1=====================================
2Filesystem-level encryption (fscrypt)
3=====================================
4
5Introduction
6============
7
8fscrypt is a library which filesystems can hook into to support
9transparent encryption of files and directories.
10
11Note: "fscrypt" in this document refers to the kernel-level portion,
12implemented in ``fs/crypto/``, as opposed to the userspace tool
13`fscrypt <https://github.com/google/fscrypt>`_.  This document only
14covers the kernel-level portion.  For command-line examples of how to
15use encryption, see the documentation for the userspace tool `fscrypt
16<https://github.com/google/fscrypt>`_.  Also, it is recommended to use
17the fscrypt userspace tool, or other existing userspace tools such as
18`fscryptctl <https://github.com/google/fscryptctl>`_ or `Android's key
19management system
20<https://source.android.com/security/encryption/file-based>`_, over
21using the kernel's API directly.  Using existing tools reduces the
22chance of introducing your own security bugs.  (Nevertheless, for
23completeness this documentation covers the kernel's API anyway.)
24
25Unlike dm-crypt, fscrypt operates at the filesystem level rather than
26at the block device level.  This allows it to encrypt different files
27with different keys and to have unencrypted files on the same
28filesystem.  This is useful for multi-user systems where each user's
29data-at-rest needs to be cryptographically isolated from the others.
30However, except for filenames, fscrypt does not encrypt filesystem
31metadata.
32
33Unlike eCryptfs, which is a stacked filesystem, fscrypt is integrated
34directly into supported filesystems --- currently ext4, F2FS, and
35UBIFS.  This allows encrypted files to be read and written without
36caching both the decrypted and encrypted pages in the pagecache,
37thereby nearly halving the memory used and bringing it in line with
38unencrypted files.  Similarly, half as many dentries and inodes are
39needed.  eCryptfs also limits encrypted filenames to 143 bytes,
40causing application compatibility issues; fscrypt allows the full 255
41bytes (NAME_MAX).  Finally, unlike eCryptfs, the fscrypt API can be
42used by unprivileged users, with no need to mount anything.
43
44fscrypt does not support encrypting files in-place.  Instead, it
45supports marking an empty directory as encrypted.  Then, after
46userspace provides the key, all regular files, directories, and
47symbolic links created in that directory tree are transparently
48encrypted.
49
50Threat model
51============
52
53Offline attacks
54---------------
55
56Provided that userspace chooses a strong encryption key, fscrypt
57protects the confidentiality of file contents and filenames in the
58event of a single point-in-time permanent offline compromise of the
59block device content.  fscrypt does not protect the confidentiality of
60non-filename metadata, e.g. file sizes, file permissions, file
61timestamps, and extended attributes.  Also, the existence and location
62of holes (unallocated blocks which logically contain all zeroes) in
63files is not protected.
64
65fscrypt is not guaranteed to protect confidentiality or authenticity
66if an attacker is able to manipulate the filesystem offline prior to
67an authorized user later accessing the filesystem.
68
69Online attacks
70--------------
71
72fscrypt (and storage encryption in general) can only provide limited
73protection, if any at all, against online attacks.  In detail:
74
75Side-channel attacks
76~~~~~~~~~~~~~~~~~~~~
77
78fscrypt is only resistant to side-channel attacks, such as timing or
79electromagnetic attacks, to the extent that the underlying Linux
80Cryptographic API algorithms or inline encryption hardware are.  If a
81vulnerable algorithm is used, such as a table-based implementation of
82AES, it may be possible for an attacker to mount a side channel attack
83against the online system.  Side channel attacks may also be mounted
84against applications consuming decrypted data.
85
86Unauthorized file access
87~~~~~~~~~~~~~~~~~~~~~~~~
88
89After an encryption key has been added, fscrypt does not hide the
90plaintext file contents or filenames from other users on the same
91system.  Instead, existing access control mechanisms such as file mode
92bits, POSIX ACLs, LSMs, or namespaces should be used for this purpose.
93
94(For the reasoning behind this, understand that while the key is
95added, the confidentiality of the data, from the perspective of the
96system itself, is *not* protected by the mathematical properties of
97encryption but rather only by the correctness of the kernel.
98Therefore, any encryption-specific access control checks would merely
99be enforced by kernel *code* and therefore would be largely redundant
100with the wide variety of access control mechanisms already available.)
101
102Kernel memory compromise
103~~~~~~~~~~~~~~~~~~~~~~~~
104
105An attacker who compromises the system enough to read from arbitrary
106memory, e.g. by mounting a physical attack or by exploiting a kernel
107security vulnerability, can compromise all encryption keys that are
108currently in use.
109
110However, fscrypt allows encryption keys to be removed from the kernel,
111which may protect them from later compromise.
112
113In more detail, the FS_IOC_REMOVE_ENCRYPTION_KEY ioctl (or the
114FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS ioctl) can wipe a master
115encryption key from kernel memory.  If it does so, it will also try to
116evict all cached inodes which had been "unlocked" using the key,
117thereby wiping their per-file keys and making them once again appear
118"locked", i.e. in ciphertext or encrypted form.
119
120However, these ioctls have some limitations:
121
122- Per-file keys for in-use files will *not* be removed or wiped.
123  Therefore, for maximum effect, userspace should close the relevant
124  encrypted files and directories before removing a master key, as
125  well as kill any processes whose working directory is in an affected
126  encrypted directory.
127
128- The kernel cannot magically wipe copies of the master key(s) that
129  userspace might have as well.  Therefore, userspace must wipe all
130  copies of the master key(s) it makes as well; normally this should
131  be done immediately after FS_IOC_ADD_ENCRYPTION_KEY, without waiting
132  for FS_IOC_REMOVE_ENCRYPTION_KEY.  Naturally, the same also applies
133  to all higher levels in the key hierarchy.  Userspace should also
134  follow other security precautions such as mlock()ing memory
135  containing keys to prevent it from being swapped out.
136
137- In general, decrypted contents and filenames in the kernel VFS
138  caches are freed but not wiped.  Therefore, portions thereof may be
139  recoverable from freed memory, even after the corresponding key(s)
140  were wiped.  To partially solve this, you can set
141  CONFIG_PAGE_POISONING=y in your kernel config and add page_poison=1
142  to your kernel command line.  However, this has a performance cost.
143
144- Secret keys might still exist in CPU registers, in crypto
145  accelerator hardware (if used by the crypto API to implement any of
146  the algorithms), or in other places not explicitly considered here.
147
148Limitations of v1 policies
149~~~~~~~~~~~~~~~~~~~~~~~~~~
150
151v1 encryption policies have some weaknesses with respect to online
152attacks:
153
154- There is no verification that the provided master key is correct.
155  Therefore, a malicious user can temporarily associate the wrong key
156  with another user's encrypted files to which they have read-only
157  access.  Because of filesystem caching, the wrong key will then be
158  used by the other user's accesses to those files, even if the other
159  user has the correct key in their own keyring.  This violates the
160  meaning of "read-only access".
161
162- A compromise of a per-file key also compromises the master key from
163  which it was derived.
164
165- Non-root users cannot securely remove encryption keys.
166
167All the above problems are fixed with v2 encryption policies.  For
168this reason among others, it is recommended to use v2 encryption
169policies on all new encrypted directories.
170
171Key hierarchy
172=============
173
174Master Keys
175-----------
176
177Each encrypted directory tree is protected by a *master key*.  Master
178keys can be up to 64 bytes long, and must be at least as long as the
179greater of the security strength of the contents and filenames
180encryption modes being used.  For example, if any AES-256 mode is
181used, the master key must be at least 256 bits, i.e. 32 bytes.  A
182stricter requirement applies if the key is used by a v1 encryption
183policy and AES-256-XTS is used; such keys must be 64 bytes.
184
185To "unlock" an encrypted directory tree, userspace must provide the
186appropriate master key.  There can be any number of master keys, each
187of which protects any number of directory trees on any number of
188filesystems.
189
190Master keys must be real cryptographic keys, i.e. indistinguishable
191from random bytestrings of the same length.  This implies that users
192**must not** directly use a password as a master key, zero-pad a
193shorter key, or repeat a shorter key.  Security cannot be guaranteed
194if userspace makes any such error, as the cryptographic proofs and
195analysis would no longer apply.
196
197Instead, users should generate master keys either using a
198cryptographically secure random number generator, or by using a KDF
199(Key Derivation Function).  The kernel does not do any key stretching;
200therefore, if userspace derives the key from a low-entropy secret such
201as a passphrase, it is critical that a KDF designed for this purpose
202be used, such as scrypt, PBKDF2, or Argon2.
203
204Key derivation function
205-----------------------
206
207With one exception, fscrypt never uses the master key(s) for
208encryption directly.  Instead, they are only used as input to a KDF
209(Key Derivation Function) to derive the actual keys.
210
211The KDF used for a particular master key differs depending on whether
212the key is used for v1 encryption policies or for v2 encryption
213policies.  Users **must not** use the same key for both v1 and v2
214encryption policies.  (No real-world attack is currently known on this
215specific case of key reuse, but its security cannot be guaranteed
216since the cryptographic proofs and analysis would no longer apply.)
217
218For v1 encryption policies, the KDF only supports deriving per-file
219encryption keys.  It works by encrypting the master key with
220AES-128-ECB, using the file's 16-byte nonce as the AES key.  The
221resulting ciphertext is used as the derived key.  If the ciphertext is
222longer than needed, then it is truncated to the needed length.
223
224For v2 encryption policies, the KDF is HKDF-SHA512.  The master key is
225passed as the "input keying material", no salt is used, and a distinct
226"application-specific information string" is used for each distinct
227key to be derived.  For example, when a per-file encryption key is
228derived, the application-specific information string is the file's
229nonce prefixed with "fscrypt\\0" and a context byte.  Different
230context bytes are used for other types of derived keys.
231
232HKDF-SHA512 is preferred to the original AES-128-ECB based KDF because
233HKDF is more flexible, is nonreversible, and evenly distributes
234entropy from the master key.  HKDF is also standardized and widely
235used by other software, whereas the AES-128-ECB based KDF is ad-hoc.
236
237Per-file encryption keys
238------------------------
239
240Since each master key can protect many files, it is necessary to
241"tweak" the encryption of each file so that the same plaintext in two
242files doesn't map to the same ciphertext, or vice versa.  In most
243cases, fscrypt does this by deriving per-file keys.  When a new
244encrypted inode (regular file, directory, or symlink) is created,
245fscrypt randomly generates a 16-byte nonce and stores it in the
246inode's encryption xattr.  Then, it uses a KDF (as described in `Key
247derivation function`_) to derive the file's key from the master key
248and nonce.
249
250Key derivation was chosen over key wrapping because wrapped keys would
251require larger xattrs which would be less likely to fit in-line in the
252filesystem's inode table, and there didn't appear to be any
253significant advantages to key wrapping.  In particular, currently
254there is no requirement to support unlocking a file with multiple
255alternative master keys or to support rotating master keys.  Instead,
256the master keys may be wrapped in userspace, e.g. as is done by the
257`fscrypt <https://github.com/google/fscrypt>`_ tool.
258
259DIRECT_KEY policies
260-------------------
261
262The Adiantum encryption mode (see `Encryption modes and usage`_) is
263suitable for both contents and filenames encryption, and it accepts
264long IVs --- long enough to hold both an 8-byte logical block number
265and a 16-byte per-file nonce.  Also, the overhead of each Adiantum key
266is greater than that of an AES-256-XTS key.
267
268Therefore, to improve performance and save memory, for Adiantum a
269"direct key" configuration is supported.  When the user has enabled
270this by setting FSCRYPT_POLICY_FLAG_DIRECT_KEY in the fscrypt policy,
271per-file encryption keys are not used.  Instead, whenever any data
272(contents or filenames) is encrypted, the file's 16-byte nonce is
273included in the IV.  Moreover:
274
275- For v1 encryption policies, the encryption is done directly with the
276  master key.  Because of this, users **must not** use the same master
277  key for any other purpose, even for other v1 policies.
278
279- For v2 encryption policies, the encryption is done with a per-mode
280  key derived using the KDF.  Users may use the same master key for
281  other v2 encryption policies.
282
283IV_INO_LBLK_64 policies
284-----------------------
285
286When FSCRYPT_POLICY_FLAG_IV_INO_LBLK_64 is set in the fscrypt policy,
287the encryption keys are derived from the master key, encryption mode
288number, and filesystem UUID.  This normally results in all files
289protected by the same master key sharing a single contents encryption
290key and a single filenames encryption key.  To still encrypt different
291files' data differently, inode numbers are included in the IVs.
292Consequently, shrinking the filesystem may not be allowed.
293
294This format is optimized for use with inline encryption hardware
295compliant with the UFS standard, which supports only 64 IV bits per
296I/O request and may have only a small number of keyslots.
297
298IV_INO_LBLK_32 policies
299-----------------------
300
301IV_INO_LBLK_32 policies work like IV_INO_LBLK_64, except that for
302IV_INO_LBLK_32, the inode number is hashed with SipHash-2-4 (where the
303SipHash key is derived from the master key) and added to the file
304logical block number mod 2^32 to produce a 32-bit IV.
305
306This format is optimized for use with inline encryption hardware
307compliant with the eMMC v5.2 standard, which supports only 32 IV bits
308per I/O request and may have only a small number of keyslots.  This
309format results in some level of IV reuse, so it should only be used
310when necessary due to hardware limitations.
311
312Key identifiers
313---------------
314
315For master keys used for v2 encryption policies, a unique 16-byte "key
316identifier" is also derived using the KDF.  This value is stored in
317the clear, since it is needed to reliably identify the key itself.
318
319Dirhash keys
320------------
321
322For directories that are indexed using a secret-keyed dirhash over the
323plaintext filenames, the KDF is also used to derive a 128-bit
324SipHash-2-4 key per directory in order to hash filenames.  This works
325just like deriving a per-file encryption key, except that a different
326KDF context is used.  Currently, only casefolded ("case-insensitive")
327encrypted directories use this style of hashing.
328
329Encryption modes and usage
330==========================
331
332fscrypt allows one encryption mode to be specified for file contents
333and one encryption mode to be specified for filenames.  Different
334directory trees are permitted to use different encryption modes.
335Currently, the following pairs of encryption modes are supported:
336
337- AES-256-XTS for contents and AES-256-CTS-CBC for filenames
338- AES-128-CBC for contents and AES-128-CTS-CBC for filenames
339- Adiantum for both contents and filenames
340
341If unsure, you should use the (AES-256-XTS, AES-256-CTS-CBC) pair.
342
343AES-128-CBC was added only for low-powered embedded devices with
344crypto accelerators such as CAAM or CESA that do not support XTS.  To
345use AES-128-CBC, CONFIG_CRYPTO_ESSIV and CONFIG_CRYPTO_SHA256 (or
346another SHA-256 implementation) must be enabled so that ESSIV can be
347used.
348
349Adiantum is a (primarily) stream cipher-based mode that is fast even
350on CPUs without dedicated crypto instructions.  It's also a true
351wide-block mode, unlike XTS.  It can also eliminate the need to derive
352per-file encryption keys.  However, it depends on the security of two
353primitives, XChaCha12 and AES-256, rather than just one.  See the
354paper "Adiantum: length-preserving encryption for entry-level
355processors" (https://eprint.iacr.org/2018/720.pdf) for more details.
356To use Adiantum, CONFIG_CRYPTO_ADIANTUM must be enabled.  Also, fast
357implementations of ChaCha and NHPoly1305 should be enabled, e.g.
358CONFIG_CRYPTO_CHACHA20_NEON and CONFIG_CRYPTO_NHPOLY1305_NEON for ARM.
359
360New encryption modes can be added relatively easily, without changes
361to individual filesystems.  However, authenticated encryption (AE)
362modes are not currently supported because of the difficulty of dealing
363with ciphertext expansion.
364
365Contents encryption
366-------------------
367
368For file contents, each filesystem block is encrypted independently.
369Starting from Linux kernel 5.5, encryption of filesystems with block
370size less than system's page size is supported.
371
372Each block's IV is set to the logical block number within the file as
373a little endian number, except that:
374
375- With CBC mode encryption, ESSIV is also used.  Specifically, each IV
376  is encrypted with AES-256 where the AES-256 key is the SHA-256 hash
377  of the file's data encryption key.
378
379- With `DIRECT_KEY policies`_, the file's nonce is appended to the IV.
380  Currently this is only allowed with the Adiantum encryption mode.
381
382- With `IV_INO_LBLK_64 policies`_, the logical block number is limited
383  to 32 bits and is placed in bits 0-31 of the IV.  The inode number
384  (which is also limited to 32 bits) is placed in bits 32-63.
385
386- With `IV_INO_LBLK_32 policies`_, the logical block number is limited
387  to 32 bits and is placed in bits 0-31 of the IV.  The inode number
388  is then hashed and added mod 2^32.
389
390Note that because file logical block numbers are included in the IVs,
391filesystems must enforce that blocks are never shifted around within
392encrypted files, e.g. via "collapse range" or "insert range".
393
394Filenames encryption
395--------------------
396
397For filenames, each full filename is encrypted at once.  Because of
398the requirements to retain support for efficient directory lookups and
399filenames of up to 255 bytes, the same IV is used for every filename
400in a directory.
401
402However, each encrypted directory still uses a unique key, or
403alternatively has the file's nonce (for `DIRECT_KEY policies`_) or
404inode number (for `IV_INO_LBLK_64 policies`_) included in the IVs.
405Thus, IV reuse is limited to within a single directory.
406
407With CTS-CBC, the IV reuse means that when the plaintext filenames
408share a common prefix at least as long as the cipher block size (16
409bytes for AES), the corresponding encrypted filenames will also share
410a common prefix.  This is undesirable.  Adiantum does not have this
411weakness, as it is a wide-block encryption mode.
412
413All supported filenames encryption modes accept any plaintext length
414>= 16 bytes; cipher block alignment is not required.  However,
415filenames shorter than 16 bytes are NUL-padded to 16 bytes before
416being encrypted.  In addition, to reduce leakage of filename lengths
417via their ciphertexts, all filenames are NUL-padded to the next 4, 8,
41816, or 32-byte boundary (configurable).  32 is recommended since this
419provides the best confidentiality, at the cost of making directory
420entries consume slightly more space.  Note that since NUL (``\0``) is
421not otherwise a valid character in filenames, the padding will never
422produce duplicate plaintexts.
423
424Symbolic link targets are considered a type of filename and are
425encrypted in the same way as filenames in directory entries, except
426that IV reuse is not a problem as each symlink has its own inode.
427
428User API
429========
430
431Setting an encryption policy
432----------------------------
433
434FS_IOC_SET_ENCRYPTION_POLICY
435~~~~~~~~~~~~~~~~~~~~~~~~~~~~
436
437The FS_IOC_SET_ENCRYPTION_POLICY ioctl sets an encryption policy on an
438empty directory or verifies that a directory or regular file already
439has the specified encryption policy.  It takes in a pointer to
440struct fscrypt_policy_v1 or struct fscrypt_policy_v2, defined as
441follows::
442
443    #define FSCRYPT_POLICY_V1               0
444    #define FSCRYPT_KEY_DESCRIPTOR_SIZE     8
445    struct fscrypt_policy_v1 {
446            __u8 version;
447            __u8 contents_encryption_mode;
448            __u8 filenames_encryption_mode;
449            __u8 flags;
450            __u8 master_key_descriptor[FSCRYPT_KEY_DESCRIPTOR_SIZE];
451    };
452    #define fscrypt_policy  fscrypt_policy_v1
453
454    #define FSCRYPT_POLICY_V2               2
455    #define FSCRYPT_KEY_IDENTIFIER_SIZE     16
456    struct fscrypt_policy_v2 {
457            __u8 version;
458            __u8 contents_encryption_mode;
459            __u8 filenames_encryption_mode;
460            __u8 flags;
461            __u8 __reserved[4];
462            __u8 master_key_identifier[FSCRYPT_KEY_IDENTIFIER_SIZE];
463    };
464
465This structure must be initialized as follows:
466
467- ``version`` must be FSCRYPT_POLICY_V1 (0) if
468  struct fscrypt_policy_v1 is used or FSCRYPT_POLICY_V2 (2) if
469  struct fscrypt_policy_v2 is used. (Note: we refer to the original
470  policy version as "v1", though its version code is really 0.)
471  For new encrypted directories, use v2 policies.
472
473- ``contents_encryption_mode`` and ``filenames_encryption_mode`` must
474  be set to constants from ``<linux/fscrypt.h>`` which identify the
475  encryption modes to use.  If unsure, use FSCRYPT_MODE_AES_256_XTS
476  (1) for ``contents_encryption_mode`` and FSCRYPT_MODE_AES_256_CTS
477  (4) for ``filenames_encryption_mode``.
478
479- ``flags`` contains optional flags from ``<linux/fscrypt.h>``:
480
481  - FSCRYPT_POLICY_FLAGS_PAD_*: The amount of NUL padding to use when
482    encrypting filenames.  If unsure, use FSCRYPT_POLICY_FLAGS_PAD_32
483    (0x3).
484  - FSCRYPT_POLICY_FLAG_DIRECT_KEY: See `DIRECT_KEY policies`_.
485  - FSCRYPT_POLICY_FLAG_IV_INO_LBLK_64: See `IV_INO_LBLK_64
486    policies`_.
487  - FSCRYPT_POLICY_FLAG_IV_INO_LBLK_32: See `IV_INO_LBLK_32
488    policies`_.
489
490  v1 encryption policies only support the PAD_* and DIRECT_KEY flags.
491  The other flags are only supported by v2 encryption policies.
492
493  The DIRECT_KEY, IV_INO_LBLK_64, and IV_INO_LBLK_32 flags are
494  mutually exclusive.
495
496- For v2 encryption policies, ``__reserved`` must be zeroed.
497
498- For v1 encryption policies, ``master_key_descriptor`` specifies how
499  to find the master key in a keyring; see `Adding keys`_.  It is up
500  to userspace to choose a unique ``master_key_descriptor`` for each
501  master key.  The e4crypt and fscrypt tools use the first 8 bytes of
502  ``SHA-512(SHA-512(master_key))``, but this particular scheme is not
503  required.  Also, the master key need not be in the keyring yet when
504  FS_IOC_SET_ENCRYPTION_POLICY is executed.  However, it must be added
505  before any files can be created in the encrypted directory.
506
507  For v2 encryption policies, ``master_key_descriptor`` has been
508  replaced with ``master_key_identifier``, which is longer and cannot
509  be arbitrarily chosen.  Instead, the key must first be added using
510  `FS_IOC_ADD_ENCRYPTION_KEY`_.  Then, the ``key_spec.u.identifier``
511  the kernel returned in the struct fscrypt_add_key_arg must
512  be used as the ``master_key_identifier`` in
513  struct fscrypt_policy_v2.
514
515If the file is not yet encrypted, then FS_IOC_SET_ENCRYPTION_POLICY
516verifies that the file is an empty directory.  If so, the specified
517encryption policy is assigned to the directory, turning it into an
518encrypted directory.  After that, and after providing the
519corresponding master key as described in `Adding keys`_, all regular
520files, directories (recursively), and symlinks created in the
521directory will be encrypted, inheriting the same encryption policy.
522The filenames in the directory's entries will be encrypted as well.
523
524Alternatively, if the file is already encrypted, then
525FS_IOC_SET_ENCRYPTION_POLICY validates that the specified encryption
526policy exactly matches the actual one.  If they match, then the ioctl
527returns 0.  Otherwise, it fails with EEXIST.  This works on both
528regular files and directories, including nonempty directories.
529
530When a v2 encryption policy is assigned to a directory, it is also
531required that either the specified key has been added by the current
532user or that the caller has CAP_FOWNER in the initial user namespace.
533(This is needed to prevent a user from encrypting their data with
534another user's key.)  The key must remain added while
535FS_IOC_SET_ENCRYPTION_POLICY is executing.  However, if the new
536encrypted directory does not need to be accessed immediately, then the
537key can be removed right away afterwards.
538
539Note that the ext4 filesystem does not allow the root directory to be
540encrypted, even if it is empty.  Users who want to encrypt an entire
541filesystem with one key should consider using dm-crypt instead.
542
543FS_IOC_SET_ENCRYPTION_POLICY can fail with the following errors:
544
545- ``EACCES``: the file is not owned by the process's uid, nor does the
546  process have the CAP_FOWNER capability in a namespace with the file
547  owner's uid mapped
548- ``EEXIST``: the file is already encrypted with an encryption policy
549  different from the one specified
550- ``EINVAL``: an invalid encryption policy was specified (invalid
551  version, mode(s), or flags; or reserved bits were set); or a v1
552  encryption policy was specified but the directory has the casefold
553  flag enabled (casefolding is incompatible with v1 policies).
554- ``ENOKEY``: a v2 encryption policy was specified, but the key with
555  the specified ``master_key_identifier`` has not been added, nor does
556  the process have the CAP_FOWNER capability in the initial user
557  namespace
558- ``ENOTDIR``: the file is unencrypted and is a regular file, not a
559  directory
560- ``ENOTEMPTY``: the file is unencrypted and is a nonempty directory
561- ``ENOTTY``: this type of filesystem does not implement encryption
562- ``EOPNOTSUPP``: the kernel was not configured with encryption
563  support for filesystems, or the filesystem superblock has not
564  had encryption enabled on it.  (For example, to use encryption on an
565  ext4 filesystem, CONFIG_FS_ENCRYPTION must be enabled in the
566  kernel config, and the superblock must have had the "encrypt"
567  feature flag enabled using ``tune2fs -O encrypt`` or ``mkfs.ext4 -O
568  encrypt``.)
569- ``EPERM``: this directory may not be encrypted, e.g. because it is
570  the root directory of an ext4 filesystem
571- ``EROFS``: the filesystem is readonly
572
573Getting an encryption policy
574----------------------------
575
576Two ioctls are available to get a file's encryption policy:
577
578- `FS_IOC_GET_ENCRYPTION_POLICY_EX`_
579- `FS_IOC_GET_ENCRYPTION_POLICY`_
580
581The extended (_EX) version of the ioctl is more general and is
582recommended to use when possible.  However, on older kernels only the
583original ioctl is available.  Applications should try the extended
584version, and if it fails with ENOTTY fall back to the original
585version.
586
587FS_IOC_GET_ENCRYPTION_POLICY_EX
588~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
589
590The FS_IOC_GET_ENCRYPTION_POLICY_EX ioctl retrieves the encryption
591policy, if any, for a directory or regular file.  No additional
592permissions are required beyond the ability to open the file.  It
593takes in a pointer to struct fscrypt_get_policy_ex_arg,
594defined as follows::
595
596    struct fscrypt_get_policy_ex_arg {
597            __u64 policy_size; /* input/output */
598            union {
599                    __u8 version;
600                    struct fscrypt_policy_v1 v1;
601                    struct fscrypt_policy_v2 v2;
602            } policy; /* output */
603    };
604
605The caller must initialize ``policy_size`` to the size available for
606the policy struct, i.e. ``sizeof(arg.policy)``.
607
608On success, the policy struct is returned in ``policy``, and its
609actual size is returned in ``policy_size``.  ``policy.version`` should
610be checked to determine the version of policy returned.  Note that the
611version code for the "v1" policy is actually 0 (FSCRYPT_POLICY_V1).
612
613FS_IOC_GET_ENCRYPTION_POLICY_EX can fail with the following errors:
614
615- ``EINVAL``: the file is encrypted, but it uses an unrecognized
616  encryption policy version
617- ``ENODATA``: the file is not encrypted
618- ``ENOTTY``: this type of filesystem does not implement encryption,
619  or this kernel is too old to support FS_IOC_GET_ENCRYPTION_POLICY_EX
620  (try FS_IOC_GET_ENCRYPTION_POLICY instead)
621- ``EOPNOTSUPP``: the kernel was not configured with encryption
622  support for this filesystem, or the filesystem superblock has not
623  had encryption enabled on it
624- ``EOVERFLOW``: the file is encrypted and uses a recognized
625  encryption policy version, but the policy struct does not fit into
626  the provided buffer
627
628Note: if you only need to know whether a file is encrypted or not, on
629most filesystems it is also possible to use the FS_IOC_GETFLAGS ioctl
630and check for FS_ENCRYPT_FL, or to use the statx() system call and
631check for STATX_ATTR_ENCRYPTED in stx_attributes.
632
633FS_IOC_GET_ENCRYPTION_POLICY
634~~~~~~~~~~~~~~~~~~~~~~~~~~~~
635
636The FS_IOC_GET_ENCRYPTION_POLICY ioctl can also retrieve the
637encryption policy, if any, for a directory or regular file.  However,
638unlike `FS_IOC_GET_ENCRYPTION_POLICY_EX`_,
639FS_IOC_GET_ENCRYPTION_POLICY only supports the original policy
640version.  It takes in a pointer directly to struct fscrypt_policy_v1
641rather than struct fscrypt_get_policy_ex_arg.
642
643The error codes for FS_IOC_GET_ENCRYPTION_POLICY are the same as those
644for FS_IOC_GET_ENCRYPTION_POLICY_EX, except that
645FS_IOC_GET_ENCRYPTION_POLICY also returns ``EINVAL`` if the file is
646encrypted using a newer encryption policy version.
647
648Getting the per-filesystem salt
649-------------------------------
650
651Some filesystems, such as ext4 and F2FS, also support the deprecated
652ioctl FS_IOC_GET_ENCRYPTION_PWSALT.  This ioctl retrieves a randomly
653generated 16-byte value stored in the filesystem superblock.  This
654value is intended to used as a salt when deriving an encryption key
655from a passphrase or other low-entropy user credential.
656
657FS_IOC_GET_ENCRYPTION_PWSALT is deprecated.  Instead, prefer to
658generate and manage any needed salt(s) in userspace.
659
660Getting a file's encryption nonce
661---------------------------------
662
663Since Linux v5.7, the ioctl FS_IOC_GET_ENCRYPTION_NONCE is supported.
664On encrypted files and directories it gets the inode's 16-byte nonce.
665On unencrypted files and directories, it fails with ENODATA.
666
667This ioctl can be useful for automated tests which verify that the
668encryption is being done correctly.  It is not needed for normal use
669of fscrypt.
670
671Adding keys
672-----------
673
674FS_IOC_ADD_ENCRYPTION_KEY
675~~~~~~~~~~~~~~~~~~~~~~~~~
676
677The FS_IOC_ADD_ENCRYPTION_KEY ioctl adds a master encryption key to
678the filesystem, making all files on the filesystem which were
679encrypted using that key appear "unlocked", i.e. in plaintext form.
680It can be executed on any file or directory on the target filesystem,
681but using the filesystem's root directory is recommended.  It takes in
682a pointer to struct fscrypt_add_key_arg, defined as follows::
683
684    struct fscrypt_add_key_arg {
685            struct fscrypt_key_specifier key_spec;
686            __u32 raw_size;
687            __u32 key_id;
688            __u32 __reserved[8];
689            __u8 raw[];
690    };
691
692    #define FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR        1
693    #define FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER        2
694
695    struct fscrypt_key_specifier {
696            __u32 type;     /* one of FSCRYPT_KEY_SPEC_TYPE_* */
697            __u32 __reserved;
698            union {
699                    __u8 __reserved[32]; /* reserve some extra space */
700                    __u8 descriptor[FSCRYPT_KEY_DESCRIPTOR_SIZE];
701                    __u8 identifier[FSCRYPT_KEY_IDENTIFIER_SIZE];
702            } u;
703    };
704
705    struct fscrypt_provisioning_key_payload {
706            __u32 type;
707            __u32 __reserved;
708            __u8 raw[];
709    };
710
711struct fscrypt_add_key_arg must be zeroed, then initialized
712as follows:
713
714- If the key is being added for use by v1 encryption policies, then
715  ``key_spec.type`` must contain FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR, and
716  ``key_spec.u.descriptor`` must contain the descriptor of the key
717  being added, corresponding to the value in the
718  ``master_key_descriptor`` field of struct fscrypt_policy_v1.
719  To add this type of key, the calling process must have the
720  CAP_SYS_ADMIN capability in the initial user namespace.
721
722  Alternatively, if the key is being added for use by v2 encryption
723  policies, then ``key_spec.type`` must contain
724  FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER, and ``key_spec.u.identifier`` is
725  an *output* field which the kernel fills in with a cryptographic
726  hash of the key.  To add this type of key, the calling process does
727  not need any privileges.  However, the number of keys that can be
728  added is limited by the user's quota for the keyrings service (see
729  ``Documentation/security/keys/core.rst``).
730
731- ``raw_size`` must be the size of the ``raw`` key provided, in bytes.
732  Alternatively, if ``key_id`` is nonzero, this field must be 0, since
733  in that case the size is implied by the specified Linux keyring key.
734
735- ``key_id`` is 0 if the raw key is given directly in the ``raw``
736  field.  Otherwise ``key_id`` is the ID of a Linux keyring key of
737  type "fscrypt-provisioning" whose payload is
738  struct fscrypt_provisioning_key_payload whose ``raw`` field contains
739  the raw key and whose ``type`` field matches ``key_spec.type``.
740  Since ``raw`` is variable-length, the total size of this key's
741  payload must be ``sizeof(struct fscrypt_provisioning_key_payload)``
742  plus the raw key size.  The process must have Search permission on
743  this key.
744
745  Most users should leave this 0 and specify the raw key directly.
746  The support for specifying a Linux keyring key is intended mainly to
747  allow re-adding keys after a filesystem is unmounted and re-mounted,
748  without having to store the raw keys in userspace memory.
749
750- ``raw`` is a variable-length field which must contain the actual
751  key, ``raw_size`` bytes long.  Alternatively, if ``key_id`` is
752  nonzero, then this field is unused.
753
754For v2 policy keys, the kernel keeps track of which user (identified
755by effective user ID) added the key, and only allows the key to be
756removed by that user --- or by "root", if they use
757`FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS`_.
758
759However, if another user has added the key, it may be desirable to
760prevent that other user from unexpectedly removing it.  Therefore,
761FS_IOC_ADD_ENCRYPTION_KEY may also be used to add a v2 policy key
762*again*, even if it's already added by other user(s).  In this case,
763FS_IOC_ADD_ENCRYPTION_KEY will just install a claim to the key for the
764current user, rather than actually add the key again (but the raw key
765must still be provided, as a proof of knowledge).
766
767FS_IOC_ADD_ENCRYPTION_KEY returns 0 if either the key or a claim to
768the key was either added or already exists.
769
770FS_IOC_ADD_ENCRYPTION_KEY can fail with the following errors:
771
772- ``EACCES``: FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR was specified, but the
773  caller does not have the CAP_SYS_ADMIN capability in the initial
774  user namespace; or the raw key was specified by Linux key ID but the
775  process lacks Search permission on the key.
776- ``EDQUOT``: the key quota for this user would be exceeded by adding
777  the key
778- ``EINVAL``: invalid key size or key specifier type, or reserved bits
779  were set
780- ``EKEYREJECTED``: the raw key was specified by Linux key ID, but the
781  key has the wrong type
782- ``ENOKEY``: the raw key was specified by Linux key ID, but no key
783  exists with that ID
784- ``ENOTTY``: this type of filesystem does not implement encryption
785- ``EOPNOTSUPP``: the kernel was not configured with encryption
786  support for this filesystem, or the filesystem superblock has not
787  had encryption enabled on it
788
789Legacy method
790~~~~~~~~~~~~~
791
792For v1 encryption policies, a master encryption key can also be
793provided by adding it to a process-subscribed keyring, e.g. to a
794session keyring, or to a user keyring if the user keyring is linked
795into the session keyring.
796
797This method is deprecated (and not supported for v2 encryption
798policies) for several reasons.  First, it cannot be used in
799combination with FS_IOC_REMOVE_ENCRYPTION_KEY (see `Removing keys`_),
800so for removing a key a workaround such as keyctl_unlink() in
801combination with ``sync; echo 2 > /proc/sys/vm/drop_caches`` would
802have to be used.  Second, it doesn't match the fact that the
803locked/unlocked status of encrypted files (i.e. whether they appear to
804be in plaintext form or in ciphertext form) is global.  This mismatch
805has caused much confusion as well as real problems when processes
806running under different UIDs, such as a ``sudo`` command, need to
807access encrypted files.
808
809Nevertheless, to add a key to one of the process-subscribed keyrings,
810the add_key() system call can be used (see:
811``Documentation/security/keys/core.rst``).  The key type must be
812"logon"; keys of this type are kept in kernel memory and cannot be
813read back by userspace.  The key description must be "fscrypt:"
814followed by the 16-character lower case hex representation of the
815``master_key_descriptor`` that was set in the encryption policy.  The
816key payload must conform to the following structure::
817
818    #define FSCRYPT_MAX_KEY_SIZE            64
819
820    struct fscrypt_key {
821            __u32 mode;
822            __u8 raw[FSCRYPT_MAX_KEY_SIZE];
823            __u32 size;
824    };
825
826``mode`` is ignored; just set it to 0.  The actual key is provided in
827``raw`` with ``size`` indicating its size in bytes.  That is, the
828bytes ``raw[0..size-1]`` (inclusive) are the actual key.
829
830The key description prefix "fscrypt:" may alternatively be replaced
831with a filesystem-specific prefix such as "ext4:".  However, the
832filesystem-specific prefixes are deprecated and should not be used in
833new programs.
834
835Removing keys
836-------------
837
838Two ioctls are available for removing a key that was added by
839`FS_IOC_ADD_ENCRYPTION_KEY`_:
840
841- `FS_IOC_REMOVE_ENCRYPTION_KEY`_
842- `FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS`_
843
844These two ioctls differ only in cases where v2 policy keys are added
845or removed by non-root users.
846
847These ioctls don't work on keys that were added via the legacy
848process-subscribed keyrings mechanism.
849
850Before using these ioctls, read the `Kernel memory compromise`_
851section for a discussion of the security goals and limitations of
852these ioctls.
853
854FS_IOC_REMOVE_ENCRYPTION_KEY
855~~~~~~~~~~~~~~~~~~~~~~~~~~~~
856
857The FS_IOC_REMOVE_ENCRYPTION_KEY ioctl removes a claim to a master
858encryption key from the filesystem, and possibly removes the key
859itself.  It can be executed on any file or directory on the target
860filesystem, but using the filesystem's root directory is recommended.
861It takes in a pointer to struct fscrypt_remove_key_arg, defined
862as follows::
863
864    struct fscrypt_remove_key_arg {
865            struct fscrypt_key_specifier key_spec;
866    #define FSCRYPT_KEY_REMOVAL_STATUS_FLAG_FILES_BUSY      0x00000001
867    #define FSCRYPT_KEY_REMOVAL_STATUS_FLAG_OTHER_USERS     0x00000002
868            __u32 removal_status_flags;     /* output */
869            __u32 __reserved[5];
870    };
871
872This structure must be zeroed, then initialized as follows:
873
874- The key to remove is specified by ``key_spec``:
875
876    - To remove a key used by v1 encryption policies, set
877      ``key_spec.type`` to FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR and fill
878      in ``key_spec.u.descriptor``.  To remove this type of key, the
879      calling process must have the CAP_SYS_ADMIN capability in the
880      initial user namespace.
881
882    - To remove a key used by v2 encryption policies, set
883      ``key_spec.type`` to FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER and fill
884      in ``key_spec.u.identifier``.
885
886For v2 policy keys, this ioctl is usable by non-root users.  However,
887to make this possible, it actually just removes the current user's
888claim to the key, undoing a single call to FS_IOC_ADD_ENCRYPTION_KEY.
889Only after all claims are removed is the key really removed.
890
891For example, if FS_IOC_ADD_ENCRYPTION_KEY was called with uid 1000,
892then the key will be "claimed" by uid 1000, and
893FS_IOC_REMOVE_ENCRYPTION_KEY will only succeed as uid 1000.  Or, if
894both uids 1000 and 2000 added the key, then for each uid
895FS_IOC_REMOVE_ENCRYPTION_KEY will only remove their own claim.  Only
896once *both* are removed is the key really removed.  (Think of it like
897unlinking a file that may have hard links.)
898
899If FS_IOC_REMOVE_ENCRYPTION_KEY really removes the key, it will also
900try to "lock" all files that had been unlocked with the key.  It won't
901lock files that are still in-use, so this ioctl is expected to be used
902in cooperation with userspace ensuring that none of the files are
903still open.  However, if necessary, this ioctl can be executed again
904later to retry locking any remaining files.
905
906FS_IOC_REMOVE_ENCRYPTION_KEY returns 0 if either the key was removed
907(but may still have files remaining to be locked), the user's claim to
908the key was removed, or the key was already removed but had files
909remaining to be the locked so the ioctl retried locking them.  In any
910of these cases, ``removal_status_flags`` is filled in with the
911following informational status flags:
912
913- ``FSCRYPT_KEY_REMOVAL_STATUS_FLAG_FILES_BUSY``: set if some file(s)
914  are still in-use.  Not guaranteed to be set in the case where only
915  the user's claim to the key was removed.
916- ``FSCRYPT_KEY_REMOVAL_STATUS_FLAG_OTHER_USERS``: set if only the
917  user's claim to the key was removed, not the key itself
918
919FS_IOC_REMOVE_ENCRYPTION_KEY can fail with the following errors:
920
921- ``EACCES``: The FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR key specifier type
922  was specified, but the caller does not have the CAP_SYS_ADMIN
923  capability in the initial user namespace
924- ``EINVAL``: invalid key specifier type, or reserved bits were set
925- ``ENOKEY``: the key object was not found at all, i.e. it was never
926  added in the first place or was already fully removed including all
927  files locked; or, the user does not have a claim to the key (but
928  someone else does).
929- ``ENOTTY``: this type of filesystem does not implement encryption
930- ``EOPNOTSUPP``: the kernel was not configured with encryption
931  support for this filesystem, or the filesystem superblock has not
932  had encryption enabled on it
933
934FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS
935~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
936
937FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS is exactly the same as
938`FS_IOC_REMOVE_ENCRYPTION_KEY`_, except that for v2 policy keys, the
939ALL_USERS version of the ioctl will remove all users' claims to the
940key, not just the current user's.  I.e., the key itself will always be
941removed, no matter how many users have added it.  This difference is
942only meaningful if non-root users are adding and removing keys.
943
944Because of this, FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS also requires
945"root", namely the CAP_SYS_ADMIN capability in the initial user
946namespace.  Otherwise it will fail with EACCES.
947
948Getting key status
949------------------
950
951FS_IOC_GET_ENCRYPTION_KEY_STATUS
952~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
953
954The FS_IOC_GET_ENCRYPTION_KEY_STATUS ioctl retrieves the status of a
955master encryption key.  It can be executed on any file or directory on
956the target filesystem, but using the filesystem's root directory is
957recommended.  It takes in a pointer to
958struct fscrypt_get_key_status_arg, defined as follows::
959
960    struct fscrypt_get_key_status_arg {
961            /* input */
962            struct fscrypt_key_specifier key_spec;
963            __u32 __reserved[6];
964
965            /* output */
966    #define FSCRYPT_KEY_STATUS_ABSENT               1
967    #define FSCRYPT_KEY_STATUS_PRESENT              2
968    #define FSCRYPT_KEY_STATUS_INCOMPLETELY_REMOVED 3
969            __u32 status;
970    #define FSCRYPT_KEY_STATUS_FLAG_ADDED_BY_SELF   0x00000001
971            __u32 status_flags;
972            __u32 user_count;
973            __u32 __out_reserved[13];
974    };
975
976The caller must zero all input fields, then fill in ``key_spec``:
977
978    - To get the status of a key for v1 encryption policies, set
979      ``key_spec.type`` to FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR and fill
980      in ``key_spec.u.descriptor``.
981
982    - To get the status of a key for v2 encryption policies, set
983      ``key_spec.type`` to FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER and fill
984      in ``key_spec.u.identifier``.
985
986On success, 0 is returned and the kernel fills in the output fields:
987
988- ``status`` indicates whether the key is absent, present, or
989  incompletely removed.  Incompletely removed means that the master
990  secret has been removed, but some files are still in use; i.e.,
991  `FS_IOC_REMOVE_ENCRYPTION_KEY`_ returned 0 but set the informational
992  status flag FSCRYPT_KEY_REMOVAL_STATUS_FLAG_FILES_BUSY.
993
994- ``status_flags`` can contain the following flags:
995
996    - ``FSCRYPT_KEY_STATUS_FLAG_ADDED_BY_SELF`` indicates that the key
997      has added by the current user.  This is only set for keys
998      identified by ``identifier`` rather than by ``descriptor``.
999
1000- ``user_count`` specifies the number of users who have added the key.
1001  This is only set for keys identified by ``identifier`` rather than
1002  by ``descriptor``.
1003
1004FS_IOC_GET_ENCRYPTION_KEY_STATUS can fail with the following errors:
1005
1006- ``EINVAL``: invalid key specifier type, or reserved bits were set
1007- ``ENOTTY``: this type of filesystem does not implement encryption
1008- ``EOPNOTSUPP``: the kernel was not configured with encryption
1009  support for this filesystem, or the filesystem superblock has not
1010  had encryption enabled on it
1011
1012Among other use cases, FS_IOC_GET_ENCRYPTION_KEY_STATUS can be useful
1013for determining whether the key for a given encrypted directory needs
1014to be added before prompting the user for the passphrase needed to
1015derive the key.
1016
1017FS_IOC_GET_ENCRYPTION_KEY_STATUS can only get the status of keys in
1018the filesystem-level keyring, i.e. the keyring managed by
1019`FS_IOC_ADD_ENCRYPTION_KEY`_ and `FS_IOC_REMOVE_ENCRYPTION_KEY`_.  It
1020cannot get the status of a key that has only been added for use by v1
1021encryption policies using the legacy mechanism involving
1022process-subscribed keyrings.
1023
1024Access semantics
1025================
1026
1027With the key
1028------------
1029
1030With the encryption key, encrypted regular files, directories, and
1031symlinks behave very similarly to their unencrypted counterparts ---
1032after all, the encryption is intended to be transparent.  However,
1033astute users may notice some differences in behavior:
1034
1035- Unencrypted files, or files encrypted with a different encryption
1036  policy (i.e. different key, modes, or flags), cannot be renamed or
1037  linked into an encrypted directory; see `Encryption policy
1038  enforcement`_.  Attempts to do so will fail with EXDEV.  However,
1039  encrypted files can be renamed within an encrypted directory, or
1040  into an unencrypted directory.
1041
1042  Note: "moving" an unencrypted file into an encrypted directory, e.g.
1043  with the `mv` program, is implemented in userspace by a copy
1044  followed by a delete.  Be aware that the original unencrypted data
1045  may remain recoverable from free space on the disk; prefer to keep
1046  all files encrypted from the very beginning.  The `shred` program
1047  may be used to overwrite the source files but isn't guaranteed to be
1048  effective on all filesystems and storage devices.
1049
1050- Direct I/O is supported on encrypted files only under some
1051  circumstances.  For details, see `Direct I/O support`_.
1052
1053- The fallocate operations FALLOC_FL_COLLAPSE_RANGE and
1054  FALLOC_FL_INSERT_RANGE are not supported on encrypted files and will
1055  fail with EOPNOTSUPP.
1056
1057- Online defragmentation of encrypted files is not supported.  The
1058  EXT4_IOC_MOVE_EXT and F2FS_IOC_MOVE_RANGE ioctls will fail with
1059  EOPNOTSUPP.
1060
1061- The ext4 filesystem does not support data journaling with encrypted
1062  regular files.  It will fall back to ordered data mode instead.
1063
1064- DAX (Direct Access) is not supported on encrypted files.
1065
1066- The maximum length of an encrypted symlink is 2 bytes shorter than
1067  the maximum length of an unencrypted symlink.  For example, on an
1068  EXT4 filesystem with a 4K block size, unencrypted symlinks can be up
1069  to 4095 bytes long, while encrypted symlinks can only be up to 4093
1070  bytes long (both lengths excluding the terminating null).
1071
1072Note that mmap *is* supported.  This is possible because the pagecache
1073for an encrypted file contains the plaintext, not the ciphertext.
1074
1075Without the key
1076---------------
1077
1078Some filesystem operations may be performed on encrypted regular
1079files, directories, and symlinks even before their encryption key has
1080been added, or after their encryption key has been removed:
1081
1082- File metadata may be read, e.g. using stat().
1083
1084- Directories may be listed, in which case the filenames will be
1085  listed in an encoded form derived from their ciphertext.  The
1086  current encoding algorithm is described in `Filename hashing and
1087  encoding`_.  The algorithm is subject to change, but it is
1088  guaranteed that the presented filenames will be no longer than
1089  NAME_MAX bytes, will not contain the ``/`` or ``\0`` characters, and
1090  will uniquely identify directory entries.
1091
1092  The ``.`` and ``..`` directory entries are special.  They are always
1093  present and are not encrypted or encoded.
1094
1095- Files may be deleted.  That is, nondirectory files may be deleted
1096  with unlink() as usual, and empty directories may be deleted with
1097  rmdir() as usual.  Therefore, ``rm`` and ``rm -r`` will work as
1098  expected.
1099
1100- Symlink targets may be read and followed, but they will be presented
1101  in encrypted form, similar to filenames in directories.  Hence, they
1102  are unlikely to point to anywhere useful.
1103
1104Without the key, regular files cannot be opened or truncated.
1105Attempts to do so will fail with ENOKEY.  This implies that any
1106regular file operations that require a file descriptor, such as
1107read(), write(), mmap(), fallocate(), and ioctl(), are also forbidden.
1108
1109Also without the key, files of any type (including directories) cannot
1110be created or linked into an encrypted directory, nor can a name in an
1111encrypted directory be the source or target of a rename, nor can an
1112O_TMPFILE temporary file be created in an encrypted directory.  All
1113such operations will fail with ENOKEY.
1114
1115It is not currently possible to backup and restore encrypted files
1116without the encryption key.  This would require special APIs which
1117have not yet been implemented.
1118
1119Encryption policy enforcement
1120=============================
1121
1122After an encryption policy has been set on a directory, all regular
1123files, directories, and symbolic links created in that directory
1124(recursively) will inherit that encryption policy.  Special files ---
1125that is, named pipes, device nodes, and UNIX domain sockets --- will
1126not be encrypted.
1127
1128Except for those special files, it is forbidden to have unencrypted
1129files, or files encrypted with a different encryption policy, in an
1130encrypted directory tree.  Attempts to link or rename such a file into
1131an encrypted directory will fail with EXDEV.  This is also enforced
1132during ->lookup() to provide limited protection against offline
1133attacks that try to disable or downgrade encryption in known locations
1134where applications may later write sensitive data.  It is recommended
1135that systems implementing a form of "verified boot" take advantage of
1136this by validating all top-level encryption policies prior to access.
1137
1138Inline encryption support
1139=========================
1140
1141By default, fscrypt uses the kernel crypto API for all cryptographic
1142operations (other than HKDF, which fscrypt partially implements
1143itself).  The kernel crypto API supports hardware crypto accelerators,
1144but only ones that work in the traditional way where all inputs and
1145outputs (e.g. plaintexts and ciphertexts) are in memory.  fscrypt can
1146take advantage of such hardware, but the traditional acceleration
1147model isn't particularly efficient and fscrypt hasn't been optimized
1148for it.
1149
1150Instead, many newer systems (especially mobile SoCs) have *inline
1151encryption hardware* that can encrypt/decrypt data while it is on its
1152way to/from the storage device.  Linux supports inline encryption
1153through a set of extensions to the block layer called *blk-crypto*.
1154blk-crypto allows filesystems to attach encryption contexts to bios
1155(I/O requests) to specify how the data will be encrypted or decrypted
1156in-line.  For more information about blk-crypto, see
1157:ref:`Documentation/block/inline-encryption.rst <inline_encryption>`.
1158
1159On supported filesystems (currently ext4 and f2fs), fscrypt can use
1160blk-crypto instead of the kernel crypto API to encrypt/decrypt file
1161contents.  To enable this, set CONFIG_FS_ENCRYPTION_INLINE_CRYPT=y in
1162the kernel configuration, and specify the "inlinecrypt" mount option
1163when mounting the filesystem.
1164
1165Note that the "inlinecrypt" mount option just specifies to use inline
1166encryption when possible; it doesn't force its use.  fscrypt will
1167still fall back to using the kernel crypto API on files where the
1168inline encryption hardware doesn't have the needed crypto capabilities
1169(e.g. support for the needed encryption algorithm and data unit size)
1170and where blk-crypto-fallback is unusable.  (For blk-crypto-fallback
1171to be usable, it must be enabled in the kernel configuration with
1172CONFIG_BLK_INLINE_ENCRYPTION_FALLBACK=y.)
1173
1174Currently fscrypt always uses the filesystem block size (which is
1175usually 4096 bytes) as the data unit size.  Therefore, it can only use
1176inline encryption hardware that supports that data unit size.
1177
1178Inline encryption doesn't affect the ciphertext or other aspects of
1179the on-disk format, so users may freely switch back and forth between
1180using "inlinecrypt" and not using "inlinecrypt".
1181
1182Direct I/O support
1183==================
1184
1185For direct I/O on an encrypted file to work, the following conditions
1186must be met (in addition to the conditions for direct I/O on an
1187unencrypted file):
1188
1189* The file must be using inline encryption.  Usually this means that
1190  the filesystem must be mounted with ``-o inlinecrypt`` and inline
1191  encryption hardware must be present.  However, a software fallback
1192  is also available.  For details, see `Inline encryption support`_.
1193
1194* The I/O request must be fully aligned to the filesystem block size.
1195  This means that the file position the I/O is targeting, the lengths
1196  of all I/O segments, and the memory addresses of all I/O buffers
1197  must be multiples of this value.  Note that the filesystem block
1198  size may be greater than the logical block size of the block device.
1199
1200If either of the above conditions is not met, then direct I/O on the
1201encrypted file will fall back to buffered I/O.
1202
1203Implementation details
1204======================
1205
1206Encryption context
1207------------------
1208
1209An encryption policy is represented on-disk by
1210struct fscrypt_context_v1 or struct fscrypt_context_v2.  It is up to
1211individual filesystems to decide where to store it, but normally it
1212would be stored in a hidden extended attribute.  It should *not* be
1213exposed by the xattr-related system calls such as getxattr() and
1214setxattr() because of the special semantics of the encryption xattr.
1215(In particular, there would be much confusion if an encryption policy
1216were to be added to or removed from anything other than an empty
1217directory.)  These structs are defined as follows::
1218
1219    #define FSCRYPT_FILE_NONCE_SIZE 16
1220
1221    #define FSCRYPT_KEY_DESCRIPTOR_SIZE  8
1222    struct fscrypt_context_v1 {
1223            u8 version;
1224            u8 contents_encryption_mode;
1225            u8 filenames_encryption_mode;
1226            u8 flags;
1227            u8 master_key_descriptor[FSCRYPT_KEY_DESCRIPTOR_SIZE];
1228            u8 nonce[FSCRYPT_FILE_NONCE_SIZE];
1229    };
1230
1231    #define FSCRYPT_KEY_IDENTIFIER_SIZE  16
1232    struct fscrypt_context_v2 {
1233            u8 version;
1234            u8 contents_encryption_mode;
1235            u8 filenames_encryption_mode;
1236            u8 flags;
1237            u8 __reserved[4];
1238            u8 master_key_identifier[FSCRYPT_KEY_IDENTIFIER_SIZE];
1239            u8 nonce[FSCRYPT_FILE_NONCE_SIZE];
1240    };
1241
1242The context structs contain the same information as the corresponding
1243policy structs (see `Setting an encryption policy`_), except that the
1244context structs also contain a nonce.  The nonce is randomly generated
1245by the kernel and is used as KDF input or as a tweak to cause
1246different files to be encrypted differently; see `Per-file encryption
1247keys`_ and `DIRECT_KEY policies`_.
1248
1249Data path changes
1250-----------------
1251
1252When inline encryption is used, filesystems just need to associate
1253encryption contexts with bios to specify how the block layer or the
1254inline encryption hardware will encrypt/decrypt the file contents.
1255
1256When inline encryption isn't used, filesystems must encrypt/decrypt
1257the file contents themselves, as described below:
1258
1259For the read path (->read_folio()) of regular files, filesystems can
1260read the ciphertext into the page cache and decrypt it in-place.  The
1261page lock must be held until decryption has finished, to prevent the
1262page from becoming visible to userspace prematurely.
1263
1264For the write path (->writepage()) of regular files, filesystems
1265cannot encrypt data in-place in the page cache, since the cached
1266plaintext must be preserved.  Instead, filesystems must encrypt into a
1267temporary buffer or "bounce page", then write out the temporary
1268buffer.  Some filesystems, such as UBIFS, already use temporary
1269buffers regardless of encryption.  Other filesystems, such as ext4 and
1270F2FS, have to allocate bounce pages specially for encryption.
1271
1272Filename hashing and encoding
1273-----------------------------
1274
1275Modern filesystems accelerate directory lookups by using indexed
1276directories.  An indexed directory is organized as a tree keyed by
1277filename hashes.  When a ->lookup() is requested, the filesystem
1278normally hashes the filename being looked up so that it can quickly
1279find the corresponding directory entry, if any.
1280
1281With encryption, lookups must be supported and efficient both with and
1282without the encryption key.  Clearly, it would not work to hash the
1283plaintext filenames, since the plaintext filenames are unavailable
1284without the key.  (Hashing the plaintext filenames would also make it
1285impossible for the filesystem's fsck tool to optimize encrypted
1286directories.)  Instead, filesystems hash the ciphertext filenames,
1287i.e. the bytes actually stored on-disk in the directory entries.  When
1288asked to do a ->lookup() with the key, the filesystem just encrypts
1289the user-supplied name to get the ciphertext.
1290
1291Lookups without the key are more complicated.  The raw ciphertext may
1292contain the ``\0`` and ``/`` characters, which are illegal in
1293filenames.  Therefore, readdir() must base64url-encode the ciphertext
1294for presentation.  For most filenames, this works fine; on ->lookup(),
1295the filesystem just base64url-decodes the user-supplied name to get
1296back to the raw ciphertext.
1297
1298However, for very long filenames, base64url encoding would cause the
1299filename length to exceed NAME_MAX.  To prevent this, readdir()
1300actually presents long filenames in an abbreviated form which encodes
1301a strong "hash" of the ciphertext filename, along with the optional
1302filesystem-specific hash(es) needed for directory lookups.  This
1303allows the filesystem to still, with a high degree of confidence, map
1304the filename given in ->lookup() back to a particular directory entry
1305that was previously listed by readdir().  See
1306struct fscrypt_nokey_name in the source for more details.
1307
1308Note that the precise way that filenames are presented to userspace
1309without the key is subject to change in the future.  It is only meant
1310as a way to temporarily present valid filenames so that commands like
1311``rm -r`` work as expected on encrypted directories.
1312
1313Tests
1314=====
1315
1316To test fscrypt, use xfstests, which is Linux's de facto standard
1317filesystem test suite.  First, run all the tests in the "encrypt"
1318group on the relevant filesystem(s).  One can also run the tests
1319with the 'inlinecrypt' mount option to test the implementation for
1320inline encryption support.  For example, to test ext4 and
1321f2fs encryption using `kvm-xfstests
1322<https://github.com/tytso/xfstests-bld/blob/master/Documentation/kvm-quickstart.md>`_::
1323
1324    kvm-xfstests -c ext4,f2fs -g encrypt
1325    kvm-xfstests -c ext4,f2fs -g encrypt -m inlinecrypt
1326
1327UBIFS encryption can also be tested this way, but it should be done in
1328a separate command, and it takes some time for kvm-xfstests to set up
1329emulated UBI volumes::
1330
1331    kvm-xfstests -c ubifs -g encrypt
1332
1333No tests should fail.  However, tests that use non-default encryption
1334modes (e.g. generic/549 and generic/550) will be skipped if the needed
1335algorithms were not built into the kernel's crypto API.  Also, tests
1336that access the raw block device (e.g. generic/399, generic/548,
1337generic/549, generic/550) will be skipped on UBIFS.
1338
1339Besides running the "encrypt" group tests, for ext4 and f2fs it's also
1340possible to run most xfstests with the "test_dummy_encryption" mount
1341option.  This option causes all new files to be automatically
1342encrypted with a dummy key, without having to make any API calls.
1343This tests the encrypted I/O paths more thoroughly.  To do this with
1344kvm-xfstests, use the "encrypt" filesystem configuration::
1345
1346    kvm-xfstests -c ext4/encrypt,f2fs/encrypt -g auto
1347    kvm-xfstests -c ext4/encrypt,f2fs/encrypt -g auto -m inlinecrypt
1348
1349Because this runs many more tests than "-g encrypt" does, it takes
1350much longer to run; so also consider using `gce-xfstests
1351<https://github.com/tytso/xfstests-bld/blob/master/Documentation/gce-xfstests.md>`_
1352instead of kvm-xfstests::
1353
1354    gce-xfstests -c ext4/encrypt,f2fs/encrypt -g auto
1355    gce-xfstests -c ext4/encrypt,f2fs/encrypt -g auto -m inlinecrypt
1356