEWPID;        break;
 409         case \(aqu\(aq: flags |= CLONE_NEWUTS;        break;
 410         case \(aqU\(aq: flags |= CLONE_NEWUSER;       break;
 411         default:  usage(argv[0]);
 412         }
 413     }
 414
 415     if (optind >= argc)
 416         usage(argv[0]);
 417
 418     if (unshare(flags) == \-1)
 419         errExit("unshare");
 420
 421     execvp(argv[optind], &argv[optind]);
 422     errExit("execvp");
 423 }
 424 .fi
 425 .SH SEE ALSO
 426 .BR unshare (1),
 427 .BR clone (2),
 428 .BR fork (2),
 429 .BR kcmp (2),
 430 .BR setns (2),
 431 .BR vfork (2),
 432 .BR namespaces (7)
 433
 434 .I Documentation/unshare.txt
 435 in the Linux kernel source tree