1 /*
2  * This is <linux/capability.h>
3  *
4  * Andrew G. Morgan <morgan@kernel.org>
5  * Alexander Kjeldaas <astor@guardian.no>
6  * with help from Aleph1, Roland Buresund and Andrew Main.
7  *
8  * See here for the libcap library ("POSIX draft" compliance):
9  *
10  * ftp://www.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  /* deprecated - use v3 */
35 #define _LINUX_CAPABILITY_U32S_2     2
36 
37 #define _LINUX_CAPABILITY_VERSION_3  0x20080522
38 #define _LINUX_CAPABILITY_U32S_3     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 VFS_CAP_REVISION_MASK	0xFF000000
53 #define VFS_CAP_REVISION_SHIFT	24
54 #define VFS_CAP_FLAGS_MASK	~VFS_CAP_REVISION_MASK
55 #define VFS_CAP_FLAGS_EFFECTIVE	0x000001
56 
57 #define VFS_CAP_REVISION_1	0x01000000
58 #define VFS_CAP_U32_1           1
59 #define XATTR_CAPS_SZ_1         (sizeof(__le32)*(1 + 2*VFS_CAP_U32_1))
60 
61 #define VFS_CAP_REVISION_2	0x02000000
62 #define VFS_CAP_U32_2           2
63 #define XATTR_CAPS_SZ_2         (sizeof(__le32)*(1 + 2*VFS_CAP_U32_2))
64 
65 #define XATTR_CAPS_SZ           XATTR_CAPS_SZ_2
66 #define VFS_CAP_U32             VFS_CAP_U32_2
67 #define VFS_CAP_REVISION	VFS_CAP_REVISION_2
68 
69 struct vfs_cap_data {
70 	__le32 magic_etc;            /* Little endian */
71 	struct {
72 		__le32 permitted;    /* Little endian */
73 		__le32 inheritable;  /* Little endian */
74 	} data[VFS_CAP_U32];
75 };
76 
77 #ifndef __KERNEL__
78 
79 /*
80  * Backwardly compatible definition for source code - trapped in a
81  * 32-bit world. If you find you need this, please consider using
82  * libcap to untrap yourself...
83  */
84 #define _LINUX_CAPABILITY_VERSION  _LINUX_CAPABILITY_VERSION_1
85 #define _LINUX_CAPABILITY_U32S     _LINUX_CAPABILITY_U32S_1
86 
87 #else
88 
89 #define _KERNEL_CAPABILITY_VERSION _LINUX_CAPABILITY_VERSION_3
90 #define _KERNEL_CAPABILITY_U32S    _LINUX_CAPABILITY_U32S_3
91 
92 extern int file_caps_enabled;
93 
94 typedef struct kernel_cap_struct {
95 	__u32 cap[_KERNEL_CAPABILITY_U32S];
96 } kernel_cap_t;
97 
98 /* exact same as vfs_cap_data but in cpu endian and always filled completely */
99 struct cpu_vfs_cap_data {
100 	__u32 magic_etc;
101 	kernel_cap_t permitted;
102 	kernel_cap_t inheritable;
103 };
104 
105 #define _USER_CAP_HEADER_SIZE  (sizeof(struct __user_cap_header_struct))
106 #define _KERNEL_CAP_T_SIZE     (sizeof(kernel_cap_t))
107 
108 #endif
109 
110 
111 /**
112  ** POSIX-draft defined capabilities.
113  **/
114 
115 /* In a system with the [_POSIX_CHOWN_RESTRICTED] option defined, this
116    overrides the restriction of changing file ownership and group
117    ownership. */
118 
119 #define CAP_CHOWN            0
120 
121 /* Override all DAC access, including ACL execute access if
122    [_POSIX_ACL] is defined. Excluding DAC access covered by
123    CAP_LINUX_IMMUTABLE. */
124 
125 #define CAP_DAC_OVERRIDE     1
126 
127 /* Overrides all DAC restrictions regarding read and search on files
128    and directories, including ACL restrictions if [_POSIX_ACL] is
129    defined. Excluding DAC access covered by CAP_LINUX_IMMUTABLE. */
130 
131 #define CAP_DAC_READ_SEARCH  2
132 
133 /* Overrides all restrictions about allowed operations on files, where
134    file owner ID must be equal to the user ID, except where CAP_FSETID
135    is applicable. It doesn't override MAC and DAC restrictions. */
136 
137 #define CAP_FOWNER           3
138 
139 /* Overrides the following restrictions that the effective user ID
140    shall match the file owner ID when setting the S_ISUID and S_ISGID
141    bits on that file; that the effective group ID (or one of the
142    supplementary group IDs) shall match the file owner ID when setting
143    the S_ISGID bit on that file; that the S_ISUID and S_ISGID bits are
144    cleared on successful return from chown(2) (not implemented). */
145 
146 #define CAP_FSETID           4
147 
148 /* Overrides the restriction that the real or effective user ID of a
149    process sending a signal must match the real or effective user ID
150    of the process receiving the signal. */
151 
152 #define CAP_KILL             5
153 
154 /* Allows setgid(2) manipulation */
155 /* Allows setgroups(2) */
156 /* Allows forged gids on socket credentials passing. */
157 
158 #define CAP_SETGID           6
159 
160 /* Allows set*uid(2) manipulation (including fsuid). */
161 /* Allows forged pids on socket credentials passing. */
162 
163 #define CAP_SETUID           7
164 
165 
166 /**
167  ** Linux-specific capabilities
168  **/
169 
170 /* Without VFS support for capabilities:
171  *   Transfer any capability in your permitted set to any pid,
172  *   remove any capability in your permitted set from any pid
173  * With VFS support for capabilities (neither of above, but)
174  *   Add any capability from current's capability bounding set
175  *       to the current process' inheritable set
176  *   Allow taking bits out of capability bounding set
177  *   Allow modification of the securebits for a process
178  */
179 
180 #define CAP_SETPCAP          8
181 
182 /* Allow modification of S_IMMUTABLE and S_APPEND file attributes */
183 
184 #define CAP_LINUX_IMMUTABLE  9
185 
186 /* Allows binding to TCP/UDP sockets below 1024 */
187 /* Allows binding to ATM VCIs below 32 */
188 
189 #define CAP_NET_BIND_SERVICE 10
190 
191 /* Allow broadcasting, listen to multicast */
192 
193 #define CAP_NET_BROADCAST    11
194 
195 /* Allow interface configuration */
196 /* Allow administration of IP firewall, masquerading and accounting */
197 /* Allow setting debug option on sockets */
198 /* Allow modification of routing tables */
199 /* Allow setting arbitrary process / process group ownership on
200    sockets */
201 /* Allow binding to any address for transparent proxying (also via NET_RAW) */
202 /* Allow setting TOS (type of service) */
203 /* Allow setting promiscuous mode */
204 /* Allow clearing driver statistics */
205 /* Allow multicasting */
206 /* Allow read/write of device-specific registers */
207 /* Allow activation of ATM control sockets */
208 
209 #define CAP_NET_ADMIN        12
210 
211 /* Allow use of RAW sockets */
212 /* Allow use of PACKET sockets */
213 /* Allow binding to any address for transparent proxying (also via NET_ADMIN) */
214 
215 #define CAP_NET_RAW          13
216 
217 /* Allow locking of shared memory segments */
218 /* Allow mlock and mlockall (which doesn't really have anything to do
219    with IPC) */
220 
221 #define CAP_IPC_LOCK         14
222 
223 /* Override IPC ownership checks */
224 
225 #define CAP_IPC_OWNER        15
226 
227 /* Insert and remove kernel modules - modify kernel without limit */
228 #define CAP_SYS_MODULE       16
229 
230 /* Allow ioperm/iopl access */
231 /* Allow sending USB messages to any device via /proc/bus/usb */
232 
233 #define CAP_SYS_RAWIO        17
234 
235 /* Allow use of chroot() */
236 
237 #define CAP_SYS_CHROOT       18
238 
239 /* Allow ptrace() of any process */
240 
241 #define CAP_SYS_PTRACE       19
242 
243 /* Allow configuration of process accounting */
244 
245 #define CAP_SYS_PACCT        20
246 
247 /* Allow configuration of the secure attention key */
248 /* Allow administration of the random device */
249 /* Allow examination and configuration of disk quotas */
250 /* Allow setting the domainname */
251 /* Allow setting the hostname */
252 /* Allow calling bdflush() */
253 /* Allow mount() and umount(), setting up new smb connection */
254 /* Allow some autofs root ioctls */
255 /* Allow nfsservctl */
256 /* Allow VM86_REQUEST_IRQ */
257 /* Allow to read/write pci config on alpha */
258 /* Allow irix_prctl on mips (setstacksize) */
259 /* Allow flushing all cache on m68k (sys_cacheflush) */
260 /* Allow removing semaphores */
261 /* Used instead of CAP_CHOWN to "chown" IPC message queues, semaphores
262    and shared memory */
263 /* Allow locking/unlocking of shared memory segment */
264 /* Allow turning swap on/off */
265 /* Allow forged pids on socket credentials passing */
266 /* Allow setting readahead and flushing buffers on block devices */
267 /* Allow setting geometry in floppy driver */
268 /* Allow turning DMA on/off in xd driver */
269 /* Allow administration of md devices (mostly the above, but some
270    extra ioctls) */
271 /* Allow tuning the ide driver */
272 /* Allow access to the nvram device */
273 /* Allow administration of apm_bios, serial and bttv (TV) device */
274 /* Allow manufacturer commands in isdn CAPI support driver */
275 /* Allow reading non-standardized portions of pci configuration space */
276 /* Allow DDI debug ioctl on sbpcd driver */
277 /* Allow setting up serial ports */
278 /* Allow sending raw qic-117 commands */
279 /* Allow enabling/disabling tagged queuing on SCSI controllers and sending
280    arbitrary SCSI commands */
281 /* Allow setting encryption key on loopback filesystem */
282 /* Allow setting zone reclaim policy */
283 
284 #define CAP_SYS_ADMIN        21
285 
286 /* Allow use of reboot() */
287 
288 #define CAP_SYS_BOOT         22
289 
290 /* Allow raising priority and setting priority on other (different
291    UID) processes */
292 /* Allow use of FIFO and round-robin (realtime) scheduling on own
293    processes and setting the scheduling algorithm used by another
294    process. */
295 /* Allow setting cpu affinity on other processes */
296 
297 #define CAP_SYS_NICE         23
298 
299 /* Override resource limits. Set resource limits. */
300 /* Override quota limits. */
301 /* Override reserved space on ext2 filesystem */
302 /* Modify data journaling mode on ext3 filesystem (uses journaling
303    resources) */
304 /* NOTE: ext2 honors fsuid when checking for resource overrides, so
305    you can override using fsuid too */
306 /* Override size restrictions on IPC message queues */
307 /* Allow more than 64hz interrupts from the real-time clock */
308 /* Override max number of consoles on console allocation */
309 /* Override max number of keymaps */
310 
311 #define CAP_SYS_RESOURCE     24
312 
313 /* Allow manipulation of system clock */
314 /* Allow irix_stime on mips */
315 /* Allow setting the real-time clock */
316 
317 #define CAP_SYS_TIME         25
318 
319 /* Allow configuration of tty devices */
320 /* Allow vhangup() of tty */
321 
322 #define CAP_SYS_TTY_CONFIG   26
323 
324 /* Allow the privileged aspects of mknod() */
325 
326 #define CAP_MKNOD            27
327 
328 /* Allow taking of leases on files */
329 
330 #define CAP_LEASE            28
331 
332 #define CAP_AUDIT_WRITE      29
333 
334 #define CAP_AUDIT_CONTROL    30
335 
336 #define CAP_SETFCAP	     31
337 
338 /* Override MAC access.
339    The base kernel enforces no MAC policy.
340    An LSM may enforce a MAC policy, and if it does and it chooses
341    to implement capability based overrides of that policy, this is
342    the capability it should use to do so. */
343 
344 #define CAP_MAC_OVERRIDE     32
345 
346 /* Allow MAC configuration or state changes.
347    The base kernel requires no MAC configuration.
348    An LSM may enforce a MAC policy, and if it does and it chooses
349    to implement capability based checks on modifications to that
350    policy or the data required to maintain it, this is the
351    capability it should use to do so. */
352 
353 #define CAP_MAC_ADMIN        33
354 
355 /* Allow configuring the kernel's syslog (printk behaviour) */
356 
357 #define CAP_SYSLOG           34
358 
359 /* Allow triggering something that will wake the system */
360 
361 #define CAP_WAKE_ALARM            35
362 
363 
364 #define CAP_LAST_CAP         CAP_WAKE_ALARM
365 
366 #define cap_valid(x) ((x) >= 0 && (x) <= CAP_LAST_CAP)
367 
368 /*
369  * Bit location of each capability (used by user-space library and kernel)
370  */
371 
372 #define CAP_TO_INDEX(x)     ((x) >> 5)        /* 1 << 5 == bits in __u32 */
373 #define CAP_TO_MASK(x)      (1 << ((x) & 31)) /* mask for indexed __u32 */
374 
375 #ifdef __KERNEL__
376 
377 struct dentry;
378 struct user_namespace;
379 
380 struct user_namespace *current_user_ns(void);
381 
382 extern const kernel_cap_t __cap_empty_set;
383 extern const kernel_cap_t __cap_init_eff_set;
384 
385 /*
386  * Internal kernel functions only
387  */
388 
389 #define CAP_FOR_EACH_U32(__capi)  \
390 	for (__capi = 0; __capi < _KERNEL_CAPABILITY_U32S; ++__capi)
391 
392 /*
393  * CAP_FS_MASK and CAP_NFSD_MASKS:
394  *
395  * The fs mask is all the privileges that fsuid==0 historically meant.
396  * At one time in the past, that included CAP_MKNOD and CAP_LINUX_IMMUTABLE.
397  *
398  * It has never meant setting security.* and trusted.* xattrs.
399  *
400  * We could also define fsmask as follows:
401  *   1. CAP_FS_MASK is the privilege to bypass all fs-related DAC permissions
402  *   2. The security.* and trusted.* xattrs are fs-related MAC permissions
403  */
404 
405 # define CAP_FS_MASK_B0     (CAP_TO_MASK(CAP_CHOWN)		\
406 			    | CAP_TO_MASK(CAP_MKNOD)		\
407 			    | CAP_TO_MASK(CAP_DAC_OVERRIDE)	\
408 			    | CAP_TO_MASK(CAP_DAC_READ_SEARCH)	\
409 			    | CAP_TO_MASK(CAP_FOWNER)		\
410 			    | CAP_TO_MASK(CAP_FSETID))
411 
412 # define CAP_FS_MASK_B1     (CAP_TO_MASK(CAP_MAC_OVERRIDE))
413 
414 #if _KERNEL_CAPABILITY_U32S != 2
415 # error Fix up hand-coded capability macro initializers
416 #else /* HAND-CODED capability initializers */
417 
418 # define CAP_EMPTY_SET    ((kernel_cap_t){{ 0, 0 }})
419 # define CAP_FULL_SET     ((kernel_cap_t){{ ~0, ~0 }})
420 # define CAP_FS_SET       ((kernel_cap_t){{ CAP_FS_MASK_B0 \
421 				    | CAP_TO_MASK(CAP_LINUX_IMMUTABLE), \
422 				    CAP_FS_MASK_B1 } })
423 # define CAP_NFSD_SET     ((kernel_cap_t){{ CAP_FS_MASK_B0 \
424 				    | CAP_TO_MASK(CAP_SYS_RESOURCE), \
425 				    CAP_FS_MASK_B1 } })
426 
427 #endif /* _KERNEL_CAPABILITY_U32S != 2 */
428 
429 # define cap_clear(c)         do { (c) = __cap_empty_set; } while (0)
430 
431 #define cap_raise(c, flag)  ((c).cap[CAP_TO_INDEX(flag)] |= CAP_TO_MASK(flag))
432 #define cap_lower(c, flag)  ((c).cap[CAP_TO_INDEX(flag)] &= ~CAP_TO_MASK(flag))
433 #define cap_raised(c, flag) ((c).cap[CAP_TO_INDEX(flag)] & CAP_TO_MASK(flag))
434 
435 #define CAP_BOP_ALL(c, a, b, OP)                                    \
436 do {                                                                \
437 	unsigned __capi;                                            \
438 	CAP_FOR_EACH_U32(__capi) {                                  \
439 		c.cap[__capi] = a.cap[__capi] OP b.cap[__capi];     \
440 	}                                                           \
441 } while (0)
442 
443 #define CAP_UOP_ALL(c, a, OP)                                       \
444 do {                                                                \
445 	unsigned __capi;                                            \
446 	CAP_FOR_EACH_U32(__capi) {                                  \
447 		c.cap[__capi] = OP a.cap[__capi];                   \
448 	}                                                           \
449 } while (0)
450 
cap_combine(const kernel_cap_t a,const kernel_cap_t b)451 static inline kernel_cap_t cap_combine(const kernel_cap_t a,
452 				       const kernel_cap_t b)
453 {
454 	kernel_cap_t dest;
455 	CAP_BOP_ALL(dest, a, b, |);
456 	return dest;
457 }
458 
cap_intersect(const kernel_cap_t a,const kernel_cap_t b)459 static inline kernel_cap_t cap_intersect(const kernel_cap_t a,
460 					 const kernel_cap_t b)
461 {
462 	kernel_cap_t dest;
463 	CAP_BOP_ALL(dest, a, b, &);
464 	return dest;
465 }
466 
cap_drop(const kernel_cap_t a,const kernel_cap_t drop)467 static inline kernel_cap_t cap_drop(const kernel_cap_t a,
468 				    const kernel_cap_t drop)
469 {
470 	kernel_cap_t dest;
471 	CAP_BOP_ALL(dest, a, drop, &~);
472 	return dest;
473 }
474 
cap_invert(const kernel_cap_t c)475 static inline kernel_cap_t cap_invert(const kernel_cap_t c)
476 {
477 	kernel_cap_t dest;
478 	CAP_UOP_ALL(dest, c, ~);
479 	return dest;
480 }
481 
cap_isclear(const kernel_cap_t a)482 static inline int cap_isclear(const kernel_cap_t a)
483 {
484 	unsigned __capi;
485 	CAP_FOR_EACH_U32(__capi) {
486 		if (a.cap[__capi] != 0)
487 			return 0;
488 	}
489 	return 1;
490 }
491 
492 /*
493  * Check if "a" is a subset of "set".
494  * return 1 if ALL of the capabilities in "a" are also in "set"
495  *	cap_issubset(0101, 1111) will return 1
496  * return 0 if ANY of the capabilities in "a" are not in "set"
497  *	cap_issubset(1111, 0101) will return 0
498  */
cap_issubset(const kernel_cap_t a,const kernel_cap_t set)499 static inline int cap_issubset(const kernel_cap_t a, const kernel_cap_t set)
500 {
501 	kernel_cap_t dest;
502 	dest = cap_drop(a, set);
503 	return cap_isclear(dest);
504 }
505 
506 /* Used to decide between falling back on the old suser() or fsuser(). */
507 
cap_is_fs_cap(int cap)508 static inline int cap_is_fs_cap(int cap)
509 {
510 	const kernel_cap_t __cap_fs_set = CAP_FS_SET;
511 	return !!(CAP_TO_MASK(cap) & __cap_fs_set.cap[CAP_TO_INDEX(cap)]);
512 }
513 
cap_drop_fs_set(const kernel_cap_t a)514 static inline kernel_cap_t cap_drop_fs_set(const kernel_cap_t a)
515 {
516 	const kernel_cap_t __cap_fs_set = CAP_FS_SET;
517 	return cap_drop(a, __cap_fs_set);
518 }
519 
cap_raise_fs_set(const kernel_cap_t a,const kernel_cap_t permitted)520 static inline kernel_cap_t cap_raise_fs_set(const kernel_cap_t a,
521 					    const kernel_cap_t permitted)
522 {
523 	const kernel_cap_t __cap_fs_set = CAP_FS_SET;
524 	return cap_combine(a,
525 			   cap_intersect(permitted, __cap_fs_set));
526 }
527 
cap_drop_nfsd_set(const kernel_cap_t a)528 static inline kernel_cap_t cap_drop_nfsd_set(const kernel_cap_t a)
529 {
530 	const kernel_cap_t __cap_fs_set = CAP_NFSD_SET;
531 	return cap_drop(a, __cap_fs_set);
532 }
533 
cap_raise_nfsd_set(const kernel_cap_t a,const kernel_cap_t permitted)534 static inline kernel_cap_t cap_raise_nfsd_set(const kernel_cap_t a,
535 					      const kernel_cap_t permitted)
536 {
537 	const kernel_cap_t __cap_nfsd_set = CAP_NFSD_SET;
538 	return cap_combine(a,
539 			   cap_intersect(permitted, __cap_nfsd_set));
540 }
541 
542 extern bool has_capability(struct task_struct *t, int cap);
543 extern bool has_ns_capability(struct task_struct *t,
544 			      struct user_namespace *ns, int cap);
545 extern bool has_capability_noaudit(struct task_struct *t, int cap);
546 extern bool has_ns_capability_noaudit(struct task_struct *t,
547 				      struct user_namespace *ns, int cap);
548 extern bool capable(int cap);
549 extern bool ns_capable(struct user_namespace *ns, int cap);
550 extern bool nsown_capable(int cap);
551 
552 /* audit system wants to get cap info from files as well */
553 extern int get_vfs_caps_from_disk(const struct dentry *dentry, struct cpu_vfs_cap_data *cpu_caps);
554 
555 #endif /* __KERNEL__ */
556 
557 #endif /* !_LINUX_CAPABILITY_H */
558