xref: /linux-6.15/include/linux/capability.h (revision 43f5b308)
1 /*
2  * This is <linux/capability.h>
3  *
4  * Andrew G. Morgan <[email protected]>
5  * Alexander Kjeldaas <[email protected]>
6  * with help from Aleph1, Roland Buresund and Andrew Main.
7  *
8  * See here for the libcap library ("POSIX draft" compliance):
9  *
10  * ftp://linux.kernel.org/pub/linux/libs/security/linux-privs/kernel-2.6/
11  */
12 
13 #ifndef _LINUX_CAPABILITY_H
14 #define _LINUX_CAPABILITY_H
15 
16 #include <linux/types.h>
17 
18 struct task_struct;
19 
20 /* User-level do most of the mapping between kernel and user
21    capabilities based on the version tag given by the kernel. The
22    kernel might be somewhat backwards compatible, but don't bet on
23    it. */
24 
25 /* Note, cap_t, is defined by POSIX (draft) to be an "opaque" pointer to
26    a set of three capability sets.  The transposition of 3*the
27    following structure to such a composite is better handled in a user
28    library since the draft standard requires the use of malloc/free
29    etc.. */
30 
31 #define _LINUX_CAPABILITY_VERSION_1  0x19980330
32 #define _LINUX_CAPABILITY_U32S_1     1
33 
34 #define _LINUX_CAPABILITY_VERSION_2  0x20071026
35 #define _LINUX_CAPABILITY_U32S_2     2
36 
37 #define _LINUX_CAPABILITY_VERSION    _LINUX_CAPABILITY_VERSION_2
38 #define _LINUX_CAPABILITY_U32S       _LINUX_CAPABILITY_U32S_2
39 
40 typedef struct __user_cap_header_struct {
41 	__u32 version;
42 	int pid;
43 } __user *cap_user_header_t;
44 
45 typedef struct __user_cap_data_struct {
46         __u32 effective;
47         __u32 permitted;
48         __u32 inheritable;
49 } __user *cap_user_data_t;
50 
51 
52 #define XATTR_CAPS_SUFFIX "capability"
53 #define XATTR_NAME_CAPS XATTR_SECURITY_PREFIX XATTR_CAPS_SUFFIX
54 
55 #define VFS_CAP_REVISION_MASK	0xFF000000
56 #define VFS_CAP_FLAGS_MASK	~VFS_CAP_REVISION_MASK
57 #define VFS_CAP_FLAGS_EFFECTIVE	0x000001
58 
59 #define VFS_CAP_REVISION_1	0x01000000
60 #define VFS_CAP_U32_1           1
61 #define XATTR_CAPS_SZ_1         (sizeof(__le32)*(1 + 2*VFS_CAP_U32_1))
62 
63 #define VFS_CAP_REVISION_2	0x02000000
64 #define VFS_CAP_U32_2           2
65 #define XATTR_CAPS_SZ_2         (sizeof(__le32)*(1 + 2*VFS_CAP_U32_2))
66 
67 #define XATTR_CAPS_SZ           XATTR_CAPS_SZ_2
68 #define VFS_CAP_U32             VFS_CAP_U32_2
69 #define VFS_CAP_REVISION	VFS_CAP_REVISION_2
70 
71 
72 struct vfs_cap_data {
73 	__le32 magic_etc;            /* Little endian */
74 	struct {
75 		__le32 permitted;    /* Little endian */
76 		__le32 inheritable;  /* Little endian */
77 	} data[VFS_CAP_U32];
78 };
79 
80 #ifdef __KERNEL__
81 
82 typedef struct kernel_cap_struct {
83 	__u32 cap[_LINUX_CAPABILITY_U32S];
84 } kernel_cap_t;
85 
86 #define _USER_CAP_HEADER_SIZE  (sizeof(struct __user_cap_header_struct))
87 #define _KERNEL_CAP_T_SIZE     (sizeof(kernel_cap_t))
88 
89 #endif
90 
91 
92 /**
93  ** POSIX-draft defined capabilities.
94  **/
95 
96 /* In a system with the [_POSIX_CHOWN_RESTRICTED] option defined, this
97    overrides the restriction of changing file ownership and group
98    ownership. */
99 
100 #define CAP_CHOWN            0
101 
102 /* Override all DAC access, including ACL execute access if
103    [_POSIX_ACL] is defined. Excluding DAC access covered by
104    CAP_LINUX_IMMUTABLE. */
105 
106 #define CAP_DAC_OVERRIDE     1
107 
108 /* Overrides all DAC restrictions regarding read and search on files
109    and directories, including ACL restrictions if [_POSIX_ACL] is
110    defined. Excluding DAC access covered by CAP_LINUX_IMMUTABLE. */
111 
112 #define CAP_DAC_READ_SEARCH  2
113 
114 /* Overrides all restrictions about allowed operations on files, where
115    file owner ID must be equal to the user ID, except where CAP_FSETID
116    is applicable. It doesn't override MAC and DAC restrictions. */
117 
118 #define CAP_FOWNER           3
119 
120 /* Overrides the following restrictions that the effective user ID
121    shall match the file owner ID when setting the S_ISUID and S_ISGID
122    bits on that file; that the effective group ID (or one of the
123    supplementary group IDs) shall match the file owner ID when setting
124    the S_ISGID bit on that file; that the S_ISUID and S_ISGID bits are
125    cleared on successful return from chown(2) (not implemented). */
126 
127 #define CAP_FSETID           4
128 
129 /* Overrides the restriction that the real or effective user ID of a
130    process sending a signal must match the real or effective user ID
131    of the process receiving the signal. */
132 
133 #define CAP_KILL             5
134 
135 /* Allows setgid(2) manipulation */
136 /* Allows setgroups(2) */
137 /* Allows forged gids on socket credentials passing. */
138 
139 #define CAP_SETGID           6
140 
141 /* Allows set*uid(2) manipulation (including fsuid). */
142 /* Allows forged pids on socket credentials passing. */
143 
144 #define CAP_SETUID           7
145 
146 
147 /**
148  ** Linux-specific capabilities
149  **/
150 
151 /* Without VFS support for capabilities:
152  *   Transfer any capability in your permitted set to any pid,
153  *   remove any capability in your permitted set from any pid
154  * With VFS support for capabilities (neither of above, but)
155  *   Add any capability from current's capability bounding set
156  *       to the current process' inheritable set
157  *   Allow taking bits out of capability bounding set
158  *   Allow modification of the securebits for a process
159  */
160 
161 #define CAP_SETPCAP          8
162 
163 /* Allow modification of S_IMMUTABLE and S_APPEND file attributes */
164 
165 #define CAP_LINUX_IMMUTABLE  9
166 
167 /* Allows binding to TCP/UDP sockets below 1024 */
168 /* Allows binding to ATM VCIs below 32 */
169 
170 #define CAP_NET_BIND_SERVICE 10
171 
172 /* Allow broadcasting, listen to multicast */
173 
174 #define CAP_NET_BROADCAST    11
175 
176 /* Allow interface configuration */
177 /* Allow administration of IP firewall, masquerading and accounting */
178 /* Allow setting debug option on sockets */
179 /* Allow modification of routing tables */
180 /* Allow setting arbitrary process / process group ownership on
181    sockets */
182 /* Allow binding to any address for transparent proxying */
183 /* Allow setting TOS (type of service) */
184 /* Allow setting promiscuous mode */
185 /* Allow clearing driver statistics */
186 /* Allow multicasting */
187 /* Allow read/write of device-specific registers */
188 /* Allow activation of ATM control sockets */
189 
190 #define CAP_NET_ADMIN        12
191 
192 /* Allow use of RAW sockets */
193 /* Allow use of PACKET sockets */
194 
195 #define CAP_NET_RAW          13
196 
197 /* Allow locking of shared memory segments */
198 /* Allow mlock and mlockall (which doesn't really have anything to do
199    with IPC) */
200 
201 #define CAP_IPC_LOCK         14
202 
203 /* Override IPC ownership checks */
204 
205 #define CAP_IPC_OWNER        15
206 
207 /* Insert and remove kernel modules - modify kernel without limit */
208 #define CAP_SYS_MODULE       16
209 
210 /* Allow ioperm/iopl access */
211 /* Allow sending USB messages to any device via /proc/bus/usb */
212 
213 #define CAP_SYS_RAWIO        17
214 
215 /* Allow use of chroot() */
216 
217 #define CAP_SYS_CHROOT       18
218 
219 /* Allow ptrace() of any process */
220 
221 #define CAP_SYS_PTRACE       19
222 
223 /* Allow configuration of process accounting */
224 
225 #define CAP_SYS_PACCT        20
226 
227 /* Allow configuration of the secure attention key */
228 /* Allow administration of the random device */
229 /* Allow examination and configuration of disk quotas */
230 /* Allow configuring the kernel's syslog (printk behaviour) */
231 /* Allow setting the domainname */
232 /* Allow setting the hostname */
233 /* Allow calling bdflush() */
234 /* Allow mount() and umount(), setting up new smb connection */
235 /* Allow some autofs root ioctls */
236 /* Allow nfsservctl */
237 /* Allow VM86_REQUEST_IRQ */
238 /* Allow to read/write pci config on alpha */
239 /* Allow irix_prctl on mips (setstacksize) */
240 /* Allow flushing all cache on m68k (sys_cacheflush) */
241 /* Allow removing semaphores */
242 /* Used instead of CAP_CHOWN to "chown" IPC message queues, semaphores
243    and shared memory */
244 /* Allow locking/unlocking of shared memory segment */
245 /* Allow turning swap on/off */
246 /* Allow forged pids on socket credentials passing */
247 /* Allow setting readahead and flushing buffers on block devices */
248 /* Allow setting geometry in floppy driver */
249 /* Allow turning DMA on/off in xd driver */
250 /* Allow administration of md devices (mostly the above, but some
251    extra ioctls) */
252 /* Allow tuning the ide driver */
253 /* Allow access to the nvram device */
254 /* Allow administration of apm_bios, serial and bttv (TV) device */
255 /* Allow manufacturer commands in isdn CAPI support driver */
256 /* Allow reading non-standardized portions of pci configuration space */
257 /* Allow DDI debug ioctl on sbpcd driver */
258 /* Allow setting up serial ports */
259 /* Allow sending raw qic-117 commands */
260 /* Allow enabling/disabling tagged queuing on SCSI controllers and sending
261    arbitrary SCSI commands */
262 /* Allow setting encryption key on loopback filesystem */
263 /* Allow setting zone reclaim policy */
264 
265 #define CAP_SYS_ADMIN        21
266 
267 /* Allow use of reboot() */
268 
269 #define CAP_SYS_BOOT         22
270 
271 /* Allow raising priority and setting priority on other (different
272    UID) processes */
273 /* Allow use of FIFO and round-robin (realtime) scheduling on own
274    processes and setting the scheduling algorithm used by another
275    process. */
276 /* Allow setting cpu affinity on other processes */
277 
278 #define CAP_SYS_NICE         23
279 
280 /* Override resource limits. Set resource limits. */
281 /* Override quota limits. */
282 /* Override reserved space on ext2 filesystem */
283 /* Modify data journaling mode on ext3 filesystem (uses journaling
284    resources) */
285 /* NOTE: ext2 honors fsuid when checking for resource overrides, so
286    you can override using fsuid too */
287 /* Override size restrictions on IPC message queues */
288 /* Allow more than 64hz interrupts from the real-time clock */
289 /* Override max number of consoles on console allocation */
290 /* Override max number of keymaps */
291 
292 #define CAP_SYS_RESOURCE     24
293 
294 /* Allow manipulation of system clock */
295 /* Allow irix_stime on mips */
296 /* Allow setting the real-time clock */
297 
298 #define CAP_SYS_TIME         25
299 
300 /* Allow configuration of tty devices */
301 /* Allow vhangup() of tty */
302 
303 #define CAP_SYS_TTY_CONFIG   26
304 
305 /* Allow the privileged aspects of mknod() */
306 
307 #define CAP_MKNOD            27
308 
309 /* Allow taking of leases on files */
310 
311 #define CAP_LEASE            28
312 
313 #define CAP_AUDIT_WRITE      29
314 
315 #define CAP_AUDIT_CONTROL    30
316 
317 #define CAP_SETFCAP	     31
318 
319 /* Override MAC access.
320    The base kernel enforces no MAC policy.
321    An LSM may enforce a MAC policy, and if it does and it chooses
322    to implement capability based overrides of that policy, this is
323    the capability it should use to do so. */
324 
325 #define CAP_MAC_OVERRIDE     32
326 
327 /* Allow MAC configuration or state changes.
328    The base kernel requires no MAC configuration.
329    An LSM may enforce a MAC policy, and if it does and it chooses
330    to implement capability based checks on modifications to that
331    policy or the data required to maintain it, this is the
332    capability it should use to do so. */
333 
334 #define CAP_MAC_ADMIN        33
335 
336 #define CAP_LAST_CAP         CAP_MAC_ADMIN
337 
338 #define cap_valid(x) ((x) >= 0 && (x) <= CAP_LAST_CAP)
339 
340 /*
341  * Bit location of each capability (used by user-space library and kernel)
342  */
343 
344 #define CAP_TO_INDEX(x)     ((x) >> 5)        /* 1 << 5 == bits in __u32 */
345 #define CAP_TO_MASK(x)      (1 << ((x) & 31)) /* mask for indexed __u32 */
346 
347 #ifdef __KERNEL__
348 
349 /*
350  * Internal kernel functions only
351  */
352 
353 #define CAP_FOR_EACH_U32(__capi)  \
354 	for (__capi = 0; __capi < _LINUX_CAPABILITY_U32S; ++__capi)
355 
356 # define CAP_FS_MASK_B0     (CAP_TO_MASK(CAP_CHOWN)		\
357 			    | CAP_TO_MASK(CAP_DAC_OVERRIDE)	\
358 			    | CAP_TO_MASK(CAP_DAC_READ_SEARCH)	\
359 			    | CAP_TO_MASK(CAP_FOWNER)		\
360 			    | CAP_TO_MASK(CAP_FSETID))
361 
362 # define CAP_FS_MASK_B1     (CAP_TO_MASK(CAP_MAC_OVERRIDE))
363 
364 #if _LINUX_CAPABILITY_U32S != 2
365 # error Fix up hand-coded capability macro initializers
366 #else /* HAND-CODED capability initializers */
367 
368 # define CAP_EMPTY_SET    ((kernel_cap_t){{ 0, 0 }})
369 # define CAP_FULL_SET     ((kernel_cap_t){{ ~0, ~0 }})
370 # define CAP_INIT_EFF_SET ((kernel_cap_t){{ ~CAP_TO_MASK(CAP_SETPCAP), ~0 }})
371 # define CAP_FS_SET       ((kernel_cap_t){{ CAP_FS_MASK_B0, CAP_FS_MASK_B1 } })
372 # define CAP_NFSD_SET     ((kernel_cap_t){{ CAP_FS_MASK_B0|CAP_TO_MASK(CAP_SYS_RESOURCE), \
373 					CAP_FS_MASK_B1 } })
374 
375 #endif /* _LINUX_CAPABILITY_U32S != 2 */
376 
377 #define CAP_INIT_INH_SET    CAP_EMPTY_SET
378 
379 # define cap_clear(c)         do { (c) = __cap_empty_set; } while (0)
380 # define cap_set_full(c)      do { (c) = __cap_full_set; } while (0)
381 # define cap_set_init_eff(c)  do { (c) = __cap_init_eff_set; } while (0)
382 
383 #define cap_raise(c, flag)  ((c).cap[CAP_TO_INDEX(flag)] |= CAP_TO_MASK(flag))
384 #define cap_lower(c, flag)  ((c).cap[CAP_TO_INDEX(flag)] &= ~CAP_TO_MASK(flag))
385 #define cap_raised(c, flag) ((c).cap[CAP_TO_INDEX(flag)] & CAP_TO_MASK(flag))
386 
387 #define CAP_BOP_ALL(c, a, b, OP)                                    \
388 do {                                                                \
389 	unsigned __capi;                                            \
390 	CAP_FOR_EACH_U32(__capi) {                                  \
391 		c.cap[__capi] = a.cap[__capi] OP b.cap[__capi];     \
392 	}                                                           \
393 } while (0)
394 
395 #define CAP_UOP_ALL(c, a, OP)                                       \
396 do {                                                                \
397 	unsigned __capi;                                            \
398 	CAP_FOR_EACH_U32(__capi) {                                  \
399 		c.cap[__capi] = OP a.cap[__capi];                   \
400 	}                                                           \
401 } while (0)
402 
403 static inline kernel_cap_t cap_combine(const kernel_cap_t a,
404 				       const kernel_cap_t b)
405 {
406 	kernel_cap_t dest;
407 	CAP_BOP_ALL(dest, a, b, |);
408 	return dest;
409 }
410 
411 static inline kernel_cap_t cap_intersect(const kernel_cap_t a,
412 					 const kernel_cap_t b)
413 {
414 	kernel_cap_t dest;
415 	CAP_BOP_ALL(dest, a, b, &);
416 	return dest;
417 }
418 
419 static inline kernel_cap_t cap_drop(const kernel_cap_t a,
420 				    const kernel_cap_t drop)
421 {
422 	kernel_cap_t dest;
423 	CAP_BOP_ALL(dest, a, drop, &~);
424 	return dest;
425 }
426 
427 static inline kernel_cap_t cap_invert(const kernel_cap_t c)
428 {
429 	kernel_cap_t dest;
430 	CAP_UOP_ALL(dest, c, ~);
431 	return dest;
432 }
433 
434 static inline int cap_isclear(const kernel_cap_t a)
435 {
436 	unsigned __capi;
437 	CAP_FOR_EACH_U32(__capi) {
438 		if (a.cap[__capi] != 0)
439 			return 0;
440 	}
441 	return 1;
442 }
443 
444 static inline int cap_issubset(const kernel_cap_t a, const kernel_cap_t set)
445 {
446 	kernel_cap_t dest;
447 	dest = cap_drop(a, set);
448 	return cap_isclear(dest);
449 }
450 
451 /* Used to decide between falling back on the old suser() or fsuser(). */
452 
453 static inline int cap_is_fs_cap(int cap)
454 {
455 	const kernel_cap_t __cap_fs_set = CAP_FS_SET;
456 	return !!(CAP_TO_MASK(cap) & __cap_fs_set.cap[CAP_TO_INDEX(cap)]);
457 }
458 
459 static inline kernel_cap_t cap_drop_fs_set(const kernel_cap_t a)
460 {
461 	const kernel_cap_t __cap_fs_set = CAP_FS_SET;
462 	return cap_drop(a, __cap_fs_set);
463 }
464 
465 static inline kernel_cap_t cap_raise_fs_set(const kernel_cap_t a,
466 					    const kernel_cap_t permitted)
467 {
468 	const kernel_cap_t __cap_fs_set = CAP_FS_SET;
469 	return cap_combine(a,
470 			   cap_intersect(permitted, __cap_fs_set));
471 }
472 
473 static inline kernel_cap_t cap_drop_nfsd_set(const kernel_cap_t a)
474 {
475 	const kernel_cap_t __cap_fs_set = CAP_NFSD_SET;
476 	return cap_drop(a, __cap_fs_set);
477 }
478 
479 static inline kernel_cap_t cap_raise_nfsd_set(const kernel_cap_t a,
480 					      const kernel_cap_t permitted)
481 {
482 	const kernel_cap_t __cap_nfsd_set = CAP_NFSD_SET;
483 	return cap_combine(a,
484 			   cap_intersect(permitted, __cap_nfsd_set));
485 }
486 
487 extern const kernel_cap_t __cap_empty_set;
488 extern const kernel_cap_t __cap_full_set;
489 extern const kernel_cap_t __cap_init_eff_set;
490 
491 int capable(int cap);
492 int __capable(struct task_struct *t, int cap);
493 
494 #endif /* __KERNEL__ */
495 
496 #endif /* !_LINUX_CAPABILITY_H */
497