blob: aaa01b0e3ee94251476d15f2ae6574844af9bb16 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001 ============================
2 KERNEL KEY RETENTION SERVICE
3 ============================
4
5This service allows cryptographic keys, authentication tokens, cross-domain
6user mappings, and similar to be cached in the kernel for the use of
7filesystems other kernel services.
8
9Keyrings are permitted; these are a special type of key that can hold links to
10other keys. Processes each have three standard keyring subscriptions that a
11kernel service can search for relevant keys.
12
13The key service can be configured on by enabling:
14
15 "Security options"/"Enable access key retention support" (CONFIG_KEYS)
16
17This document has the following sections:
18
19 - Key overview
20 - Key service overview
21 - Key access permissions
22 - New procfs files
23 - Userspace system call interface
24 - Kernel services
David Howells76d8aea2005-06-23 22:00:49 -070025 - Notes on accessing payload contents
Linus Torvalds1da177e2005-04-16 15:20:36 -070026 - Defining a key type
27 - Request-key callback service
28 - Key access filesystem
29
30
31============
32KEY OVERVIEW
33============
34
35In this context, keys represent units of cryptographic data, authentication
36tokens, keyrings, etc.. These are represented in the kernel by struct key.
37
38Each key has a number of attributes:
39
40 - A serial number.
41 - A type.
42 - A description (for matching a key in a search).
43 - Access control information.
44 - An expiry time.
45 - A payload.
46 - State.
47
48
David Howells76d8aea2005-06-23 22:00:49 -070049 (*) Each key is issued a serial number of type key_serial_t that is unique for
50 the lifetime of that key. All serial numbers are positive non-zero 32-bit
51 integers.
Linus Torvalds1da177e2005-04-16 15:20:36 -070052
53 Userspace programs can use a key's serial numbers as a way to gain access
54 to it, subject to permission checking.
55
56 (*) Each key is of a defined "type". Types must be registered inside the
David Howells76d8aea2005-06-23 22:00:49 -070057 kernel by a kernel service (such as a filesystem) before keys of that type
58 can be added or used. Userspace programs cannot define new types directly.
Linus Torvalds1da177e2005-04-16 15:20:36 -070059
David Howells76d8aea2005-06-23 22:00:49 -070060 Key types are represented in the kernel by struct key_type. This defines a
61 number of operations that can be performed on a key of that type.
Linus Torvalds1da177e2005-04-16 15:20:36 -070062
63 Should a type be removed from the system, all the keys of that type will
64 be invalidated.
65
66 (*) Each key has a description. This should be a printable string. The key
David Howells76d8aea2005-06-23 22:00:49 -070067 type provides an operation to perform a match between the description on a
68 key and a criterion string.
Linus Torvalds1da177e2005-04-16 15:20:36 -070069
70 (*) Each key has an owner user ID, a group ID and a permissions mask. These
71 are used to control what a process may do to a key from userspace, and
72 whether a kernel service will be able to find the key.
73
74 (*) Each key can be set to expire at a specific time by the key type's
75 instantiation function. Keys can also be immortal.
76
David Howells76d8aea2005-06-23 22:00:49 -070077 (*) Each key can have a payload. This is a quantity of data that represent the
78 actual "key". In the case of a keyring, this is a list of keys to which
79 the keyring links; in the case of a user-defined key, it's an arbitrary
80 blob of data.
Linus Torvalds1da177e2005-04-16 15:20:36 -070081
82 Having a payload is not required; and the payload can, in fact, just be a
83 value stored in the struct key itself.
84
85 When a key is instantiated, the key type's instantiation function is
86 called with a blob of data, and that then creates the key's payload in
87 some way.
88
89 Similarly, when userspace wants to read back the contents of the key, if
90 permitted, another key type operation will be called to convert the key's
91 attached payload back into a blob of data.
92
93 (*) Each key can be in one of a number of basic states:
94
David Howells76d8aea2005-06-23 22:00:49 -070095 (*) Uninstantiated. The key exists, but does not have any data attached.
96 Keys being requested from userspace will be in this state.
Linus Torvalds1da177e2005-04-16 15:20:36 -070097
98 (*) Instantiated. This is the normal state. The key is fully formed, and
99 has data attached.
100
101 (*) Negative. This is a relatively short-lived state. The key acts as a
102 note saying that a previous call out to userspace failed, and acts as
103 a throttle on key lookups. A negative key can be updated to a normal
104 state.
105
106 (*) Expired. Keys can have lifetimes set. If their lifetime is exceeded,
107 they traverse to this state. An expired key can be updated back to a
108 normal state.
109
110 (*) Revoked. A key is put in this state by userspace action. It can't be
111 found or operated upon (apart from by unlinking it).
112
113 (*) Dead. The key's type was unregistered, and so the key is now useless.
114
115
116====================
117KEY SERVICE OVERVIEW
118====================
119
120The key service provides a number of features besides keys:
121
122 (*) The key service defines two special key types:
123
124 (+) "keyring"
125
126 Keyrings are special keys that contain a list of other keys. Keyring
127 lists can be modified using various system calls. Keyrings should not
128 be given a payload when created.
129
130 (+) "user"
131
132 A key of this type has a description and a payload that are arbitrary
133 blobs of data. These can be created, updated and read by userspace,
134 and aren't intended for use by kernel services.
135
136 (*) Each process subscribes to three keyrings: a thread-specific keyring, a
137 process-specific keyring, and a session-specific keyring.
138
139 The thread-specific keyring is discarded from the child when any sort of
140 clone, fork, vfork or execve occurs. A new keyring is created only when
141 required.
142
David Howells76d8aea2005-06-23 22:00:49 -0700143 The process-specific keyring is replaced with an empty one in the child on
144 clone, fork, vfork unless CLONE_THREAD is supplied, in which case it is
145 shared. execve also discards the process's process keyring and creates a
146 new one.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147
148 The session-specific keyring is persistent across clone, fork, vfork and
149 execve, even when the latter executes a set-UID or set-GID binary. A
150 process can, however, replace its current session keyring with a new one
151 by using PR_JOIN_SESSION_KEYRING. It is permitted to request an anonymous
152 new one, or to attempt to create or join one of a specific name.
153
154 The ownership of the thread keyring changes when the real UID and GID of
155 the thread changes.
156
157 (*) Each user ID resident in the system holds two special keyrings: a user
158 specific keyring and a default user session keyring. The default session
159 keyring is initialised with a link to the user-specific keyring.
160
161 When a process changes its real UID, if it used to have no session key, it
162 will be subscribed to the default session key for the new UID.
163
164 If a process attempts to access its session key when it doesn't have one,
165 it will be subscribed to the default for its current UID.
166
167 (*) Each user has two quotas against which the keys they own are tracked. One
168 limits the total number of keys and keyrings, the other limits the total
169 amount of description and payload space that can be consumed.
170
171 The user can view information on this and other statistics through procfs
172 files.
173
174 Process-specific and thread-specific keyrings are not counted towards a
175 user's quota.
176
177 If a system call that modifies a key or keyring in some way would put the
178 user over quota, the operation is refused and error EDQUOT is returned.
179
David Howells76d8aea2005-06-23 22:00:49 -0700180 (*) There's a system call interface by which userspace programs can create and
181 manipulate keys and keyrings.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700182
David Howells76d8aea2005-06-23 22:00:49 -0700183 (*) There's a kernel interface by which services can register types and search
184 for keys.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700185
186 (*) There's a way for the a search done from the kernel to call back to
187 userspace to request a key that can't be found in a process's keyrings.
188
189 (*) An optional filesystem is available through which the key database can be
190 viewed and manipulated.
191
192
193======================
194KEY ACCESS PERMISSIONS
195======================
196
David Howells76d8aea2005-06-23 22:00:49 -0700197Keys have an owner user ID, a group access ID, and a permissions mask. The mask
David Howells664cceb2005-09-28 17:03:15 +0100198has up to eight bits each for possessor, user, group and other access. Only
David Howells29db9192005-10-30 15:02:44 -0800199six of each set of eight bits are defined. These permissions granted are:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700200
201 (*) View
202
203 This permits a key or keyring's attributes to be viewed - including key
204 type and description.
205
206 (*) Read
207
208 This permits a key's payload to be viewed or a keyring's list of linked
209 keys.
210
211 (*) Write
212
David Howells76d8aea2005-06-23 22:00:49 -0700213 This permits a key's payload to be instantiated or updated, or it allows a
214 link to be added to or removed from a keyring.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700215
216 (*) Search
217
218 This permits keyrings to be searched and keys to be found. Searches can
219 only recurse into nested keyrings that have search permission set.
220
221 (*) Link
222
223 This permits a key or keyring to be linked to. To create a link from a
224 keyring to a key, a process must have Write permission on the keyring and
225 Link permission on the key.
226
David Howells29db9192005-10-30 15:02:44 -0800227 (*) Set Attribute
228
229 This permits a key's UID, GID and permissions mask to be changed.
230
Linus Torvalds1da177e2005-04-16 15:20:36 -0700231For changing the ownership, group ID or permissions mask, being the owner of
232the key or having the sysadmin capability is sufficient.
233
234
235================
236NEW PROCFS FILES
237================
238
239Two files have been added to procfs by which an administrator can find out
240about the status of the key service:
241
242 (*) /proc/keys
243
244 This lists all the keys on the system, giving information about their
David Howells76d8aea2005-06-23 22:00:49 -0700245 type, description and permissions. The payload of the key is not available
246 this way:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700247
David Howells664cceb2005-09-28 17:03:15 +0100248 SERIAL FLAGS USAGE EXPY PERM UID GID TYPE DESCRIPTION: SUMMARY
David Howells29db9192005-10-30 15:02:44 -0800249 00000001 I----- 39 perm 1f3f0000 0 0 keyring _uid_ses.0: 1/4
250 00000002 I----- 2 perm 1f3f0000 0 0 keyring _uid.0: empty
251 00000007 I----- 1 perm 1f3f0000 0 0 keyring _pid.1: empty
252 0000018d I----- 1 perm 1f3f0000 0 0 keyring _pid.412: empty
253 000004d2 I--Q-- 1 perm 1f3f0000 32 -1 keyring _uid.32: 1/4
254 000004d3 I--Q-- 3 perm 1f3f0000 32 -1 keyring _uid_ses.32: empty
David Howells664cceb2005-09-28 17:03:15 +0100255 00000892 I--QU- 1 perm 1f000000 0 0 user metal:copper: 0
David Howells29db9192005-10-30 15:02:44 -0800256 00000893 I--Q-N 1 35s 1f3f0000 0 0 user metal:silver: 0
257 00000894 I--Q-- 1 10h 003f0000 0 0 user metal:gold: 0
Linus Torvalds1da177e2005-04-16 15:20:36 -0700258
259 The flags are:
260
261 I Instantiated
262 R Revoked
263 D Dead
264 Q Contributes to user's quota
265 U Under contruction by callback to userspace
266 N Negative key
267
268 This file must be enabled at kernel configuration time as it allows anyone
269 to list the keys database.
270
271 (*) /proc/key-users
272
273 This file lists the tracking data for each user that has at least one key
274 on the system. Such data includes quota information and statistics:
275
276 [root@andromeda root]# cat /proc/key-users
277 0: 46 45/45 1/100 13/10000
278 29: 2 2/2 2/100 40/10000
279 32: 2 2/2 2/100 40/10000
280 38: 2 2/2 2/100 40/10000
281
282 The format of each line is
283 <UID>: User ID to which this applies
284 <usage> Structure refcount
285 <inst>/<keys> Total number of keys and number instantiated
286 <keys>/<max> Key count quota
287 <bytes>/<max> Key size quota
288
289
290===============================
291USERSPACE SYSTEM CALL INTERFACE
292===============================
293
294Userspace can manipulate keys directly through three new syscalls: add_key,
295request_key and keyctl. The latter provides a number of functions for
296manipulating keys.
297
298When referring to a key directly, userspace programs should use the key's
299serial number (a positive 32-bit integer). However, there are some special
300values available for referring to special keys and keyrings that relate to the
301process making the call:
302
303 CONSTANT VALUE KEY REFERENCED
304 ============================== ====== ===========================
305 KEY_SPEC_THREAD_KEYRING -1 thread-specific keyring
306 KEY_SPEC_PROCESS_KEYRING -2 process-specific keyring
307 KEY_SPEC_SESSION_KEYRING -3 session-specific keyring
308 KEY_SPEC_USER_KEYRING -4 UID-specific keyring
309 KEY_SPEC_USER_SESSION_KEYRING -5 UID-session keyring
310 KEY_SPEC_GROUP_KEYRING -6 GID-specific keyring
David Howellsb5f545c2006-01-08 01:02:47 -0800311 KEY_SPEC_REQKEY_AUTH_KEY -7 assumed request_key()
312 authorisation key
Linus Torvalds1da177e2005-04-16 15:20:36 -0700313
314
315The main syscalls are:
316
317 (*) Create a new key of given type, description and payload and add it to the
318 nominated keyring:
319
320 key_serial_t add_key(const char *type, const char *desc,
321 const void *payload, size_t plen,
322 key_serial_t keyring);
323
324 If a key of the same type and description as that proposed already exists
325 in the keyring, this will try to update it with the given payload, or it
326 will return error EEXIST if that function is not supported by the key
David Howells76d8aea2005-06-23 22:00:49 -0700327 type. The process must also have permission to write to the key to be able
328 to update it. The new key will have all user permissions granted and no
329 group or third party permissions.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700330
David Howells76d8aea2005-06-23 22:00:49 -0700331 Otherwise, this will attempt to create a new key of the specified type and
332 description, and to instantiate it with the supplied payload and attach it
333 to the keyring. In this case, an error will be generated if the process
334 does not have permission to write to the keyring.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700335
336 The payload is optional, and the pointer can be NULL if not required by
337 the type. The payload is plen in size, and plen can be zero for an empty
338 payload.
339
David Howells76d8aea2005-06-23 22:00:49 -0700340 A new keyring can be generated by setting type "keyring", the keyring name
341 as the description (or NULL) and setting the payload to NULL.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700342
343 User defined keys can be created by specifying type "user". It is
344 recommended that a user defined key's description by prefixed with a type
345 ID and a colon, such as "krb5tgt:" for a Kerberos 5 ticket granting
346 ticket.
347
348 Any other type must have been registered with the kernel in advance by a
349 kernel service such as a filesystem.
350
351 The ID of the new or updated key is returned if successful.
352
353
354 (*) Search the process's keyrings for a key, potentially calling out to
355 userspace to create it.
356
357 key_serial_t request_key(const char *type, const char *description,
358 const char *callout_info,
359 key_serial_t dest_keyring);
360
361 This function searches all the process's keyrings in the order thread,
362 process, session for a matching key. This works very much like
363 KEYCTL_SEARCH, including the optional attachment of the discovered key to
364 a keyring.
365
366 If a key cannot be found, and if callout_info is not NULL, then
367 /sbin/request-key will be invoked in an attempt to obtain a key. The
368 callout_info string will be passed as an argument to the program.
369
David Howellsf1a9bad2005-10-07 15:04:52 +0100370 See also Documentation/keys-request-key.txt.
371
Linus Torvalds1da177e2005-04-16 15:20:36 -0700372
373The keyctl syscall functions are:
374
375 (*) Map a special key ID to a real key ID for this process:
376
377 key_serial_t keyctl(KEYCTL_GET_KEYRING_ID, key_serial_t id,
378 int create);
379
David Howells76d8aea2005-06-23 22:00:49 -0700380 The special key specified by "id" is looked up (with the key being created
381 if necessary) and the ID of the key or keyring thus found is returned if
382 it exists.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700383
384 If the key does not yet exist, the key will be created if "create" is
385 non-zero; and the error ENOKEY will be returned if "create" is zero.
386
387
388 (*) Replace the session keyring this process subscribes to with a new one:
389
390 key_serial_t keyctl(KEYCTL_JOIN_SESSION_KEYRING, const char *name);
391
392 If name is NULL, an anonymous keyring is created attached to the process
393 as its session keyring, displacing the old session keyring.
394
395 If name is not NULL, if a keyring of that name exists, the process
396 attempts to attach it as the session keyring, returning an error if that
397 is not permitted; otherwise a new keyring of that name is created and
398 attached as the session keyring.
399
400 To attach to a named keyring, the keyring must have search permission for
401 the process's ownership.
402
403 The ID of the new session keyring is returned if successful.
404
405
406 (*) Update the specified key:
407
408 long keyctl(KEYCTL_UPDATE, key_serial_t key, const void *payload,
409 size_t plen);
410
411 This will try to update the specified key with the given payload, or it
412 will return error EOPNOTSUPP if that function is not supported by the key
David Howells76d8aea2005-06-23 22:00:49 -0700413 type. The process must also have permission to write to the key to be able
414 to update it.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700415
416 The payload is of length plen, and may be absent or empty as for
417 add_key().
418
419
420 (*) Revoke a key:
421
422 long keyctl(KEYCTL_REVOKE, key_serial_t key);
423
424 This makes a key unavailable for further operations. Further attempts to
425 use the key will be met with error EKEYREVOKED, and the key will no longer
426 be findable.
427
428
429 (*) Change the ownership of a key:
430
431 long keyctl(KEYCTL_CHOWN, key_serial_t key, uid_t uid, gid_t gid);
432
David Howells76d8aea2005-06-23 22:00:49 -0700433 This function permits a key's owner and group ID to be changed. Either one
434 of uid or gid can be set to -1 to suppress that change.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700435
436 Only the superuser can change a key's owner to something other than the
437 key's current owner. Similarly, only the superuser can change a key's
438 group ID to something other than the calling process's group ID or one of
439 its group list members.
440
441
442 (*) Change the permissions mask on a key:
443
444 long keyctl(KEYCTL_SETPERM, key_serial_t key, key_perm_t perm);
445
446 This function permits the owner of a key or the superuser to change the
447 permissions mask on a key.
448
449 Only bits the available bits are permitted; if any other bits are set,
450 error EINVAL will be returned.
451
452
453 (*) Describe a key:
454
455 long keyctl(KEYCTL_DESCRIBE, key_serial_t key, char *buffer,
456 size_t buflen);
457
458 This function returns a summary of the key's attributes (but not its
459 payload data) as a string in the buffer provided.
460
461 Unless there's an error, it always returns the amount of data it could
462 produce, even if that's too big for the buffer, but it won't copy more
463 than requested to userspace. If the buffer pointer is NULL then no copy
464 will take place.
465
466 A process must have view permission on the key for this function to be
467 successful.
468
469 If successful, a string is placed in the buffer in the following format:
470
471 <type>;<uid>;<gid>;<perm>;<description>
472
473 Where type and description are strings, uid and gid are decimal, and perm
474 is hexadecimal. A NUL character is included at the end of the string if
475 the buffer is sufficiently big.
476
477 This can be parsed with
478
479 sscanf(buffer, "%[^;];%d;%d;%o;%s", type, &uid, &gid, &mode, desc);
480
481
482 (*) Clear out a keyring:
483
484 long keyctl(KEYCTL_CLEAR, key_serial_t keyring);
485
486 This function clears the list of keys attached to a keyring. The calling
487 process must have write permission on the keyring, and it must be a
488 keyring (or else error ENOTDIR will result).
489
490
491 (*) Link a key into a keyring:
492
493 long keyctl(KEYCTL_LINK, key_serial_t keyring, key_serial_t key);
494
David Howells76d8aea2005-06-23 22:00:49 -0700495 This function creates a link from the keyring to the key. The process must
496 have write permission on the keyring and must have link permission on the
497 key.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700498
David Howells76d8aea2005-06-23 22:00:49 -0700499 Should the keyring not be a keyring, error ENOTDIR will result; and if the
500 keyring is full, error ENFILE will result.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700501
502 The link procedure checks the nesting of the keyrings, returning ELOOP if
David Howells017679c2006-01-08 01:02:43 -0800503 it appears too deep or EDEADLK if the link would introduce a cycle.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700504
David Howellscab8eb52006-01-08 01:02:45 -0800505 Any links within the keyring to keys that match the new key in terms of
506 type and description will be discarded from the keyring as the new one is
507 added.
508
Linus Torvalds1da177e2005-04-16 15:20:36 -0700509
510 (*) Unlink a key or keyring from another keyring:
511
512 long keyctl(KEYCTL_UNLINK, key_serial_t keyring, key_serial_t key);
513
514 This function looks through the keyring for the first link to the
515 specified key, and removes it if found. Subsequent links to that key are
516 ignored. The process must have write permission on the keyring.
517
David Howells76d8aea2005-06-23 22:00:49 -0700518 If the keyring is not a keyring, error ENOTDIR will result; and if the key
519 is not present, error ENOENT will be the result.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700520
521
522 (*) Search a keyring tree for a key:
523
524 key_serial_t keyctl(KEYCTL_SEARCH, key_serial_t keyring,
525 const char *type, const char *description,
526 key_serial_t dest_keyring);
527
David Howells76d8aea2005-06-23 22:00:49 -0700528 This searches the keyring tree headed by the specified keyring until a key
529 is found that matches the type and description criteria. Each keyring is
530 checked for keys before recursion into its children occurs.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700531
532 The process must have search permission on the top level keyring, or else
533 error EACCES will result. Only keyrings that the process has search
534 permission on will be recursed into, and only keys and keyrings for which
535 a process has search permission can be matched. If the specified keyring
536 is not a keyring, ENOTDIR will result.
537
538 If the search succeeds, the function will attempt to link the found key
539 into the destination keyring if one is supplied (non-zero ID). All the
540 constraints applicable to KEYCTL_LINK apply in this case too.
541
542 Error ENOKEY, EKEYREVOKED or EKEYEXPIRED will be returned if the search
543 fails. On success, the resulting key ID will be returned.
544
545
546 (*) Read the payload data from a key:
547
David Howellsf1a9bad2005-10-07 15:04:52 +0100548 long keyctl(KEYCTL_READ, key_serial_t keyring, char *buffer,
549 size_t buflen);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700550
551 This function attempts to read the payload data from the specified key
552 into the buffer. The process must have read permission on the key to
553 succeed.
554
555 The returned data will be processed for presentation by the key type. For
556 instance, a keyring will return an array of key_serial_t entries
557 representing the IDs of all the keys to which it is subscribed. The user
558 defined key type will return its data as is. If a key type does not
559 implement this function, error EOPNOTSUPP will result.
560
561 As much of the data as can be fitted into the buffer will be copied to
562 userspace if the buffer pointer is not NULL.
563
David Howells76d8aea2005-06-23 22:00:49 -0700564 On a successful return, the function will always return the amount of data
565 available rather than the amount copied.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700566
567
568 (*) Instantiate a partially constructed key.
569
David Howellsf1a9bad2005-10-07 15:04:52 +0100570 long keyctl(KEYCTL_INSTANTIATE, key_serial_t key,
571 const void *payload, size_t plen,
572 key_serial_t keyring);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700573
574 If the kernel calls back to userspace to complete the instantiation of a
575 key, userspace should use this call to supply data for the key before the
576 invoked process returns, or else the key will be marked negative
577 automatically.
578
579 The process must have write access on the key to be able to instantiate
580 it, and the key must be uninstantiated.
581
582 If a keyring is specified (non-zero), the key will also be linked into
David Howells76d8aea2005-06-23 22:00:49 -0700583 that keyring, however all the constraints applying in KEYCTL_LINK apply in
584 this case too.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700585
586 The payload and plen arguments describe the payload data as for add_key().
587
588
589 (*) Negatively instantiate a partially constructed key.
590
David Howellsf1a9bad2005-10-07 15:04:52 +0100591 long keyctl(KEYCTL_NEGATE, key_serial_t key,
592 unsigned timeout, key_serial_t keyring);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700593
594 If the kernel calls back to userspace to complete the instantiation of a
595 key, userspace should use this call mark the key as negative before the
596 invoked process returns if it is unable to fulfil the request.
597
598 The process must have write access on the key to be able to instantiate
599 it, and the key must be uninstantiated.
600
601 If a keyring is specified (non-zero), the key will also be linked into
David Howells76d8aea2005-06-23 22:00:49 -0700602 that keyring, however all the constraints applying in KEYCTL_LINK apply in
603 this case too.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700604
605
David Howells3e301482005-06-23 22:00:56 -0700606 (*) Set the default request-key destination keyring.
607
608 long keyctl(KEYCTL_SET_REQKEY_KEYRING, int reqkey_defl);
609
610 This sets the default keyring to which implicitly requested keys will be
611 attached for this thread. reqkey_defl should be one of these constants:
612
613 CONSTANT VALUE NEW DEFAULT KEYRING
614 ====================================== ====== =======================
615 KEY_REQKEY_DEFL_NO_CHANGE -1 No change
616 KEY_REQKEY_DEFL_DEFAULT 0 Default[1]
617 KEY_REQKEY_DEFL_THREAD_KEYRING 1 Thread keyring
618 KEY_REQKEY_DEFL_PROCESS_KEYRING 2 Process keyring
619 KEY_REQKEY_DEFL_SESSION_KEYRING 3 Session keyring
620 KEY_REQKEY_DEFL_USER_KEYRING 4 User keyring
621 KEY_REQKEY_DEFL_USER_SESSION_KEYRING 5 User session keyring
622 KEY_REQKEY_DEFL_GROUP_KEYRING 6 Group keyring
623
624 The old default will be returned if successful and error EINVAL will be
625 returned if reqkey_defl is not one of the above values.
626
627 The default keyring can be overridden by the keyring indicated to the
628 request_key() system call.
629
630 Note that this setting is inherited across fork/exec.
631
632 [1] The default default is: the thread keyring if there is one, otherwise
633 the process keyring if there is one, otherwise the session keyring if
634 there is one, otherwise the user default session keyring.
635
636
David Howells017679c2006-01-08 01:02:43 -0800637 (*) Set the timeout on a key.
638
639 long keyctl(KEYCTL_SET_TIMEOUT, key_serial_t key, unsigned timeout);
640
641 This sets or clears the timeout on a key. The timeout can be 0 to clear
642 the timeout or a number of seconds to set the expiry time that far into
643 the future.
644
645 The process must have attribute modification access on a key to set its
646 timeout. Timeouts may not be set with this function on negative, revoked
647 or expired keys.
648
649
David Howellsb5f545c2006-01-08 01:02:47 -0800650 (*) Assume the authority granted to instantiate a key
651
652 long keyctl(KEYCTL_ASSUME_AUTHORITY, key_serial_t key);
653
654 This assumes or divests the authority required to instantiate the
655 specified key. Authority can only be assumed if the thread has the
656 authorisation key associated with the specified key in its keyrings
657 somewhere.
658
659 Once authority is assumed, searches for keys will also search the
660 requester's keyrings using the requester's security label, UID, GID and
661 groups.
662
663 If the requested authority is unavailable, error EPERM will be returned,
664 likewise if the authority has been revoked because the target key is
665 already instantiated.
666
667 If the specified key is 0, then any assumed authority will be divested.
668
669 The assumed authorititive key is inherited across fork and exec.
670
671
Linus Torvalds1da177e2005-04-16 15:20:36 -0700672===============
673KERNEL SERVICES
674===============
675
676The kernel services for key managment are fairly simple to deal with. They can
677be broken down into two areas: keys and key types.
678
679Dealing with keys is fairly straightforward. Firstly, the kernel service
680registers its type, then it searches for a key of that type. It should retain
681the key as long as it has need of it, and then it should release it. For a
David Howells76d8aea2005-06-23 22:00:49 -0700682filesystem or device file, a search would probably be performed during the open
683call, and the key released upon close. How to deal with conflicting keys due to
684two different users opening the same file is left to the filesystem author to
685solve.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700686
David Howells664cceb2005-09-28 17:03:15 +0100687Note that there are two different types of pointers to keys that may be
688encountered:
689
690 (*) struct key *
691
692 This simply points to the key structure itself. Key structures will be at
693 least four-byte aligned.
694
695 (*) key_ref_t
696
697 This is equivalent to a struct key *, but the least significant bit is set
698 if the caller "possesses" the key. By "possession" it is meant that the
699 calling processes has a searchable link to the key from one of its
700 keyrings. There are three functions for dealing with these:
701
702 key_ref_t make_key_ref(const struct key *key,
703 unsigned long possession);
704
705 struct key *key_ref_to_ptr(const key_ref_t key_ref);
706
707 unsigned long is_key_possessed(const key_ref_t key_ref);
708
709 The first function constructs a key reference from a key pointer and
710 possession information (which must be 0 or 1 and not any other value).
711
712 The second function retrieves the key pointer from a reference and the
713 third retrieves the possession flag.
714
David Howells76d8aea2005-06-23 22:00:49 -0700715When accessing a key's payload contents, certain precautions must be taken to
716prevent access vs modification races. See the section "Notes on accessing
717payload contents" for more information.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700718
719(*) To search for a key, call:
720
721 struct key *request_key(const struct key_type *type,
722 const char *description,
723 const char *callout_string);
724
725 This is used to request a key or keyring with a description that matches
726 the description specified according to the key type's match function. This
727 permits approximate matching to occur. If callout_string is not NULL, then
728 /sbin/request-key will be invoked in an attempt to obtain the key from
729 userspace. In that case, callout_string will be passed as an argument to
730 the program.
731
732 Should the function fail error ENOKEY, EKEYEXPIRED or EKEYREVOKED will be
733 returned.
734
David Howells3e301482005-06-23 22:00:56 -0700735 If successful, the key will have been attached to the default keyring for
736 implicitly obtained request-key keys, as set by KEYCTL_SET_REQKEY_KEYRING.
737
David Howellsf1a9bad2005-10-07 15:04:52 +0100738 See also Documentation/keys-request-key.txt.
739
Linus Torvalds1da177e2005-04-16 15:20:36 -0700740
741(*) When it is no longer required, the key should be released using:
742
743 void key_put(struct key *key);
744
David Howells664cceb2005-09-28 17:03:15 +0100745 Or:
746
747 void key_ref_put(key_ref_t key_ref);
748
749 These can be called from interrupt context. If CONFIG_KEYS is not set then
Linus Torvalds1da177e2005-04-16 15:20:36 -0700750 the argument will not be parsed.
751
752
753(*) Extra references can be made to a key by calling the following function:
754
755 struct key *key_get(struct key *key);
756
757 These need to be disposed of by calling key_put() when they've been
758 finished with. The key pointer passed in will be returned. If the pointer
759 is NULL or CONFIG_KEYS is not set then the key will not be dereferenced and
760 no increment will take place.
761
762
763(*) A key's serial number can be obtained by calling:
764
765 key_serial_t key_serial(struct key *key);
766
767 If key is NULL or if CONFIG_KEYS is not set then 0 will be returned (in the
768 latter case without parsing the argument).
769
770
771(*) If a keyring was found in the search, this can be further searched by:
772
David Howells664cceb2005-09-28 17:03:15 +0100773 key_ref_t keyring_search(key_ref_t keyring_ref,
774 const struct key_type *type,
775 const char *description)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776
777 This searches the keyring tree specified for a matching key. Error ENOKEY
David Howells664cceb2005-09-28 17:03:15 +0100778 is returned upon failure (use IS_ERR/PTR_ERR to determine). If successful,
779 the returned key will need to be released.
780
781 The possession attribute from the keyring reference is used to control
782 access through the permissions mask and is propagated to the returned key
783 reference pointer if successful.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700784
785
786(*) To check the validity of a key, this function can be called:
787
788 int validate_key(struct key *key);
789
790 This checks that the key in question hasn't expired or and hasn't been
791 revoked. Should the key be invalid, error EKEYEXPIRED or EKEYREVOKED will
792 be returned. If the key is NULL or if CONFIG_KEYS is not set then 0 will be
793 returned (in the latter case without parsing the argument).
794
795
796(*) To register a key type, the following function should be called:
797
798 int register_key_type(struct key_type *type);
799
800 This will return error EEXIST if a type of the same name is already
801 present.
802
803
804(*) To unregister a key type, call:
805
806 void unregister_key_type(struct key_type *type);
807
808
David Howells76d8aea2005-06-23 22:00:49 -0700809===================================
810NOTES ON ACCESSING PAYLOAD CONTENTS
811===================================
812
813The simplest payload is just a number in key->payload.value. In this case,
814there's no need to indulge in RCU or locking when accessing the payload.
815
816More complex payload contents must be allocated and a pointer to them set in
817key->payload.data. One of the following ways must be selected to access the
818data:
819
David Howells664cceb2005-09-28 17:03:15 +0100820 (1) Unmodifiable key type.
David Howells76d8aea2005-06-23 22:00:49 -0700821
822 If the key type does not have a modify method, then the key's payload can
823 be accessed without any form of locking, provided that it's known to be
824 instantiated (uninstantiated keys cannot be "found").
825
826 (2) The key's semaphore.
827
828 The semaphore could be used to govern access to the payload and to control
829 the payload pointer. It must be write-locked for modifications and would
830 have to be read-locked for general access. The disadvantage of doing this
831 is that the accessor may be required to sleep.
832
833 (3) RCU.
834
835 RCU must be used when the semaphore isn't already held; if the semaphore
836 is held then the contents can't change under you unexpectedly as the
837 semaphore must still be used to serialise modifications to the key. The
838 key management code takes care of this for the key type.
839
840 However, this means using:
841
842 rcu_read_lock() ... rcu_dereference() ... rcu_read_unlock()
843
844 to read the pointer, and:
845
846 rcu_dereference() ... rcu_assign_pointer() ... call_rcu()
847
848 to set the pointer and dispose of the old contents after a grace period.
849 Note that only the key type should ever modify a key's payload.
850
851 Furthermore, an RCU controlled payload must hold a struct rcu_head for the
852 use of call_rcu() and, if the payload is of variable size, the length of
853 the payload. key->datalen cannot be relied upon to be consistent with the
854 payload just dereferenced if the key's semaphore is not held.
855
856
Linus Torvalds1da177e2005-04-16 15:20:36 -0700857===================
858DEFINING A KEY TYPE
859===================
860
861A kernel service may want to define its own key type. For instance, an AFS
862filesystem might want to define a Kerberos 5 ticket key type. To do this, it
863author fills in a struct key_type and registers it with the system.
864
865The structure has a number of fields, some of which are mandatory:
866
867 (*) const char *name
868
869 The name of the key type. This is used to translate a key type name
870 supplied by userspace into a pointer to the structure.
871
872
873 (*) size_t def_datalen
874
875 This is optional - it supplies the default payload data length as
876 contributed to the quota. If the key type's payload is always or almost
877 always the same size, then this is a more efficient way to do things.
878
879 The data length (and quota) on a particular key can always be changed
880 during instantiation or update by calling:
881
882 int key_payload_reserve(struct key *key, size_t datalen);
883
David Howells76d8aea2005-06-23 22:00:49 -0700884 With the revised data length. Error EDQUOT will be returned if this is not
885 viable.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700886
887
888 (*) int (*instantiate)(struct key *key, const void *data, size_t datalen);
889
890 This method is called to attach a payload to a key during construction.
David Howells76d8aea2005-06-23 22:00:49 -0700891 The payload attached need not bear any relation to the data passed to this
892 function.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700893
894 If the amount of data attached to the key differs from the size in
895 keytype->def_datalen, then key_payload_reserve() should be called.
896
897 This method does not have to lock the key in order to attach a payload.
898 The fact that KEY_FLAG_INSTANTIATED is not set in key->flags prevents
899 anything else from gaining access to the key.
900
David Howells76d8aea2005-06-23 22:00:49 -0700901 It is safe to sleep in this method.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902
903
Linus Torvalds1da177e2005-04-16 15:20:36 -0700904 (*) int (*update)(struct key *key, const void *data, size_t datalen);
905
David Howells76d8aea2005-06-23 22:00:49 -0700906 If this type of key can be updated, then this method should be provided.
907 It is called to update a key's payload from the blob of data provided.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700908
909 key_payload_reserve() should be called if the data length might change
David Howells76d8aea2005-06-23 22:00:49 -0700910 before any changes are actually made. Note that if this succeeds, the type
911 is committed to changing the key because it's already been altered, so all
912 memory allocation must be done first.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913
David Howells76d8aea2005-06-23 22:00:49 -0700914 The key will have its semaphore write-locked before this method is called,
915 but this only deters other writers; any changes to the key's payload must
916 be made under RCU conditions, and call_rcu() must be used to dispose of
917 the old payload.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700918
David Howells76d8aea2005-06-23 22:00:49 -0700919 key_payload_reserve() should be called before the changes are made, but
920 after all allocations and other potentially failing function calls are
921 made.
922
923 It is safe to sleep in this method.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700924
925
926 (*) int (*match)(const struct key *key, const void *desc);
927
928 This method is called to match a key against a description. It should
929 return non-zero if the two match, zero if they don't.
930
931 This method should not need to lock the key in any way. The type and
932 description can be considered invariant, and the payload should not be
933 accessed (the key may not yet be instantiated).
934
935 It is not safe to sleep in this method; the caller may hold spinlocks.
936
937
938 (*) void (*destroy)(struct key *key);
939
David Howells76d8aea2005-06-23 22:00:49 -0700940 This method is optional. It is called to discard the payload data on a key
941 when it is being destroyed.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700942
David Howells76d8aea2005-06-23 22:00:49 -0700943 This method does not need to lock the key to access the payload; it can
944 consider the key as being inaccessible at this time. Note that the key's
945 type may have been changed before this function is called.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700946
947 It is not safe to sleep in this method; the caller may hold spinlocks.
948
949
950 (*) void (*describe)(const struct key *key, struct seq_file *p);
951
952 This method is optional. It is called during /proc/keys reading to
953 summarise a key's description and payload in text form.
954
David Howells76d8aea2005-06-23 22:00:49 -0700955 This method will be called with the RCU read lock held. rcu_dereference()
956 should be used to read the payload pointer if the payload is to be
957 accessed. key->datalen cannot be trusted to stay consistent with the
958 contents of the payload.
959
960 The description will not change, though the key's state may.
961
962 It is not safe to sleep in this method; the RCU read lock is held by the
963 caller.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700964
965
966 (*) long (*read)(const struct key *key, char __user *buffer, size_t buflen);
967
968 This method is optional. It is called by KEYCTL_READ to translate the
David Howells76d8aea2005-06-23 22:00:49 -0700969 key's payload into something a blob of data for userspace to deal with.
970 Ideally, the blob should be in the same format as that passed in to the
971 instantiate and update methods.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700972
973 If successful, the blob size that could be produced should be returned
974 rather than the size copied.
975
David Howells76d8aea2005-06-23 22:00:49 -0700976 This method will be called with the key's semaphore read-locked. This will
977 prevent the key's payload changing. It is not necessary to use RCU locking
978 when accessing the key's payload. It is safe to sleep in this method, such
979 as might happen when the userspace buffer is accessed.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700980
981
982============================
983REQUEST-KEY CALLBACK SERVICE
984============================
985
986To create a new key, the kernel will attempt to execute the following command
987line:
988
989 /sbin/request-key create <key> <uid> <gid> \
990 <threadring> <processring> <sessionring> <callout_info>
991
992<key> is the key being constructed, and the three keyrings are the process
993keyrings from the process that caused the search to be issued. These are
994included for two reasons:
995
996 (1) There may be an authentication token in one of the keyrings that is
997 required to obtain the key, eg: a Kerberos Ticket-Granting Ticket.
998
999 (2) The new key should probably be cached in one of these rings.
1000
1001This program should set it UID and GID to those specified before attempting to
1002access any more keys. It may then look around for a user specific process to
1003hand the request off to (perhaps a path held in placed in another key by, for
1004example, the KDE desktop manager).
1005
1006The program (or whatever it calls) should finish construction of the key by
1007calling KEYCTL_INSTANTIATE, which also permits it to cache the key in one of
1008the keyrings (probably the session ring) before returning. Alternatively, the
1009key can be marked as negative with KEYCTL_NEGATE; this also permits the key to
1010be cached in one of the keyrings.
1011
1012If it returns with the key remaining in the unconstructed state, the key will
1013be marked as being negative, it will be added to the session keyring, and an
1014error will be returned to the key requestor.
1015
David Howells76d8aea2005-06-23 22:00:49 -07001016Supplementary information may be provided from whoever or whatever invoked this
1017service. This will be passed as the <callout_info> parameter. If no such
Linus Torvalds1da177e2005-04-16 15:20:36 -07001018information was made available, then "-" will be passed as this parameter
1019instead.
1020
1021
1022Similarly, the kernel may attempt to update an expired or a soon to expire key
1023by executing:
1024
1025 /sbin/request-key update <key> <uid> <gid> \
1026 <threadring> <processring> <sessionring>
1027
1028In this case, the program isn't required to actually attach the key to a ring;
1029the rings are provided for reference.