1.. SPDX-License-Identifier: GPL-2.0 2 3.. _fsverity: 4 5======================================================= 6fs-verity: read-only file-based authenticity protection 7======================================================= 8 9Introduction 10============ 11 12fs-verity (``fs/verity/``) is a support layer that filesystems can 13hook into to support transparent integrity and authenticity protection 14of read-only files. Currently, it is supported by the ext4, f2fs, and 15btrfs filesystems. Like fscrypt, not too much filesystem-specific 16code is needed to support fs-verity. 17 18fs-verity is similar to `dm-verity 19<https://www.kernel.org/doc/Documentation/device-mapper/verity.txt>`_ 20but works on files rather than block devices. On regular files on 21filesystems supporting fs-verity, userspace can execute an ioctl that 22causes the filesystem to build a Merkle tree for the file and persist 23it to a filesystem-specific location associated with the file. 24 25After this, the file is made readonly, and all reads from the file are 26automatically verified against the file's Merkle tree. Reads of any 27corrupted data, including mmap reads, will fail. 28 29Userspace can use another ioctl to retrieve the root hash (actually 30the "fs-verity file digest", which is a hash that includes the Merkle 31tree root hash) that fs-verity is enforcing for the file. This ioctl 32executes in constant time, regardless of the file size. 33 34fs-verity is essentially a way to hash a file in constant time, 35subject to the caveat that reads which would violate the hash will 36fail at runtime. 37 38Use cases 39========= 40 41By itself, fs-verity only provides integrity protection, i.e. 42detection of accidental (non-malicious) corruption. 43 44However, because fs-verity makes retrieving the file hash extremely 45efficient, it's primarily meant to be used as a tool to support 46authentication (detection of malicious modifications) or auditing 47(logging file hashes before use). 48 49A standard file hash could be used instead of fs-verity. However, 50this is inefficient if the file is large and only a small portion may 51be accessed. This is often the case for Android application package 52(APK) files, for example. These typically contain many translations, 53classes, and other resources that are infrequently or even never 54accessed on a particular device. It would be slow and wasteful to 55read and hash the entire file before starting the application. 56 57Unlike an ahead-of-time hash, fs-verity also re-verifies data each 58time it's paged in. This ensures that malicious disk firmware can't 59undetectably change the contents of the file at runtime. 60 61fs-verity does not replace or obsolete dm-verity. dm-verity should 62still be used on read-only filesystems. fs-verity is for files that 63must live on a read-write filesystem because they are independently 64updated and potentially user-installed, so dm-verity cannot be used. 65 66fs-verity does not mandate a particular scheme for authenticating its 67file hashes. (Similarly, dm-verity does not mandate a particular 68scheme for authenticating its block device root hashes.) Options for 69authenticating fs-verity file hashes include: 70 71- Trusted userspace code. Often, the userspace code that accesses 72 files can be trusted to authenticate them. Consider e.g. an 73 application that wants to authenticate data files before using them, 74 or an application loader that is part of the operating system (which 75 is already authenticated in a different way, such as by being loaded 76 from a read-only partition that uses dm-verity) and that wants to 77 authenticate applications before loading them. In these cases, this 78 trusted userspace code can authenticate a file's contents by 79 retrieving its fs-verity digest using `FS_IOC_MEASURE_VERITY`_, then 80 verifying a signature of it using any userspace cryptographic 81 library that supports digital signatures. 82 83- Integrity Measurement Architecture (IMA). IMA supports fs-verity 84 file digests as an alternative to its traditional full file digests. 85 "IMA appraisal" enforces that files contain a valid, matching 86 signature in their "security.ima" extended attribute, as controlled 87 by the IMA policy. For more information, see the IMA documentation. 88 89- Integrity Policy Enforcement (IPE). IPE supports enforcing access 90 control decisions based on immutable security properties of files, 91 including those protected by fs-verity's built-in signatures. 92 "IPE policy" specifically allows for the authorization of fs-verity 93 files using properties ``fsverity_digest`` for identifying 94 files by their verity digest, and ``fsverity_signature`` to authorize 95 files with a verified fs-verity's built-in signature. For 96 details on configuring IPE policies and understanding its operational 97 modes, please refer to :doc:`IPE admin guide </admin-guide/LSM/ipe>`. 98 99- Trusted userspace code in combination with `Built-in signature 100 verification`_. This approach should be used only with great care. 101 102User API 103======== 104 105FS_IOC_ENABLE_VERITY 106-------------------- 107 108The FS_IOC_ENABLE_VERITY ioctl enables fs-verity on a file. It takes 109in a pointer to a struct fsverity_enable_arg, defined as 110follows:: 111 112 struct fsverity_enable_arg { 113 __u32 version; 114 __u32 hash_algorithm; 115 __u32 block_size; 116 __u32 salt_size; 117 __u64 salt_ptr; 118 __u32 sig_size; 119 __u32 __reserved1; 120 __u64 sig_ptr; 121 __u64 __reserved2[11]; 122 }; 123 124This structure contains the parameters of the Merkle tree to build for 125the file. It must be initialized as follows: 126 127- ``version`` must be 1. 128- ``hash_algorithm`` must be the identifier for the hash algorithm to 129 use for the Merkle tree, such as FS_VERITY_HASH_ALG_SHA256. See 130 ``include/uapi/linux/fsverity.h`` for the list of possible values. 131- ``block_size`` is the Merkle tree block size, in bytes. In Linux 132 v6.3 and later, this can be any power of 2 between (inclusively) 133 1024 and the minimum of the system page size and the filesystem 134 block size. In earlier versions, the page size was the only allowed 135 value. 136- ``salt_size`` is the size of the salt in bytes, or 0 if no salt is 137 provided. The salt is a value that is prepended to every hashed 138 block; it can be used to personalize the hashing for a particular 139 file or device. Currently the maximum salt size is 32 bytes. 140- ``salt_ptr`` is the pointer to the salt, or NULL if no salt is 141 provided. 142- ``sig_size`` is the size of the builtin signature in bytes, or 0 if no 143 builtin signature is provided. Currently the builtin signature is 144 (somewhat arbitrarily) limited to 16128 bytes. 145- ``sig_ptr`` is the pointer to the builtin signature, or NULL if no 146 builtin signature is provided. A builtin signature is only needed 147 if the `Built-in signature verification`_ feature is being used. It 148 is not needed for IMA appraisal, and it is not needed if the file 149 signature is being handled entirely in userspace. 150- All reserved fields must be zeroed. 151 152FS_IOC_ENABLE_VERITY causes the filesystem to build a Merkle tree for 153the file and persist it to a filesystem-specific location associated 154with the file, then mark the file as a verity file. This ioctl may 155take a long time to execute on large files, and it is interruptible by 156fatal signals. 157 158FS_IOC_ENABLE_VERITY checks for write access to the inode. However, 159it must be executed on an O_RDONLY file descriptor and no processes 160can have the file open for writing. Attempts to open the file for 161writing while this ioctl is executing will fail with ETXTBSY. (This 162is necessary to guarantee that no writable file descriptors will exist 163after verity is enabled, and to guarantee that the file's contents are 164stable while the Merkle tree is being built over it.) 165 166On success, FS_IOC_ENABLE_VERITY returns 0, and the file becomes a 167verity file. On failure (including the case of interruption by a 168fatal signal), no changes are made to the file. 169 170FS_IOC_ENABLE_VERITY can fail with the following errors: 171 172- ``EACCES``: the process does not have write access to the file 173- ``EBADMSG``: the builtin signature is malformed 174- ``EBUSY``: this ioctl is already running on the file 175- ``EEXIST``: the file already has verity enabled 176- ``EFAULT``: the caller provided inaccessible memory 177- ``EFBIG``: the file is too large to enable verity on 178- ``EINTR``: the operation was interrupted by a fatal signal 179- ``EINVAL``: unsupported version, hash algorithm, or block size; or 180 reserved bits are set; or the file descriptor refers to neither a 181 regular file nor a directory. 182- ``EISDIR``: the file descriptor refers to a directory 183- ``EKEYREJECTED``: the builtin signature doesn't match the file 184- ``EMSGSIZE``: the salt or builtin signature is too long 185- ``ENOKEY``: the ".fs-verity" keyring doesn't contain the certificate 186 needed to verify the builtin signature 187- ``ENOPKG``: fs-verity recognizes the hash algorithm, but it's not 188 available in the kernel's crypto API as currently configured (e.g. 189 for SHA-512, missing CONFIG_CRYPTO_SHA512). 190- ``ENOTTY``: this type of filesystem does not implement fs-verity 191- ``EOPNOTSUPP``: the kernel was not configured with fs-verity 192 support; or the filesystem superblock has not had the 'verity' 193 feature enabled on it; or the filesystem does not support fs-verity 194 on this file. (See `Filesystem support`_.) 195- ``EPERM``: the file is append-only; or, a builtin signature is 196 required and one was not provided. 197- ``EROFS``: the filesystem is read-only 198- ``ETXTBSY``: someone has the file open for writing. This can be the 199 caller's file descriptor, another open file descriptor, or the file 200 reference held by a writable memory map. 201 202FS_IOC_MEASURE_VERITY 203--------------------- 204 205The FS_IOC_MEASURE_VERITY ioctl retrieves the digest of a verity file. 206The fs-verity file digest is a cryptographic digest that identifies 207the file contents that are being enforced on reads; it is computed via 208a Merkle tree and is different from a traditional full-file digest. 209 210This ioctl takes in a pointer to a variable-length structure:: 211 212 struct fsverity_digest { 213 __u16 digest_algorithm; 214 __u16 digest_size; /* input/output */ 215 __u8 digest[]; 216 }; 217 218``digest_size`` is an input/output field. On input, it must be 219initialized to the number of bytes allocated for the variable-length 220``digest`` field. 221 222On success, 0 is returned and the kernel fills in the structure as 223follows: 224 225- ``digest_algorithm`` will be the hash algorithm used for the file 226 digest. It will match ``fsverity_enable_arg::hash_algorithm``. 227- ``digest_size`` will be the size of the digest in bytes, e.g. 32 228 for SHA-256. (This can be redundant with ``digest_algorithm``.) 229- ``digest`` will be the actual bytes of the digest. 230 231FS_IOC_MEASURE_VERITY is guaranteed to execute in constant time, 232regardless of the size of the file. 233 234FS_IOC_MEASURE_VERITY can fail with the following errors: 235 236- ``EFAULT``: the caller provided inaccessible memory 237- ``ENODATA``: the file is not a verity file 238- ``ENOTTY``: this type of filesystem does not implement fs-verity 239- ``EOPNOTSUPP``: the kernel was not configured with fs-verity 240 support, or the filesystem superblock has not had the 'verity' 241 feature enabled on it. (See `Filesystem support`_.) 242- ``EOVERFLOW``: the digest is longer than the specified 243 ``digest_size`` bytes. Try providing a larger buffer. 244 245FS_IOC_READ_VERITY_METADATA 246--------------------------- 247 248The FS_IOC_READ_VERITY_METADATA ioctl reads verity metadata from a 249verity file. This ioctl is available since Linux v5.12. 250 251This ioctl allows writing a server program that takes a verity file 252and serves it to a client program, such that the client can do its own 253fs-verity compatible verification of the file. This only makes sense 254if the client doesn't trust the server and if the server needs to 255provide the storage for the client. 256 257This is a fairly specialized use case, and most fs-verity users won't 258need this ioctl. 259 260This ioctl takes in a pointer to the following structure:: 261 262 #define FS_VERITY_METADATA_TYPE_MERKLE_TREE 1 263 #define FS_VERITY_METADATA_TYPE_DESCRIPTOR 2 264 #define FS_VERITY_METADATA_TYPE_SIGNATURE 3 265 266 struct fsverity_read_metadata_arg { 267 __u64 metadata_type; 268 __u64 offset; 269 __u64 length; 270 __u64 buf_ptr; 271 __u64 __reserved; 272 }; 273 274``metadata_type`` specifies the type of metadata to read: 275 276- ``FS_VERITY_METADATA_TYPE_MERKLE_TREE`` reads the blocks of the 277 Merkle tree. The blocks are returned in order from the root level 278 to the leaf level. Within each level, the blocks are returned in 279 the same order that their hashes are themselves hashed. 280 See `Merkle tree`_ for more information. 281 282- ``FS_VERITY_METADATA_TYPE_DESCRIPTOR`` reads the fs-verity 283 descriptor. See `fs-verity descriptor`_. 284 285- ``FS_VERITY_METADATA_TYPE_SIGNATURE`` reads the builtin signature 286 which was passed to FS_IOC_ENABLE_VERITY, if any. See `Built-in 287 signature verification`_. 288 289The semantics are similar to those of ``pread()``. ``offset`` 290specifies the offset in bytes into the metadata item to read from, and 291``length`` specifies the maximum number of bytes to read from the 292metadata item. ``buf_ptr`` is the pointer to the buffer to read into, 293cast to a 64-bit integer. ``__reserved`` must be 0. On success, the 294number of bytes read is returned. 0 is returned at the end of the 295metadata item. The returned length may be less than ``length``, for 296example if the ioctl is interrupted. 297 298The metadata returned by FS_IOC_READ_VERITY_METADATA isn't guaranteed 299to be authenticated against the file digest that would be returned by 300`FS_IOC_MEASURE_VERITY`_, as the metadata is expected to be used to 301implement fs-verity compatible verification anyway (though absent a 302malicious disk, the metadata will indeed match). E.g. to implement 303this ioctl, the filesystem is allowed to just read the Merkle tree 304blocks from disk without actually verifying the path to the root node. 305 306FS_IOC_READ_VERITY_METADATA can fail with the following errors: 307 308- ``EFAULT``: the caller provided inaccessible memory 309- ``EINTR``: the ioctl was interrupted before any data was read 310- ``EINVAL``: reserved fields were set, or ``offset + length`` 311 overflowed 312- ``ENODATA``: the file is not a verity file, or 313 FS_VERITY_METADATA_TYPE_SIGNATURE was requested but the file doesn't 314 have a builtin signature 315- ``ENOTTY``: this type of filesystem does not implement fs-verity, or 316 this ioctl is not yet implemented on it 317- ``EOPNOTSUPP``: the kernel was not configured with fs-verity 318 support, or the filesystem superblock has not had the 'verity' 319 feature enabled on it. (See `Filesystem support`_.) 320 321FS_IOC_GETFLAGS 322--------------- 323 324The existing ioctl FS_IOC_GETFLAGS (which isn't specific to fs-verity) 325can also be used to check whether a file has fs-verity enabled or not. 326To do so, check for FS_VERITY_FL (0x00100000) in the returned flags. 327 328The verity flag is not settable via FS_IOC_SETFLAGS. You must use 329FS_IOC_ENABLE_VERITY instead, since parameters must be provided. 330 331statx 332----- 333 334Since Linux v5.5, the statx() system call sets STATX_ATTR_VERITY if 335the file has fs-verity enabled. This can perform better than 336FS_IOC_GETFLAGS and FS_IOC_MEASURE_VERITY because it doesn't require 337opening the file, and opening verity files can be expensive. 338 339.. _accessing_verity_files: 340 341Accessing verity files 342====================== 343 344Applications can transparently access a verity file just like a 345non-verity one, with the following exceptions: 346 347- Verity files are readonly. They cannot be opened for writing or 348 truncate()d, even if the file mode bits allow it. Attempts to do 349 one of these things will fail with EPERM. However, changes to 350 metadata such as owner, mode, timestamps, and xattrs are still 351 allowed, since these are not measured by fs-verity. Verity files 352 can also still be renamed, deleted, and linked to. 353 354- Direct I/O is not supported on verity files. Attempts to use direct 355 I/O on such files will fall back to buffered I/O. 356 357- DAX (Direct Access) is not supported on verity files, because this 358 would circumvent the data verification. 359 360- Reads of data that doesn't match the verity Merkle tree will fail 361 with EIO (for read()) or SIGBUS (for mmap() reads). 362 363- If the sysctl "fs.verity.require_signatures" is set to 1 and the 364 file is not signed by a key in the ".fs-verity" keyring, then 365 opening the file will fail. See `Built-in signature verification`_. 366 367Direct access to the Merkle tree is not supported. Therefore, if a 368verity file is copied, or is backed up and restored, then it will lose 369its "verity"-ness. fs-verity is primarily meant for files like 370executables that are managed by a package manager. 371 372File digest computation 373======================= 374 375This section describes how fs-verity hashes the file contents using a 376Merkle tree to produce the digest which cryptographically identifies 377the file contents. This algorithm is the same for all filesystems 378that support fs-verity. 379 380Userspace only needs to be aware of this algorithm if it needs to 381compute fs-verity file digests itself, e.g. in order to sign files. 382 383.. _fsverity_merkle_tree: 384 385Merkle tree 386----------- 387 388The file contents is divided into blocks, where the block size is 389configurable but is usually 4096 bytes. The end of the last block is 390zero-padded if needed. Each block is then hashed, producing the first 391level of hashes. Then, the hashes in this first level are grouped 392into 'blocksize'-byte blocks (zero-padding the ends as needed) and 393these blocks are hashed, producing the second level of hashes. This 394proceeds up the tree until only a single block remains. The hash of 395this block is the "Merkle tree root hash". 396 397If the file fits in one block and is nonempty, then the "Merkle tree 398root hash" is simply the hash of the single data block. If the file 399is empty, then the "Merkle tree root hash" is all zeroes. 400 401The "blocks" here are not necessarily the same as "filesystem blocks". 402 403If a salt was specified, then it's zero-padded to the closest multiple 404of the input size of the hash algorithm's compression function, e.g. 40564 bytes for SHA-256 or 128 bytes for SHA-512. The padded salt is 406prepended to every data or Merkle tree block that is hashed. 407 408The purpose of the block padding is to cause every hash to be taken 409over the same amount of data, which simplifies the implementation and 410keeps open more possibilities for hardware acceleration. The purpose 411of the salt padding is to make the salting "free" when the salted hash 412state is precomputed, then imported for each hash. 413 414Example: in the recommended configuration of SHA-256 and 4K blocks, 415128 hash values fit in each block. Thus, each level of the Merkle 416tree is approximately 128 times smaller than the previous, and for 417large files the Merkle tree's size converges to approximately 1/127 of 418the original file size. However, for small files, the padding is 419significant, making the space overhead proportionally more. 420 421.. _fsverity_descriptor: 422 423fs-verity descriptor 424-------------------- 425 426By itself, the Merkle tree root hash is ambiguous. For example, it 427can't a distinguish a large file from a small second file whose data 428is exactly the top-level hash block of the first file. Ambiguities 429also arise from the convention of padding to the next block boundary. 430 431To solve this problem, the fs-verity file digest is actually computed 432as a hash of the following structure, which contains the Merkle tree 433root hash as well as other fields such as the file size:: 434 435 struct fsverity_descriptor { 436 __u8 version; /* must be 1 */ 437 __u8 hash_algorithm; /* Merkle tree hash algorithm */ 438 __u8 log_blocksize; /* log2 of size of data and tree blocks */ 439 __u8 salt_size; /* size of salt in bytes; 0 if none */ 440 __le32 __reserved_0x04; /* must be 0 */ 441 __le64 data_size; /* size of file the Merkle tree is built over */ 442 __u8 root_hash[64]; /* Merkle tree root hash */ 443 __u8 salt[32]; /* salt prepended to each hashed block */ 444 __u8 __reserved[144]; /* must be 0's */ 445 }; 446 447Built-in signature verification 448=============================== 449 450CONFIG_FS_VERITY_BUILTIN_SIGNATURES=y adds supports for in-kernel 451verification of fs-verity builtin signatures. 452 453**IMPORTANT**! Please take great care before using this feature. 454It is not the only way to do signatures with fs-verity, and the 455alternatives (such as userspace signature verification, and IMA 456appraisal) can be much better. It's also easy to fall into a trap 457of thinking this feature solves more problems than it actually does. 458 459Enabling this option adds the following: 460 4611. At boot time, the kernel creates a keyring named ".fs-verity". The 462 root user can add trusted X.509 certificates to this keyring using 463 the add_key() system call. 464 4652. `FS_IOC_ENABLE_VERITY`_ accepts a pointer to a PKCS#7 formatted 466 detached signature in DER format of the file's fs-verity digest. 467 On success, the ioctl persists the signature alongside the Merkle 468 tree. Then, any time the file is opened, the kernel verifies the 469 file's actual digest against this signature, using the certificates 470 in the ".fs-verity" keyring. This verification happens as long as the 471 file's signature exists, regardless of the state of the sysctl variable 472 "fs.verity.require_signatures" described in the next item. The IPE LSM 473 relies on this behavior to recognize and label fsverity files 474 that contain a verified built-in fsverity signature. 475 4763. A new sysctl "fs.verity.require_signatures" is made available. 477 When set to 1, the kernel requires that all verity files have a 478 correctly signed digest as described in (2). 479 480The data that the signature as described in (2) must be a signature of 481is the fs-verity file digest in the following format:: 482 483 struct fsverity_formatted_digest { 484 char magic[8]; /* must be "FSVerity" */ 485 __le16 digest_algorithm; 486 __le16 digest_size; 487 __u8 digest[]; 488 }; 489 490That's it. It should be emphasized again that fs-verity builtin 491signatures are not the only way to do signatures with fs-verity. See 492`Use cases`_ for an overview of ways in which fs-verity can be used. 493fs-verity builtin signatures have some major limitations that should 494be carefully considered before using them: 495 496- Builtin signature verification does *not* make the kernel enforce 497 that any files actually have fs-verity enabled. Thus, it is not a 498 complete authentication policy. Currently, if it is used, one 499 way to complete the authentication policy is for trusted userspace 500 code to explicitly check whether files have fs-verity enabled with a 501 signature before they are accessed. (With 502 fs.verity.require_signatures=1, just checking whether fs-verity is 503 enabled suffices.) But, in this case the trusted userspace code 504 could just store the signature alongside the file and verify it 505 itself using a cryptographic library, instead of using this feature. 506 507- Another approach is to utilize fs-verity builtin signature 508 verification in conjunction with the IPE LSM, which supports defining 509 a kernel-enforced, system-wide authentication policy that allows only 510 files with a verified fs-verity builtin signature to perform certain 511 operations, such as execution. Note that IPE doesn't require 512 fs.verity.require_signatures=1. 513 Please refer to :doc:`IPE admin guide </admin-guide/LSM/ipe>` for 514 more details. 515 516- A file's builtin signature can only be set at the same time that 517 fs-verity is being enabled on the file. Changing or deleting the 518 builtin signature later requires re-creating the file. 519 520- Builtin signature verification uses the same set of public keys for 521 all fs-verity enabled files on the system. Different keys cannot be 522 trusted for different files; each key is all or nothing. 523 524- The sysctl fs.verity.require_signatures applies system-wide. 525 Setting it to 1 only works when all users of fs-verity on the system 526 agree that it should be set to 1. This limitation can prevent 527 fs-verity from being used in cases where it would be helpful. 528 529- Builtin signature verification can only use signature algorithms 530 that are supported by the kernel. For example, the kernel does not 531 yet support Ed25519, even though this is often the signature 532 algorithm that is recommended for new cryptographic designs. 533 534- fs-verity builtin signatures are in PKCS#7 format, and the public 535 keys are in X.509 format. These formats are commonly used, 536 including by some other kernel features (which is why the fs-verity 537 builtin signatures use them), and are very feature rich. 538 Unfortunately, history has shown that code that parses and handles 539 these formats (which are from the 1990s and are based on ASN.1) 540 often has vulnerabilities as a result of their complexity. This 541 complexity is not inherent to the cryptography itself. 542 543 fs-verity users who do not need advanced features of X.509 and 544 PKCS#7 should strongly consider using simpler formats, such as plain 545 Ed25519 keys and signatures, and verifying signatures in userspace. 546 547 fs-verity users who choose to use X.509 and PKCS#7 anyway should 548 still consider that verifying those signatures in userspace is more 549 flexible (for other reasons mentioned earlier in this document) and 550 eliminates the need to enable CONFIG_FS_VERITY_BUILTIN_SIGNATURES 551 and its associated increase in kernel attack surface. In some cases 552 it can even be necessary, since advanced X.509 and PKCS#7 features 553 do not always work as intended with the kernel. For example, the 554 kernel does not check X.509 certificate validity times. 555 556 Note: IMA appraisal, which supports fs-verity, does not use PKCS#7 557 for its signatures, so it partially avoids the issues discussed 558 here. IMA appraisal does use X.509. 559 560Filesystem support 561================== 562 563fs-verity is supported by several filesystems, described below. The 564CONFIG_FS_VERITY kconfig option must be enabled to use fs-verity on 565any of these filesystems. 566 567``include/linux/fsverity.h`` declares the interface between the 568``fs/verity/`` support layer and filesystems. Briefly, filesystems 569must provide an ``fsverity_operations`` structure that provides 570methods to read and write the verity metadata to a filesystem-specific 571location, including the Merkle tree blocks and 572``fsverity_descriptor``. Filesystems must also call functions in 573``fs/verity/`` at certain times, such as when a file is opened or when 574pages have been read into the pagecache. (See `Verifying data`_.) 575 576ext4 577---- 578 579ext4 supports fs-verity since Linux v5.4 and e2fsprogs v1.45.2. 580 581To create verity files on an ext4 filesystem, the filesystem must have 582been formatted with ``-O verity`` or had ``tune2fs -O verity`` run on 583it. "verity" is an RO_COMPAT filesystem feature, so once set, old 584kernels will only be able to mount the filesystem readonly, and old 585versions of e2fsck will be unable to check the filesystem. 586 587Originally, an ext4 filesystem with the "verity" feature could only be 588mounted when its block size was equal to the system page size 589(typically 4096 bytes). In Linux v6.3, this limitation was removed. 590 591ext4 sets the EXT4_VERITY_FL on-disk inode flag on verity files. It 592can only be set by `FS_IOC_ENABLE_VERITY`_, and it cannot be cleared. 593 594ext4 also supports encryption, which can be used simultaneously with 595fs-verity. In this case, the plaintext data is verified rather than 596the ciphertext. This is necessary in order to make the fs-verity file 597digest meaningful, since every file is encrypted differently. 598 599ext4 stores the verity metadata (Merkle tree and fsverity_descriptor) 600past the end of the file, starting at the first 64K boundary beyond 601i_size. This approach works because (a) verity files are readonly, 602and (b) pages fully beyond i_size aren't visible to userspace but can 603be read/written internally by ext4 with only some relatively small 604changes to ext4. This approach avoids having to depend on the 605EA_INODE feature and on rearchitecturing ext4's xattr support to 606support paging multi-gigabyte xattrs into memory, and to support 607encrypting xattrs. Note that the verity metadata *must* be encrypted 608when the file is, since it contains hashes of the plaintext data. 609 610ext4 only allows verity on extent-based files. 611 612f2fs 613---- 614 615f2fs supports fs-verity since Linux v5.4 and f2fs-tools v1.11.0. 616 617To create verity files on an f2fs filesystem, the filesystem must have 618been formatted with ``-O verity``. 619 620f2fs sets the FADVISE_VERITY_BIT on-disk inode flag on verity files. 621It can only be set by `FS_IOC_ENABLE_VERITY`_, and it cannot be 622cleared. 623 624Like ext4, f2fs stores the verity metadata (Merkle tree and 625fsverity_descriptor) past the end of the file, starting at the first 62664K boundary beyond i_size. See explanation for ext4 above. 627Moreover, f2fs supports at most 4096 bytes of xattr entries per inode 628which usually wouldn't be enough for even a single Merkle tree block. 629 630f2fs doesn't support enabling verity on files that currently have 631atomic or volatile writes pending. 632 633btrfs 634----- 635 636btrfs supports fs-verity since Linux v5.15. Verity-enabled inodes are 637marked with a RO_COMPAT inode flag, and the verity metadata is stored 638in separate btree items. 639 640Implementation details 641====================== 642 643Verifying data 644-------------- 645 646fs-verity ensures that all reads of a verity file's data are verified, 647regardless of which syscall is used to do the read (e.g. mmap(), 648read(), pread()) and regardless of whether it's the first read or a 649later read (unless the later read can return cached data that was 650already verified). Below, we describe how filesystems implement this. 651 652Pagecache 653~~~~~~~~~ 654 655For filesystems using Linux's pagecache, the ``->read_folio()`` and 656``->readahead()`` methods must be modified to verify folios before 657they are marked Uptodate. Merely hooking ``->read_iter()`` would be 658insufficient, since ``->read_iter()`` is not used for memory maps. 659 660Therefore, fs/verity/ provides the function fsverity_verify_blocks() 661which verifies data that has been read into the pagecache of a verity 662inode. The containing folio must still be locked and not Uptodate, so 663it's not yet readable by userspace. As needed to do the verification, 664fsverity_verify_blocks() will call back into the filesystem to read 665hash blocks via fsverity_operations::read_merkle_tree_page(). 666 667fsverity_verify_blocks() returns false if verification failed; in this 668case, the filesystem must not set the folio Uptodate. Following this, 669as per the usual Linux pagecache behavior, attempts by userspace to 670read() from the part of the file containing the folio will fail with 671EIO, and accesses to the folio within a memory map will raise SIGBUS. 672 673In principle, verifying a data block requires verifying the entire 674path in the Merkle tree from the data block to the root hash. 675However, for efficiency the filesystem may cache the hash blocks. 676Therefore, fsverity_verify_blocks() only ascends the tree reading hash 677blocks until an already-verified hash block is seen. It then verifies 678the path to that block. 679 680This optimization, which is also used by dm-verity, results in 681excellent sequential read performance. This is because usually (e.g. 682127 in 128 times for 4K blocks and SHA-256) the hash block from the 683bottom level of the tree will already be cached and checked from 684reading a previous data block. However, random reads perform worse. 685 686Block device based filesystems 687~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 688 689Block device based filesystems (e.g. ext4 and f2fs) in Linux also use 690the pagecache, so the above subsection applies too. However, they 691also usually read many data blocks from a file at once, grouped into a 692structure called a "bio". To make it easier for these types of 693filesystems to support fs-verity, fs/verity/ also provides a function 694fsverity_verify_bio() which verifies all data blocks in a bio. 695 696ext4 and f2fs also support encryption. If a verity file is also 697encrypted, the data must be decrypted before being verified. To 698support this, these filesystems allocate a "post-read context" for 699each bio and store it in ``->bi_private``:: 700 701 struct bio_post_read_ctx { 702 struct bio *bio; 703 struct work_struct work; 704 unsigned int cur_step; 705 unsigned int enabled_steps; 706 }; 707 708``enabled_steps`` is a bitmask that specifies whether decryption, 709verity, or both is enabled. After the bio completes, for each needed 710postprocessing step the filesystem enqueues the bio_post_read_ctx on a 711workqueue, and then the workqueue work does the decryption or 712verification. Finally, folios where no decryption or verity error 713occurred are marked Uptodate, and the folios are unlocked. 714 715On many filesystems, files can contain holes. Normally, 716``->readahead()`` simply zeroes hole blocks and considers the 717corresponding data to be up-to-date; no bios are issued. To prevent 718this case from bypassing fs-verity, filesystems use 719fsverity_verify_blocks() to verify hole blocks. 720 721Filesystems also disable direct I/O on verity files, since otherwise 722direct I/O would bypass fs-verity. 723 724Userspace utility 725================= 726 727This document focuses on the kernel, but a userspace utility for 728fs-verity can be found at: 729 730 https://git.kernel.org/pub/scm/fs/fsverity/fsverity-utils.git 731 732See the README.md file in the fsverity-utils source tree for details, 733including examples of setting up fs-verity protected files. 734 735Tests 736===== 737 738To test fs-verity, use xfstests. For example, using `kvm-xfstests 739<https://github.com/tytso/xfstests-bld/blob/master/Documentation/kvm-quickstart.md>`_:: 740 741 kvm-xfstests -c ext4,f2fs,btrfs -g verity 742 743FAQ 744=== 745 746This section answers frequently asked questions about fs-verity that 747weren't already directly answered in other parts of this document. 748 749:Q: Why isn't fs-verity part of IMA? 750:A: fs-verity and IMA (Integrity Measurement Architecture) have 751 different focuses. fs-verity is a filesystem-level mechanism for 752 hashing individual files using a Merkle tree. In contrast, IMA 753 specifies a system-wide policy that specifies which files are 754 hashed and what to do with those hashes, such as log them, 755 authenticate them, or add them to a measurement list. 756 757 IMA supports the fs-verity hashing mechanism as an alternative 758 to full file hashes, for those who want the performance and 759 security benefits of the Merkle tree based hash. However, it 760 doesn't make sense to force all uses of fs-verity to be through 761 IMA. fs-verity already meets many users' needs even as a 762 standalone filesystem feature, and it's testable like other 763 filesystem features e.g. with xfstests. 764 765:Q: Isn't fs-verity useless because the attacker can just modify the 766 hashes in the Merkle tree, which is stored on-disk? 767:A: To verify the authenticity of an fs-verity file you must verify 768 the authenticity of the "fs-verity file digest", which 769 incorporates the root hash of the Merkle tree. See `Use cases`_. 770 771:Q: Isn't fs-verity useless because the attacker can just replace a 772 verity file with a non-verity one? 773:A: See `Use cases`_. In the initial use case, it's really trusted 774 userspace code that authenticates the files; fs-verity is just a 775 tool to do this job efficiently and securely. The trusted 776 userspace code will consider non-verity files to be inauthentic. 777 778:Q: Why does the Merkle tree need to be stored on-disk? Couldn't you 779 store just the root hash? 780:A: If the Merkle tree wasn't stored on-disk, then you'd have to 781 compute the entire tree when the file is first accessed, even if 782 just one byte is being read. This is a fundamental consequence of 783 how Merkle tree hashing works. To verify a leaf node, you need to 784 verify the whole path to the root hash, including the root node 785 (the thing which the root hash is a hash of). But if the root 786 node isn't stored on-disk, you have to compute it by hashing its 787 children, and so on until you've actually hashed the entire file. 788 789 That defeats most of the point of doing a Merkle tree-based hash, 790 since if you have to hash the whole file ahead of time anyway, 791 then you could simply do sha256(file) instead. That would be much 792 simpler, and a bit faster too. 793 794 It's true that an in-memory Merkle tree could still provide the 795 advantage of verification on every read rather than just on the 796 first read. However, it would be inefficient because every time a 797 hash page gets evicted (you can't pin the entire Merkle tree into 798 memory, since it may be very large), in order to restore it you 799 again need to hash everything below it in the tree. This again 800 defeats most of the point of doing a Merkle tree-based hash, since 801 a single block read could trigger re-hashing gigabytes of data. 802 803:Q: But couldn't you store just the leaf nodes and compute the rest? 804:A: See previous answer; this really just moves up one level, since 805 one could alternatively interpret the data blocks as being the 806 leaf nodes of the Merkle tree. It's true that the tree can be 807 computed much faster if the leaf level is stored rather than just 808 the data, but that's only because each level is less than 1% the 809 size of the level below (assuming the recommended settings of 810 SHA-256 and 4K blocks). For the exact same reason, by storing 811 "just the leaf nodes" you'd already be storing over 99% of the 812 tree, so you might as well simply store the whole tree. 813 814:Q: Can the Merkle tree be built ahead of time, e.g. distributed as 815 part of a package that is installed to many computers? 816:A: This isn't currently supported. It was part of the original 817 design, but was removed to simplify the kernel UAPI and because it 818 wasn't a critical use case. Files are usually installed once and 819 used many times, and cryptographic hashing is somewhat fast on 820 most modern processors. 821 822:Q: Why doesn't fs-verity support writes? 823:A: Write support would be very difficult and would require a 824 completely different design, so it's well outside the scope of 825 fs-verity. Write support would require: 826 827 - A way to maintain consistency between the data and hashes, 828 including all levels of hashes, since corruption after a crash 829 (especially of potentially the entire file!) is unacceptable. 830 The main options for solving this are data journalling, 831 copy-on-write, and log-structured volume. But it's very hard to 832 retrofit existing filesystems with new consistency mechanisms. 833 Data journalling is available on ext4, but is very slow. 834 835 - Rebuilding the Merkle tree after every write, which would be 836 extremely inefficient. Alternatively, a different authenticated 837 dictionary structure such as an "authenticated skiplist" could 838 be used. However, this would be far more complex. 839 840 Compare it to dm-verity vs. dm-integrity. dm-verity is very 841 simple: the kernel just verifies read-only data against a 842 read-only Merkle tree. In contrast, dm-integrity supports writes 843 but is slow, is much more complex, and doesn't actually support 844 full-device authentication since it authenticates each sector 845 independently, i.e. there is no "root hash". It doesn't really 846 make sense for the same device-mapper target to support these two 847 very different cases; the same applies to fs-verity. 848 849:Q: Since verity files are immutable, why isn't the immutable bit set? 850:A: The existing "immutable" bit (FS_IMMUTABLE_FL) already has a 851 specific set of semantics which not only make the file contents 852 read-only, but also prevent the file from being deleted, renamed, 853 linked to, or having its owner or mode changed. These extra 854 properties are unwanted for fs-verity, so reusing the immutable 855 bit isn't appropriate. 856 857:Q: Why does the API use ioctls instead of setxattr() and getxattr()? 858:A: Abusing the xattr interface for basically arbitrary syscalls is 859 heavily frowned upon by most of the Linux filesystem developers. 860 An xattr should really just be an xattr on-disk, not an API to 861 e.g. magically trigger construction of a Merkle tree. 862 863:Q: Does fs-verity support remote filesystems? 864:A: So far all filesystems that have implemented fs-verity support are 865 local filesystems, but in principle any filesystem that can store 866 per-file verity metadata can support fs-verity, regardless of 867 whether it's local or remote. Some filesystems may have fewer 868 options of where to store the verity metadata; one possibility is 869 to store it past the end of the file and "hide" it from userspace 870 by manipulating i_size. The data verification functions provided 871 by ``fs/verity/`` also assume that the filesystem uses the Linux 872 pagecache, but both local and remote filesystems normally do so. 873 874:Q: Why is anything filesystem-specific at all? Shouldn't fs-verity 875 be implemented entirely at the VFS level? 876:A: There are many reasons why this is not possible or would be very 877 difficult, including the following: 878 879 - To prevent bypassing verification, folios must not be marked 880 Uptodate until they've been verified. Currently, each 881 filesystem is responsible for marking folios Uptodate via 882 ``->readahead()``. Therefore, currently it's not possible for 883 the VFS to do the verification on its own. Changing this would 884 require significant changes to the VFS and all filesystems. 885 886 - It would require defining a filesystem-independent way to store 887 the verity metadata. Extended attributes don't work for this 888 because (a) the Merkle tree may be gigabytes, but many 889 filesystems assume that all xattrs fit into a single 4K 890 filesystem block, and (b) ext4 and f2fs encryption doesn't 891 encrypt xattrs, yet the Merkle tree *must* be encrypted when the 892 file contents are, because it stores hashes of the plaintext 893 file contents. 894 895 So the verity metadata would have to be stored in an actual 896 file. Using a separate file would be very ugly, since the 897 metadata is fundamentally part of the file to be protected, and 898 it could cause problems where users could delete the real file 899 but not the metadata file or vice versa. On the other hand, 900 having it be in the same file would break applications unless 901 filesystems' notion of i_size were divorced from the VFS's, 902 which would be complex and require changes to all filesystems. 903 904 - It's desirable that FS_IOC_ENABLE_VERITY uses the filesystem's 905 transaction mechanism so that either the file ends up with 906 verity enabled, or no changes were made. Allowing intermediate 907 states to occur after a crash may cause problems. 908