linux/Documentation/security/keys.txt
<<
>>
Prefs
   1                         ============================
   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 and 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        - SELinux support
  23        - New procfs files
  24        - Userspace system call interface
  25        - Kernel services
  26        - Notes on accessing payload contents
  27        - Defining a key type
  28        - Request-key callback service
  29        - Garbage collection
  30
  31
  32============
  33KEY OVERVIEW
  34============
  35
  36In this context, keys represent units of cryptographic data, authentication
  37tokens, keyrings, etc.. These are represented in the kernel by struct key.
  38
  39Each key has a number of attributes:
  40
  41        - A serial number.
  42        - A type.
  43        - A description (for matching a key in a search).
  44        - Access control information.
  45        - An expiry time.
  46        - A payload.
  47        - State.
  48
  49
  50 (*) Each key is issued a serial number of type key_serial_t that is unique for
  51     the lifetime of that key. All serial numbers are positive non-zero 32-bit
  52     integers.
  53
  54     Userspace programs can use a key's serial numbers as a way to gain access
  55     to it, subject to permission checking.
  56
  57 (*) Each key is of a defined "type". Types must be registered inside the
  58     kernel by a kernel service (such as a filesystem) before keys of that type
  59     can be added or used. Userspace programs cannot define new types directly.
  60
  61     Key types are represented in the kernel by struct key_type. This defines a
  62     number of operations that can be performed on a key of that type.
  63
  64     Should a type be removed from the system, all the keys of that type will
  65     be invalidated.
  66
  67 (*) Each key has a description. This should be a printable string. The key
  68     type provides an operation to perform a match between the description on a
  69     key and a criterion string.
  70
  71 (*) Each key has an owner user ID, a group ID and a permissions mask. These
  72     are used to control what a process may do to a key from userspace, and
  73     whether a kernel service will be able to find the key.
  74
  75 (*) Each key can be set to expire at a specific time by the key type's
  76     instantiation function. Keys can also be immortal.
  77
  78 (*) Each key can have a payload. This is a quantity of data that represent the
  79     actual "key". In the case of a keyring, this is a list of keys to which
  80     the keyring links; in the case of a user-defined key, it's an arbitrary
  81     blob of data.
  82
  83     Having a payload is not required; and the payload can, in fact, just be a
  84     value stored in the struct key itself.
  85
  86     When a key is instantiated, the key type's instantiation function is
  87     called with a blob of data, and that then creates the key's payload in
  88     some way.
  89
  90     Similarly, when userspace wants to read back the contents of the key, if
  91     permitted, another key type operation will be called to convert the key's
  92     attached payload back into a blob of data.
  93
  94 (*) Each key can be in one of a number of basic states:
  95
  96     (*) Uninstantiated. The key exists, but does not have any data attached.
  97         Keys being requested from userspace will be in this state.
  98
  99     (*) Instantiated. This is the normal state. The key is fully formed, and
 100         has data attached.
 101
 102     (*) Negative. This is a relatively short-lived state. The key acts as a
 103         note saying that a previous call out to userspace failed, and acts as
 104         a throttle on key lookups. A negative key can be updated to a normal
 105         state.
 106
 107     (*) Expired. Keys can have lifetimes set. If their lifetime is exceeded,
 108         they traverse to this state. An expired key can be updated back to a
 109         normal state.
 110
 111     (*) Revoked. A key is put in this state by userspace action. It can't be
 112         found or operated upon (apart from by unlinking it).
 113
 114     (*) Dead. The key's type was unregistered, and so the key is now useless.
 115
 116Keys in the last three states are subject to garbage collection.  See the
 117section on "Garbage collection".
 118
 119
 120====================
 121KEY SERVICE OVERVIEW
 122====================
 123
 124The key service provides a number of features besides keys:
 125
 126 (*) The key service defines three special key types:
 127
 128     (+) "keyring"
 129
 130         Keyrings are special keys that contain a list of other keys. Keyring
 131         lists can be modified using various system calls. Keyrings should not
 132         be given a payload when created.
 133
 134     (+) "user"
 135
 136         A key of this type has a description and a payload that are arbitrary
 137         blobs of data. These can be created, updated and read by userspace,
 138         and aren't intended for use by kernel services.
 139
 140     (+) "logon"
 141
 142         Like a "user" key, a "logon" key has a payload that is an arbitrary
 143         blob of data. It is intended as a place to store secrets which are
 144         accessible to the kernel but not to userspace programs.
 145
 146         The description can be arbitrary, but must be prefixed with a non-zero
 147         length string that describes the key "subclass". The subclass is
 148         separated from the rest of the description by a ':'. "logon" keys can
 149         be created and updated from userspace, but the payload is only
 150         readable from kernel space.
 151
 152 (*) Each process subscribes to three keyrings: a thread-specific keyring, a
 153     process-specific keyring, and a session-specific keyring.
 154
 155     The thread-specific keyring is discarded from the child when any sort of
 156     clone, fork, vfork or execve occurs. A new keyring is created only when
 157     required.
 158
 159     The process-specific keyring is replaced with an empty one in the child on
 160     clone, fork, vfork unless CLONE_THREAD is supplied, in which case it is
 161     shared. execve also discards the process's process keyring and creates a
 162     new one.
 163
 164     The session-specific keyring is persistent across clone, fork, vfork and
 165     execve, even when the latter executes a set-UID or set-GID binary. A
 166     process can, however, replace its current session keyring with a new one
 167     by using PR_JOIN_SESSION_KEYRING. It is permitted to request an anonymous
 168     new one, or to attempt to create or join one of a specific name.
 169
 170     The ownership of the thread keyring changes when the real UID and GID of
 171     the thread changes.
 172
 173 (*) Each user ID resident in the system holds two special keyrings: a user
 174     specific keyring and a default user session keyring. The default session
 175     keyring is initialised with a link to the user-specific keyring.
 176
 177     When a process changes its real UID, if it used to have no session key, it
 178     will be subscribed to the default session key for the new UID.
 179
 180     If a process attempts to access its session key when it doesn't have one,
 181     it will be subscribed to the default for its current UID.
 182
 183 (*) Each user has two quotas against which the keys they own are tracked. One
 184     limits the total number of keys and keyrings, the other limits the total
 185     amount of description and payload space that can be consumed.
 186
 187     The user can view information on this and other statistics through procfs
 188     files.  The root user may also alter the quota limits through sysctl files
 189     (see the section "New procfs files").
 190
 191     Process-specific and thread-specific keyrings are not counted towards a
 192     user's quota.
 193
 194     If a system call that modifies a key or keyring in some way would put the
 195     user over quota, the operation is refused and error EDQUOT is returned.
 196
 197 (*) There's a system call interface by which userspace programs can create and
 198     manipulate keys and keyrings.
 199
 200 (*) There's a kernel interface by which services can register types and search
 201     for keys.
 202
 203 (*) There's a way for the a search done from the kernel to call back to
 204     userspace to request a key that can't be found in a process's keyrings.
 205
 206 (*) An optional filesystem is available through which the key database can be
 207     viewed and manipulated.
 208
 209
 210======================
 211KEY ACCESS PERMISSIONS
 212======================
 213
 214Keys have an owner user ID, a group access ID, and a permissions mask. The mask
 215has up to eight bits each for possessor, user, group and other access. Only
 216six of each set of eight bits are defined. These permissions granted are:
 217
 218 (*) View
 219
 220     This permits a key or keyring's attributes to be viewed - including key
 221     type and description.
 222
 223 (*) Read
 224
 225     This permits a key's payload to be viewed or a keyring's list of linked
 226     keys.
 227
 228 (*) Write
 229
 230     This permits a key's payload to be instantiated or updated, or it allows a
 231     link to be added to or removed from a keyring.
 232
 233 (*) Search
 234
 235     This permits keyrings to be searched and keys to be found. Searches can
 236     only recurse into nested keyrings that have search permission set.
 237
 238 (*) Link
 239
 240     This permits a key or keyring to be linked to. To create a link from a
 241     keyring to a key, a process must have Write permission on the keyring and
 242     Link permission on the key.
 243
 244 (*) Set Attribute
 245
 246     This permits a key's UID, GID and permissions mask to be changed.
 247
 248For changing the ownership, group ID or permissions mask, being the owner of
 249the key or having the sysadmin capability is sufficient.
 250
 251
 252===============
 253SELINUX SUPPORT
 254===============
 255
 256The security class "key" has been added to SELinux so that mandatory access
 257controls can be applied to keys created within various contexts.  This support
 258is preliminary, and is likely to change quite significantly in the near future.
 259Currently, all of the basic permissions explained above are provided in SELinux
 260as well; SELinux is simply invoked after all basic permission checks have been
 261performed.
 262
 263The value of the file /proc/self/attr/keycreate influences the labeling of
 264newly-created keys.  If the contents of that file correspond to an SELinux
 265security context, then the key will be assigned that context.  Otherwise, the
 266key will be assigned the current context of the task that invoked the key
 267creation request.  Tasks must be granted explicit permission to assign a
 268particular context to newly-created keys, using the "create" permission in the
 269key security class.
 270
 271The default keyrings associated with users will be labeled with the default
 272context of the user if and only if the login programs have been instrumented to
 273properly initialize keycreate during the login process.  Otherwise, they will
 274be labeled with the context of the login program itself.
 275
 276Note, however, that the default keyrings associated with the root user are
 277labeled with the default kernel context, since they are created early in the
 278boot process, before root has a chance to log in.
 279
 280The keyrings associated with new threads are each labeled with the context of
 281their associated thread, and both session and process keyrings are handled
 282similarly.
 283
 284
 285================
 286NEW PROCFS FILES
 287================
 288
 289Two files have been added to procfs by which an administrator can find out
 290about the status of the key service:
 291
 292 (*) /proc/keys
 293
 294     This lists the keys that are currently viewable by the task reading the
 295     file, giving information about their type, description and permissions.
 296     It is not possible to view the payload of the key this way, though some
 297     information about it may be given.
 298
 299     The only keys included in the list are those that grant View permission to
 300     the reading process whether or not it possesses them.  Note that LSM
 301     security checks are still performed, and may further filter out keys that
 302     the current process is not authorised to view.
 303
 304     The contents of the file look like this:
 305
 306        SERIAL   FLAGS  USAGE EXPY PERM     UID   GID   TYPE      DESCRIPTION: SUMMARY
 307        00000001 I-----    39 perm 1f3f0000     0     0 keyring   _uid_ses.0: 1/4
 308        00000002 I-----     2 perm 1f3f0000     0     0 keyring   _uid.0: empty
 309        00000007 I-----     1 perm 1f3f0000     0     0 keyring   _pid.1: empty
 310        0000018d I-----     1 perm 1f3f0000     0     0 keyring   _pid.412: empty
 311        000004d2 I--Q--     1 perm 1f3f0000    32    -1 keyring   _uid.32: 1/4
 312        000004d3 I--Q--     3 perm 1f3f0000    32    -1 keyring   _uid_ses.32: empty
 313        00000892 I--QU-     1 perm 1f000000     0     0 user      metal:copper: 0
 314        00000893 I--Q-N     1  35s 1f3f0000     0     0 user      metal:silver: 0
 315        00000894 I--Q--     1  10h 003f0000     0     0 user      metal:gold: 0
 316
 317     The flags are:
 318
 319        I       Instantiated
 320        R       Revoked
 321        D       Dead
 322        Q       Contributes to user's quota
 323        U       Under construction by callback to userspace
 324        N       Negative key
 325
 326     This file must be enabled at kernel configuration time as it allows anyone
 327     to list the keys database.
 328
 329 (*) /proc/key-users
 330
 331     This file lists the tracking data for each user that has at least one key
 332     on the system.  Such data includes quota information and statistics:
 333
 334        [root@andromeda root]# cat /proc/key-users
 335        0:     46 45/45 1/100 13/10000
 336        29:     2 2/2 2/100 40/10000
 337        32:     2 2/2 2/100 40/10000
 338        38:     2 2/2 2/100 40/10000
 339
 340     The format of each line is
 341        <UID>:                  User ID to which this applies
 342        <usage>                 Structure refcount
 343        <inst>/<keys>           Total number of keys and number instantiated
 344        <keys>/<max>            Key count quota
 345        <bytes>/<max>           Key size quota
 346
 347
 348Four new sysctl files have been added also for the purpose of controlling the
 349quota limits on keys:
 350
 351 (*) /proc/sys/kernel/keys/root_maxkeys
 352     /proc/sys/kernel/keys/root_maxbytes
 353
 354     These files hold the maximum number of keys that root may have and the
 355     maximum total number of bytes of data that root may have stored in those
 356     keys.
 357
 358 (*) /proc/sys/kernel/keys/maxkeys
 359     /proc/sys/kernel/keys/maxbytes
 360
 361     These files hold the maximum number of keys that each non-root user may
 362     have and the maximum total number of bytes of data that each of those
 363     users may have stored in their keys.
 364
 365Root may alter these by writing each new limit as a decimal number string to
 366the appropriate file.
 367
 368
 369===============================
 370USERSPACE SYSTEM CALL INTERFACE
 371===============================
 372
 373Userspace can manipulate keys directly through three new syscalls: add_key,
 374request_key and keyctl. The latter provides a number of functions for
 375manipulating keys.
 376
 377When referring to a key directly, userspace programs should use the key's
 378serial number (a positive 32-bit integer). However, there are some special
 379values available for referring to special keys and keyrings that relate to the
 380process making the call:
 381
 382        CONSTANT                        VALUE   KEY REFERENCED
 383        ==============================  ======  ===========================
 384        KEY_SPEC_THREAD_KEYRING         -1      thread-specific keyring
 385        KEY_SPEC_PROCESS_KEYRING        -2      process-specific keyring
 386        KEY_SPEC_SESSION_KEYRING        -3      session-specific keyring
 387        KEY_SPEC_USER_KEYRING           -4      UID-specific keyring
 388        KEY_SPEC_USER_SESSION_KEYRING   -5      UID-session keyring
 389        KEY_SPEC_GROUP_KEYRING          -6      GID-specific keyring
 390        KEY_SPEC_REQKEY_AUTH_KEY        -7      assumed request_key()
 391                                                  authorisation key
 392
 393
 394The main syscalls are:
 395
 396 (*) Create a new key of given type, description and payload and add it to the
 397     nominated keyring:
 398
 399        key_serial_t add_key(const char *type, const char *desc,
 400                             const void *payload, size_t plen,
 401                             key_serial_t keyring);
 402
 403     If a key of the same type and description as that proposed already exists
 404     in the keyring, this will try to update it with the given payload, or it
 405     will return error EEXIST if that function is not supported by the key
 406     type. The process must also have permission to write to the key to be able
 407     to update it. The new key will have all user permissions granted and no
 408     group or third party permissions.
 409
 410     Otherwise, this will attempt to create a new key of the specified type and
 411     description, and to instantiate it with the supplied payload and attach it
 412     to the keyring. In this case, an error will be generated if the process
 413     does not have permission to write to the keyring.
 414
 415     The payload is optional, and the pointer can be NULL if not required by
 416     the type. The payload is plen in size, and plen can be zero for an empty
 417     payload.
 418
 419     A new keyring can be generated by setting type "keyring", the keyring name
 420     as the description (or NULL) and setting the payload to NULL.
 421
 422     User defined keys can be created by specifying type "user". It is
 423     recommended that a user defined key's description by prefixed with a type
 424     ID and a colon, such as "krb5tgt:" for a Kerberos 5 ticket granting
 425     ticket.
 426
 427     Any other type must have been registered with the kernel in advance by a
 428     kernel service such as a filesystem.
 429
 430     The ID of the new or updated key is returned if successful.
 431
 432
 433 (*) Search the process's keyrings for a key, potentially calling out to
 434     userspace to create it.
 435
 436        key_serial_t request_key(const char *type, const char *description,
 437                                 const char *callout_info,
 438                                 key_serial_t dest_keyring);
 439
 440     This function searches all the process's keyrings in the order thread,
 441     process, session for a matching key. This works very much like
 442     KEYCTL_SEARCH, including the optional attachment of the discovered key to
 443     a keyring.
 444
 445     If a key cannot be found, and if callout_info is not NULL, then
 446     /sbin/request-key will be invoked in an attempt to obtain a key. The
 447     callout_info string will be passed as an argument to the program.
 448
 449     See also Documentation/security/keys-request-key.txt.
 450
 451
 452The keyctl syscall functions are:
 453
 454 (*) Map a special key ID to a real key ID for this process:
 455
 456        key_serial_t keyctl(KEYCTL_GET_KEYRING_ID, key_serial_t id,
 457                            int create);
 458
 459     The special key specified by "id" is looked up (with the key being created
 460     if necessary) and the ID of the key or keyring thus found is returned if
 461     it exists.
 462
 463     If the key does not yet exist, the key will be created if "create" is
 464     non-zero; and the error ENOKEY will be returned if "create" is zero.
 465
 466
 467 (*) Replace the session keyring this process subscribes to with a new one:
 468
 469        key_serial_t keyctl(KEYCTL_JOIN_SESSION_KEYRING, const char *name);
 470
 471     If name is NULL, an anonymous keyring is created attached to the process
 472     as its session keyring, displacing the old session keyring.
 473
 474     If name is not NULL, if a keyring of that name exists, the process
 475     attempts to attach it as the session keyring, returning an error if that
 476     is not permitted; otherwise a new keyring of that name is created and
 477     attached as the session keyring.
 478
 479     To attach to a named keyring, the keyring must have search permission for
 480     the process's ownership.
 481
 482     The ID of the new session keyring is returned if successful.
 483
 484
 485 (*) Update the specified key:
 486
 487        long keyctl(KEYCTL_UPDATE, key_serial_t key, const void *payload,
 488                    size_t plen);
 489
 490     This will try to update the specified key with the given payload, or it
 491     will return error EOPNOTSUPP if that function is not supported by the key
 492     type. The process must also have permission to write to the key to be able
 493     to update it.
 494
 495     The payload is of length plen, and may be absent or empty as for
 496     add_key().
 497
 498
 499 (*) Revoke a key:
 500
 501        long keyctl(KEYCTL_REVOKE, key_serial_t key);
 502
 503     This makes a key unavailable for further operations. Further attempts to
 504     use the key will be met with error EKEYREVOKED, and the key will no longer
 505     be findable.
 506
 507
 508 (*) Change the ownership of a key:
 509
 510        long keyctl(KEYCTL_CHOWN, key_serial_t key, uid_t uid, gid_t gid);
 511
 512     This function permits a key's owner and group ID to be changed. Either one
 513     of uid or gid can be set to -1 to suppress that change.
 514
 515     Only the superuser can change a key's owner to something other than the
 516     key's current owner. Similarly, only the superuser can change a key's
 517     group ID to something other than the calling process's group ID or one of
 518     its group list members.
 519
 520
 521 (*) Change the permissions mask on a key:
 522
 523        long keyctl(KEYCTL_SETPERM, key_serial_t key, key_perm_t perm);
 524
 525     This function permits the owner of a key or the superuser to change the
 526     permissions mask on a key.
 527
 528     Only bits the available bits are permitted; if any other bits are set,
 529     error EINVAL will be returned.
 530
 531
 532 (*) Describe a key:
 533
 534        long keyctl(KEYCTL_DESCRIBE, key_serial_t key, char *buffer,
 535                    size_t buflen);
 536
 537     This function returns a summary of the key's attributes (but not its
 538     payload data) as a string in the buffer provided.
 539
 540     Unless there's an error, it always returns the amount of data it could
 541     produce, even if that's too big for the buffer, but it won't copy more
 542     than requested to userspace. If the buffer pointer is NULL then no copy
 543     will take place.
 544
 545     A process must have view permission on the key for this function to be
 546     successful.
 547
 548     If successful, a string is placed in the buffer in the following format:
 549
 550        <type>;<uid>;<gid>;<perm>;<description>
 551
 552     Where type and description are strings, uid and gid are decimal, and perm
 553     is hexadecimal. A NUL character is included at the end of the string if
 554     the buffer is sufficiently big.
 555
 556     This can be parsed with
 557
 558        sscanf(buffer, "%[^;];%d;%d;%o;%s", type, &uid, &gid, &mode, desc);
 559
 560
 561 (*) Clear out a keyring:
 562
 563        long keyctl(KEYCTL_CLEAR, key_serial_t keyring);
 564
 565     This function clears the list of keys attached to a keyring. The calling
 566     process must have write permission on the keyring, and it must be a
 567     keyring (or else error ENOTDIR will result).
 568
 569     This function can also be used to clear special kernel keyrings if they
 570     are appropriately marked if the user has CAP_SYS_ADMIN capability.  The
 571     DNS resolver cache keyring is an example of this.
 572
 573
 574 (*) Link a key into a keyring:
 575
 576        long keyctl(KEYCTL_LINK, key_serial_t keyring, key_serial_t key);
 577
 578     This function creates a link from the keyring to the key. The process must
 579     have write permission on the keyring and must have link permission on the
 580     key.
 581
 582     Should the keyring not be a keyring, error ENOTDIR will result; and if the
 583     keyring is full, error ENFILE will result.
 584
 585     The link procedure checks the nesting of the keyrings, returning ELOOP if
 586     it appears too deep or EDEADLK if the link would introduce a cycle.
 587
 588     Any links within the keyring to keys that match the new key in terms of
 589     type and description will be discarded from the keyring as the new one is
 590     added.
 591
 592
 593 (*) Unlink a key or keyring from another keyring:
 594
 595        long keyctl(KEYCTL_UNLINK, key_serial_t keyring, key_serial_t key);
 596
 597     This function looks through the keyring for the first link to the
 598     specified key, and removes it if found. Subsequent links to that key are
 599     ignored. The process must have write permission on the keyring.
 600
 601     If the keyring is not a keyring, error ENOTDIR will result; and if the key
 602     is not present, error ENOENT will be the result.
 603
 604
 605 (*) Search a keyring tree for a key:
 606
 607        key_serial_t keyctl(KEYCTL_SEARCH, key_serial_t keyring,
 608                            const char *type, const char *description,
 609                            key_serial_t dest_keyring);
 610
 611     This searches the keyring tree headed by the specified keyring until a key
 612     is found that matches the type and description criteria. Each keyring is
 613     checked for keys before recursion into its children occurs.
 614
 615     The process must have search permission on the top level keyring, or else
 616     error EACCES will result. Only keyrings that the process has search
 617     permission on will be recursed into, and only keys and keyrings for which
 618     a process has search permission can be matched. If the specified keyring
 619     is not a keyring, ENOTDIR will result.
 620
 621     If the search succeeds, the function will attempt to link the found key
 622     into the destination keyring if one is supplied (non-zero ID). All the
 623     constraints applicable to KEYCTL_LINK apply in this case too.
 624
 625     Error ENOKEY, EKEYREVOKED or EKEYEXPIRED will be returned if the search
 626     fails. On success, the resulting key ID will be returned.
 627
 628
 629 (*) Read the payload data from a key:
 630
 631        long keyctl(KEYCTL_READ, key_serial_t keyring, char *buffer,
 632                    size_t buflen);
 633
 634     This function attempts to read the payload data from the specified key
 635     into the buffer. The process must have read permission on the key to
 636     succeed.
 637
 638     The returned data will be processed for presentation by the key type. For
 639     instance, a keyring will return an array of key_serial_t entries
 640     representing the IDs of all the keys to which it is subscribed. The user
 641     defined key type will return its data as is. If a key type does not
 642     implement this function, error EOPNOTSUPP will result.
 643
 644     As much of the data as can be fitted into the buffer will be copied to
 645     userspace if the buffer pointer is not NULL.
 646
 647     On a successful return, the function will always return the amount of data
 648     available rather than the amount copied.
 649
 650
 651 (*) Instantiate a partially constructed key.
 652
 653        long keyctl(KEYCTL_INSTANTIATE, key_serial_t key,
 654                    const void *payload, size_t plen,
 655                    key_serial_t keyring);
 656        long keyctl(KEYCTL_INSTANTIATE_IOV, key_serial_t key,
 657                    const struct iovec *payload_iov, unsigned ioc,
 658                    key_serial_t keyring);
 659
 660     If the kernel calls back to userspace to complete the instantiation of a
 661     key, userspace should use this call to supply data for the key before the
 662     invoked process returns, or else the key will be marked negative
 663     automatically.
 664
 665     The process must have write access on the key to be able to instantiate
 666     it, and the key must be uninstantiated.
 667
 668     If a keyring is specified (non-zero), the key will also be linked into
 669     that keyring, however all the constraints applying in KEYCTL_LINK apply in
 670     this case too.
 671
 672     The payload and plen arguments describe the payload data as for add_key().
 673
 674     The payload_iov and ioc arguments describe the payload data in an iovec
 675     array instead of a single buffer.
 676
 677
 678 (*) Negatively instantiate a partially constructed key.
 679
 680        long keyctl(KEYCTL_NEGATE, key_serial_t key,
 681                    unsigned timeout, key_serial_t keyring);
 682        long keyctl(KEYCTL_REJECT, key_serial_t key,
 683                    unsigned timeout, unsigned error, key_serial_t keyring);
 684
 685     If the kernel calls back to userspace to complete the instantiation of a
 686     key, userspace should use this call mark the key as negative before the
 687     invoked process returns if it is unable to fulfill the request.
 688
 689     The process must have write access on the key to be able to instantiate
 690     it, and the key must be uninstantiated.
 691
 692     If a keyring is specified (non-zero), the key will also be linked into
 693     that keyring, however all the constraints applying in KEYCTL_LINK apply in
 694     this case too.
 695
 696     If the key is rejected, future searches for it will return the specified
 697     error code until the rejected key expires.  Negating the key is the same
 698     as rejecting the key with ENOKEY as the error code.
 699
 700
 701 (*) Set the default request-key destination keyring.
 702
 703        long keyctl(KEYCTL_SET_REQKEY_KEYRING, int reqkey_defl);
 704
 705     This sets the default keyring to which implicitly requested keys will be
 706     attached for this thread. reqkey_defl should be one of these constants:
 707
 708        CONSTANT                                VALUE   NEW DEFAULT KEYRING
 709        ======================================  ======  =======================
 710        KEY_REQKEY_DEFL_NO_CHANGE               -1      No change
 711        KEY_REQKEY_DEFL_DEFAULT                 0       Default[1]
 712        KEY_REQKEY_DEFL_THREAD_KEYRING          1       Thread keyring
 713        KEY_REQKEY_DEFL_PROCESS_KEYRING         2       Process keyring
 714        KEY_REQKEY_DEFL_SESSION_KEYRING         3       Session keyring
 715        KEY_REQKEY_DEFL_USER_KEYRING            4       User keyring
 716        KEY_REQKEY_DEFL_USER_SESSION_KEYRING    5       User session keyring
 717        KEY_REQKEY_DEFL_GROUP_KEYRING           6       Group keyring
 718
 719     The old default will be returned if successful and error EINVAL will be
 720     returned if reqkey_defl is not one of the above values.
 721
 722     The default keyring can be overridden by the keyring indicated to the
 723     request_key() system call.
 724
 725     Note that this setting is inherited across fork/exec.
 726
 727     [1] The default is: the thread keyring if there is one, otherwise
 728     the process keyring if there is one, otherwise the session keyring if
 729     there is one, otherwise the user default session keyring.
 730
 731
 732 (*) Set the timeout on a key.
 733
 734        long keyctl(KEYCTL_SET_TIMEOUT, key_serial_t key, unsigned timeout);
 735
 736     This sets or clears the timeout on a key. The timeout can be 0 to clear
 737     the timeout or a number of seconds to set the expiry time that far into
 738     the future.
 739
 740     The process must have attribute modification access on a key to set its
 741     timeout. Timeouts may not be set with this function on negative, revoked
 742     or expired keys.
 743
 744
 745 (*) Assume the authority granted to instantiate a key
 746
 747        long keyctl(KEYCTL_ASSUME_AUTHORITY, key_serial_t key);
 748
 749     This assumes or divests the authority required to instantiate the
 750     specified key. Authority can only be assumed if the thread has the
 751     authorisation key associated with the specified key in its keyrings
 752     somewhere.
 753
 754     Once authority is assumed, searches for keys will also search the
 755     requester's keyrings using the requester's security label, UID, GID and
 756     groups.
 757
 758     If the requested authority is unavailable, error EPERM will be returned,
 759     likewise if the authority has been revoked because the target key is
 760     already instantiated.
 761
 762     If the specified key is 0, then any assumed authority will be divested.
 763
 764     The assumed authoritative key is inherited across fork and exec.
 765
 766
 767 (*) Get the LSM security context attached to a key.
 768
 769        long keyctl(KEYCTL_GET_SECURITY, key_serial_t key, char *buffer,
 770                    size_t buflen)
 771
 772     This function returns a string that represents the LSM security context
 773     attached to a key in the buffer provided.
 774
 775     Unless there's an error, it always returns the amount of data it could
 776     produce, even if that's too big for the buffer, but it won't copy more
 777     than requested to userspace. If the buffer pointer is NULL then no copy
 778     will take place.
 779
 780     A NUL character is included at the end of the string if the buffer is
 781     sufficiently big.  This is included in the returned count.  If no LSM is
 782     in force then an empty string will be returned.
 783
 784     A process must have view permission on the key for this function to be
 785     successful.
 786
 787
 788 (*) Install the calling process's session keyring on its parent.
 789
 790        long keyctl(KEYCTL_SESSION_TO_PARENT);
 791
 792     This functions attempts to install the calling process's session keyring
 793     on to the calling process's parent, replacing the parent's current session
 794     keyring.
 795
 796     The calling process must have the same ownership as its parent, the
 797     keyring must have the same ownership as the calling process, the calling
 798     process must have LINK permission on the keyring and the active LSM module
 799     mustn't deny permission, otherwise error EPERM will be returned.
 800
 801     Error ENOMEM will be returned if there was insufficient memory to complete
 802     the operation, otherwise 0 will be returned to indicate success.
 803
 804     The keyring will be replaced next time the parent process leaves the
 805     kernel and resumes executing userspace.
 806
 807
 808 (*) Invalidate a key.
 809
 810        long keyctl(KEYCTL_INVALIDATE, key_serial_t key);
 811
 812     This function marks a key as being invalidated and then wakes up the
 813     garbage collector.  The garbage collector immediately removes invalidated
 814     keys from all keyrings and deletes the key when its reference count
 815     reaches zero.
 816
 817     Keys that are marked invalidated become invisible to normal key operations
 818     immediately, though they are still visible in /proc/keys until deleted
 819     (they're marked with an 'i' flag).
 820
 821     A process must have search permission on the key for this function to be
 822     successful.
 823
 824
 825===============
 826KERNEL SERVICES
 827===============
 828
 829The kernel services for key management are fairly simple to deal with. They can
 830be broken down into two areas: keys and key types.
 831
 832Dealing with keys is fairly straightforward. Firstly, the kernel service
 833registers its type, then it searches for a key of that type. It should retain
 834the key as long as it has need of it, and then it should release it. For a
 835filesystem or device file, a search would probably be performed during the open
 836call, and the key released upon close. How to deal with conflicting keys due to
 837two different users opening the same file is left to the filesystem author to
 838solve.
 839
 840To access the key manager, the following header must be #included:
 841
 842        <linux/key.h>
 843
 844Specific key types should have a header file under include/keys/ that should be
 845used to access that type.  For keys of type "user", for example, that would be:
 846
 847        <keys/user-type.h>
 848
 849Note that there are two different types of pointers to keys that may be
 850encountered:
 851
 852 (*) struct key *
 853
 854     This simply points to the key structure itself. Key structures will be at
 855     least four-byte aligned.
 856
 857 (*) key_ref_t
 858
 859     This is equivalent to a struct key *, but the least significant bit is set
 860     if the caller "possesses" the key. By "possession" it is meant that the
 861     calling processes has a searchable link to the key from one of its
 862     keyrings. There are three functions for dealing with these:
 863
 864        key_ref_t make_key_ref(const struct key *key,
 865                               unsigned long possession);
 866
 867        struct key *key_ref_to_ptr(const key_ref_t key_ref);
 868
 869        unsigned long is_key_possessed(const key_ref_t key_ref);
 870
 871     The first function constructs a key reference from a key pointer and
 872     possession information (which must be 0 or 1 and not any other value).
 873
 874     The second function retrieves the key pointer from a reference and the
 875     third retrieves the possession flag.
 876
 877When accessing a key's payload contents, certain precautions must be taken to
 878prevent access vs modification races. See the section "Notes on accessing
 879payload contents" for more information.
 880
 881(*) To search for a key, call:
 882
 883        struct key *request_key(const struct key_type *type,
 884                                const char *description,
 885                                const char *callout_info);
 886
 887    This is used to request a key or keyring with a description that matches
 888    the description specified according to the key type's match function. This
 889    permits approximate matching to occur. If callout_string is not NULL, then
 890    /sbin/request-key will be invoked in an attempt to obtain the key from
 891    userspace. In that case, callout_string will be passed as an argument to
 892    the program.
 893
 894    Should the function fail error ENOKEY, EKEYEXPIRED or EKEYREVOKED will be
 895    returned.
 896
 897    If successful, the key will have been attached to the default keyring for
 898    implicitly obtained request-key keys, as set by KEYCTL_SET_REQKEY_KEYRING.
 899
 900    See also Documentation/security/keys-request-key.txt.
 901
 902
 903(*) To search for a key, passing auxiliary data to the upcaller, call:
 904
 905        struct key *request_key_with_auxdata(const struct key_type *type,
 906                                             const char *description,
 907                                             const void *callout_info,
 908                                             size_t callout_len,
 909                                             void *aux);
 910
 911    This is identical to request_key(), except that the auxiliary data is
 912    passed to the key_type->request_key() op if it exists, and the callout_info
 913    is a blob of length callout_len, if given (the length may be 0).
 914
 915
 916(*) A key can be requested asynchronously by calling one of:
 917
 918        struct key *request_key_async(const struct key_type *type,
 919                                      const char *description,
 920                                      const void *callout_info,
 921                                      size_t callout_len);
 922
 923    or:
 924
 925        struct key *request_key_async_with_auxdata(const struct key_type *type,
 926                                                   const char *description,
 927                                                   const char *callout_info,
 928                                                   size_t callout_len,
 929                                                   void *aux);
 930
 931    which are asynchronous equivalents of request_key() and
 932    request_key_with_auxdata() respectively.
 933
 934    These two functions return with the key potentially still under
 935    construction.  To wait for construction completion, the following should be
 936    called:
 937
 938        int wait_for_key_construction(struct key *key, bool intr);
 939
 940    The function will wait for the key to finish being constructed and then
 941    invokes key_validate() to return an appropriate value to indicate the state
 942    of the key (0 indicates the key is usable).
 943
 944    If intr is true, then the wait can be interrupted by a signal, in which
 945    case error ERESTARTSYS will be returned.
 946
 947
 948(*) When it is no longer required, the key should be released using:
 949
 950        void key_put(struct key *key);
 951
 952    Or:
 953
 954        void key_ref_put(key_ref_t key_ref);
 955
 956    These can be called from interrupt context. If CONFIG_KEYS is not set then
 957    the argument will not be parsed.
 958
 959
 960(*) Extra references can be made to a key by calling the following function:
 961
 962        struct key *key_get(struct key *key);
 963
 964    These need to be disposed of by calling key_put() when they've been
 965    finished with. The key pointer passed in will be returned. If the pointer
 966    is NULL or CONFIG_KEYS is not set then the key will not be dereferenced and
 967    no increment will take place.
 968
 969
 970(*) A key's serial number can be obtained by calling:
 971
 972        key_serial_t key_serial(struct key *key);
 973
 974    If key is NULL or if CONFIG_KEYS is not set then 0 will be returned (in the
 975    latter case without parsing the argument).
 976
 977
 978(*) If a keyring was found in the search, this can be further searched by:
 979
 980        key_ref_t keyring_search(key_ref_t keyring_ref,
 981                                 const struct key_type *type,
 982                                 const char *description)
 983
 984    This searches the keyring tree specified for a matching key. Error ENOKEY
 985    is returned upon failure (use IS_ERR/PTR_ERR to determine). If successful,
 986    the returned key will need to be released.
 987
 988    The possession attribute from the keyring reference is used to control
 989    access through the permissions mask and is propagated to the returned key
 990    reference pointer if successful.
 991
 992
 993(*) To check the validity of a key, this function can be called:
 994
 995        int validate_key(struct key *key);
 996
 997    This checks that the key in question hasn't expired or and hasn't been
 998    revoked. Should the key be invalid, error EKEYEXPIRED or EKEYREVOKED will
 999    be returned. If the key is NULL or if CONFIG_KEYS is not set then 0 will be
1000    returned (in the latter case without parsing the argument).
1001
1002
1003(*) To register a key type, the following function should be called:
1004
1005        int register_key_type(struct key_type *type);
1006
1007    This will return error EEXIST if a type of the same name is already
1008    present.
1009
1010
1011(*) To unregister a key type, call:
1012
1013        void unregister_key_type(struct key_type *type);
1014
1015
1016Under some circumstances, it may be desirable to deal with a bundle of keys.
1017The facility provides access to the keyring type for managing such a bundle:
1018
1019        struct key_type key_type_keyring;
1020
1021This can be used with a function such as request_key() to find a specific
1022keyring in a process's keyrings.  A keyring thus found can then be searched
1023with keyring_search().  Note that it is not possible to use request_key() to
1024search a specific keyring, so using keyrings in this way is of limited utility.
1025
1026
1027===================================
1028NOTES ON ACCESSING PAYLOAD CONTENTS
1029===================================
1030
1031The simplest payload is just a number in key->payload.value. In this case,
1032there's no need to indulge in RCU or locking when accessing the payload.
1033
1034More complex payload contents must be allocated and a pointer to them set in
1035key->payload.data. One of the following ways must be selected to access the
1036data:
1037
1038 (1) Unmodifiable key type.
1039
1040     If the key type does not have a modify method, then the key's payload can
1041     be accessed without any form of locking, provided that it's known to be
1042     instantiated (uninstantiated keys cannot be "found").
1043
1044 (2) The key's semaphore.
1045
1046     The semaphore could be used to govern access to the payload and to control
1047     the payload pointer. It must be write-locked for modifications and would
1048     have to be read-locked for general access. The disadvantage of doing this
1049     is that the accessor may be required to sleep.
1050
1051 (3) RCU.
1052
1053     RCU must be used when the semaphore isn't already held; if the semaphore
1054     is held then the contents can't change under you unexpectedly as the
1055     semaphore must still be used to serialise modifications to the key. The
1056     key management code takes care of this for the key type.
1057
1058     However, this means using:
1059
1060        rcu_read_lock() ... rcu_dereference() ... rcu_read_unlock()
1061
1062     to read the pointer, and:
1063
1064        rcu_dereference() ... rcu_assign_pointer() ... call_rcu()
1065
1066     to set the pointer and dispose of the old contents after a grace period.
1067     Note that only the key type should ever modify a key's payload.
1068
1069     Furthermore, an RCU controlled payload must hold a struct rcu_head for the
1070     use of call_rcu() and, if the payload is of variable size, the length of
1071     the payload. key->datalen cannot be relied upon to be consistent with the
1072     payload just dereferenced if the key's semaphore is not held.
1073
1074
1075===================
1076DEFINING A KEY TYPE
1077===================
1078
1079A kernel service may want to define its own key type. For instance, an AFS
1080filesystem might want to define a Kerberos 5 ticket key type. To do this, it
1081author fills in a key_type struct and registers it with the system.
1082
1083Source files that implement key types should include the following header file:
1084
1085        <linux/key-type.h>
1086
1087The structure has a number of fields, some of which are mandatory:
1088
1089 (*) const char *name
1090
1091     The name of the key type. This is used to translate a key type name
1092     supplied by userspace into a pointer to the structure.
1093
1094
1095 (*) size_t def_datalen
1096
1097     This is optional - it supplies the default payload data length as
1098     contributed to the quota. If the key type's payload is always or almost
1099     always the same size, then this is a more efficient way to do things.
1100
1101     The data length (and quota) on a particular key can always be changed
1102     during instantiation or update by calling:
1103
1104        int key_payload_reserve(struct key *key, size_t datalen);
1105
1106     With the revised data length. Error EDQUOT will be returned if this is not
1107     viable.
1108
1109
1110 (*) int (*vet_description)(const char *description);
1111
1112     This optional method is called to vet a key description.  If the key type
1113     doesn't approve of the key description, it may return an error, otherwise
1114     it should return 0.
1115
1116
1117 (*) int (*instantiate)(struct key *key, const void *data, size_t datalen);
1118
1119     This method is called to attach a payload to a key during construction.
1120     The payload attached need not bear any relation to the data passed to this
1121     function.
1122
1123     If the amount of data attached to the key differs from the size in
1124     keytype->def_datalen, then key_payload_reserve() should be called.
1125
1126     This method does not have to lock the key in order to attach a payload.
1127     The fact that KEY_FLAG_INSTANTIATED is not set in key->flags prevents
1128     anything else from gaining access to the key.
1129
1130     It is safe to sleep in this method.
1131
1132
1133 (*) int (*update)(struct key *key, const void *data, size_t datalen);
1134
1135     If this type of key can be updated, then this method should be provided.
1136     It is called to update a key's payload from the blob of data provided.
1137
1138     key_payload_reserve() should be called if the data length might change
1139     before any changes are actually made. Note that if this succeeds, the type
1140     is committed to changing the key because it's already been altered, so all
1141     memory allocation must be done first.
1142
1143     The key will have its semaphore write-locked before this method is called,
1144     but this only deters other writers; any changes to the key's payload must
1145     be made under RCU conditions, and call_rcu() must be used to dispose of
1146     the old payload.
1147
1148     key_payload_reserve() should be called before the changes are made, but
1149     after all allocations and other potentially failing function calls are
1150     made.
1151
1152     It is safe to sleep in this method.
1153
1154
1155 (*) int (*match)(const struct key *key, const void *desc);
1156
1157     This method is called to match a key against a description. It should
1158     return non-zero if the two match, zero if they don't.
1159
1160     This method should not need to lock the key in any way. The type and
1161     description can be considered invariant, and the payload should not be
1162     accessed (the key may not yet be instantiated).
1163
1164     It is not safe to sleep in this method; the caller may hold spinlocks.
1165
1166
1167 (*) void (*revoke)(struct key *key);
1168
1169     This method is optional.  It is called to discard part of the payload
1170     data upon a key being revoked.  The caller will have the key semaphore
1171     write-locked.
1172
1173     It is safe to sleep in this method, though care should be taken to avoid
1174     a deadlock against the key semaphore.
1175
1176
1177 (*) void (*destroy)(struct key *key);
1178
1179     This method is optional. It is called to discard the payload data on a key
1180     when it is being destroyed.
1181
1182     This method does not need to lock the key to access the payload; it can
1183     consider the key as being inaccessible at this time. Note that the key's
1184     type may have been changed before this function is called.
1185
1186     It is not safe to sleep in this method; the caller may hold spinlocks.
1187
1188
1189 (*) void (*describe)(const struct key *key, struct seq_file *p);
1190
1191     This method is optional. It is called during /proc/keys reading to
1192     summarise a key's description and payload in text form.
1193
1194     This method will be called with the RCU read lock held. rcu_dereference()
1195     should be used to read the payload pointer if the payload is to be
1196     accessed. key->datalen cannot be trusted to stay consistent with the
1197     contents of the payload.
1198
1199     The description will not change, though the key's state may.
1200
1201     It is not safe to sleep in this method; the RCU read lock is held by the
1202     caller.
1203
1204
1205 (*) long (*read)(const struct key *key, char __user *buffer, size_t buflen);
1206
1207     This method is optional. It is called by KEYCTL_READ to translate the
1208     key's payload into something a blob of data for userspace to deal with.
1209     Ideally, the blob should be in the same format as that passed in to the
1210     instantiate and update methods.
1211
1212     If successful, the blob size that could be produced should be returned
1213     rather than the size copied.
1214
1215     This method will be called with the key's semaphore read-locked. This will
1216     prevent the key's payload changing. It is not necessary to use RCU locking
1217     when accessing the key's payload. It is safe to sleep in this method, such
1218     as might happen when the userspace buffer is accessed.
1219
1220
1221 (*) int (*request_key)(struct key_construction *cons, const char *op,
1222                        void *aux);
1223
1224     This method is optional.  If provided, request_key() and friends will
1225     invoke this function rather than upcalling to /sbin/request-key to operate
1226     upon a key of this type.
1227
1228     The aux parameter is as passed to request_key_async_with_auxdata() and
1229     similar or is NULL otherwise.  Also passed are the construction record for
1230     the key to be operated upon and the operation type (currently only
1231     "create").
1232
1233     This method is permitted to return before the upcall is complete, but the
1234     following function must be called under all circumstances to complete the
1235     instantiation process, whether or not it succeeds, whether or not there's
1236     an error:
1237
1238        void complete_request_key(struct key_construction *cons, int error);
1239
1240     The error parameter should be 0 on success, -ve on error.  The
1241     construction record is destroyed by this action and the authorisation key
1242     will be revoked.  If an error is indicated, the key under construction
1243     will be negatively instantiated if it wasn't already instantiated.
1244
1245     If this method returns an error, that error will be returned to the
1246     caller of request_key*().  complete_request_key() must be called prior to
1247     returning.
1248
1249     The key under construction and the authorisation key can be found in the
1250     key_construction struct pointed to by cons:
1251
1252     (*) struct key *key;
1253
1254         The key under construction.
1255
1256     (*) struct key *authkey;
1257
1258         The authorisation key.
1259
1260
1261============================
1262REQUEST-KEY CALLBACK SERVICE
1263============================
1264
1265To create a new key, the kernel will attempt to execute the following command
1266line:
1267
1268        /sbin/request-key create <key> <uid> <gid> \
1269                <threadring> <processring> <sessionring> <callout_info>
1270
1271<key> is the key being constructed, and the three keyrings are the process
1272keyrings from the process that caused the search to be issued. These are
1273included for two reasons:
1274
1275  (1) There may be an authentication token in one of the keyrings that is
1276      required to obtain the key, eg: a Kerberos Ticket-Granting Ticket.
1277
1278  (2) The new key should probably be cached in one of these rings.
1279
1280This program should set it UID and GID to those specified before attempting to
1281access any more keys. It may then look around for a user specific process to
1282hand the request off to (perhaps a path held in placed in another key by, for
1283example, the KDE desktop manager).
1284
1285The program (or whatever it calls) should finish construction of the key by
1286calling KEYCTL_INSTANTIATE or KEYCTL_INSTANTIATE_IOV, which also permits it to
1287cache the key in one of the keyrings (probably the session ring) before
1288returning.  Alternatively, the key can be marked as negative with KEYCTL_NEGATE
1289or KEYCTL_REJECT; this also permits the key to be cached in one of the
1290keyrings.
1291
1292If it returns with the key remaining in the unconstructed state, the key will
1293be marked as being negative, it will be added to the session keyring, and an
1294error will be returned to the key requestor.
1295
1296Supplementary information may be provided from whoever or whatever invoked this
1297service. This will be passed as the <callout_info> parameter. If no such
1298information was made available, then "-" will be passed as this parameter
1299instead.
1300
1301
1302Similarly, the kernel may attempt to update an expired or a soon to expire key
1303by executing:
1304
1305        /sbin/request-key update <key> <uid> <gid> \
1306                <threadring> <processring> <sessionring>
1307
1308In this case, the program isn't required to actually attach the key to a ring;
1309the rings are provided for reference.
1310
1311
1312==================
1313GARBAGE COLLECTION
1314==================
1315
1316Dead keys (for which the type has been removed) will be automatically unlinked
1317from those keyrings that point to them and deleted as soon as possible by a
1318background garbage collector.
1319
1320Similarly, revoked and expired keys will be garbage collected, but only after a
1321certain amount of time has passed.  This time is set as a number of seconds in:
1322
1323        /proc/sys/kernel/keys/gc_delay
1324
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.