1 .\" Copyright (c) 1992 Drew Eckhardt <drew@cs.colorado.edu>, March 28, 1992
2 .\" and Copyright (c) Michael Kerrisk, 2001, 2002, 2005, 2013
4 .\" %%%LICENSE_START(GPL_NOVERSION_ONELINE)
5 .\" May be distributed under the GNU General Public License.
8 .\" Modified by Michael Haardt <michael@moria.de>
9 .\" Modified 24 Jul 1993 by Rik Faith <faith@cs.unc.edu>
10 .\" Modified 21 Aug 1994 by Michael Chastain <mec@shell.portal.com>:
11 .\" New man page (copied from 'fork.2').
12 .\" Modified 10 June 1995 by Andries Brouwer <aeb@cwi.nl>
13 .\" Modified 25 April 1998 by Xavier Leroy <Xavier.Leroy@inria.fr>
14 .\" Modified 26 Jun 2001 by Michael Kerrisk
15 .\" Mostly upgraded to 2.4.x
16 .\" Added prototype for sys_clone() plus description
17 .\" Added CLONE_THREAD with a brief description of thread groups
18 .\" Added CLONE_PARENT and revised entire page remove ambiguity
19 .\" between "calling process" and "parent process"
20 .\" Added CLONE_PTRACE and CLONE_VFORK
21 .\" Added EPERM and EINVAL error codes
22 .\" Renamed "__clone" to "clone" (which is the prototype in <sched.h>)
23 .\" various other minor tidy ups and clarifications.
24 .\" Modified 26 Jun 2001 by Michael Kerrisk <mtk.manpages@gmail.com>
25 .\" Updated notes for 2.4.7+ behavior of CLONE_THREAD
26 .\" Modified 15 Oct 2002 by Michael Kerrisk <mtk.manpages@gmail.com>
27 .\" Added description for CLONE_NEWNS, which was added in 2.4.19
28 .\" Slightly rephrased, aeb.
29 .\" Modified 1 Feb 2003 - added CLONE_SIGHAND restriction, aeb.
30 .\" Modified 1 Jan 2004 - various updates, aeb
31 .\" Modified 2004-09-10 - added CLONE_PARENT_SETTID etc. - aeb.
32 .\" 2005-04-12, mtk, noted the PID caching behavior of NPTL's getpid()
33 .\" wrapper under BUGS.
34 .\" 2005-05-10, mtk, added CLONE_SYSVSEM, CLONE_UNTRACED, CLONE_STOPPED.
35 .\" 2005-05-17, mtk, Substantially enhanced discussion of CLONE_THREAD.
36 .\" 2008-11-18, mtk, order CLONE_* flags alphabetically
37 .\" 2008-11-18, mtk, document CLONE_NEWPID
38 .\" 2008-11-19, mtk, document CLONE_NEWUTS
39 .\" 2008-11-19, mtk, document CLONE_NEWIPC
40 .\" 2008-11-19, Jens Axboe, mtk, document CLONE_IO
42 .TH CLONE 2 2019-08-02 "Linux" "Linux Programmer's Manual"
44 clone, __clone2 \- create a child process
47 /* Prototype for the glibc wrapper function */
49 .B #define _GNU_SOURCE
52 .BI "int clone(int (*" "fn" ")(void *), void *" child_stack ,
53 .BI " int " flags ", void *" "arg" ", ... "
54 .BI " /* pid_t *" ptid ", void *" newtls \
55 ", pid_t *" ctid " */ );"
57 /* For the prototype of the raw system call, see NOTES */
61 creates a new process, in a manner similar to
64 This page describes both the glibc
66 wrapper function and the underlying system call on which it is based.
67 The main text describes the wrapper function;
68 the differences for the raw system call
69 are described toward the end of this page.
74 allows the child process to share parts of its execution context with
75 the calling process, such as the virtual address space, the table of file
76 descriptors, and the table of signal handlers.
77 (Note that on this manual
78 page, "calling process" normally corresponds to "parent process".
79 But see the description of
85 is to implement threads: multiple flows of control in a program that
86 run concurrently in a shared address space.
88 When the child process is created with
90 it commences execution by calling the function pointed to by the argument
94 where execution continues in the child from the point
100 argument is passed as the argument of the function
105 function returns, the child process terminates.
106 The integer returned by
108 is the exit status for the child process.
109 The child process may also terminate explicitly by calling
111 or after receiving a fatal signal.
115 argument specifies the location of the stack used by the child process.
116 Since the child and calling process may share memory,
117 it is not possible for the child process to execute in the
118 same stack as the calling process.
119 The calling process must therefore
120 set up memory space for the child stack and pass a pointer to this
123 Stacks grow downward on all processors that run Linux
124 (except the HP PA processors), so
126 usually points to the topmost address of the memory space set up for
131 contains the number of the
132 .I "termination signal"
133 sent to the parent when the child dies.
134 If this signal is specified as anything other than
136 then the parent process must specify the
140 options when waiting for the child with
142 If no signal is specified, then the parent process is not signaled
143 when the child terminates.
146 may also be bitwise-ORed with zero or more of the following constants,
147 in order to specify what is shared between the calling process
148 and the child process:
150 .BR CLONE_CHILD_CLEARTID " (since Linux 2.5.49)"
151 Clear (zero) the child thread ID at the location
153 in child memory when the child exits, and do a wakeup on the futex
155 The address involved may be changed by the
156 .BR set_tid_address (2)
158 This is used by threading libraries.
160 .BR CLONE_CHILD_SETTID " (since Linux 2.5.49)"
161 Store the child thread ID at the location
163 in the child's memory.
164 The store operation completes before
166 returns control to user space in the child process.
167 (Note that the store operation may not have completed before
169 returns in the parent process, which will be relevant if the
171 flag is also employed.)
173 .BR CLONE_FILES " (since Linux 2.0)"
176 is set, the calling process and the child process share the same file
178 Any file descriptor created by the calling process or by the child
179 process is also valid in the other process.
180 Similarly, if one of the processes closes a file descriptor,
181 or changes its associated flags (using the
184 operation), the other process is also affected.
185 If a process sharing a file descriptor table calls
187 its file descriptor table is duplicated (unshared).
191 is not set, the child process inherits a copy of all file descriptors
192 opened in the calling process at the time of
194 Subsequent operations that open or close file descriptors,
195 or change file descriptor flags,
196 performed by either the calling
197 process or the child process do not affect the other process.
199 that the duplicated file descriptors in the child refer to the same
200 open file descriptions as the corresponding file descriptors
201 in the calling process,
202 and thus share file offsets and file status flags (see
205 .BR CLONE_FS " (since Linux 2.0)"
208 is set, the caller and the child process share the same filesystem
210 This includes the root of the filesystem, the current
211 working directory, and the umask.
217 performed by the calling process or the child process also affects the
222 is not set, the child process works on a copy of the filesystem
223 information of the calling process at the time of the
231 performed later by one of the processes do not affect the other process.
233 .BR CLONE_IO " (since Linux 2.6.25)"
236 is set, then the new process shares an I/O context with
238 If this flag is not set, then (as with
240 the new process has its own I/O context.
242 .\" The following based on text from Jens Axboe
243 The I/O context is the I/O scope of the disk scheduler (i.e.,
244 what the I/O scheduler uses to model scheduling of a process's I/O).
245 If processes share the same I/O context,
246 they are treated as one by the I/O scheduler.
247 As a consequence, they get to share disk time.
248 For some I/O schedulers,
249 .\" the anticipatory and CFQ scheduler
250 if two processes share an I/O context,
251 they will be allowed to interleave their disk access.
252 If several threads are doing I/O on behalf of the same process
254 for instance), they should employ
256 to get better I/O performance.
259 If the kernel is not configured with the
261 option, this flag is a no-op.
263 .BR CLONE_NEWCGROUP " (since Linux 4.6)"
264 Create the process in a new cgroup namespace.
265 If this flag is not set, then (as with
267 the process is created in the same cgroup namespaces as the calling process.
268 This flag is intended for the implementation of containers.
270 For further information on cgroup namespaces, see
271 .BR cgroup_namespaces (7).
273 Only a privileged process
274 .RB ( CAP_SYS_ADMIN )
276 .BR CLONE_NEWCGROUP .
279 .BR CLONE_NEWIPC " (since Linux 2.6.19)"
282 is set, then create the process in a new IPC namespace.
283 If this flag is not set, then (as with
285 the process is created in the same IPC namespace as
287 This flag is intended for the implementation of containers.
289 An IPC namespace provides an isolated view of System\ V IPC objects (see
291 and (since Linux 2.6.30)
292 .\" commit 7eafd7c74c3f2e67c27621b987b28397110d643f
293 .\" https://lwn.net/Articles/312232/
296 .BR mq_overview (7)).
297 The common characteristic of these IPC mechanisms is that IPC
298 objects are identified by mechanisms other than filesystem
301 Objects created in an IPC namespace are visible to all other processes
302 that are members of that namespace,
303 but are not visible to processes in other IPC namespaces.
305 When an IPC namespace is destroyed
306 (i.e., when the last process that is a member of the namespace terminates),
307 all IPC objects in the namespace are automatically destroyed.
309 Only a privileged process
310 .RB ( CAP_SYS_ADMIN )
313 This flag can't be specified in conjunction with
316 For further information on IPC namespaces, see
319 .BR CLONE_NEWNET " (since Linux 2.6.24)"
320 (The implementation of this flag was completed only
321 by about kernel version 2.6.29.)
325 is set, then create the process in a new network namespace.
326 If this flag is not set, then (as with
328 the process is created in the same network namespace as
330 This flag is intended for the implementation of containers.
332 A network namespace provides an isolated view of the networking stack
333 (network device interfaces, IPv4 and IPv6 protocol stacks,
334 IP routing tables, firewall rules, the
338 directory trees, sockets, etc.).
339 A physical network device can live in exactly one
343 device pair provides a pipe-like abstraction
344 that can be used to create tunnels between network namespaces,
345 and can be used to create a bridge to a physical network device
346 in another namespace.
348 When a network namespace is freed
349 (i.e., when the last process in the namespace terminates),
350 its physical network devices are moved back to the
351 initial network namespace (not to the parent of the process).
352 For further information on network namespaces, see
355 Only a privileged process
356 .RB ( CAP_SYS_ADMIN )
360 .BR CLONE_NEWNS " (since Linux 2.4.19)"
363 is set, the cloned child is started in a new mount namespace,
364 initialized with a copy of the namespace of the parent.
367 is not set, the child lives in the same mount
368 namespace as the parent.
370 Only a privileged process
371 .RB ( CAP_SYS_ADMIN )
374 It is not permitted to specify both
378 .\" See https://lwn.net/Articles/543273/
383 For further information on mount namespaces, see
386 .BR mount_namespaces (7).
388 .BR CLONE_NEWPID " (since Linux 2.6.24)"
389 .\" This explanation draws a lot of details from
390 .\" http://lwn.net/Articles/259217/
391 .\" Authors: Pavel Emelyanov <xemul@openvz.org>
392 .\" and Kir Kolyshkin <kir@openvz.org>
394 .\" The primary kernel commit is 30e49c263e36341b60b735cbef5ca37912549264
395 .\" Author: Pavel Emelyanov <xemul@openvz.org>
398 is set, then create the process in a new PID namespace.
399 If this flag is not set, then (as with
401 the process is created in the same PID namespace as
403 This flag is intended for the implementation of containers.
405 For further information on PID namespaces, see
408 .BR pid_namespaces (7).
410 Only a privileged process
411 .RB ( CAP_SYS_ADMIN )
414 This flag can't be specified in conjunction with
420 (This flag first became meaningful for
425 semantics were merged in Linux 3.5,
426 and the final pieces to make the user namespaces completely usable were
427 merged in Linux 3.8.)
431 is set, then create the process in a new user namespace.
432 If this flag is not set, then (as with
434 the process is created in the same user namespace as the calling process.
436 Before Linux 3.8, use of
438 required that the caller have three capabilities:
443 .\" Before Linux 2.6.29, it appears that only CAP_SYS_ADMIN was needed
444 Starting with Linux 3.8,
445 no privileges are needed to create a user namespace.
447 This flag can't be specified in conjunction with
451 For security reasons,
452 .\" commit e66eded8309ebf679d3d3c1f5820d1f2ca332c71
453 .\" https://lwn.net/Articles/543273/
454 .\" The fix actually went into 3.9 and into 3.8.3. However, user namespaces
455 .\" were, for practical purposes, unusable in earlier 3.8.x because of the
456 .\" various filesystems that didn't support userns.
458 cannot be specified in conjunction with
461 For further information on user namespaces, see
464 .BR user_namespaces (7).
466 .BR CLONE_NEWUTS " (since Linux 2.6.19)"
469 is set, then create the process in a new UTS namespace,
470 whose identifiers are initialized by duplicating the identifiers
471 from the UTS namespace of the calling process.
472 If this flag is not set, then (as with
474 the process is created in the same UTS namespace as
476 This flag is intended for the implementation of containers.
478 A UTS namespace is the set of identifiers returned by
480 among these, the domain name and the hostname can be modified by
481 .BR setdomainname (2)
485 Changes made to the identifiers in a UTS namespace
486 are visible to all other processes in the same namespace,
487 but are not visible to processes in other UTS namespaces.
489 Only a privileged process
490 .RB ( CAP_SYS_ADMIN )
494 For further information on UTS namespaces, see
497 .BR CLONE_PARENT " (since Linux 2.3.12)"
500 is set, then the parent of the new child (as returned by
502 will be the same as that of the calling process.
506 is not set, then (as with
508 the child's parent is the calling process.
510 Note that it is the parent process, as returned by
512 which is signaled when the child terminates, so that
515 is set, then the parent of the calling process, rather than the
516 calling process itself, will be signaled.
518 .BR CLONE_PARENT_SETTID " (since Linux 2.5.49)"
519 Store the child thread ID at the location
521 in the parent's memory.
522 (In Linux 2.5.32-2.5.48 there was a flag
525 The store operation completes before
527 returns control to user space.
529 .BR CLONE_PID " (Linux 2.0 to 2.5.15)"
532 is set, the child process is created with the same process ID as
534 This is good for hacking the system, but otherwise
536 From Linux 2.3.21 onward, this flag could be
537 specified only by the system boot process (PID 0).
538 The flag disappeared completely from the kernel sources in Linux 2.5.16.
539 Since then, the kernel silently ignores this bit if it is specified in
542 .BR CLONE_PTRACE " (since Linux 2.2)"
545 is specified, and the calling process is being traced,
546 then trace the child also (see
549 .BR CLONE_SETTLS " (since Linux 2.5.32)"
550 The TLS (Thread Local Storage) descriptor is set to
553 The interpretation of
555 and the resulting effect is architecture dependent.
559 .IR "struct user_desc\ *"
561 .BR set_thread_area (2)).
562 On x86-64 it is the new value to be set for the %fs base register
567 On architectures with a dedicated TLS register, it is the new value
570 .BR CLONE_SIGHAND " (since Linux 2.0)"
573 is set, the calling process and the child process share the same table of
575 If the calling process or child process calls
577 to change the behavior associated with a signal, the behavior is
578 changed in the other process as well.
579 However, the calling process and child
580 processes still have distinct signal masks and sets of pending
582 So, one of them may block or unblock signals using
584 without affecting the other process.
588 is not set, the child process inherits a copy of the signal handlers
589 of the calling process at the time
594 performed later by one of the processes have no effect on the other
598 .\" Precisely: Linux 2.6.0-test6
606 .BR CLONE_STOPPED " (since Linux 2.6.0)"
607 .\" Precisely: Linux 2.6.0-test2
610 is set, then the child is initially stopped (as though it was sent a
612 signal), and must be resumed by sending it a
618 from Linux 2.6.25 onward,
621 altogether in Linux 2.6.38.
622 Since then, the kernel silently ignores it without error.
623 .\" glibc 2.8 removed this defn from bits/sched.h
624 Starting with Linux 4.6, the same bit was reused for the
628 .BR CLONE_SYSVSEM " (since Linux 2.5.10)"
631 is set, then the child and the calling process share
632 a single list of System V semaphore adjustment
636 In this case, the shared list accumulates
638 values across all processes sharing the list,
639 and semaphore adjustments are performed only when the last process
640 that is sharing the list terminates (or ceases sharing the list using
642 If this flag is not set, then the child has a separate
644 list that is initially empty.
646 .BR CLONE_THREAD " (since Linux 2.4.0)"
647 .\" Precisely: Linux 2.6.0-test8
650 is set, the child is placed in the same thread group as the calling process.
651 To make the remainder of the discussion of
653 more readable, the term "thread" is used to refer to the
654 processes within a thread group.
656 Thread groups were a feature added in Linux 2.4 to support the
657 POSIX threads notion of a set of threads that share a single PID.
658 Internally, this shared PID is the so-called
659 thread group identifier (TGID) for the thread group.
660 Since Linux 2.4, calls to
662 return the TGID of the caller.
664 The threads within a group can be distinguished by their (system-wide)
665 unique thread IDs (TID).
666 A new thread's TID is available as the function result
667 returned to the caller of
669 and a thread can obtain
673 When a call is made to
677 then the resulting thread is placed in a new thread group
678 whose TGID is the same as the thread's TID.
681 of the new thread group.
683 A new thread created with
685 has the same parent process as the caller of
691 return the same value for all of the threads in a thread group.
694 thread terminates, the thread that created it using
698 (or other termination) signal;
699 nor can the status of such a thread be obtained
702 (The thread is said to be
705 After all of the threads in a thread group terminate
706 the parent process of the thread group is sent a
708 (or other termination) signal.
710 If any of the threads in a thread group performs an
712 then all threads other than the thread group leader are terminated,
713 and the new program is executed in the thread group leader.
715 If one of the threads in a thread group creates a child using
717 then any thread in the group can
728 (and note that, since Linux 2.6.0,
729 .\" Precisely: Linux 2.6.0-test6
735 Signal dispositions and actions are process-wide:
736 if an unhandled signal is delivered to a thread, then
737 it will affect (terminate, stop, continue, be ignored in)
738 all members of the thread group.
740 Each thread has its own signal mask, as set by
743 A signal may be process-directed or thread-directed.
744 A process-directed signal is targeted at a thread group (i.e., a TGID),
745 and is delivered to an arbitrarily selected thread from among those
746 that are not blocking the signal.
747 A signal may be process directed because it was generated by the kernel
748 for reasons other than a hardware exception, or because it was sent using
752 A thread-directed signal is targeted at (i.e., delivered to)
754 A signal may be thread directed because it was sent using
757 .BR pthread_sigqueue (3),
758 or because the thread executed a machine language instruction that triggered
760 (e.g., invalid memory access triggering
762 or a floating-point exception triggering
767 returns a signal set that is the union of the pending process-directed
768 signals and the signals that are pending for the calling thread.
770 If a process-directed signal is delivered to a thread group,
771 and the thread group has installed a handler for the signal, then
772 the handler will be invoked in exactly one, arbitrarily selected
773 member of the thread group that has not blocked the signal.
774 If multiple threads in a group are waiting to accept the same signal using
776 the kernel will arbitrarily select one of these threads
777 to receive the signal.
779 .BR CLONE_UNTRACED " (since Linux 2.5.46)"
782 is specified, then a tracing process cannot force
784 on this child process.
786 .BR CLONE_VFORK " (since Linux 2.2)"
789 is set, the execution of the calling process is suspended
790 until the child releases its virtual memory
791 resources via a call to
800 is not set, then both the calling process and the child are schedulable
801 after the call, and an application should not rely on execution occurring
802 in any particular order.
804 .BR CLONE_VM " (since Linux 2.0)"
807 is set, the calling process and the child process run in the same memory
809 In particular, memory writes performed by the calling process
810 or by the child process are also visible in the other process.
811 Moreover, any memory mapping or unmapping performed with
815 by the child or calling process also affects the other process.
819 is not set, the child process runs in a separate copy of the memory
820 space of the calling process at the time of
822 Memory writes or file mappings/unmappings performed by one of the
823 processes do not affect the other, as with
828 wrapper function makes some changes
829 in the memory pointed to by
831 (changes required to set the stack up correctly for the child)
838 is used to recursively create children,
839 do not use the buffer employed for the parent's stack
840 as the stack of the child.
842 .SS C library/kernel differences
845 system call corresponds more closely to
847 in that execution in the child continues from the point of the
855 wrapper function are omitted.
857 Another difference for the raw
859 system call is that the
861 argument may be NULL,
862 in which case the child uses a duplicate of the parent's stack.
863 (Copy-on-write semantics ensure that the child gets separate copies
864 of stack pages when either process modifies the stack.)
865 In this case, for correct operation, the
867 option should not be specified.
870 the parent's memory because of the use of the
873 then no copy-on-write duplication occurs and chaos is likely to result.)
875 The order of the arguments also differs in the raw system call,
876 and there are variations in the arguments across architectures,
877 as detailed in the following paragraphs.
879 The raw system call interface on x86-64 and some other architectures
880 (including sh, tile, ia-64, and alpha) is:
884 .BI "long clone(unsigned long " flags ", void *" child_stack ,
885 .BI " int *" ptid ", int *" ctid ,
886 .BI " unsigned long " newtls );
890 On x86-32, and several other common architectures
891 (including score, ARM, ARM 64, PA-RISC, arc, Power PC, xtensa,
893 .\" CONFIG_CLONE_BACKWARDS
894 the order of the last two arguments is reversed:
898 .BI "long clone(unsigned long " flags ", void *" child_stack ,
899 .BI " int *" ptid ", unsigned long " newtls ,
904 On the cris and s390 architectures,
905 .\" CONFIG_CLONE_BACKWARDS2
906 the order of the first two arguments is reversed:
910 .BI "long clone(void *" child_stack ", unsigned long " flags ,
911 .BI " int *" ptid ", int *" ctid ,
912 .BI " unsigned long " newtls );
916 On the microblaze architecture,
917 .\" CONFIG_CLONE_BACKWARDS3
918 an additional argument is supplied:
922 .BI "long clone(unsigned long " flags ", void *" child_stack ,
923 .BI " int " stack_size , "\fR /* Size of stack */"
924 .BI " int *" ptid ", int *" ctid ,
925 .BI " unsigned long " newtls );
929 .SS blackfin, m68k, and sparc
930 .\" Mike Frysinger noted in a 2013 mail:
931 .\" these arches don't define __ARCH_WANT_SYS_CLONE:
932 .\" blackfin ia64 m68k sparc
933 The argument-passing conventions on
934 blackfin, m68k, and sparc are different from the descriptions above.
935 For details, see the kernel (and glibc) source.
937 On ia64, a different interface is used:
941 .BI "int __clone2(int (*" "fn" ")(void *), "
942 .BI " void *" child_stack_base ", size_t " stack_size ,
943 .BI " int " flags ", void *" "arg" ", ... "
944 .BI " /* pid_t *" ptid ", struct user_desc *" tls \
945 ", pid_t *" ctid " */ );"
949 The prototype shown above is for the glibc wrapper function;
950 for the system call itself,
951 the prototype can be described as follows (it is identical to the
953 prototype on microblaze):
957 .BI "long clone2(unsigned long " flags ", void *" child_stack_base ,
958 .BI " int " stack_size , "\fR /* Size of stack */"
959 .BI " int *" ptid ", int *" ctid ,
960 .BI " unsigned long " tls );
965 operates in the same way as
969 points to the lowest address of the child's stack area,
972 specifies the size of the stack pointed to by
973 .IR child_stack_base .
974 .SS Linux 2.4 and earlier
975 In Linux 2.4 and earlier,
977 does not take arguments
983 .\" gettid(2) returns current->pid;
984 .\" getpid(2) returns current->tgid;
985 On success, the thread ID of the child process is returned
986 in the caller's thread of execution.
987 On failure, \-1 is returned
988 in the caller's context, no child process will be created, and
990 will be set appropriately.
994 Too many processes are already running; see
1002 (Since Linux 2.6.0.)
1003 .\" Precisely: Linux 2.6.0-test6
1010 (Since Linux 2.5.35.)
1013 .\" Precisely one of
1014 .\" .B CLONE_DETACHED
1018 .\" (Since Linux 2.6.0-test6.)
1022 was specified, but the current process previously called
1028 to reassociate itself with a PID namespace.
1031 .\" commit e66eded8309ebf679d3d3c1f5820d1f2ca332c71
1039 .BR EINVAL " (since Linux 3.9)"
1060 and one (or both) of
1068 Returned by the glibc
1070 wrapper function when
1074 is specified as NULL.
1080 but the kernel was not configured with the
1090 but the kernel was not configured with the
1098 but the kernel was not configured with the
1106 but the kernel was not configured with the
1114 but the kernel was not configured with the
1120 is not aligned to a suitable boundary for this architecture.
1121 For example, on aarch64,
1123 must be a multiple of 16.
1126 Cannot allocate sufficient memory to allocate a task structure for the
1127 child, or to copy those parts of the caller's context that need to be
1130 .BR ENOSPC " (since Linux 3.7)"
1131 .\" commit f2302505775fd13ba93f034206f1e2a587017929
1133 was specified in flags,
1134 but the limit on the nesting depth of PID namespaces
1135 would have been exceeded; see
1136 .BR pid_namespaces (7).
1138 .BR ENOSPC " (since Linux 4.9; beforehand " EUSERS )
1142 and the call would cause the limit on the number of
1143 nested user namespaces to be exceeded.
1145 .BR user_namespaces (7).
1147 From Linux 3.11 to Linux 4.8, the error diagnosed in this case was
1150 .BR ENOSPC " (since Linux 4.9)"
1151 One of the values in
1153 specified the creation of a new user namespace,
1154 but doing so would have caused the limit defined by the corresponding file in
1157 For further details, see
1161 .BR CLONE_NEWCGROUP ,
1168 was specified by an unprivileged process (process without \fBCAP_SYS_ADMIN\fP).
1172 was specified by a process other than process 0.
1173 (This error occurs only on Linux 2.5.15 and earlier.)
1179 but either the effective user ID or the effective group ID of the caller
1180 does not have a mapping in the parent namespace (see
1181 .BR user_namespaces (7)).
1183 .BR EPERM " (since Linux 3.9)"
1184 .\" commit 3151527ee007b73a0ebd296010f1c0454a919c7d
1188 and the caller is in a chroot environment
1189 .\" FIXME What is the rationale for this restriction?
1190 (i.e., the caller's root directory does not match the root directory
1191 of the mount namespace in which it resides).
1193 .BR ERESTARTNOINTR " (since Linux 2.6.17)"
1194 .\" commit 4a2c7a7837da1b91468e50426066d988050e4d56
1195 System call was interrupted by a signal and will be restarted.
1196 (This can be seen only during a trace.)
1198 .BR EUSERS " (Linux 3.11 to Linux 4.8)"
1202 and the limit on the number of nested user namespaces would be exceeded.
1203 See the discussion of the
1207 .\" There is no entry for
1212 .\" as described in this manual page.
1215 is Linux-specific and should not be used in programs
1216 intended to be portable.
1220 system call can be used to test whether two processes share various
1221 resources such as a file descriptor table,
1222 System V semaphore undo operations, or a virtual address space.
1225 Handlers registered using
1226 .BR pthread_atfork (3)
1227 are not executed during a call to
1230 In the Linux 2.4.x series,
1232 generally does not make the parent of the new thread the same
1233 as the parent of the calling process.
1234 However, for kernel versions 2.4.7 to 2.4.18 the
1238 flag (as in Linux 2.6.0 and later).
1240 For a while there was
1242 (introduced in 2.5.32):
1243 parent wants no child-exit signal.
1244 In Linux 2.6.2, the need to give this flag together with
1247 This flag is still defined, but has no effect.
1251 should not be called through vsyscall, but directly through
1254 GNU C library versions 2.3.4 up to and including 2.24
1255 contained a wrapper function for
1257 that performed caching of PIDs.
1258 This caching relied on support in the glibc wrapper for
1260 but limitations in the implementation
1261 meant that the cache was not up to date in some circumstances.
1263 if a signal was delivered to the child immediately after the
1265 call, then a call to
1267 in a handler for the signal could return the PID
1268 of the calling process ("the parent"),
1269 if the clone wrapper had not yet had a chance to update the PID
1271 (This discussion ignores the case where the child was created using
1276 return the same value in the child and in the process that called
1278 since the caller and the child are in the same thread group.
1279 The stale-cache problem also does not occur if the
1283 To get the truth, it was sometimes necessary to use code such as the following:
1287 #include <syscall.h>
1291 mypid = syscall(SYS_getpid);
1294 .\" See also the following bug reports
1295 .\" https://bugzilla.redhat.com/show_bug.cgi?id=417521
1296 .\" http://sourceware.org/bugzilla/show_bug.cgi?id=6910
1298 Because of the stale-cache problem, as well as other problems noted in
1300 the PID caching feature was removed in glibc 2.25.
1302 The following program demonstrates the use of
1304 to create a child process that executes in a separate UTS namespace.
1305 The child changes the hostname in its UTS namespace.
1306 Both parent and child then display the system hostname,
1307 making it possible to see that the hostname
1308 differs in the UTS namespaces of the parent and child.
1309 For an example of the use of this program, see
1314 #include <sys/wait.h>
1315 #include <sys/utsname.h>
1322 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \e
1325 static int /* Start function for cloned child */
1326 childFunc(void *arg)
1330 /* Change hostname in UTS namespace of child */
1332 if (sethostname(arg, strlen(arg)) == \-1)
1333 errExit("sethostname");
1335 /* Retrieve and display hostname */
1337 if (uname(&uts) == \-1)
1339 printf("uts.nodename in child: %s\en", uts.nodename);
1341 /* Keep the namespace open for a while, by sleeping.
1342 This allows some experimentation\-\-for example, another
1343 process might join the namespace. */
1347 return 0; /* Child terminates now */
1350 #define STACK_SIZE (1024 * 1024) /* Stack size for cloned child */
1353 main(int argc, char *argv[])
1355 char *stack; /* Start of stack buffer */
1356 char *stackTop; /* End of stack buffer */
1361 fprintf(stderr, "Usage: %s <child\-hostname>\en", argv[0]);
1365 /* Allocate stack for child */
1367 stack = malloc(STACK_SIZE);
1370 stackTop = stack + STACK_SIZE; /* Assume stack grows downward */
1372 /* Create child that has its own UTS namespace;
1373 child commences execution in childFunc() */
1375 pid = clone(childFunc, stackTop, CLONE_NEWUTS | SIGCHLD, argv[1]);
1378 printf("clone() returned %ld\en", (long) pid);
1380 /* Parent falls through to here */
1382 sleep(1); /* Give child time to change its hostname */
1384 /* Display hostname in parent\(aqs UTS namespace. This will be
1385 different from hostname in child\(aqs UTS namespace. */
1387 if (uname(&uts) == \-1)
1389 printf("uts.nodename in parent: %s\en", uts.nodename);
1391 if (waitpid(pid, NULL, 0) == \-1) /* Wait for child */
1393 printf("child has terminated\en");
1404 .BR set_thread_area (2),
1405 .BR set_tid_address (2),
1410 .BR capabilities (7),