1 .\" Copyright (c) 2013, 2014 by Michael Kerrisk <mtk.manpages@gmail.com>
2 .\" and Copyright (c) 2012, 2014 by Eric W. Biederman <ebiederm@xmission.com>
4 .\" %%%LICENSE_START(VERBATIM)
5 .\" Permission is granted to make and distribute verbatim copies of this
6 .\" manual provided the copyright notice and this permission notice are
7 .\" preserved on all copies.
9 .\" Permission is granted to copy and distribute modified versions of this
10 .\" manual under the conditions for verbatim copying, provided that the
11 .\" entire resulting derived work is distributed under the terms of a
12 .\" permission notice identical to this one.
14 .\" Since the Linux kernel and libraries are constantly changing, this
15 .\" manual page may be incorrect or out-of-date. The author(s) assume no
16 .\" responsibility for errors or omissions, or for damages resulting from
17 .\" the use of the information contained herein. The author(s) may not
18 .\" have taken the same level of care in the production of this manual,
19 .\" which is licensed free of charge, as they might when working
22 .\" Formatted or processed versions of this manual, if unaccompanied by
23 .\" the source, must acknowledge the copyright and authors of this work.
27 .TH USER_NAMESPACES 7 2014-09-21 "Linux" "Linux Programmer's Manual"
29 user_namespaces \- overview of Linux user namespaces
31 For an overview of namespaces, see
34 User namespaces isolate security-related identifiers and attributes,
36 user IDs and group IDs (see
41 .\" FIXME: This page says very little about the interaction
42 .\" of user namespaces and keys. Add something on this topic.
44 .BR capabilities (7)).
45 A process's user and group IDs can be different
46 inside and outside a user namespace.
48 a process can have a normal unprivileged user ID outside a user namespace
49 while at the same time having a user ID of 0 inside the namespace;
51 the process has full privileges for operations inside the user namespace,
52 but is unprivileged for operations outside the namespace.
54 .\" ============================================================
56 .SS Nested namespaces, namespace membership
57 User namespaces can be nested;
58 that is, each user namespace\(emexcept the initial ("root")
59 namespace\(emhas a parent user namespace,
60 and can have zero or more child user namespaces.
61 The parent user namespace is the user namespace
62 of the process that creates the user namespace via a call to
70 The kernel imposes (since version 3.11) a limit of 32 nested levels of
71 .\" commit 8742f229b635bf1c1c84a3dfe5e47c814c20b5c8
73 .\" FIXME Explain the rationale for this limit. (What is the rationale?)
78 that would cause this limit to be exceeded fail with the error
81 Each process is a member of exactly one user namespace.
88 flag is a member of the same user namespace as its parent.
89 A single-threaded process can join another user namespace with
94 upon doing so, it gains a full set of capabilities in that namespace.
102 flag makes the new child process (for
106 a member of the new user namespace created by the call.
108 .\" ============================================================
111 The child process created by
115 flag starts out with a complete set
116 of capabilities in the new user namespace.
117 Likewise, a process that creates a new user namespace using
119 or joins an existing user namespace using
121 gains a full set of capabilities in that namespace.
123 that process has no capabilities in the parent (in the case of
125 or previous (in the case of
130 even if the new namespace is created or joined by the root user
131 (i.e., a process with user ID 0 in the root namespace).
135 will cause a process's capabilities to be recalculated in the usual way (see
136 .BR capabilities (7)),
138 unless it has a user ID of 0 within the namespace or the executable file
139 has a nonempty inheritable capabilities mask,
140 it will lose all capabilities.
141 See the discussion of user and group ID mappings, below.
150 flag sets the "securebits" flags
152 .BR capabilities (7))
153 to their default values (all flags disabled) in the child (for
159 Note that because the caller no longer has capabilities
160 in its original user namespace after a call to
162 it is not possible for a process to reset its "securebits" flags while
163 retaining its user namespace membership by using a pair of
165 calls to move to another user namespace and then return to
166 its original user namespace.
168 Having a capability inside a user namespace
169 permits a process to perform operations (that require privilege)
170 only on resources governed by that namespace.
171 The rules for determining whether or not a process has a capability
172 in a particular user namespace are as follows:
174 A process has a capability inside a user namespace
175 if it is a member of that namespace and
176 it has the capability in its effective capability set.
177 A process can gain capabilities in its effective capability
179 For example, it may execute a set-user-ID program or an
180 executable with associated file capabilities.
182 a process may gain capabilities via the effect of
187 as already described.
188 .\" In the 3.8 sources, see security/commoncap.c::cap_capable():
190 If a process has a capability in a user namespace,
191 then it has that capability in all child (and further removed descendant)
194 .\" * The owner of the user namespace in the parent of the
195 .\" * user namespace has all caps.
196 When a user namespace is created, the kernel records the effective
197 user ID of the creating process as being the "owner" of the namespace.
198 .\" (and likewise associates the effective group ID of the creating process
199 .\" with the namespace).
200 A process that resides
201 in the parent of the user namespace
202 .\" See kernel commit 520d9eabce18edfef76a60b7b839d54facafe1f9 for a fix
204 and whose effective user ID matches the owner of the namespace
205 has all capabilities in the namespace.
206 .\" This includes the case where the process executes a set-user-ID
207 .\" program that confers the effective UID of the creator of the namespace.
208 By virtue of the previous rule,
209 this means that the process has all capabilities in all
210 further removed descendant user namespaces as well.
212 .\" ============================================================
214 .SS Interaction of user namespaces and other types of namespaces
215 Starting in Linux 3.8, unprivileged processes can create user namespaces,
216 and mount, PID, IPC, network, and UTS namespaces can be created with just the
218 capability in the caller's user namespace.
220 When a non-user-namespace is created,
221 it is owned by the user namespace in which the creating process
222 was a member at the time of the creation of the namespace.
223 Actions on the non-user-namespace
224 require capabilities in the corresponding user namespace.
228 is specified along with other
234 call, the user namespace is guaranteed to be created first,
239 privileges over the remaining namespaces created by the call.
240 Thus, it is possible for an unprivileged caller to specify this combination
243 When a new IPC, mount, network, PID, or UTS namespace is created via
247 the kernel records the user namespace of the creating process against
249 (This association can't be changed.)
250 When a process in the new namespace subsequently performs
251 privileged operations that operate on global
252 resources isolated by the namespace,
253 the permission checks are performed according to the process's capabilities
254 in the user namespace that the kernel associated with the new namespace.
256 .\" ============================================================
258 .SS Restrictions on mount namespaces
260 Note the following points with respect to mount namespaces:
262 A mount namespace has an owner user namespace.
263 A mount namespace whose owner user namespace is different from
264 the owner user namespace of its parent mount namespace is
265 considered a less privileged mount namespace.
267 When creating a less privileged mount namespace,
268 shared mounts are reduced to slave mounts.
269 This ensures that mappings performed in less
270 privileged mount namespaces will not propagate to more privileged
274 .\" What does "come as a single unit from more privileged mount" mean?
275 Mounts that come as a single unit from more privileged mount are
276 locked together and may not be separated in a less privileged mount
281 operation brings across all of the mounts from the original
282 mount namespace as a single unit,
283 and recursive mounts that propagate between
284 mount namespaces propagate as a single unit.)
292 and the "atime" flags
296 settings become locked
297 .\" commit 9566d6742852c527bf5af38af5cbb878dad75705
298 .\" Author: Eric W. Biederman <ebiederm@xmission.com>
299 .\" Date: Mon Jul 28 17:26:07 2014 -0700
301 .\" mnt: Correct permission checks in do_remount
303 when propagated from a more privileged to
304 a less privileged mount namespace,
305 and may not be changed in the less privileged mount namespace.
307 .\" (As of 3.18-rc1 (in Al Viro's 2014-08-30 vfs.git#for-next tree))
308 A file or directory that is a mount point in one namespace that is not
309 a mount point in another namespace, may be renamed, unlinked, or removed
311 in the mount namespace in which it is not a mount point
312 (subject to the usual permission checks).
314 Previously, attempting to unlink, rename, or remove a file or directory
315 that was a mount point in another mount namespace would result in the error
317 That behavior had technical problems of enforcement (e.g., for NFS)
318 and permitted denial-of-service attacks against more privileged users.
319 (i.e., preventing individual files from being updated
320 by bind mounting on top of them).
322 .\" ============================================================
324 .SS User and group ID mappings: uid_map and gid_map
325 When a user namespace is created,
326 it starts out without a mapping of user IDs (group IDs)
327 to the parent user namespace.
329 .IR /proc/[pid]/uid_map
331 .IR /proc/[pid]/gid_map
332 files (available since Linux 3.5)
333 .\" commit 22d917d80e842829d0ca0a561967d728eb1d6303
334 expose the mappings for user and group IDs
335 inside the user namespace for the process
337 These files can be read to view the mappings in a user namespace and
338 written to (once) to define the mappings.
340 The description in the following paragraphs explains the details for
344 but each instance of "user ID" is replaced by "group ID".
348 file exposes the mapping of user IDs from the user namespace
351 to the user namespace of the process that opened
353 (but see a qualification to this point below).
354 In other words, processes that are in different user namespaces
355 will potentially see different values when reading from a particular
357 file, depending on the user ID mappings for the user namespaces
358 of the reading processes.
362 file specifies a 1-to-1 mapping of a range of contiguous
363 user IDs between two user namespaces.
364 (When a user namespace is first created, this file is empty.)
365 The specification in each line takes the form of
366 three numbers delimited by white space.
367 The first two numbers specify the starting user ID in
368 each of the two user namespaces.
369 The third number specifies the length of the mapped range.
370 In detail, the fields are interpreted as follows:
372 The start of the range of user IDs in
373 the user namespace of the process
376 The start of the range of user
377 IDs to which the user IDs specified by field one map.
378 How field two is interpreted depends on whether the process that opened
382 are in the same user namespace, as follows:
385 If the two processes are in different user namespaces:
386 field two is the start of a range of
387 user IDs in the user namespace of the process that opened
390 If the two processes are in the same user namespace:
391 field two is the start of the range of
392 user IDs in the parent user namespace of the process
394 This case enables the opener of
396 (the common case here is opening
397 .IR /proc/self/uid_map )
398 to see the mapping of user IDs into the user namespace of the process
399 that created this user namespace.
402 The length of the range of user IDs that is mapped between the two
405 System calls that return user IDs (group IDs)\(emfor example,
408 and the credential fields in the structure returned by
409 .BR stat (2)\(emreturn
410 the user ID (group ID) mapped into the caller's user namespace.
412 When a process accesses a file, its user and group IDs
413 are mapped into the initial user namespace for the purpose of permission
414 checking and assigning IDs when creating a file.
415 When a process retrieves file user and group IDs via
417 the IDs are mapped in the opposite direction,
418 to produce values relative to the process user and group ID mappings.
420 The initial user namespace has no parent namespace,
421 but, for consistency, the kernel provides dummy user and group
422 ID mapping files for this namespace.
427 is the same) from a shell in the initial namespace shows:
431 $ \fBcat /proc/$$/uid_map\fP
436 This mapping tells us
437 that the range starting at user ID 0 in this namespace
438 maps to a range starting at 0 in the (nonexistent) parent namespace,
439 and the length of the range is the largest 32-bit unsigned integer.
440 This leaves 4294967295 (the 32-bit signed \-1 value) unmapped.
443 is used in several interfaces (e.g.,
445 as a way to specify "no user ID".
448 unmapped and unusable guarantees that there will be no
449 confusion when using these interfaces.
451 .\" ============================================================
453 .SS Defining user and group ID mappings: writing to uid_map and gid_map
455 After the creation of a new user namespace, the
459 of the processes in the namespace may be written to
461 to define the mapping of user IDs in the new user namespace.
462 An attempt to write more than once to a
464 file in a user namespace fails with the error
466 Similar rules apply for
473 must conform to the following rules:
475 The three fields must be valid numbers,
476 and the last field must be greater than 0.
478 Lines are terminated by newline characters.
480 There is an (arbitrary) limit on the number of lines in the file.
481 As at Linux 3.18, the limit is five lines.
482 In addition, the number of bytes written to
483 the file must be less than the system page size,
484 .\" FIXME(Eric): the restriction "less than" rather than "less than or equal"
485 .\" seems strangely arbitrary. Furthermore, the comment does not agree
486 .\" with the code in kernel/user_namespace.c. Which is correct?
487 and the write must be performed at the start of the file (i.e.,
491 can't be used to write to nonzero offsets in the file).
493 The range of user IDs (group IDs)
494 specified in each line cannot overlap with the ranges
496 In the initial implementation (Linux 3.8), this requirement was
497 satisfied by a simplistic implementation that imposed the further
499 the values in both field 1 and field 2 of successive lines must be
500 in ascending numerical order,
501 which prevented some otherwise valid maps from being created.
503 .\" commit 0bd14b4fd72afd5df41e9fd59f356740f22fceba
504 fix this limitation, allowing any valid set of nonoverlapping maps.
506 At least one line must be written to the file.
508 Writes that violate the above rules fail with the error
511 In order for a process to write to the
512 .I /proc/[pid]/uid_map
513 .RI ( /proc/[pid]/gid_map )
514 file, all of the following requirements must be met:
516 The writing process must have the
519 capability in the user namespace of the process
522 The writing process must either be in the user namespace of the process
524 or be in the parent user namespace of the process
527 The mapped user IDs (group IDs) must in turn have a mapping
528 in the parent user namespace.
530 One of the following two cases applies:
534 the writing process has the
542 No further restrictions apply:
543 the process can make mappings to arbitrary user IDs (group IDs)
544 in the parent user namespace.
548 otherwise all of the following restrictions apply:
554 must consist of a single line that maps
555 the writing process's effective user ID
556 (group ID) in the parent user namespace to a user ID (group ID)
557 in the user namespace.
559 The writing process must have the same effective user ID as the process
560 that created the user namespace.
566 system call must first be denied by writing
569 .I /proc/[pid]/setgroups
570 file (see below) before writing to
575 Writes that violate the above rules fail with the error
578 .\" ============================================================
580 .SS Interaction with system calls that change process UIDs or GIDs
581 In a user namespace where the
583 file has not been written, the system calls that change user IDs will fail.
586 file has not been written, the system calls that change group IDs will fail.
591 files have been written, only the mapped values may be used in
592 system calls that change user and group IDs.
594 For user IDs, the relevant system calls include
600 For group IDs, the relevant system calls include
611 .I /proc/[pid]/setgroups
612 file before writing to
613 .I /proc/[pid]/gid_map
614 .\" Things changed in Linux 3.19
615 .\" commit 9cc46516ddf497ea16e8d7cb986ae03a0f6b92f8
616 .\" commit 66d2f338ee4c449396b6f99f5e75cd18eb6df272
617 .\" http://lwn.net/Articles/626665/
618 will permanently disable
620 in a user namespace and allow writing to
621 .I /proc/[pid]/gid_map
624 capability in the parent user namespace.
626 .\" ============================================================
628 .SS The /proc/[pid]/setgroups file
630 .\" commit 9cc46516ddf497ea16e8d7cb986ae03a0f6b92f8
631 .\" commit 66d2f338ee4c449396b6f99f5e75cd18eb6df272
632 .\" http://lwn.net/Articles/626665/
633 .\" http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-8989
636 .I /proc/[pid]/setgroups
637 file displays the string
639 if processes in the user namespace that contains the process
641 are permitted to employ the
643 system call; it displays
647 is not permitted in that user namespace.
649 A privileged process (one with the
651 capability in the namespace) may write either of the strings
657 writing a group ID mapping
658 for this user namespace to the file
659 .IR /proc/[pid]/gid_map .
662 prevents any process in the user namespace from employing
664 Note that regardless of the value in the
665 .I /proc/[pid]/setgroups
668 are also not permitted if
669 .IR /proc/[pid]/gid_map
670 has not yet been set.
672 The essence of the restrictions described in the preceding
673 paragraph is that it is permitted to write to
674 .I /proc/[pid]/setgroups
675 only so long as calling
677 is disallowed because
678 .I /proc/[pid]gid_map
680 This ensures that a process cannot transition from a state where
682 is allowed to a state where
685 a process can only transition from
691 The default value of this file in the initial user namespace is
695 .IR /proc/[pid]/gid_map
697 (which has the effect of enabling
699 in the user namespace),
700 it is no longer possible to deny
703 .IR /proc/[pid]/setgroups .
705 A child user namespace inherits the
706 .IR /proc/[pid]/gid_map
707 setting from its parent.
715 system call can't subsequently be reenabled (by writing
717 to the file) in this user namespace.
718 (Attempts to do so will fail with the error
720 This restriction also propagates down to all child user namespaces of
724 .I /proc/[pid]/setgroups
725 file was added in Linux 3.19,
726 but was backported to many earlier stable kernel series,
727 because it addresses a security issue.
728 The issue concerned files with permissions such as "rwx\-\-\-rwx".
729 Such files give fewer permissions to "group" than they do to "other".
730 This means that dropping groups using
732 might allow a process file access that it did not formerly have.
733 Before the existence of user namespaces this was not a concern,
734 since only a privileged process (one with the
736 capability) could call
738 However, with the introduction of user namespaces,
739 it became possible for an unprivileged process to create
740 a new namespace in which the user had all privileges.
741 This then allowed formerly unprivileged
742 users to drop groups and thus gain file access
743 that they did not previously have.
745 .I /proc/[pid]/setgroups
746 file was added to address this security issue,
747 by denying any pathway for an unprivileged process to drop groups with
750 .\" /proc/PID/setgroups
751 .\" [allow == setgroups() is allowed, "deny" == setgroups() is disallowed]
752 .\" * Can write if have CAP_SYS_ADMIN in NS
753 .\" * Must write BEFORE writing to /proc/PID/gid_map
756 .\" * Must already have written to gid_maps
757 .\" * /proc/PID/setgroups must be "allow"
759 .\" /proc/PID/gid_map -- writing
760 .\" * Must already have written "deny" to /proc/PID/setgroups
762 .\" ============================================================
764 .SS Unmapped user and group IDs
766 There are various places where an unmapped user ID (group ID)
767 may be exposed to user space.
768 For example, the first process in a new user namespace may call
770 before a user ID mapping has been defined for the namespace.
771 In most such cases, an unmapped user ID is converted
772 .\" from_kuid_munged(), from_kgid_munged()
773 to the overflow user ID (group ID);
774 the default value for the overflow user ID (group ID) is 65534.
775 See the descriptions of
776 .IR /proc/sys/kernel/overflowuid
778 .IR /proc/sys/kernel/overflowgid
782 The cases where unmapped IDs are mapped in this fashion include
783 system calls that return user IDs
787 credentials passed over a UNIX domain socket,
789 credentials returned by
792 and the System V IPC "ctl"
795 credentials exposed by
798 .IR /proc/sysvipc/* ,
799 credentials returned via the
803 received with a signal (see
805 credentials written to the process accounting file (see
807 and credentials returned with POSIX message queue notifications (see
810 There is one notable case where unmapped user and group IDs are
812 .\" from_kuid(), from_kgid()
813 .\" Also F_GETOWNER_UIDS is an exception
814 converted to the corresponding overflow ID value.
819 file in which there is no mapping for the second field,
820 that field is displayed as 4294967295 (\-1 as an unsigned integer);
822 .\" ============================================================
824 .SS Set-user-ID and set-group-ID programs
826 When a process inside a user namespace executes
827 a set-user-ID (set-group-ID) program,
828 the process's effective user (group) ID inside the namespace is changed
829 to whatever value is mapped for the user (group) ID of the file.
830 However, if either the user
832 the group ID of the file has no mapping inside the namespace,
833 the set-user-ID (set-group-ID) bit is silently ignored:
834 the new program is executed,
835 but the process's effective user (group) ID is left unchanged.
836 (This mirrors the semantics of executing a set-user-ID or set-group-ID
837 program that resides on a filesystem that was mounted with the
839 flag, as described in
842 .\" ============================================================
846 When a process's user and group IDs are passed over a UNIX domain socket
847 to a process in a different user namespace (see the description of
851 they are translated into the corresponding values as per the
852 receiving process's user and group ID mappings.
855 Namespaces are a Linux-specific feature.
858 Over the years, there have been a lot of features that have been added
859 to the Linux kernel that have been made available only to privileged users
860 because of their potential to confuse set-user-ID-root applications.
861 In general, it becomes safe to allow the root user in a user namespace to
862 use those features because it is impossible, while in a user namespace,
863 to gain more privilege than the root user of a user namespace has.
865 .\" ============================================================
868 Use of user namespaces requires a kernel that is configured with the
871 User namespaces require support in a range of subsystems across
873 When an unsupported subsystem is configured into the kernel,
874 it is not possible to configure user namespaces support.
876 As at Linux 3.8, most relevant subsystems supported user namespaces,
877 but a number of filesystems did not have the infrastructure needed
878 to map user and group IDs between user namespaces.
879 Linux 3.9 added the required infrastructure support for many of
880 the remaining unsupported filesystems
881 (Plan 9 (9P), Andrew File System (AFS), Ceph, CIFS, CODA, NFS, and OCFS2).
882 Linux 3.11 added support the last of the unsupported major filesystems,
883 .\" commit d6970d4b726cea6d7a9bc4120814f95c09571fc3
887 The program below is designed to allow experimenting with
888 user namespaces, as well as other types of namespaces.
889 It creates namespaces as specified by command-line options and then executes
890 a command inside those namespaces.
893 function inside the program provide a full explanation of the program.
894 The following shell session demonstrates its use.
896 First, we look at the run-time environment:
900 $ \fBuname -rs\fP # Need Linux 3.8 or later
902 $ \fBid -u\fP # Running as unprivileged user
909 Now start a new shell in new user
915 namespaces, with user ID
919 1000 mapped to 0 inside the user namespace:
923 $ \fB./userns_child_exec -p -m -U -M '0 1000 1' -G '0 1000 1' bash\fP
927 The shell has PID 1, because it is the first process in the new
937 Inside the user namespace, the shell has user and group ID 0,
938 and a full set of permitted and effective capabilities:
942 bash$ \fBcat /proc/$$/status | egrep '^[UG]id'\fP
945 bash$ \fBcat /proc/$$/status | egrep '^Cap(Prm|Inh|Eff)'\fP
946 CapInh: 0000000000000000
947 CapPrm: 0000001fffffffff
948 CapEff: 0000001fffffffff
954 filesystem and listing all of the processes visible
955 in the new PID namespace shows that the shell can't see
956 any processes outside the PID namespace:
960 bash$ \fBmount -t proc proc /proc\fP
962 PID TTY STAT TIME COMMAND
964 22 pts/3 R+ 0:00 ps ax
970 /* userns_child_exec.c
972 Licensed under GNU General Public License v2 or later
974 Create a child process that executes a shell command in new
975 namespace(s); allow UID and GID mappings to be specified when
976 creating a user namespace.
982 #include <sys/wait.h>
990 /* A simple error\-handling function: print an error message based
991 on the value in \(aqerrno\(aq and terminate the calling process */
993 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \\
997 char **argv; /* Command to be executed by child, with args */
998 int pipe_fd[2]; /* Pipe used to synchronize parent and child */
1006 fprintf(stderr, "Usage: %s [options] cmd [arg...]\\n\\n", pname);
1007 fprintf(stderr, "Create a child process that executes a shell "
1008 "command in a new user namespace,\\n"
1009 "and possibly also other new namespace(s).\\n\\n");
1010 fprintf(stderr, "Options can be:\\n\\n");
1011 #define fpe(str) fprintf(stderr, " %s", str);
1012 fpe("\-i New IPC namespace\\n");
1013 fpe("\-m New mount namespace\\n");
1014 fpe("\-n New network namespace\\n");
1015 fpe("\-p New PID namespace\\n");
1016 fpe("\-u New UTS namespace\\n");
1017 fpe("\-U New user namespace\\n");
1018 fpe("\-M uid_map Specify UID map for user namespace\\n");
1019 fpe("\-G gid_map Specify GID map for user namespace\\n");
1020 fpe("\-z Map user\(aqs UID and GID to 0 in user namespace\\n");
1021 fpe(" (equivalent to: \-M \(aq0 <uid> 1\(aq \-G \(aq0 <gid> 1\(aq)\\n");
1022 fpe("\-v Display verbose messages\\n");
1024 fpe("If \-z, \-M, or \-G is specified, \-U is required.\\n");
1025 fpe("It is not permitted to specify both \-z and either \-M or \-G.\\n");
1027 fpe("Map strings for \-M and \-G consist of records of the form:\\n");
1029 fpe(" ID\-inside\-ns ID\-outside\-ns len\\n");
1031 fpe("A map string can contain multiple records, separated"
1033 fpe("the commas are replaced by newlines before writing"
1034 " to map files.\\n");
1039 /* Update the mapping file \(aqmap_file\(aq, with the value provided in
1040 \(aqmapping\(aq, a string that defines a UID or GID mapping. A UID or
1041 GID mapping consists of one or more newline\-delimited records
1044 ID_inside\-ns ID\-outside\-ns length
1046 Requiring the user to supply a string that contains newlines is
1047 of course inconvenient for command\-line use. Thus, we permit the
1048 use of commas to delimit records in this string, and replace them
1049 with newlines before writing the string to the file. */
1052 update_map(char *mapping, char *map_file)
1055 size_t map_len; /* Length of \(aqmapping\(aq */
1057 /* Replace commas in mapping string with newlines */
1059 map_len = strlen(mapping);
1060 for (j = 0; j < map_len; j++)
1061 if (mapping[j] == \(aq,\(aq)
1062 mapping[j] = \(aq\\n\(aq;
1064 fd = open(map_file, O_RDWR);
1066 fprintf(stderr, "ERROR: open %s: %s\\n", map_file,
1071 if (write(fd, mapping, map_len) != map_len) {
1072 fprintf(stderr, "ERROR: write %s: %s\\n", map_file,
1080 /* Linux 3.19 made a change in the handling of setgroups(2) and the
1081 \(aqgid_map\(aq file to address a security issue. The issue allowed
1082 *unprivileged* users to employ user namespaces in order to drop
1083 The upshot of the 3.19 changes is that in order to update the
1084 \(aqgid_maps\(aq file, use of the setgroups() system call in this
1085 user namespace must first be disabled by writing "deny" to one of
1086 the /proc/PID/setgroups files for this namespace. That is the
1087 purpose of the following function. */
1090 proc_setgroups_write(pid_t child_pid, char *str)
1092 char setgroups_path[PATH_MAX];
1095 snprintf(setgroups_path, PATH_MAX, "/proc/%ld/setgroups",
1098 fd = open(setgroups_path, O_RDWR);
1101 /* We may be on a system that doesn\(aqt support
1102 /proc/PID/setgroups. In that case, the file won\(aqt exist,
1103 and the system won\(aqt impose the restrictions that Linux 3.19
1104 added. That\(aqs fine: we don\(aqt need to do anything in order
1105 to permit \(aqgid_map\(aq to be updated.
1107 However, if the error from open() was something other than
1108 the ENOENT error that is expected for that case, let the
1111 if (errno != ENOENT)
1112 fprintf(stderr, "ERROR: open %s: %s\\n", setgroups_path,
1117 if (write(fd, str, strlen(str)) == \-1)
1118 fprintf(stderr, "ERROR: write %s: %s\\n", setgroups_path,
1124 static int /* Start function for cloned child */
1125 childFunc(void *arg)
1127 struct child_args *args = (struct child_args *) arg;
1130 /* Wait until the parent has updated the UID and GID mappings.
1131 See the comment in main(). We wait for end of file on a
1132 pipe that will be closed by the parent process once it has
1133 updated the mappings. */
1135 close(args\->pipe_fd[1]); /* Close our descriptor for the write
1136 end of the pipe so that we see EOF
1137 when parent closes its descriptor */
1138 if (read(args\->pipe_fd[0], &ch, 1) != 0) {
1140 "Failure in child: read from pipe returned != 0\\n");
1144 /* Execute a shell command */
1146 printf("About to exec %s\\n", args\->argv[0]);
1147 execvp(args\->argv[0], args\->argv);
1151 #define STACK_SIZE (1024 * 1024)
1153 static char child_stack[STACK_SIZE]; /* Space for child\(aqs stack */
1156 main(int argc, char *argv[])
1158 int flags, opt, map_zero;
1160 struct child_args args;
1161 char *uid_map, *gid_map;
1162 const int MAP_BUF_SIZE = 100;
1163 char map_buf[MAP_BUF_SIZE];
1164 char map_path[PATH_MAX];
1166 /* Parse command\-line options. The initial \(aq+\(aq character in
1167 the final getopt() argument prevents GNU\-style permutation
1168 of command\-line options. That\(aqs useful, since sometimes
1169 the \(aqcommand\(aq to be executed by this program itself
1170 has command\-line options. We don\(aqt want getopt() to treat
1171 those as options to this program. */
1178 while ((opt = getopt(argc, argv, "+imnpuUM:G:zv")) != \-1) {
1180 case \(aqi\(aq: flags |= CLONE_NEWIPC; break;
1181 case \(aqm\(aq: flags |= CLONE_NEWNS; break;
1182 case \(aqn\(aq: flags |= CLONE_NEWNET; break;
1183 case \(aqp\(aq: flags |= CLONE_NEWPID; break;
1184 case \(aqu\(aq: flags |= CLONE_NEWUTS; break;
1185 case \(aqv\(aq: verbose = 1; break;
1186 case \(aqz\(aq: map_zero = 1; break;
1187 case \(aqM\(aq: uid_map = optarg; break;
1188 case \(aqG\(aq: gid_map = optarg; break;
1189 case \(aqU\(aq: flags |= CLONE_NEWUSER; break;
1190 default: usage(argv[0]);
1194 /* \-M or \-G without \-U is nonsensical */
1196 if (((uid_map != NULL || gid_map != NULL || map_zero) &&
1197 !(flags & CLONE_NEWUSER)) ||
1198 (map_zero && (uid_map != NULL || gid_map != NULL)))
1201 args.argv = &argv[optind];
1203 /* We use a pipe to synchronize the parent and child, in order to
1204 ensure that the parent sets the UID and GID maps before the child
1205 calls execve(). This ensures that the child maintains its
1206 capabilities during the execve() in the common case where we
1207 want to map the child\(aqs effective user ID to 0 in the new user
1208 namespace. Without this synchronization, the child would lose
1209 its capabilities if it performed an execve() with nonzero
1210 user IDs (see the capabilities(7) man page for details of the
1211 transformation of a process\(aqs capabilities during execve()). */
1213 if (pipe(args.pipe_fd) == \-1)
1216 /* Create the child in new namespace(s) */
1218 child_pid = clone(childFunc, child_stack + STACK_SIZE,
1219 flags | SIGCHLD, &args);
1220 if (child_pid == \-1)
1223 /* Parent falls through to here */
1226 printf("%s: PID of child created by clone() is %ld\\n",
1227 argv[0], (long) child_pid);
1229 /* Update the UID and GID maps in the child */
1231 if (uid_map != NULL || map_zero) {
1232 snprintf(map_path, PATH_MAX, "/proc/%ld/uid_map",
1235 snprintf(map_buf, MAP_BUF_SIZE, "0 %ld 1", (long) getuid());
1238 update_map(uid_map, map_path);
1241 if (gid_map != NULL || map_zero) {
1242 proc_setgroups_write(child_pid, "deny");
1244 snprintf(map_path, PATH_MAX, "/proc/%ld/gid_map",
1247 snprintf(map_buf, MAP_BUF_SIZE, "0 %ld 1", (long) getgid());
1250 update_map(gid_map, map_path);
1253 /* Close the write end of the pipe, to signal to the child that we
1254 have updated the UID and GID maps */
1256 close(args.pipe_fd[1]);
1258 if (waitpid(child_pid, NULL, 0) == \-1) /* Wait for child */
1262 printf("%s: terminating\\n", argv[0]);
1268 .BR newgidmap (1), \" From the shadow package
1269 .BR newuidmap (1), \" From the shadow package
1274 .BR subgid (5), \" From the shadow package
1275 .BR subuid (5), \" From the shadow package
1276 .BR credentials (7),
1277 .BR capabilities (7),
1279 .BR pid_namespaces (7)
1281 The kernel source file
1282 .IR Documentation/namespaces/resource-control.txt .