1CAPABILITIES(7) Linux Programmer's Manual CAPABILITIES(7)
2
6 capabilities - overview of Linux capabilities
7
9 For the purpose of performing permission checks, traditional Unix
10 implementations distinguish two categories of processes: privileged
11 processes (whose effective user ID is 0, referred to as superuser or
12 root), and unprivileged processes (whose effective UID is nonzero).
13 Privileged processes bypass all kernel permission checks, while unpriv‐
14 ileged processes are subject to full permission checking based on the
15 process's credentials (usually: effective UID, effective GID, and sup‐
16 plementary group list).
17 Starting with kernel 2.2, Linux divides the privileges traditionally
19 associated with superuser into distinct units, known as capabilities,
20 which can be independently enabled and disabled. Capabilities are a
21 per-thread attribute.
22 Capabilities List
24 The following list shows the capabilities implemented on Linux, and the
25 operations or behaviors that each capability permits:
26 CAP_AUDIT_CONTROL (since Linux 2.6.11)
28 Enable and disable kernel auditing; change auditing filter
29 rules; retrieve auditing status and filtering rules.
30 CAP_AUDIT_WRITE (since Linux 2.6.11)
32 Write records to kernel auditing log.
33 CAP_CHOWN
35 Make arbitrary changes to file UIDs and GIDs (see chown(2)).
36 CAP_DAC_OVERRIDE
38 Bypass file read, write, and execute permission checks. (DAC is
39 an abbreviation of "discretionary access control".)
40 CAP_DAC_READ_SEARCH
42 Bypass file read permission checks and directory read and exe‐
43 cute permission checks.
44 CAP_FOWNER
46 * Bypass permission checks on operations that normally require
47 the file system UID of the process to match the UID of the
48 file (e.g., chmod(2), utime(2)), excluding those operations
49 covered by CAP_DAC_OVERRIDE and CAP_DAC_READ_SEARCH;
50 * set extended file attributes (see chattr(1)) on arbitrary
51 files;
52 * set Access Control Lists (ACLs) on arbitrary files;
53 * ignore directory sticky bit on file deletion;
54 * specify O_NOATIME for arbitrary files in open(2) and fcntl(2).
55 CAP_FSETID
57 Don't clear set-user-ID and set-group-ID permission bits when a
58 file is modified; set the set-group-ID bit for a file whose GID
59 does not match the file system or any of the supplementary GIDs
60 of the calling process.
61 CAP_IPC_LOCK
63 Lock memory (mlock(2), mlockall(2), mmap(2), shmctl(2)).
64 CAP_IPC_OWNER
66 Bypass permission checks for operations on System V IPC objects.
67 CAP_KILL
69 Bypass permission checks for sending signals (see kill(2)).
70 This includes use of the ioctl(2) KDSIGACCEPT operation.
71 CAP_LEASE (since Linux 2.4)
73 Establish leases on arbitrary files (see fcntl(2)).
74 CAP_LINUX_IMMUTABLE
76 Set the FS_APPEND_FL and FS_IMMUTABLE_FL i-node flags (see
77 chattr(1)).
78 CAP_MAC_ADMIN (since Linux 2.6.25)
80 Override Mandatory Access Control (MAC). Implemented for the
81 Smack Linux Security Module (LSM).
82 CAP_MAC_OVERRIDE (since Linux 2.6.25)
84 Allow MAC configuration or state changes. Implemented for the
85 Smack LSM.
86 CAP_MKNOD (since Linux 2.4)
88 Create special files using mknod(2).
89 CAP_NET_ADMIN
91 Perform various network-related operations (e.g., setting privi‐
92 leged socket options, enabling multicasting, interface configu‐
93 ration, modifying routing tables).
94 CAP_NET_BIND_SERVICE
96 Bind a socket to Internet domain privileged ports (port numbers
97 less than 1024).
98 CAP_NET_BROADCAST
100 (Unused) Make socket broadcasts, and listen to multicasts.
101 CAP_NET_RAW
103 Use RAW and PACKET sockets.
104 CAP_SETGID
106 Make arbitrary manipulations of process GIDs and supplementary
107 GID list; forge GID when passing socket credentials via Unix
108 domain sockets.
109 CAP_SETFCAP (since Linux 2.6.24)
111 Set file capabilities.
112 CAP_SETPCAP
114 If file capabilities are not supported: grant or remove any
115 capability in the caller's permitted capability set to or from
116 any other process. (This property of CAP_SETPCAP is not avail‐
117 able when the kernel is configured to support file capabilities,
118 since CAP_SETPCAP has entirely different semantics for such ker‐
119 nels.)
120 If file capabilities are supported: add any capability from the
122 calling thread's bounding set to its inheritable set; drop capa‐
123 bilities from the bounding set (via prctl(2) PR_CAPBSET_DROP);
124 make changes to the securebits flags.
125 CAP_SETUID
127 Make arbitrary manipulations of process UIDs (setuid(2),
128 setreuid(2), setresuid(2), setfsuid(2)); make forged UID when
129 passing socket credentials via Unix domain sockets.
130 CAP_SYS_ADMIN
132 * Perform a range of system administration operations including:
133 quotactl(2), mount(2), umount(2), swapon(2), swapoff(2),
134 sethostname(2), and setdomainname(2);
135 * perform IPC_SET and IPC_RMID operations on arbitrary System V
136 IPC objects;
137 * perform operations on trusted and security Extended Attributes
138 (see attr(5));
139 * use lookup_dcookie(2);
140 * use ioprio_set(2) to assign IOPRIO_CLASS_RT and (before Linux
141 2.6.25) IOPRIO_CLASS_IDLE I/O scheduling classes;
142 * forge UID when passing socket credentials;
143 * exceed /proc/sys/fs/file-max, the system-wide limit on the
144 number of open files, in system calls that open files (e.g.,
145 accept(2), execve(2), open(2), pipe(2));
146 * employ CLONE_NEWNS flag with clone(2) and unshare(2);
147 * perform KEYCTL_CHOWN and KEYCTL_SETPERM keyctl(2) operations;
148 * perform madvise(2) MADV_HWPOISON operation.
149 CAP_SYS_BOOT
151 Use reboot(2) and kexec_load(2).
152 CAP_SYS_CHROOT
154 Use chroot(2).
155 CAP_SYS_MODULE
157 Load and unload kernel modules (see init_module(2) and
158 delete_module(2)); in kernels before 2.6.25: drop capabilities
159 from the system-wide capability bounding set.
160 CAP_SYS_NICE
162 * Raise process nice value (nice(2), setpriority(2)) and change
163 the nice value for arbitrary processes;
164 * set real-time scheduling policies for calling process, and set
165 scheduling policies and priorities for arbitrary processes
166 (sched_setscheduler(2), sched_setparam(2));
167 * set CPU affinity for arbitrary processes (sched_setaffin‐
168 ity(2));
169 * set I/O scheduling class and priority for arbitrary processes
170 (ioprio_set(2));
171 * apply migrate_pages(2) to arbitrary processes and allow pro‐
172 cesses to be migrated to arbitrary nodes;
173 * apply move_pages(2) to arbitrary processes;
174 * use the MPOL_MF_MOVE_ALL flag with mbind(2) and move_pages(2).
175 CAP_SYS_PACCT
177 Use acct(2).
178 CAP_SYS_PTRACE
180 Trace arbitrary processes using ptrace(2); apply
181 get_robust_list(2) to arbitrary processes.
182 CAP_SYS_RAWIO
184 Perform I/O port operations (iopl(2) and ioperm(2)); access
185 /proc/kcore.
186 CAP_SYS_RESOURCE
188 * Use reserved space on ext2 file systems;
189 * make ioctl(2) calls controlling ext3 journaling;
190 * override disk quota limits;
191 * increase resource limits (see setrlimit(2));
192 * override RLIMIT_NPROC resource limit;
193 * raise msg_qbytes limit for a System V message queue above the
194 limit in /proc/sys/kernel/msgmnb (see msgop(2) and msgctl(2)).
195 * use F_SETPIPE_SZ to increase the capacity of a pipe above the
196 limit specified by /proc/sys/fs/pipe-max-size.
197 CAP_SYS_TIME
199 Set system clock (settimeofday(2), stime(2), adjtimex(2)); set
200 real-time (hardware) clock.
201 CAP_SYS_TTY_CONFIG
203 Use vhangup(2).
204 Past and Current Implementation
206 A full implementation of capabilities requires that:
207 1. For all privileged operations, the kernel must check whether the
209 thread has the required capability in its effective set.
210 2. The kernel must provide system calls allowing a thread's capability
212 sets to be changed and retrieved.
213 3. The file system must support attaching capabilities to an executable
215 file, so that a process gains those capabilities when the file is
216 executed.
217 Before kernel 2.6.24, only the first two of these requirements are met;
219 since kernel 2.6.24, all three requirements are met.
220 Thread Capability Sets
222 Each thread has three capability sets containing zero or more of the
223 above capabilities:
224 Permitted:
226 This is a limiting superset for the effective capabilities that
227 the thread may assume. It is also a limiting superset for the
228 capabilities that may be added to the inheritable set by a
229 thread that does not have the CAP_SETPCAP capability in its
230 effective set.
231 If a thread drops a capability from its permitted set, it can
233 never reacquire that capability (unless it execve(2)s either a
234 set-user-ID-root program, or a program whose associated file
235 capabilities grant that capability).
236 Inheritable:
238 This is a set of capabilities preserved across an execve(2). It
239 provides a mechanism for a process to assign capabilities to the
240 permitted set of the new program during an execve(2).
241 Effective:
243 This is the set of capabilities used by the kernel to perform
244 permission checks for the thread.
245 A child created via fork(2) inherits copies of its parent's capability
247 sets. See below for a discussion of the treatment of capabilities dur‐
248 ing execve(2).
249 Using capset(2), a thread may manipulate its own capability sets (see
251 below).
252 File Capabilities
254 Since kernel 2.6.24, the kernel supports associating capability sets
255 with an executable file using setcap(8). The file capability sets are
256 stored in an extended attribute (see setxattr(2)) named security.capa‐
257 bility. Writing to this extended attribute requires the CAP_SETFCAP
258 capability. The file capability sets, in conjunction with the capabil‐
259 ity sets of the thread, determine the capabilities of a thread after an
260 execve(2).
261 The three file capability sets are:
263 Permitted (formerly known as forced):
265 These capabilities are automatically permitted to the thread,
266 regardless of the thread's inheritable capabilities.
267 Inheritable (formerly known as allowed):
269 This set is ANDed with the thread's inheritable set to determine
270 which inheritable capabilities are enabled in the permitted set
271 of the thread after the execve(2).
272 Effective:
274 This is not a set, but rather just a single bit. If this bit is
275 set, then during an execve(2) all of the new permitted capabili‐
276 ties for the thread are also raised in the effective set. If
277 this bit is not set, then after an execve(2), none of the new
278 permitted capabilities is in the new effective set.
279 Enabling the file effective capability bit implies that any file
281 permitted or inheritable capability that causes a thread to
282 acquire the corresponding permitted capability during an
283 execve(2) (see the transformation rules described below) will
284 also acquire that capability in its effective set. Therefore,
285 when assigning capabilities to a file (setcap(8),
286 cap_set_file(3), cap_set_fd(3)), if we specify the effective
287 flag as being enabled for any capability, then the effective
288 flag must also be specified as enabled for all other capabili‐
289 ties for which the corresponding permitted or inheritable flags
290 is enabled.
291 Transformation of Capabilities During execve()
293 During an execve(2), the kernel calculates the new capabilities of the
294 process using the following algorithm:
295 P'(permitted) = (P(inheritable) & F(inheritable)) |
297 (F(permitted) & cap_bset)
298 P'(effective) = F(effective) ? P'(permitted) : 0
300 P'(inheritable) = P(inheritable) [i.e., unchanged]
302 where:
304 P denotes the value of a thread capability set before the
306 execve(2)
307 P' denotes the value of a capability set after the execve(2)
309 F denotes a file capability set
311 cap_bset is the value of the capability bounding set (described
313 below).
314 Capabilities and execution of programs by root
316 In order to provide an all-powerful root using capability sets, during
317 an execve(2):
318 1. If a set-user-ID-root program is being executed, or the real user ID
320 of the process is 0 (root) then the file inheritable and permitted
321 sets are defined to be all ones (i.e., all capabilities enabled).
322 2. If a set-user-ID-root program is being executed, then the file
324 effective bit is defined to be one (enabled).
325 The upshot of the above rules, combined with the capabilities transfor‐
327 mations described above, is that when a process execve(2)s a set-user-
328 ID-root program, or when a process with an effective UID of 0
329 execve(2)s a program, it gains all capabilities in its permitted and
330 effective capability sets, except those masked out by the capability
331 bounding set. This provides semantics that are the same as those pro‐
332 vided by traditional Unix systems.
333 Capability bounding set
335 The capability bounding set is a security mechanism that can be used to
336 limit the capabilities that can be gained during an execve(2). The
337 bounding set is used in the following ways:
338 * During an execve(2), the capability bounding set is ANDed with the
340 file permitted capability set, and the result of this operation is
341 assigned to the thread's permitted capability set. The capability
342 bounding set thus places a limit on the permitted capabilities that
343 may be granted by an executable file.
344 * (Since Linux 2.6.25) The capability bounding set acts as a limiting
346 superset for the capabilities that a thread can add to its inherita‐
347 ble set using capset(2). This means that if a capability is not in
348 the bounding set, then a thread can't add this capability to its
349 inheritable set, even if it was in its permitted capabilities, and
350 thereby cannot have this capability preserved in its permitted set
351 when it execve(2)s a file that has the capability in its inheritable
352 set.
353 Note that the bounding set masks the file permitted capabilities, but
355 not the inherited capabilities. If a thread maintains a capability in
356 its inherited set that is not in its bounding set, then it can still
357 gain that capability in its permitted set by executing a file that has
358 the capability in its inherited set.
359 Depending on the kernel version, the capability bounding set is either
361 a system-wide attribute, or a per-process attribute.
362 Capability bounding set prior to Linux 2.6.25
364 In kernels before 2.6.25, the capability bounding set is a system-wide
366 attribute that affects all threads on the system. The bounding set is
367 accessible via the file /proc/sys/kernel/cap-bound. (Confusingly, this
368 bit mask parameter is expressed as a signed decimal number in
369 /proc/sys/kernel/cap-bound.)
370 Only the init process may set capabilities in the capability bounding
372 set; other than that, the superuser (more precisely: programs with the
373 CAP_SYS_MODULE capability) may only clear capabilities from this set.
374 On a standard system the capability bounding set always masks out the
376 CAP_SETPCAP capability. To remove this restriction (dangerous!), mod‐
377 ify the definition of CAP_INIT_EFF_SET in include/linux/capability.h
378 and rebuild the kernel.
379 The system-wide capability bounding set feature was added to Linux
381 starting with kernel version 2.2.11.
382 Capability bounding set from Linux 2.6.25 onwards
384 From Linux 2.6.25, the capability bounding set is a per-thread
386 attribute. (There is no longer a system-wide capability bounding set.)
387 The bounding set is inherited at fork(2) from the thread's parent, and
389 is preserved across an execve(2).
390 A thread may remove capabilities from its capability bounding set using
392 the prctl(2) PR_CAPBSET_DROP operation, provided it has the CAP_SETPCAP
393 capability. Once a capability has been dropped from the bounding set,
394 it cannot be restored to that set. A thread can determine if a capa‐
395 bility is in its bounding set using the prctl(2) PR_CAPBSET_READ opera‐
396 tion.
397 Removing capabilities from the bounding set is only supported if file
399 capabilities are compiled into the kernel (CONFIG_SECURITY_FILE_CAPA‐
400 BILITIES). In that case, the init process (the ancestor of all pro‐
401 cesses) begins with a full bounding set. If file capabilities are not
402 compiled into the kernel, then init begins with a full bounding set
403 minus CAP_SETPCAP, because this capability has a different meaning when
404 there are no file capabilities.
405 Removing a capability from the bounding set does not remove it from the
407 thread's inherited set. However it does prevent the capability from
408 being added back into the thread's inherited set in the future.
409 Effect of User ID Changes on Capabilities
411 To preserve the traditional semantics for transitions between 0 and
412 nonzero user IDs, the kernel makes the following changes to a thread's
413 capability sets on changes to the thread's real, effective, saved set,
414 and file system user IDs (using setuid(2), setresuid(2), or similar):
415 1. If one or more of the real, effective or saved set user IDs was pre‐
417 viously 0, and as a result of the UID changes all of these IDs have
418 a nonzero value, then all capabilities are cleared from the permit‐
419 ted and effective capability sets.
420 2. If the effective user ID is changed from 0 to nonzero, then all
422 capabilities are cleared from the effective set.
423 3. If the effective user ID is changed from nonzero to 0, then the per‐
425 mitted set is copied to the effective set.
426 4. If the file system user ID is changed from 0 to nonzero (see setf‐
428 suid(2)) then the following capabilities are cleared from the effec‐
429 tive set: CAP_CHOWN, CAP_DAC_OVERRIDE, CAP_DAC_READ_SEARCH,
430 CAP_FOWNER, CAP_FSETID, CAP_LINUX_IMMUTABLE (since Linux 2.2.30),
431 CAP_MAC_OVERRIDE, and CAP_MKNOD (since Linux 2.2.30). If the file
432 system UID is changed from nonzero to 0, then any of these capabili‐
433 ties that are enabled in the permitted set are enabled in the effec‐
434 tive set.
435 If a thread that has a 0 value for one or more of its user IDs wants to
437 prevent its permitted capability set being cleared when it resets all
438 of its user IDs to nonzero values, it can do so using the prctl(2)
439 PR_SET_KEEPCAPS operation.
440 Programmatically adjusting capability sets
442 A thread can retrieve and change its capability sets using the
443 capget(2) and capset(2) system calls. However, the use of
444 cap_get_proc(3) and cap_set_proc(3), both provided in the libcap pack‐
445 age, is preferred for this purpose. The following rules govern changes
446 to the thread capability sets:
447 1. If the caller does not have the CAP_SETPCAP capability, the new
449 inheritable set must be a subset of the combination of the existing
450 inheritable and permitted sets.
451 2. (Since kernel 2.6.25) The new inheritable set must be a subset of
453 the combination of the existing inheritable set and the capability
454 bounding set.
455 3. The new permitted set must be a subset of the existing permitted set
457 (i.e., it is not possible to acquire permitted capabilities that the
458 thread does not currently have).
459 4. The new effective set must be a subset of the new permitted set.
461 The "securebits" flags: establishing a capabilities-only environment
463 Starting with kernel 2.6.26, and with a kernel in which file capabili‐
464 ties are enabled, Linux implements a set of per-thread securebits flags
465 that can be used to disable special handling of capabilities for UID 0
466 (root). These flags are as follows:
467 SECBIT_KEEP_CAPS
469 Setting this flag allows a thread that has one or more 0 UIDs to
470 retain its capabilities when it switches all of its UIDs to a
471 nonzero value. If this flag is not set, then such a UID switch
472 causes the thread to lose all capabilities. This flag is always
473 cleared on an execve(2). (This flag provides the same function‐
474 ality as the older prctl(2) PR_SET_KEEPCAPS operation.)
475 SECBIT_NO_SETUID_FIXUP
477 Setting this flag stops the kernel from adjusting capability
478 sets when the threads's effective and file system UIDs are
479 switched between zero and nonzero values. (See the subsection
480 Effect of User ID Changes on Capabilities.)
481 SECBIT_NOROOT
483 If this bit is set, then the kernel does not grant capabilities
484 when a set-user-ID-root program is executed, or when a process
485 with an effective or real UID of 0 calls execve(2). (See the
486 subsection Capabilities and execution of programs by root.)
487 Each of the above "base" flags has a companion "locked" flag. Setting
489 any of the "locked" flags is irreversible, and has the effect of pre‐
490 venting further changes to the corresponding "base" flag. The locked
491 flags are: SECBIT_KEEP_CAPS_LOCKED, SECBIT_NO_SETUID_FIXUP_LOCKED, and
492 SECBIT_NOROOT_LOCKED.
493 The securebits flags can be modified and retrieved using the prctl(2)
495 PR_SET_SECUREBITS and PR_GET_SECUREBITS operations. The CAP_SETPCAP
496 capability is required to modify the flags.
497 The securebits flags are inherited by child processes. During an
499 execve(2), all of the flags are preserved, except SECURE_KEEP_CAPS
500 which is always cleared.
501 An application can use the following call to lock itself, and all of
503 its descendants, into an environment where the only way of gaining
504 capabilities is by executing a program with associated file capabili‐
505 ties:
506 prctl(PR_SET_SECUREBITS,
508 SECBIT_KEEP_CAPS_LOCKED |
509 SECBIT_NO_SETUID_FIXUP |
510 SECBIT_NO_SETUID_FIXUP_LOCKED |
511 SECBIT_NOROOT |
512 SECBIT_NOROOT_LOCKED);
513
515 No standards govern capabilities, but the Linux capability implementa‐
516 tion is based on the withdrawn POSIX.1e draft standard; see
517 http://wt.xpilot.org/publications/posix.1e/.
518
520 Since kernel 2.5.27, capabilities are an optional kernel component, and
521 can be enabled/disabled via the CONFIG_SECURITY_CAPABILITIES kernel
522 configuration option.
523 The /proc/PID/task/TID/status file can be used to view the capability
525 sets of a thread. The /proc/PID/status file shows the capability sets
526 of a process's main thread.
527 The libcap package provides a suite of routines for setting and getting
529 capabilities that is more comfortable and less likely to change than
530 the interface provided by capset(2) and capget(2). This package also
531 provides the setcap(8) and getcap(8) programs. It can be found at
532 http://www.kernel.org/pub/linux/libs/security/linux-privs.
533 Before kernel 2.6.24, and since kernel 2.6.24 if file capabilities are
535 not enabled, a thread with the CAP_SETPCAP capability can manipulate
536 the capabilities of threads other than itself. However, this is only
537 theoretically possible, since no thread ever has CAP_SETPCAP in either
538 of these cases:
539 * In the pre-2.6.25 implementation the system-wide capability bounding
541 set, /proc/sys/kernel/cap-bound, always masks out this capability,
542 and this can not be changed without modifying the kernel source and
543 rebuilding.
544 * If file capabilities are disabled in the current implementation, then
546 init starts out with this capability removed from its per-process
547 bounding set, and that bounding set is inherited by all other pro‐
548 cesses created on the system.
549
551 capget(2), prctl(2), setfsuid(2), cap_clear(3), cap_copy_ext(3),
552 cap_from_text(3), cap_get_file(3), cap_get_proc(3), cap_init(3),
553 capgetp(3), capsetp(3), credentials(7), pthreads(7), getcap(8), set‐
554 cap(8)
555 include/linux/capability.h in the kernel source
557
559 This page is part of release 3.25 of the Linux man-pages project. A
560 description of the project, and information about reporting bugs, can
561 be found at http://www.kernel.org/doc/man-pages/.
562Linux 2010-06-19 CAPABILITIES(7)