1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/refcount.h> 21 #include <linux/types.h> 22 #include <linux/blkdev.h> 23 #include <linux/magic.h> 24 #include <linux/jbd2.h> 25 #include <linux/quota.h> 26 #include <linux/rwsem.h> 27 #include <linux/rbtree.h> 28 #include <linux/seqlock.h> 29 #include <linux/mutex.h> 30 #include <linux/timer.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #include <linux/fiemap.h> 40 #ifdef __KERNEL__ 41 #include <linux/compat.h> 42 #endif 43 #include <uapi/linux/ext4.h> 44 45 #include <linux/fscrypt.h> 46 #include <linux/fsverity.h> 47 48 #include <linux/compiler.h> 49 50 /* 51 * The fourth extended filesystem constants/structures 52 */ 53 54 /* 55 * with AGGRESSIVE_CHECK allocator runs consistency checks over 56 * structures. these checks slow things down a lot 57 */ 58 #define AGGRESSIVE_CHECK__ 59 60 /* 61 * with DOUBLE_CHECK defined mballoc creates persistent in-core 62 * bitmaps, maintains and uses them to check for double allocations 63 */ 64 #define DOUBLE_CHECK__ 65 66 /* 67 * Define EXT4FS_DEBUG to produce debug messages 68 */ 69 #undef EXT4FS_DEBUG 70 71 /* 72 * Debug code 73 */ 74 #ifdef EXT4FS_DEBUG 75 #define ext4_debug(f, a...) \ 76 do { \ 77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 78 __FILE__, __LINE__, __func__); \ 79 printk(KERN_DEBUG f, ## a); \ 80 } while (0) 81 #else 82 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 83 #endif 84 85 /* 86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 87 */ 88 #define EXT_DEBUG__ 89 90 /* 91 * Dynamic printk for controlled extents debugging. 92 */ 93 #ifdef CONFIG_EXT4_DEBUG 94 #define ext_debug(ino, fmt, ...) \ 95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 96 current->comm, task_pid_nr(current), \ 97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 98 __func__, ##__VA_ARGS__) 99 #else 100 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 101 #endif 102 103 #define ASSERT(assert) \ 104 do { \ 105 if (unlikely(!(assert))) { \ 106 printk(KERN_EMERG \ 107 "Assertion failure in %s() at %s:%d: '%s'\n", \ 108 __func__, __FILE__, __LINE__, #assert); \ 109 BUG(); \ 110 } \ 111 } while (0) 112 113 /* data type for block offset of block group */ 114 typedef int ext4_grpblk_t; 115 116 /* data type for filesystem-wide blocks number */ 117 typedef unsigned long long ext4_fsblk_t; 118 119 /* data type for file logical block number */ 120 typedef __u32 ext4_lblk_t; 121 122 /* data type for block group number */ 123 typedef unsigned int ext4_group_t; 124 125 enum SHIFT_DIRECTION { 126 SHIFT_LEFT = 0, 127 SHIFT_RIGHT, 128 }; 129 130 /* 131 * For each criteria, mballoc has slightly different way of finding 132 * the required blocks nad usually, higher the criteria the slower the 133 * allocation. We start at lower criterias and keep falling back to 134 * higher ones if we are not able to find any blocks. Lower (earlier) 135 * criteria are faster. 136 */ 137 enum criteria { 138 /* 139 * Used when number of blocks needed is a power of 2. This 140 * doesn't trigger any disk IO except prefetch and is the 141 * fastest criteria. 142 */ 143 CR_POWER2_ALIGNED, 144 145 /* 146 * Tries to lookup in-memory data structures to find the most 147 * suitable group that satisfies goal request. No disk IO 148 * except block prefetch. 149 */ 150 CR_GOAL_LEN_FAST, 151 152 /* 153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal 154 * length to the best available length for faster allocation. 155 */ 156 CR_BEST_AVAIL_LEN, 157 158 /* 159 * Reads each block group sequentially, performing disk IO if 160 * necessary, to find find_suitable block group. Tries to 161 * allocate goal length but might trim the request if nothing 162 * is found after enough tries. 163 */ 164 CR_GOAL_LEN_SLOW, 165 166 /* 167 * Finds the first free set of blocks and allocates 168 * those. This is only used in rare cases when 169 * CR_GOAL_LEN_SLOW also fails to allocate anything. 170 */ 171 CR_ANY_FREE, 172 173 /* 174 * Number of criterias defined. 175 */ 176 EXT4_MB_NUM_CRS 177 }; 178 179 /* 180 * Flags used in mballoc's allocation_context flags field. 181 * 182 * Also used to show what's going on for debugging purposes when the 183 * flag field is exported via the traceport interface 184 */ 185 186 /* prefer goal again. length */ 187 #define EXT4_MB_HINT_MERGE 0x0001 188 /* blocks already reserved */ 189 #define EXT4_MB_HINT_RESERVED 0x0002 190 /* metadata is being allocated */ 191 #define EXT4_MB_HINT_METADATA 0x0004 192 /* first blocks in the file */ 193 #define EXT4_MB_HINT_FIRST 0x0008 194 /* search for the best chunk */ 195 #define EXT4_MB_HINT_BEST 0x0010 196 /* data is being allocated */ 197 #define EXT4_MB_HINT_DATA 0x0020 198 /* don't preallocate (for tails) */ 199 #define EXT4_MB_HINT_NOPREALLOC 0x0040 200 /* allocate for locality group */ 201 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 202 /* allocate goal blocks or none */ 203 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 204 /* goal is meaningful */ 205 #define EXT4_MB_HINT_TRY_GOAL 0x0200 206 /* blocks already pre-reserved by delayed allocation */ 207 #define EXT4_MB_DELALLOC_RESERVED 0x0400 208 /* We are doing stream allocation */ 209 #define EXT4_MB_STREAM_ALLOC 0x0800 210 /* Use reserved root blocks if needed */ 211 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 212 /* Use blocks from reserved pool */ 213 #define EXT4_MB_USE_RESERVED 0x2000 214 /* Do strict check for free blocks while retrying block allocation */ 215 #define EXT4_MB_STRICT_CHECK 0x4000 216 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 217 #define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000 218 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 219 #define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000 220 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1.5 */ 221 #define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000 222 223 struct ext4_allocation_request { 224 /* target inode for block we're allocating */ 225 struct inode *inode; 226 /* how many blocks we want to allocate */ 227 unsigned int len; 228 /* logical block in target inode */ 229 ext4_lblk_t logical; 230 /* the closest logical allocated block to the left */ 231 ext4_lblk_t lleft; 232 /* the closest logical allocated block to the right */ 233 ext4_lblk_t lright; 234 /* phys. target (a hint) */ 235 ext4_fsblk_t goal; 236 /* phys. block for the closest logical allocated block to the left */ 237 ext4_fsblk_t pleft; 238 /* phys. block for the closest logical allocated block to the right */ 239 ext4_fsblk_t pright; 240 /* flags. see above EXT4_MB_HINT_* */ 241 unsigned int flags; 242 }; 243 244 /* 245 * Logical to physical block mapping, used by ext4_map_blocks() 246 * 247 * This structure is used to pass requests into ext4_map_blocks() as 248 * well as to store the information returned by ext4_map_blocks(). It 249 * takes less room on the stack than a struct buffer_head. 250 */ 251 #define EXT4_MAP_NEW BIT(BH_New) 252 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 253 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 254 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 255 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 256 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 257 258 struct ext4_map_blocks { 259 ext4_fsblk_t m_pblk; 260 ext4_lblk_t m_lblk; 261 unsigned int m_len; 262 unsigned int m_flags; 263 }; 264 265 /* 266 * Block validity checking, system zone rbtree. 267 */ 268 struct ext4_system_blocks { 269 struct rb_root root; 270 struct rcu_head rcu; 271 }; 272 273 /* 274 * Flags for ext4_io_end->flags 275 */ 276 #define EXT4_IO_END_UNWRITTEN 0x0001 277 278 struct ext4_io_end_vec { 279 struct list_head list; /* list of io_end_vec */ 280 loff_t offset; /* offset in the file */ 281 ssize_t size; /* size of the extent */ 282 }; 283 284 /* 285 * For converting unwritten extents on a work queue. 'handle' is used for 286 * buffered writeback. 287 */ 288 typedef struct ext4_io_end { 289 struct list_head list; /* per-file finished IO list */ 290 handle_t *handle; /* handle reserved for extent 291 * conversion */ 292 struct inode *inode; /* file being written to */ 293 struct bio *bio; /* Linked list of completed 294 * bios covering the extent */ 295 unsigned int flag; /* unwritten or not */ 296 refcount_t count; /* reference counter */ 297 struct list_head list_vec; /* list of ext4_io_end_vec */ 298 } ext4_io_end_t; 299 300 struct ext4_io_submit { 301 struct writeback_control *io_wbc; 302 struct bio *io_bio; 303 ext4_io_end_t *io_end; 304 sector_t io_next_block; 305 }; 306 307 /* 308 * Special inodes numbers 309 */ 310 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 311 #define EXT4_ROOT_INO 2 /* Root inode */ 312 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 313 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 314 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 315 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 316 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 317 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 318 319 /* First non-reserved inode for old ext4 filesystems */ 320 #define EXT4_GOOD_OLD_FIRST_INO 11 321 322 /* 323 * Maximal count of links to a file 324 */ 325 #define EXT4_LINK_MAX 65000 326 327 /* 328 * Macro-instructions used to manage several block sizes 329 */ 330 #define EXT4_MIN_BLOCK_SIZE 1024 331 #define EXT4_MAX_BLOCK_SIZE 65536 332 #define EXT4_MIN_BLOCK_LOG_SIZE 10 333 #define EXT4_MAX_BLOCK_LOG_SIZE 16 334 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 335 #ifdef __KERNEL__ 336 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 337 #else 338 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 339 #endif 340 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 341 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 342 EXT4_SB(s)->s_cluster_bits) 343 #ifdef __KERNEL__ 344 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 345 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 346 #else 347 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 348 #endif 349 #ifdef __KERNEL__ 350 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 351 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 352 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 353 #else 354 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 355 EXT4_GOOD_OLD_INODE_SIZE : \ 356 (s)->s_inode_size) 357 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 358 EXT4_GOOD_OLD_FIRST_INO : \ 359 (s)->s_first_ino) 360 #endif 361 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 362 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 363 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 364 blkbits)) 365 366 /* Translate a block number to a cluster number */ 367 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 368 /* Translate a cluster number to a block number */ 369 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 370 /* Translate # of blks to # of clusters */ 371 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 372 (sbi)->s_cluster_bits) 373 /* Mask out the low bits to get the starting block of the cluster */ 374 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 375 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 376 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 377 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 378 /* Fill in the low bits to get the last block of the cluster */ 379 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 380 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 381 /* Get the cluster offset */ 382 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 383 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 384 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 385 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 386 387 /* 388 * Structure of a blocks group descriptor 389 */ 390 struct ext4_group_desc 391 { 392 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 393 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 394 __le32 bg_inode_table_lo; /* Inodes table block */ 395 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 396 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 397 __le16 bg_used_dirs_count_lo; /* Directories count */ 398 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 399 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 400 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 401 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 402 __le16 bg_itable_unused_lo; /* Unused inodes count */ 403 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 404 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 405 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 406 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 407 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 408 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 409 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 410 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 411 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 412 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 413 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 414 __u32 bg_reserved; 415 }; 416 417 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 418 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 419 sizeof(__le16)) 420 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 421 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 422 sizeof(__le16)) 423 424 /* 425 * Structure of a flex block group info 426 */ 427 428 struct flex_groups { 429 atomic64_t free_clusters; 430 atomic_t free_inodes; 431 atomic_t used_dirs; 432 }; 433 434 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 435 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 436 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 437 438 /* 439 * Macro-instructions used to manage group descriptors 440 */ 441 #define EXT4_MIN_DESC_SIZE 32 442 #define EXT4_MIN_DESC_SIZE_64BIT 64 443 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 444 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 445 #ifdef __KERNEL__ 446 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 447 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 448 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 449 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 450 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 451 #else 452 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 453 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 454 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 455 #endif 456 457 /* 458 * Constants relative to the data blocks 459 */ 460 #define EXT4_NDIR_BLOCKS 12 461 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 462 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 463 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 464 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 465 466 /* 467 * Inode flags 468 */ 469 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 470 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 471 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 472 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 473 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 474 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 475 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 476 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 477 /* Reserved for compression usage... */ 478 #define EXT4_DIRTY_FL 0x00000100 479 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 480 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 481 /* nb: was previously EXT2_ECOMPR_FL */ 482 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 483 /* End compression flags --- maybe not all used */ 484 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 485 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 486 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 487 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 488 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 489 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 490 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 491 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 492 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 493 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 494 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 495 496 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 497 498 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 499 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 500 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 501 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 502 503 /* User modifiable flags */ 504 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 505 EXT4_UNRM_FL | \ 506 EXT4_COMPR_FL | \ 507 EXT4_SYNC_FL | \ 508 EXT4_IMMUTABLE_FL | \ 509 EXT4_APPEND_FL | \ 510 EXT4_NODUMP_FL | \ 511 EXT4_NOATIME_FL | \ 512 EXT4_JOURNAL_DATA_FL | \ 513 EXT4_NOTAIL_FL | \ 514 EXT4_DIRSYNC_FL | \ 515 EXT4_TOPDIR_FL | \ 516 EXT4_EXTENTS_FL | \ 517 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 518 EXT4_DAX_FL | \ 519 EXT4_PROJINHERIT_FL | \ 520 EXT4_CASEFOLD_FL) 521 522 /* User visible flags */ 523 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 524 EXT4_DIRTY_FL | \ 525 EXT4_COMPRBLK_FL | \ 526 EXT4_NOCOMPR_FL | \ 527 EXT4_ENCRYPT_FL | \ 528 EXT4_INDEX_FL | \ 529 EXT4_VERITY_FL | \ 530 EXT4_INLINE_DATA_FL) 531 532 /* Flags that should be inherited by new inodes from their parent. */ 533 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 534 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 535 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 536 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 537 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 538 EXT4_DAX_FL) 539 540 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 541 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 542 EXT4_PROJINHERIT_FL)) 543 544 /* Flags that are appropriate for non-directories/regular files. */ 545 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 546 547 /* The only flags that should be swapped */ 548 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 549 550 /* Flags which are mutually exclusive to DAX */ 551 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 552 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 553 554 /* Mask out flags that are inappropriate for the given type of inode. */ 555 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 556 { 557 if (S_ISDIR(mode)) 558 return flags; 559 else if (S_ISREG(mode)) 560 return flags & EXT4_REG_FLMASK; 561 else 562 return flags & EXT4_OTHER_FLMASK; 563 } 564 565 /* 566 * Inode flags used for atomic set/get 567 */ 568 enum { 569 EXT4_INODE_SECRM = 0, /* Secure deletion */ 570 EXT4_INODE_UNRM = 1, /* Undelete */ 571 EXT4_INODE_COMPR = 2, /* Compress file */ 572 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 573 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 574 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 575 EXT4_INODE_NODUMP = 6, /* do not dump file */ 576 EXT4_INODE_NOATIME = 7, /* do not update atime */ 577 /* Reserved for compression usage... */ 578 EXT4_INODE_DIRTY = 8, 579 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 580 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 581 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 582 /* End compression flags --- maybe not all used */ 583 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 584 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 585 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 586 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 587 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 588 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 589 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 590 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 591 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 592 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 593 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 594 EXT4_INODE_DAX = 25, /* Inode is DAX */ 595 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 596 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 597 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 598 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 599 }; 600 601 /* 602 * Since it's pretty easy to mix up bit numbers and hex values, we use a 603 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 604 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 605 * any extra space in the compiled kernel image, otherwise, the build will fail. 606 * It's important that these values are the same, since we are using 607 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 608 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 609 * values found in ext2, ext3 and ext4 filesystems, and of course the values 610 * defined in e2fsprogs. 611 * 612 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 613 */ 614 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 615 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 616 617 static inline void ext4_check_flag_values(void) 618 { 619 CHECK_FLAG_VALUE(SECRM); 620 CHECK_FLAG_VALUE(UNRM); 621 CHECK_FLAG_VALUE(COMPR); 622 CHECK_FLAG_VALUE(SYNC); 623 CHECK_FLAG_VALUE(IMMUTABLE); 624 CHECK_FLAG_VALUE(APPEND); 625 CHECK_FLAG_VALUE(NODUMP); 626 CHECK_FLAG_VALUE(NOATIME); 627 CHECK_FLAG_VALUE(DIRTY); 628 CHECK_FLAG_VALUE(COMPRBLK); 629 CHECK_FLAG_VALUE(NOCOMPR); 630 CHECK_FLAG_VALUE(ENCRYPT); 631 CHECK_FLAG_VALUE(INDEX); 632 CHECK_FLAG_VALUE(IMAGIC); 633 CHECK_FLAG_VALUE(JOURNAL_DATA); 634 CHECK_FLAG_VALUE(NOTAIL); 635 CHECK_FLAG_VALUE(DIRSYNC); 636 CHECK_FLAG_VALUE(TOPDIR); 637 CHECK_FLAG_VALUE(HUGE_FILE); 638 CHECK_FLAG_VALUE(EXTENTS); 639 CHECK_FLAG_VALUE(VERITY); 640 CHECK_FLAG_VALUE(EA_INODE); 641 CHECK_FLAG_VALUE(INLINE_DATA); 642 CHECK_FLAG_VALUE(PROJINHERIT); 643 CHECK_FLAG_VALUE(CASEFOLD); 644 CHECK_FLAG_VALUE(RESERVED); 645 } 646 647 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 648 struct compat_ext4_new_group_input { 649 u32 group; 650 compat_u64 block_bitmap; 651 compat_u64 inode_bitmap; 652 compat_u64 inode_table; 653 u32 blocks_count; 654 u16 reserved_blocks; 655 u16 unused; 656 }; 657 #endif 658 659 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 660 struct ext4_new_group_data { 661 __u32 group; 662 __u64 block_bitmap; 663 __u64 inode_bitmap; 664 __u64 inode_table; 665 __u32 blocks_count; 666 __u16 reserved_blocks; 667 __u16 mdata_blocks; 668 __u32 free_clusters_count; 669 }; 670 671 /* Indexes used to index group tables in ext4_new_group_data */ 672 enum { 673 BLOCK_BITMAP = 0, /* block bitmap */ 674 INODE_BITMAP, /* inode bitmap */ 675 INODE_TABLE, /* inode tables */ 676 GROUP_TABLE_COUNT, 677 }; 678 679 /* 680 * Flags used by ext4_map_blocks() 681 */ 682 /* Allocate any needed blocks and/or convert an unwritten 683 extent to be an initialized ext4 */ 684 #define EXT4_GET_BLOCKS_CREATE 0x0001 685 /* Request the creation of an unwritten extent */ 686 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 687 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 688 EXT4_GET_BLOCKS_CREATE) 689 /* Caller is from the delayed allocation writeout path 690 * finally doing the actual allocation of delayed blocks */ 691 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 692 /* caller is from the direct IO path, request to creation of an 693 unwritten extents if not allocated, split the unwritten 694 extent if blocks has been preallocated already*/ 695 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 696 #define EXT4_GET_BLOCKS_CONVERT 0x0010 697 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 698 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 699 /* Convert extent to initialized after IO complete */ 700 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 701 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 702 /* Eventual metadata allocation (due to growing extent tree) 703 * should not fail, so try to use reserved blocks for that.*/ 704 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 705 /* Don't normalize allocation size (used for fallocate) */ 706 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 707 /* Convert written extents to unwritten */ 708 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 709 /* Write zeros to newly created written extents */ 710 #define EXT4_GET_BLOCKS_ZERO 0x0200 711 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 712 EXT4_GET_BLOCKS_ZERO) 713 /* Caller will submit data before dropping transaction handle. This 714 * allows jbd2 to avoid submitting data before commit. */ 715 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 716 /* Caller is in the atomic contex, find extent if it has been cached */ 717 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 718 719 /* 720 * The bit position of these flags must not overlap with any of the 721 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 722 * read_extent_tree_block(), ext4_split_extent_at(), 723 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 724 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 725 * caching the extents when reading from the extent tree while a 726 * truncate or punch hole operation is in progress. 727 */ 728 #define EXT4_EX_NOCACHE 0x40000000 729 #define EXT4_EX_FORCE_CACHE 0x20000000 730 #define EXT4_EX_NOFAIL 0x10000000 731 732 /* 733 * Flags used by ext4_free_blocks 734 */ 735 #define EXT4_FREE_BLOCKS_METADATA 0x0001 736 #define EXT4_FREE_BLOCKS_FORGET 0x0002 737 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 738 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 739 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 740 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 741 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 742 743 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 744 /* 745 * ioctl commands in 32 bit emulation 746 */ 747 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 748 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 749 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 750 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 751 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 752 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 753 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 754 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 755 #endif 756 757 /* Max physical block we can address w/o extents */ 758 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 759 760 /* Max logical block we can support */ 761 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 762 763 /* 764 * Structure of an inode on the disk 765 */ 766 struct ext4_inode { 767 __le16 i_mode; /* File mode */ 768 __le16 i_uid; /* Low 16 bits of Owner Uid */ 769 __le32 i_size_lo; /* Size in bytes */ 770 __le32 i_atime; /* Access time */ 771 __le32 i_ctime; /* Inode Change time */ 772 __le32 i_mtime; /* Modification time */ 773 __le32 i_dtime; /* Deletion Time */ 774 __le16 i_gid; /* Low 16 bits of Group Id */ 775 __le16 i_links_count; /* Links count */ 776 __le32 i_blocks_lo; /* Blocks count */ 777 __le32 i_flags; /* File flags */ 778 union { 779 struct { 780 __le32 l_i_version; 781 } linux1; 782 struct { 783 __u32 h_i_translator; 784 } hurd1; 785 struct { 786 __u32 m_i_reserved1; 787 } masix1; 788 } osd1; /* OS dependent 1 */ 789 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 790 __le32 i_generation; /* File version (for NFS) */ 791 __le32 i_file_acl_lo; /* File ACL */ 792 __le32 i_size_high; 793 __le32 i_obso_faddr; /* Obsoleted fragment address */ 794 union { 795 struct { 796 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 797 __le16 l_i_file_acl_high; 798 __le16 l_i_uid_high; /* these 2 fields */ 799 __le16 l_i_gid_high; /* were reserved2[0] */ 800 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 801 __le16 l_i_reserved; 802 } linux2; 803 struct { 804 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 805 __u16 h_i_mode_high; 806 __u16 h_i_uid_high; 807 __u16 h_i_gid_high; 808 __u32 h_i_author; 809 } hurd2; 810 struct { 811 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 812 __le16 m_i_file_acl_high; 813 __u32 m_i_reserved2[2]; 814 } masix2; 815 } osd2; /* OS dependent 2 */ 816 __le16 i_extra_isize; 817 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 818 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 819 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 820 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 821 __le32 i_crtime; /* File Creation time */ 822 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 823 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 824 __le32 i_projid; /* Project ID */ 825 }; 826 827 #define EXT4_EPOCH_BITS 2 828 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 829 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 830 831 /* 832 * Extended fields will fit into an inode if the filesystem was formatted 833 * with large inodes (-I 256 or larger) and there are not currently any EAs 834 * consuming all of the available space. For new inodes we always reserve 835 * enough space for the kernel's known extended fields, but for inodes 836 * created with an old kernel this might not have been the case. None of 837 * the extended inode fields is critical for correct filesystem operation. 838 * This macro checks if a certain field fits in the inode. Note that 839 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 840 */ 841 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 842 ((offsetof(typeof(*ext4_inode), field) + \ 843 sizeof((ext4_inode)->field)) \ 844 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 845 (einode)->i_extra_isize)) \ 846 847 /* 848 * We use an encoding that preserves the times for extra epoch "00": 849 * 850 * extra msb of adjust for signed 851 * epoch 32-bit 32-bit tv_sec to 852 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 853 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 854 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 855 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 856 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 857 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 858 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 859 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 860 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 861 * 862 * Note that previous versions of the kernel on 64-bit systems would 863 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 864 * 1970. e2fsck will correct this, assuming that it is run on the 865 * affected filesystem before 2242. 866 */ 867 868 static inline __le32 ext4_encode_extra_time(struct timespec64 ts) 869 { 870 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK; 871 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS)); 872 } 873 874 static inline struct timespec64 ext4_decode_extra_time(__le32 base, 875 __le32 extra) 876 { 877 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) }; 878 879 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 880 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 881 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 882 return ts; 883 } 884 885 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \ 886 do { \ 887 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 888 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \ 889 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \ 890 } else \ 891 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \ 892 } while (0) 893 894 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \ 895 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode)) 896 897 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \ 898 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode)) 899 900 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \ 901 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode)) 902 903 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 904 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 905 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \ 906 raw_inode, (einode)->xtime) 907 908 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \ 909 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \ 910 ext4_decode_extra_time((raw_inode)->xtime, \ 911 (raw_inode)->xtime ## _extra) : \ 912 (struct timespec64) { \ 913 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \ 914 }) 915 916 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \ 917 do { \ 918 inode_set_atime_to_ts(inode, \ 919 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \ 920 } while (0) 921 922 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \ 923 do { \ 924 inode_set_mtime_to_ts(inode, \ 925 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \ 926 } while (0) 927 928 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \ 929 do { \ 930 inode_set_ctime_to_ts(inode, \ 931 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \ 932 } while (0) 933 934 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 935 do { \ 936 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 937 (einode)->xtime = \ 938 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \ 939 raw_inode); \ 940 else \ 941 (einode)->xtime = (struct timespec64){0, 0}; \ 942 } while (0) 943 944 #define i_disk_version osd1.linux1.l_i_version 945 946 #if defined(__KERNEL__) || defined(__linux__) 947 #define i_reserved1 osd1.linux1.l_i_reserved1 948 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 949 #define i_blocks_high osd2.linux2.l_i_blocks_high 950 #define i_uid_low i_uid 951 #define i_gid_low i_gid 952 #define i_uid_high osd2.linux2.l_i_uid_high 953 #define i_gid_high osd2.linux2.l_i_gid_high 954 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 955 956 #elif defined(__GNU__) 957 958 #define i_translator osd1.hurd1.h_i_translator 959 #define i_uid_high osd2.hurd2.h_i_uid_high 960 #define i_gid_high osd2.hurd2.h_i_gid_high 961 #define i_author osd2.hurd2.h_i_author 962 963 #elif defined(__masix__) 964 965 #define i_reserved1 osd1.masix1.m_i_reserved1 966 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 967 #define i_reserved2 osd2.masix2.m_i_reserved2 968 969 #endif /* defined(__KERNEL__) || defined(__linux__) */ 970 971 #include "extents_status.h" 972 #include "fast_commit.h" 973 974 /* 975 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 976 * 977 * These are needed to avoid lockdep false positives when we need to 978 * allocate blocks to the quota inode during ext4_map_blocks(), while 979 * holding i_data_sem for a normal (non-quota) inode. Since we don't 980 * do quota tracking for the quota inode, this avoids deadlock (as 981 * well as infinite recursion, since it isn't turtles all the way 982 * down...) 983 * 984 * I_DATA_SEM_NORMAL - Used for most inodes 985 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 986 * where the second inode has larger inode number 987 * than the first 988 * I_DATA_SEM_QUOTA - Used for quota inodes only 989 * I_DATA_SEM_EA - Used for ea_inodes only 990 */ 991 enum { 992 I_DATA_SEM_NORMAL = 0, 993 I_DATA_SEM_OTHER, 994 I_DATA_SEM_QUOTA, 995 I_DATA_SEM_EA 996 }; 997 998 999 /* 1000 * fourth extended file system inode data in memory 1001 */ 1002 struct ext4_inode_info { 1003 __le32 i_data[15]; /* unconverted */ 1004 __u32 i_dtime; 1005 ext4_fsblk_t i_file_acl; 1006 1007 /* 1008 * i_block_group is the number of the block group which contains 1009 * this file's inode. Constant across the lifetime of the inode, 1010 * it is used for making block allocation decisions - we try to 1011 * place a file's data blocks near its inode block, and new inodes 1012 * near to their parent directory's inode. 1013 */ 1014 ext4_group_t i_block_group; 1015 ext4_lblk_t i_dir_start_lookup; 1016 #if (BITS_PER_LONG < 64) 1017 unsigned long i_state_flags; /* Dynamic state flags */ 1018 #endif 1019 unsigned long i_flags; 1020 1021 /* 1022 * Extended attributes can be read independently of the main file 1023 * data. Taking i_rwsem even when reading would cause contention 1024 * between readers of EAs and writers of regular file data, so 1025 * instead we synchronize on xattr_sem when reading or changing 1026 * EAs. 1027 */ 1028 struct rw_semaphore xattr_sem; 1029 1030 /* 1031 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1032 * i_orphan is used. 1033 */ 1034 union { 1035 struct list_head i_orphan; /* unlinked but open inodes */ 1036 unsigned int i_orphan_idx; /* Index in orphan file */ 1037 }; 1038 1039 /* Fast commit related info */ 1040 1041 /* For tracking dentry create updates */ 1042 struct list_head i_fc_dilist; 1043 struct list_head i_fc_list; /* 1044 * inodes that need fast commit 1045 * protected by sbi->s_fc_lock. 1046 */ 1047 1048 /* Start of lblk range that needs to be committed in this fast commit */ 1049 ext4_lblk_t i_fc_lblk_start; 1050 1051 /* End of lblk range that needs to be committed in this fast commit */ 1052 ext4_lblk_t i_fc_lblk_len; 1053 1054 /* Number of ongoing updates on this inode */ 1055 atomic_t i_fc_updates; 1056 1057 /* Fast commit wait queue for this inode */ 1058 wait_queue_head_t i_fc_wait; 1059 1060 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1061 struct mutex i_fc_lock; 1062 1063 /* 1064 * i_disksize keeps track of what the inode size is ON DISK, not 1065 * in memory. During truncate, i_size is set to the new size by 1066 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1067 * set i_disksize to 0 until the truncate is actually under way. 1068 * 1069 * The intent is that i_disksize always represents the blocks which 1070 * are used by this file. This allows recovery to restart truncate 1071 * on orphans if we crash during truncate. We actually write i_disksize 1072 * into the on-disk inode when writing inodes out, instead of i_size. 1073 * 1074 * The only time when i_disksize and i_size may be different is when 1075 * a truncate is in progress. The only things which change i_disksize 1076 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1077 */ 1078 loff_t i_disksize; 1079 1080 /* 1081 * i_data_sem is for serialising ext4_truncate() against 1082 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1083 * data tree are chopped off during truncate. We can't do that in 1084 * ext4 because whenever we perform intermediate commits during 1085 * truncate, the inode and all the metadata blocks *must* be in a 1086 * consistent state which allows truncation of the orphans to restart 1087 * during recovery. Hence we must fix the get_block-vs-truncate race 1088 * by other means, so we have i_data_sem. 1089 */ 1090 struct rw_semaphore i_data_sem; 1091 struct inode vfs_inode; 1092 struct jbd2_inode *jinode; 1093 1094 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1095 1096 /* 1097 * File creation time. Its function is same as that of 1098 * struct timespec64 i_{a,c,m}time in the generic inode. 1099 */ 1100 struct timespec64 i_crtime; 1101 1102 /* mballoc */ 1103 atomic_t i_prealloc_active; 1104 struct rb_root i_prealloc_node; 1105 rwlock_t i_prealloc_lock; 1106 1107 /* extents status tree */ 1108 struct ext4_es_tree i_es_tree; 1109 rwlock_t i_es_lock; 1110 struct list_head i_es_list; 1111 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1112 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1113 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1114 extents to shrink. Protected by 1115 i_es_lock */ 1116 1117 /* ialloc */ 1118 ext4_group_t i_last_alloc_group; 1119 1120 /* allocation reservation info for delalloc */ 1121 /* In case of bigalloc, this refer to clusters rather than blocks */ 1122 unsigned int i_reserved_data_blocks; 1123 1124 /* pending cluster reservations for bigalloc file systems */ 1125 struct ext4_pending_tree i_pending_tree; 1126 1127 /* on-disk additional length */ 1128 __u16 i_extra_isize; 1129 1130 /* Indicate the inline data space. */ 1131 u16 i_inline_off; 1132 u16 i_inline_size; 1133 1134 #ifdef CONFIG_QUOTA 1135 /* quota space reservation, managed internally by quota code */ 1136 qsize_t i_reserved_quota; 1137 #endif 1138 1139 /* Lock protecting lists below */ 1140 spinlock_t i_completed_io_lock; 1141 /* 1142 * Completed IOs that need unwritten extents handling and have 1143 * transaction reserved 1144 */ 1145 struct list_head i_rsv_conversion_list; 1146 struct work_struct i_rsv_conversion_work; 1147 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1148 1149 spinlock_t i_block_reservation_lock; 1150 1151 /* 1152 * Transactions that contain inode's metadata needed to complete 1153 * fsync and fdatasync, respectively. 1154 */ 1155 tid_t i_sync_tid; 1156 tid_t i_datasync_tid; 1157 1158 #ifdef CONFIG_QUOTA 1159 struct dquot *i_dquot[MAXQUOTAS]; 1160 #endif 1161 1162 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1163 __u32 i_csum_seed; 1164 1165 kprojid_t i_projid; 1166 }; 1167 1168 /* 1169 * File system states 1170 */ 1171 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1172 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1173 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1174 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1175 1176 /* 1177 * Misc. filesystem flags 1178 */ 1179 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1180 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1181 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1182 1183 /* 1184 * Mount flags set via mount options or defaults 1185 */ 1186 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1187 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1188 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1189 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1190 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1191 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1192 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1193 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1194 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1195 #ifdef CONFIG_FS_DAX 1196 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1197 #else 1198 #define EXT4_MOUNT_DAX_ALWAYS 0 1199 #endif 1200 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1201 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1202 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1203 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1204 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1205 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1206 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1207 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1208 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1209 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1210 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1211 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1212 * enable enforcement for hidden 1213 * quota files */ 1214 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1215 * enforcement for hidden quota 1216 * files */ 1217 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1218 * enforcement */ 1219 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1220 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1221 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1222 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1223 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1224 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1225 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1226 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1227 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1228 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1229 1230 /* 1231 * Mount flags set either automatically (could not be set by mount option) 1232 * based on per file system feature or property or in special cases such as 1233 * distinguishing between explicit mount option definition and default. 1234 */ 1235 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1236 specified delalloc */ 1237 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1238 size of blocksize * 8 1239 blocks */ 1240 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1241 file systems */ 1242 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1243 specified journal checksum */ 1244 1245 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1246 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1247 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1248 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1249 * scanning in mballoc 1250 */ 1251 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1252 1253 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1254 ~EXT4_MOUNT_##opt 1255 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1256 EXT4_MOUNT_##opt 1257 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1258 EXT4_MOUNT_##opt) 1259 1260 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1261 ~EXT4_MOUNT2_##opt 1262 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1263 EXT4_MOUNT2_##opt 1264 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1265 EXT4_MOUNT2_##opt) 1266 1267 #define ext4_test_and_set_bit __test_and_set_bit_le 1268 #define ext4_set_bit __set_bit_le 1269 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1270 #define ext4_clear_bit __clear_bit_le 1271 #define ext4_test_bit test_bit_le 1272 #define ext4_find_next_zero_bit find_next_zero_bit_le 1273 #define ext4_find_next_bit find_next_bit_le 1274 1275 extern void mb_set_bits(void *bm, int cur, int len); 1276 1277 /* 1278 * Maximal mount counts between two filesystem checks 1279 */ 1280 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1281 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1282 1283 /* 1284 * Behaviour when detecting errors 1285 */ 1286 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1287 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1288 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1289 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1290 1291 /* Metadata checksum algorithm codes */ 1292 #define EXT4_CRC32C_CHKSUM 1 1293 1294 #define EXT4_LABEL_MAX 16 1295 1296 /* 1297 * Structure of the super block 1298 */ 1299 struct ext4_super_block { 1300 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1301 __le32 s_blocks_count_lo; /* Blocks count */ 1302 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1303 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1304 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1305 __le32 s_first_data_block; /* First Data Block */ 1306 __le32 s_log_block_size; /* Block size */ 1307 __le32 s_log_cluster_size; /* Allocation cluster size */ 1308 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1309 __le32 s_clusters_per_group; /* # Clusters per group */ 1310 __le32 s_inodes_per_group; /* # Inodes per group */ 1311 __le32 s_mtime; /* Mount time */ 1312 /*30*/ __le32 s_wtime; /* Write time */ 1313 __le16 s_mnt_count; /* Mount count */ 1314 __le16 s_max_mnt_count; /* Maximal mount count */ 1315 __le16 s_magic; /* Magic signature */ 1316 __le16 s_state; /* File system state */ 1317 __le16 s_errors; /* Behaviour when detecting errors */ 1318 __le16 s_minor_rev_level; /* minor revision level */ 1319 /*40*/ __le32 s_lastcheck; /* time of last check */ 1320 __le32 s_checkinterval; /* max. time between checks */ 1321 __le32 s_creator_os; /* OS */ 1322 __le32 s_rev_level; /* Revision level */ 1323 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1324 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1325 /* 1326 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1327 * 1328 * Note: the difference between the compatible feature set and 1329 * the incompatible feature set is that if there is a bit set 1330 * in the incompatible feature set that the kernel doesn't 1331 * know about, it should refuse to mount the filesystem. 1332 * 1333 * e2fsck's requirements are more strict; if it doesn't know 1334 * about a feature in either the compatible or incompatible 1335 * feature set, it must abort and not try to meddle with 1336 * things it doesn't understand... 1337 */ 1338 __le32 s_first_ino; /* First non-reserved inode */ 1339 __le16 s_inode_size; /* size of inode structure */ 1340 __le16 s_block_group_nr; /* block group # of this superblock */ 1341 __le32 s_feature_compat; /* compatible feature set */ 1342 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1343 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1344 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1345 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1346 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1347 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1348 /* 1349 * Performance hints. Directory preallocation should only 1350 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1351 */ 1352 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1353 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1354 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1355 /* 1356 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1357 */ 1358 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1359 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1360 __le32 s_journal_dev; /* device number of journal file */ 1361 __le32 s_last_orphan; /* start of list of inodes to delete */ 1362 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1363 __u8 s_def_hash_version; /* Default hash version to use */ 1364 __u8 s_jnl_backup_type; 1365 __le16 s_desc_size; /* size of group descriptor */ 1366 /*100*/ __le32 s_default_mount_opts; 1367 __le32 s_first_meta_bg; /* First metablock block group */ 1368 __le32 s_mkfs_time; /* When the filesystem was created */ 1369 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1370 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1371 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1372 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1373 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1374 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1375 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1376 __le32 s_flags; /* Miscellaneous flags */ 1377 __le16 s_raid_stride; /* RAID stride */ 1378 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1379 __le64 s_mmp_block; /* Block for multi-mount protection */ 1380 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1381 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1382 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1383 __u8 s_encryption_level; /* versioning level for encryption */ 1384 __u8 s_reserved_pad; /* Padding to next 32bits */ 1385 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1386 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1387 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1388 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1389 snapshot's future use */ 1390 __le32 s_snapshot_list; /* inode number of the head of the 1391 on-disk snapshot list */ 1392 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1393 __le32 s_error_count; /* number of fs errors */ 1394 __le32 s_first_error_time; /* first time an error happened */ 1395 __le32 s_first_error_ino; /* inode involved in first error */ 1396 __le64 s_first_error_block; /* block involved of first error */ 1397 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1398 __le32 s_first_error_line; /* line number where error happened */ 1399 __le32 s_last_error_time; /* most recent time of an error */ 1400 __le32 s_last_error_ino; /* inode involved in last error */ 1401 __le32 s_last_error_line; /* line number where error happened */ 1402 __le64 s_last_error_block; /* block involved of last error */ 1403 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1404 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1405 __u8 s_mount_opts[64]; 1406 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1407 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1408 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1409 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1410 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1411 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1412 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1413 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1414 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1415 __u8 s_wtime_hi; 1416 __u8 s_mtime_hi; 1417 __u8 s_mkfs_time_hi; 1418 __u8 s_lastcheck_hi; 1419 __u8 s_first_error_time_hi; 1420 __u8 s_last_error_time_hi; 1421 __u8 s_first_error_errcode; 1422 __u8 s_last_error_errcode; 1423 __le16 s_encoding; /* Filename charset encoding */ 1424 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1425 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1426 __le32 s_reserved[94]; /* Padding to the end of the block */ 1427 __le32 s_checksum; /* crc32c(superblock) */ 1428 }; 1429 1430 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1431 1432 #ifdef __KERNEL__ 1433 1434 /* Number of quota types we support */ 1435 #define EXT4_MAXQUOTAS 3 1436 1437 #define EXT4_ENC_UTF8_12_1 1 1438 1439 /* Types of ext4 journal triggers */ 1440 enum ext4_journal_trigger_type { 1441 EXT4_JTR_ORPHAN_FILE, 1442 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1443 }; 1444 1445 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1446 1447 struct ext4_journal_trigger { 1448 struct jbd2_buffer_trigger_type tr_triggers; 1449 struct super_block *sb; 1450 }; 1451 1452 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1453 struct jbd2_buffer_trigger_type *trigger) 1454 { 1455 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1456 } 1457 1458 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1459 1460 /* Structure at the tail of orphan block */ 1461 struct ext4_orphan_block_tail { 1462 __le32 ob_magic; 1463 __le32 ob_checksum; 1464 }; 1465 1466 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1467 { 1468 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1469 sizeof(u32); 1470 } 1471 1472 struct ext4_orphan_block { 1473 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1474 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1475 }; 1476 1477 /* 1478 * Info about orphan file. 1479 */ 1480 struct ext4_orphan_info { 1481 int of_blocks; /* Number of orphan blocks in a file */ 1482 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1483 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1484 * file blocks */ 1485 }; 1486 1487 /* 1488 * fourth extended-fs super-block data in memory 1489 */ 1490 struct ext4_sb_info { 1491 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1492 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1493 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1494 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1495 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1496 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1497 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1498 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1499 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1500 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1501 unsigned long s_overhead; /* # of fs overhead clusters */ 1502 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1503 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1504 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1505 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1506 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1507 /* Array of bh's for the block group descriptors */ 1508 struct buffer_head * __rcu *s_group_desc; 1509 unsigned int s_mount_opt; 1510 unsigned int s_mount_opt2; 1511 unsigned long s_mount_flags; 1512 unsigned int s_def_mount_opt; 1513 unsigned int s_def_mount_opt2; 1514 ext4_fsblk_t s_sb_block; 1515 atomic64_t s_resv_clusters; 1516 kuid_t s_resuid; 1517 kgid_t s_resgid; 1518 unsigned short s_mount_state; 1519 unsigned short s_pad; 1520 int s_addr_per_block_bits; 1521 int s_desc_per_block_bits; 1522 int s_inode_size; 1523 int s_first_ino; 1524 unsigned int s_inode_readahead_blks; 1525 unsigned int s_inode_goal; 1526 u32 s_hash_seed[4]; 1527 int s_def_hash_version; 1528 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1529 struct percpu_counter s_freeclusters_counter; 1530 struct percpu_counter s_freeinodes_counter; 1531 struct percpu_counter s_dirs_counter; 1532 struct percpu_counter s_dirtyclusters_counter; 1533 struct percpu_counter s_sra_exceeded_retry_limit; 1534 struct blockgroup_lock *s_blockgroup_lock; 1535 struct proc_dir_entry *s_proc; 1536 struct kobject s_kobj; 1537 struct completion s_kobj_unregister; 1538 struct super_block *s_sb; 1539 struct buffer_head *s_mmp_bh; 1540 1541 /* Journaling */ 1542 struct journal_s *s_journal; 1543 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1544 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1545 struct list_head s_orphan; /* List of orphaned inodes in on disk 1546 list */ 1547 struct ext4_orphan_info s_orphan_info; 1548 unsigned long s_commit_interval; 1549 u32 s_max_batch_time; 1550 u32 s_min_batch_time; 1551 struct bdev_handle *s_journal_bdev_handle; 1552 #ifdef CONFIG_QUOTA 1553 /* Names of quota files with journalled quota */ 1554 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1555 int s_jquota_fmt; /* Format of quota to use */ 1556 #endif 1557 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1558 struct ext4_system_blocks __rcu *s_system_blks; 1559 1560 #ifdef EXTENTS_STATS 1561 /* ext4 extents stats */ 1562 unsigned long s_ext_min; 1563 unsigned long s_ext_max; 1564 unsigned long s_depth_max; 1565 spinlock_t s_ext_stats_lock; 1566 unsigned long s_ext_blocks; 1567 unsigned long s_ext_extents; 1568 #endif 1569 1570 /* for buddy allocator */ 1571 struct ext4_group_info ** __rcu *s_group_info; 1572 struct inode *s_buddy_cache; 1573 spinlock_t s_md_lock; 1574 unsigned short *s_mb_offsets; 1575 unsigned int *s_mb_maxs; 1576 unsigned int s_group_info_size; 1577 unsigned int s_mb_free_pending; 1578 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1579 after commit completed */ 1580 struct list_head s_discard_list; 1581 struct work_struct s_discard_work; 1582 atomic_t s_retry_alloc_pending; 1583 struct list_head *s_mb_avg_fragment_size; 1584 rwlock_t *s_mb_avg_fragment_size_locks; 1585 struct list_head *s_mb_largest_free_orders; 1586 rwlock_t *s_mb_largest_free_orders_locks; 1587 1588 /* tunables */ 1589 unsigned long s_stripe; 1590 unsigned int s_mb_max_linear_groups; 1591 unsigned int s_mb_stream_request; 1592 unsigned int s_mb_max_to_scan; 1593 unsigned int s_mb_min_to_scan; 1594 unsigned int s_mb_stats; 1595 unsigned int s_mb_order2_reqs; 1596 unsigned int s_mb_group_prealloc; 1597 unsigned int s_max_dir_size_kb; 1598 /* where last allocation was done - for stream allocation */ 1599 unsigned long s_mb_last_group; 1600 unsigned long s_mb_last_start; 1601 unsigned int s_mb_prefetch; 1602 unsigned int s_mb_prefetch_limit; 1603 unsigned int s_mb_best_avail_max_trim_order; 1604 1605 /* stats for buddy allocator */ 1606 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1607 atomic_t s_bal_success; /* we found long enough chunks */ 1608 atomic_t s_bal_allocated; /* in blocks */ 1609 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1610 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1611 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1612 atomic_t s_bal_goals; /* goal hits */ 1613 atomic_t s_bal_len_goals; /* len goal hits */ 1614 atomic_t s_bal_breaks; /* too long searches */ 1615 atomic_t s_bal_2orders; /* 2^order hits */ 1616 atomic_t s_bal_p2_aligned_bad_suggestions; 1617 atomic_t s_bal_goal_fast_bad_suggestions; 1618 atomic_t s_bal_best_avail_bad_suggestions; 1619 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1620 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1621 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1622 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1623 atomic64_t s_mb_generation_time; 1624 atomic_t s_mb_lost_chunks; 1625 atomic_t s_mb_preallocated; 1626 atomic_t s_mb_discarded; 1627 atomic_t s_lock_busy; 1628 1629 /* locality groups */ 1630 struct ext4_locality_group __percpu *s_locality_groups; 1631 1632 /* for write statistics */ 1633 unsigned long s_sectors_written_start; 1634 u64 s_kbytes_written; 1635 1636 /* the size of zero-out chunk */ 1637 unsigned int s_extent_max_zeroout_kb; 1638 1639 unsigned int s_log_groups_per_flex; 1640 struct flex_groups * __rcu *s_flex_groups; 1641 ext4_group_t s_flex_groups_allocated; 1642 1643 /* workqueue for reserved extent conversions (buffered io) */ 1644 struct workqueue_struct *rsv_conversion_wq; 1645 1646 /* timer for periodic error stats printing */ 1647 struct timer_list s_err_report; 1648 1649 /* Lazy inode table initialization info */ 1650 struct ext4_li_request *s_li_request; 1651 /* Wait multiplier for lazy initialization thread */ 1652 unsigned int s_li_wait_mult; 1653 1654 /* Kernel thread for multiple mount protection */ 1655 struct task_struct *s_mmp_tsk; 1656 1657 /* record the last minlen when FITRIM is called. */ 1658 unsigned long s_last_trim_minblks; 1659 1660 /* Reference to checksum algorithm driver via cryptoapi */ 1661 struct crypto_shash *s_chksum_driver; 1662 1663 /* Precomputed FS UUID checksum for seeding other checksums */ 1664 __u32 s_csum_seed; 1665 1666 /* Reclaim extents from extent status tree */ 1667 struct shrinker *s_es_shrinker; 1668 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1669 long s_es_nr_inode; 1670 struct ext4_es_stats s_es_stats; 1671 struct mb_cache *s_ea_block_cache; 1672 struct mb_cache *s_ea_inode_cache; 1673 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1674 1675 /* Journal triggers for checksum computation */ 1676 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1677 1678 /* Ratelimit ext4 messages. */ 1679 struct ratelimit_state s_err_ratelimit_state; 1680 struct ratelimit_state s_warning_ratelimit_state; 1681 struct ratelimit_state s_msg_ratelimit_state; 1682 atomic_t s_warning_count; 1683 atomic_t s_msg_count; 1684 1685 /* Encryption policy for '-o test_dummy_encryption' */ 1686 struct fscrypt_dummy_policy s_dummy_enc_policy; 1687 1688 /* 1689 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1690 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1691 * DIOREAD_NOLOCK mount options on remount. 1692 */ 1693 struct percpu_rw_semaphore s_writepages_rwsem; 1694 struct dax_device *s_daxdev; 1695 u64 s_dax_part_off; 1696 #ifdef CONFIG_EXT4_DEBUG 1697 unsigned long s_simulate_fail; 1698 #endif 1699 /* Record the errseq of the backing block device */ 1700 errseq_t s_bdev_wb_err; 1701 spinlock_t s_bdev_wb_lock; 1702 1703 /* Information about errors that happened during this mount */ 1704 spinlock_t s_error_lock; 1705 int s_add_error_count; 1706 int s_first_error_code; 1707 __u32 s_first_error_line; 1708 __u32 s_first_error_ino; 1709 __u64 s_first_error_block; 1710 const char *s_first_error_func; 1711 time64_t s_first_error_time; 1712 int s_last_error_code; 1713 __u32 s_last_error_line; 1714 __u32 s_last_error_ino; 1715 __u64 s_last_error_block; 1716 const char *s_last_error_func; 1717 time64_t s_last_error_time; 1718 /* 1719 * If we are in a context where we cannot update the on-disk 1720 * superblock, we queue the work here. This is used to update 1721 * the error information in the superblock, and for periodic 1722 * updates of the superblock called from the commit callback 1723 * function. 1724 */ 1725 struct work_struct s_sb_upd_work; 1726 1727 /* Ext4 fast commit sub transaction ID */ 1728 atomic_t s_fc_subtid; 1729 1730 /* 1731 * After commit starts, the main queue gets locked, and the further 1732 * updates get added in the staging queue. 1733 */ 1734 #define FC_Q_MAIN 0 1735 #define FC_Q_STAGING 1 1736 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1737 * that have data changes in them. 1738 */ 1739 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1740 unsigned int s_fc_bytes; 1741 /* 1742 * Main fast commit lock. This lock protects accesses to the 1743 * following fields: 1744 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1745 */ 1746 spinlock_t s_fc_lock; 1747 struct buffer_head *s_fc_bh; 1748 struct ext4_fc_stats s_fc_stats; 1749 tid_t s_fc_ineligible_tid; 1750 #ifdef CONFIG_EXT4_DEBUG 1751 int s_fc_debug_max_replay; 1752 #endif 1753 struct ext4_fc_replay_state s_fc_replay_state; 1754 }; 1755 1756 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1757 { 1758 return sb->s_fs_info; 1759 } 1760 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1761 { 1762 return container_of(inode, struct ext4_inode_info, vfs_inode); 1763 } 1764 1765 static inline int ext4_writepages_down_read(struct super_block *sb) 1766 { 1767 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1768 return memalloc_nofs_save(); 1769 } 1770 1771 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1772 { 1773 memalloc_nofs_restore(ctx); 1774 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1775 } 1776 1777 static inline int ext4_writepages_down_write(struct super_block *sb) 1778 { 1779 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1780 return memalloc_nofs_save(); 1781 } 1782 1783 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1784 { 1785 memalloc_nofs_restore(ctx); 1786 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1787 } 1788 1789 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1790 { 1791 return ino == EXT4_ROOT_INO || 1792 (ino >= EXT4_FIRST_INO(sb) && 1793 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1794 } 1795 1796 /* 1797 * Returns: sbi->field[index] 1798 * Used to access an array element from the following sbi fields which require 1799 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1800 * - s_group_desc 1801 * - s_group_info 1802 * - s_flex_group 1803 */ 1804 #define sbi_array_rcu_deref(sbi, field, index) \ 1805 ({ \ 1806 typeof(*((sbi)->field)) _v; \ 1807 rcu_read_lock(); \ 1808 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1809 rcu_read_unlock(); \ 1810 _v; \ 1811 }) 1812 1813 /* 1814 * run-time mount flags 1815 */ 1816 enum { 1817 EXT4_MF_MNTDIR_SAMPLED, 1818 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1819 }; 1820 1821 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1822 { 1823 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1824 } 1825 1826 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1827 { 1828 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1829 } 1830 1831 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1832 { 1833 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1834 } 1835 1836 1837 /* 1838 * Simulate_fail codes 1839 */ 1840 #define EXT4_SIM_BBITMAP_EIO 1 1841 #define EXT4_SIM_BBITMAP_CRC 2 1842 #define EXT4_SIM_IBITMAP_EIO 3 1843 #define EXT4_SIM_IBITMAP_CRC 4 1844 #define EXT4_SIM_INODE_EIO 5 1845 #define EXT4_SIM_INODE_CRC 6 1846 #define EXT4_SIM_DIRBLOCK_EIO 7 1847 #define EXT4_SIM_DIRBLOCK_CRC 8 1848 1849 static inline bool ext4_simulate_fail(struct super_block *sb, 1850 unsigned long code) 1851 { 1852 #ifdef CONFIG_EXT4_DEBUG 1853 struct ext4_sb_info *sbi = EXT4_SB(sb); 1854 1855 if (unlikely(sbi->s_simulate_fail == code)) { 1856 sbi->s_simulate_fail = 0; 1857 return true; 1858 } 1859 #endif 1860 return false; 1861 } 1862 1863 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1864 struct buffer_head *bh, 1865 unsigned long code) 1866 { 1867 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1868 clear_buffer_uptodate(bh); 1869 } 1870 1871 /* 1872 * Error number codes for s_{first,last}_error_errno 1873 * 1874 * Linux errno numbers are architecture specific, so we need to translate 1875 * them into something which is architecture independent. We don't define 1876 * codes for all errno's; just the ones which are most likely to be the cause 1877 * of an ext4_error() call. 1878 */ 1879 #define EXT4_ERR_UNKNOWN 1 1880 #define EXT4_ERR_EIO 2 1881 #define EXT4_ERR_ENOMEM 3 1882 #define EXT4_ERR_EFSBADCRC 4 1883 #define EXT4_ERR_EFSCORRUPTED 5 1884 #define EXT4_ERR_ENOSPC 6 1885 #define EXT4_ERR_ENOKEY 7 1886 #define EXT4_ERR_EROFS 8 1887 #define EXT4_ERR_EFBIG 9 1888 #define EXT4_ERR_EEXIST 10 1889 #define EXT4_ERR_ERANGE 11 1890 #define EXT4_ERR_EOVERFLOW 12 1891 #define EXT4_ERR_EBUSY 13 1892 #define EXT4_ERR_ENOTDIR 14 1893 #define EXT4_ERR_ENOTEMPTY 15 1894 #define EXT4_ERR_ESHUTDOWN 16 1895 #define EXT4_ERR_EFAULT 17 1896 1897 /* 1898 * Inode dynamic state flags 1899 */ 1900 enum { 1901 EXT4_STATE_NEW, /* inode is newly created */ 1902 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1903 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1904 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1905 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1906 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1907 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1908 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1909 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1910 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1911 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1912 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1913 }; 1914 1915 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1916 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1917 { \ 1918 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1919 } \ 1920 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1921 { \ 1922 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1923 } \ 1924 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1925 { \ 1926 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1927 } 1928 1929 /* Add these declarations here only so that these functions can be 1930 * found by name. Otherwise, they are very hard to locate. */ 1931 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1932 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1933 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1934 EXT4_INODE_BIT_FNS(flag, flags, 0) 1935 1936 /* Add these declarations here only so that these functions can be 1937 * found by name. Otherwise, they are very hard to locate. */ 1938 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1939 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1940 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1941 #if (BITS_PER_LONG < 64) 1942 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1943 1944 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1945 { 1946 (ei)->i_state_flags = 0; 1947 } 1948 #else 1949 EXT4_INODE_BIT_FNS(state, flags, 32) 1950 1951 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1952 { 1953 /* We depend on the fact that callers will set i_flags */ 1954 } 1955 #endif 1956 #else 1957 /* Assume that user mode programs are passing in an ext4fs superblock, not 1958 * a kernel struct super_block. This will allow us to call the feature-test 1959 * macros from user land. */ 1960 #define EXT4_SB(sb) (sb) 1961 #endif 1962 1963 static inline bool ext4_verity_in_progress(struct inode *inode) 1964 { 1965 return IS_ENABLED(CONFIG_FS_VERITY) && 1966 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1967 } 1968 1969 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1970 1971 /* 1972 * Codes for operating systems 1973 */ 1974 #define EXT4_OS_LINUX 0 1975 #define EXT4_OS_HURD 1 1976 #define EXT4_OS_MASIX 2 1977 #define EXT4_OS_FREEBSD 3 1978 #define EXT4_OS_LITES 4 1979 1980 /* 1981 * Revision levels 1982 */ 1983 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1984 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1985 1986 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1987 1988 #define EXT4_GOOD_OLD_INODE_SIZE 128 1989 1990 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1991 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1992 #define EXT4_TIMESTAMP_MIN S32_MIN 1993 1994 /* 1995 * Feature set definitions 1996 */ 1997 1998 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1999 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2000 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2001 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2002 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2003 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2004 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2005 /* 2006 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2007 * incompatible only if fast commit blocks are present in the FS. Since we 2008 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2009 * incompatible. We also have a JBD2 incompat feature, which gets set when 2010 * there are fast commit blocks present in the journal. 2011 */ 2012 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2013 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2014 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2015 2016 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2017 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2018 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2019 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2020 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2021 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2022 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2023 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2024 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2025 /* 2026 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2027 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2028 * all other data structures' checksums. However, the METADATA_CSUM and 2029 * GDT_CSUM bits are mutually exclusive. 2030 */ 2031 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2032 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2033 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2034 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2035 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2036 non-empty */ 2037 2038 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2039 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2040 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2041 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2042 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2043 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2044 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2045 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2046 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2047 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2048 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2049 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2050 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2051 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2052 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2053 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2054 2055 extern void ext4_update_dynamic_rev(struct super_block *sb); 2056 2057 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2058 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2059 { \ 2060 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2061 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2062 } \ 2063 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2064 { \ 2065 ext4_update_dynamic_rev(sb); \ 2066 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2067 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2068 } \ 2069 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2070 { \ 2071 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2072 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2073 } 2074 2075 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2076 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2077 { \ 2078 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2079 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2080 } \ 2081 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2082 { \ 2083 ext4_update_dynamic_rev(sb); \ 2084 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2085 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2086 } \ 2087 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2088 { \ 2089 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2090 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2091 } 2092 2093 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2094 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2095 { \ 2096 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2097 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2098 } \ 2099 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2100 { \ 2101 ext4_update_dynamic_rev(sb); \ 2102 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2103 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2104 } \ 2105 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2106 { \ 2107 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2108 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2109 } 2110 2111 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2112 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2113 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2114 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2115 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2116 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2117 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2118 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2119 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2120 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2121 2122 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2123 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2124 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2125 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2126 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2127 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2128 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2129 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2130 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2131 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2132 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2133 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2134 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2135 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2136 2137 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2138 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2139 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2140 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2141 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2142 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2143 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2144 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2145 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2146 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2147 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2148 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2149 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2150 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2151 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2152 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2153 2154 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2155 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2156 EXT4_FEATURE_INCOMPAT_META_BG) 2157 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2158 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2159 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2160 2161 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2162 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2163 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2164 EXT4_FEATURE_INCOMPAT_META_BG) 2165 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2166 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2167 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2168 2169 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2170 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2171 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2172 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2173 EXT4_FEATURE_INCOMPAT_META_BG| \ 2174 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2175 EXT4_FEATURE_INCOMPAT_64BIT| \ 2176 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2177 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2178 EXT4_FEATURE_INCOMPAT_MMP | \ 2179 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2180 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2181 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2182 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2183 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2184 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2185 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2186 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2187 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2188 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2189 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2190 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2191 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2192 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2193 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2194 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2195 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2196 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2197 2198 #define EXTN_FEATURE_FUNCS(ver) \ 2199 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2200 { \ 2201 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2202 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2203 } \ 2204 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2205 { \ 2206 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2207 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2208 } \ 2209 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2210 { \ 2211 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2212 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2213 } 2214 2215 EXTN_FEATURE_FUNCS(2) 2216 EXTN_FEATURE_FUNCS(3) 2217 EXTN_FEATURE_FUNCS(4) 2218 2219 static inline bool ext4_has_compat_features(struct super_block *sb) 2220 { 2221 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2222 } 2223 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2224 { 2225 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2226 } 2227 static inline bool ext4_has_incompat_features(struct super_block *sb) 2228 { 2229 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2230 } 2231 2232 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2233 2234 /* 2235 * Superblock flags 2236 */ 2237 #define EXT4_FLAGS_RESIZING 0 2238 #define EXT4_FLAGS_SHUTDOWN 1 2239 #define EXT4_FLAGS_BDEV_IS_DAX 2 2240 2241 static inline int ext4_forced_shutdown(struct super_block *sb) 2242 { 2243 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2244 } 2245 2246 /* 2247 * Default values for user and/or group using reserved blocks 2248 */ 2249 #define EXT4_DEF_RESUID 0 2250 #define EXT4_DEF_RESGID 0 2251 2252 /* 2253 * Default project ID 2254 */ 2255 #define EXT4_DEF_PROJID 0 2256 2257 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2258 2259 /* 2260 * Default mount options 2261 */ 2262 #define EXT4_DEFM_DEBUG 0x0001 2263 #define EXT4_DEFM_BSDGROUPS 0x0002 2264 #define EXT4_DEFM_XATTR_USER 0x0004 2265 #define EXT4_DEFM_ACL 0x0008 2266 #define EXT4_DEFM_UID16 0x0010 2267 #define EXT4_DEFM_JMODE 0x0060 2268 #define EXT4_DEFM_JMODE_DATA 0x0020 2269 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2270 #define EXT4_DEFM_JMODE_WBACK 0x0060 2271 #define EXT4_DEFM_NOBARRIER 0x0100 2272 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2273 #define EXT4_DEFM_DISCARD 0x0400 2274 #define EXT4_DEFM_NODELALLOC 0x0800 2275 2276 /* 2277 * Default journal batch times 2278 */ 2279 #define EXT4_DEF_MIN_BATCH_TIME 0 2280 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2281 2282 /* 2283 * Minimum number of groups in a flexgroup before we separate out 2284 * directories into the first block group of a flexgroup 2285 */ 2286 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2287 2288 /* 2289 * Structure of a directory entry 2290 */ 2291 #define EXT4_NAME_LEN 255 2292 /* 2293 * Base length of the ext4 directory entry excluding the name length 2294 */ 2295 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2296 2297 struct ext4_dir_entry { 2298 __le32 inode; /* Inode number */ 2299 __le16 rec_len; /* Directory entry length */ 2300 __le16 name_len; /* Name length */ 2301 char name[EXT4_NAME_LEN]; /* File name */ 2302 }; 2303 2304 2305 /* 2306 * Encrypted Casefolded entries require saving the hash on disk. This structure 2307 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2308 * boundary. 2309 */ 2310 struct ext4_dir_entry_hash { 2311 __le32 hash; 2312 __le32 minor_hash; 2313 }; 2314 2315 /* 2316 * The new version of the directory entry. Since EXT4 structures are 2317 * stored in intel byte order, and the name_len field could never be 2318 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2319 * file_type field. 2320 */ 2321 struct ext4_dir_entry_2 { 2322 __le32 inode; /* Inode number */ 2323 __le16 rec_len; /* Directory entry length */ 2324 __u8 name_len; /* Name length */ 2325 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2326 char name[EXT4_NAME_LEN]; /* File name */ 2327 }; 2328 2329 /* 2330 * Access the hashes at the end of ext4_dir_entry_2 2331 */ 2332 #define EXT4_DIRENT_HASHES(entry) \ 2333 ((struct ext4_dir_entry_hash *) \ 2334 (((void *)(entry)) + \ 2335 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2336 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2337 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2338 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2339 2340 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2341 { 2342 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2343 } 2344 2345 /* 2346 * This is a bogus directory entry at the end of each leaf block that 2347 * records checksums. 2348 */ 2349 struct ext4_dir_entry_tail { 2350 __le32 det_reserved_zero1; /* Pretend to be unused */ 2351 __le16 det_rec_len; /* 12 */ 2352 __u8 det_reserved_zero2; /* Zero name length */ 2353 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2354 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2355 }; 2356 2357 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2358 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2359 ((blocksize) - \ 2360 sizeof(struct ext4_dir_entry_tail)))) 2361 2362 /* 2363 * Ext4 directory file types. Only the low 3 bits are used. The 2364 * other bits are reserved for now. 2365 */ 2366 #define EXT4_FT_UNKNOWN 0 2367 #define EXT4_FT_REG_FILE 1 2368 #define EXT4_FT_DIR 2 2369 #define EXT4_FT_CHRDEV 3 2370 #define EXT4_FT_BLKDEV 4 2371 #define EXT4_FT_FIFO 5 2372 #define EXT4_FT_SOCK 6 2373 #define EXT4_FT_SYMLINK 7 2374 2375 #define EXT4_FT_MAX 8 2376 2377 #define EXT4_FT_DIR_CSUM 0xDE 2378 2379 /* 2380 * EXT4_DIR_PAD defines the directory entries boundaries 2381 * 2382 * NOTE: It must be a multiple of 4 2383 */ 2384 #define EXT4_DIR_PAD 4 2385 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2386 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2387 2388 /* 2389 * The rec_len is dependent on the type of directory. Directories that are 2390 * casefolded and encrypted need to store the hash as well, so we add room for 2391 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2392 * pass NULL for dir, as those entries do not use the extra fields. 2393 */ 2394 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2395 const struct inode *dir) 2396 { 2397 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2398 2399 if (dir && ext4_hash_in_dirent(dir)) 2400 rec_len += sizeof(struct ext4_dir_entry_hash); 2401 return (rec_len & ~EXT4_DIR_ROUND); 2402 } 2403 2404 /* 2405 * If we ever get support for fs block sizes > page_size, we'll need 2406 * to remove the #if statements in the next two functions... 2407 */ 2408 static inline unsigned int 2409 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2410 { 2411 unsigned len = le16_to_cpu(dlen); 2412 2413 #if (PAGE_SIZE >= 65536) 2414 if (len == EXT4_MAX_REC_LEN || len == 0) 2415 return blocksize; 2416 return (len & 65532) | ((len & 3) << 16); 2417 #else 2418 return len; 2419 #endif 2420 } 2421 2422 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2423 { 2424 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2425 #if (PAGE_SIZE >= 65536) 2426 if (len < 65536) 2427 return cpu_to_le16(len); 2428 if (len == blocksize) { 2429 if (blocksize == 65536) 2430 return cpu_to_le16(EXT4_MAX_REC_LEN); 2431 else 2432 return cpu_to_le16(0); 2433 } 2434 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2435 #else 2436 return cpu_to_le16(len); 2437 #endif 2438 } 2439 2440 /* 2441 * Hash Tree Directory indexing 2442 * (c) Daniel Phillips, 2001 2443 */ 2444 2445 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2446 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2447 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2448 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2449 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2450 2451 /* Legal values for the dx_root hash_version field: */ 2452 2453 #define DX_HASH_LEGACY 0 2454 #define DX_HASH_HALF_MD4 1 2455 #define DX_HASH_TEA 2 2456 #define DX_HASH_LEGACY_UNSIGNED 3 2457 #define DX_HASH_HALF_MD4_UNSIGNED 4 2458 #define DX_HASH_TEA_UNSIGNED 5 2459 #define DX_HASH_SIPHASH 6 2460 2461 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2462 const void *address, unsigned int length) 2463 { 2464 struct { 2465 struct shash_desc shash; 2466 char ctx[4]; 2467 } desc; 2468 2469 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2470 2471 desc.shash.tfm = sbi->s_chksum_driver; 2472 *(u32 *)desc.ctx = crc; 2473 2474 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2475 2476 return *(u32 *)desc.ctx; 2477 } 2478 2479 #ifdef __KERNEL__ 2480 2481 /* hash info structure used by the directory hash */ 2482 struct dx_hash_info 2483 { 2484 u32 hash; 2485 u32 minor_hash; 2486 int hash_version; 2487 u32 *seed; 2488 }; 2489 2490 2491 /* 32 and 64 bit signed EOF for dx directories */ 2492 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2493 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2494 2495 2496 /* 2497 * Control parameters used by ext4_htree_next_block 2498 */ 2499 #define HASH_NB_ALWAYS 1 2500 2501 struct ext4_filename { 2502 const struct qstr *usr_fname; 2503 struct fscrypt_str disk_name; 2504 struct dx_hash_info hinfo; 2505 #ifdef CONFIG_FS_ENCRYPTION 2506 struct fscrypt_str crypto_buf; 2507 #endif 2508 #if IS_ENABLED(CONFIG_UNICODE) 2509 struct fscrypt_str cf_name; 2510 #endif 2511 }; 2512 2513 #define fname_name(p) ((p)->disk_name.name) 2514 #define fname_usr_name(p) ((p)->usr_fname->name) 2515 #define fname_len(p) ((p)->disk_name.len) 2516 2517 /* 2518 * Describe an inode's exact location on disk and in memory 2519 */ 2520 struct ext4_iloc 2521 { 2522 struct buffer_head *bh; 2523 unsigned long offset; 2524 ext4_group_t block_group; 2525 }; 2526 2527 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2528 { 2529 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2530 } 2531 2532 static inline bool ext4_is_quota_file(struct inode *inode) 2533 { 2534 return IS_NOQUOTA(inode) && 2535 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2536 } 2537 2538 /* 2539 * This structure is stuffed into the struct file's private_data field 2540 * for directories. It is where we put information so that we can do 2541 * readdir operations in hash tree order. 2542 */ 2543 struct dir_private_info { 2544 struct rb_root root; 2545 struct rb_node *curr_node; 2546 struct fname *extra_fname; 2547 loff_t last_pos; 2548 __u32 curr_hash; 2549 __u32 curr_minor_hash; 2550 __u32 next_hash; 2551 }; 2552 2553 /* calculate the first block number of the group */ 2554 static inline ext4_fsblk_t 2555 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2556 { 2557 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2558 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2559 } 2560 2561 /* 2562 * Special error return code only used by dx_probe() and its callers. 2563 */ 2564 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2565 2566 /* htree levels for ext4 */ 2567 #define EXT4_HTREE_LEVEL_COMPAT 2 2568 #define EXT4_HTREE_LEVEL 3 2569 2570 static inline int ext4_dir_htree_level(struct super_block *sb) 2571 { 2572 return ext4_has_feature_largedir(sb) ? 2573 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2574 } 2575 2576 /* 2577 * Timeout and state flag for lazy initialization inode thread. 2578 */ 2579 #define EXT4_DEF_LI_WAIT_MULT 10 2580 #define EXT4_DEF_LI_MAX_START_DELAY 5 2581 #define EXT4_LAZYINIT_QUIT 0x0001 2582 #define EXT4_LAZYINIT_RUNNING 0x0002 2583 2584 /* 2585 * Lazy inode table initialization info 2586 */ 2587 struct ext4_lazy_init { 2588 unsigned long li_state; 2589 struct list_head li_request_list; 2590 struct mutex li_list_mtx; 2591 }; 2592 2593 enum ext4_li_mode { 2594 EXT4_LI_MODE_PREFETCH_BBITMAP, 2595 EXT4_LI_MODE_ITABLE, 2596 }; 2597 2598 struct ext4_li_request { 2599 struct super_block *lr_super; 2600 enum ext4_li_mode lr_mode; 2601 ext4_group_t lr_first_not_zeroed; 2602 ext4_group_t lr_next_group; 2603 struct list_head lr_request; 2604 unsigned long lr_next_sched; 2605 unsigned long lr_timeout; 2606 }; 2607 2608 struct ext4_features { 2609 struct kobject f_kobj; 2610 struct completion f_kobj_unregister; 2611 }; 2612 2613 /* 2614 * This structure will be used for multiple mount protection. It will be 2615 * written into the block number saved in the s_mmp_block field in the 2616 * superblock. Programs that check MMP should assume that if 2617 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2618 * to use the filesystem, regardless of how old the timestamp is. 2619 */ 2620 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2621 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2622 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2623 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2624 2625 struct mmp_struct { 2626 __le32 mmp_magic; /* Magic number for MMP */ 2627 __le32 mmp_seq; /* Sequence no. updated periodically */ 2628 2629 /* 2630 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2631 * purposes and do not affect the correctness of the algorithm 2632 */ 2633 __le64 mmp_time; /* Time last updated */ 2634 char mmp_nodename[64]; /* Node which last updated MMP block */ 2635 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2636 2637 /* 2638 * mmp_check_interval is used to verify if the MMP block has been 2639 * updated on the block device. The value is updated based on the 2640 * maximum time to write the MMP block during an update cycle. 2641 */ 2642 __le16 mmp_check_interval; 2643 2644 __le16 mmp_pad1; 2645 __le32 mmp_pad2[226]; 2646 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2647 }; 2648 2649 /* arguments passed to the mmp thread */ 2650 struct mmpd_data { 2651 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2652 struct super_block *sb; /* super block of the fs */ 2653 }; 2654 2655 /* 2656 * Check interval multiplier 2657 * The MMP block is written every update interval and initially checked every 2658 * update interval x the multiplier (the value is then adapted based on the 2659 * write latency). The reason is that writes can be delayed under load and we 2660 * don't want readers to incorrectly assume that the filesystem is no longer 2661 * in use. 2662 */ 2663 #define EXT4_MMP_CHECK_MULT 2UL 2664 2665 /* 2666 * Minimum interval for MMP checking in seconds. 2667 */ 2668 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2669 2670 /* 2671 * Maximum interval for MMP checking in seconds. 2672 */ 2673 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2674 2675 /* 2676 * Function prototypes 2677 */ 2678 2679 /* 2680 * Ok, these declarations are also in <linux/kernel.h> but none of the 2681 * ext4 source programs needs to include it so they are duplicated here. 2682 */ 2683 # define NORET_TYPE /**/ 2684 # define ATTRIB_NORET __attribute__((noreturn)) 2685 # define NORET_AND noreturn, 2686 2687 /* bitmap.c */ 2688 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2689 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2690 struct ext4_group_desc *gdp, 2691 struct buffer_head *bh, int sz); 2692 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2693 struct ext4_group_desc *gdp, 2694 struct buffer_head *bh, int sz); 2695 void ext4_block_bitmap_csum_set(struct super_block *sb, 2696 struct ext4_group_desc *gdp, 2697 struct buffer_head *bh); 2698 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2699 struct ext4_group_desc *gdp, 2700 struct buffer_head *bh); 2701 2702 /* balloc.c */ 2703 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2704 ext4_fsblk_t blocknr, 2705 ext4_group_t *blockgrpp, 2706 ext4_grpblk_t *offsetp); 2707 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2708 ext4_fsblk_t block); 2709 2710 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2711 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2712 ext4_group_t group); 2713 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2714 ext4_fsblk_t goal, 2715 unsigned int flags, 2716 unsigned long *count, 2717 int *errp); 2718 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2719 s64 nclusters, unsigned int flags); 2720 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2721 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2722 ext4_group_t block_group, 2723 struct buffer_head ** bh); 2724 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2725 ext4_group_t group); 2726 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2727 2728 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2729 ext4_group_t block_group, 2730 bool ignore_locked); 2731 extern int ext4_wait_block_bitmap(struct super_block *sb, 2732 ext4_group_t block_group, 2733 struct buffer_head *bh); 2734 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2735 ext4_group_t block_group); 2736 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2737 ext4_group_t block_group, 2738 struct ext4_group_desc *gdp); 2739 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2740 2741 #if IS_ENABLED(CONFIG_UNICODE) 2742 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2743 const struct qstr *iname, 2744 struct ext4_filename *fname); 2745 #endif 2746 2747 /* ext4 encryption related stuff goes here crypto.c */ 2748 #ifdef CONFIG_FS_ENCRYPTION 2749 extern const struct fscrypt_operations ext4_cryptops; 2750 2751 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2752 int lookup, struct ext4_filename *fname); 2753 2754 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2755 struct ext4_filename *fname); 2756 2757 void ext4_fname_free_filename(struct ext4_filename *fname); 2758 2759 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2760 2761 #else /* !CONFIG_FS_ENCRYPTION */ 2762 static inline int ext4_fname_setup_filename(struct inode *dir, 2763 const struct qstr *iname, 2764 int lookup, 2765 struct ext4_filename *fname) 2766 { 2767 int err = 0; 2768 fname->usr_fname = iname; 2769 fname->disk_name.name = (unsigned char *) iname->name; 2770 fname->disk_name.len = iname->len; 2771 2772 #if IS_ENABLED(CONFIG_UNICODE) 2773 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2774 #endif 2775 2776 return err; 2777 } 2778 2779 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2780 struct dentry *dentry, 2781 struct ext4_filename *fname) 2782 { 2783 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2784 } 2785 2786 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2787 { 2788 #if IS_ENABLED(CONFIG_UNICODE) 2789 kfree(fname->cf_name.name); 2790 fname->cf_name.name = NULL; 2791 #endif 2792 } 2793 2794 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2795 void __user *arg) 2796 { 2797 return -EOPNOTSUPP; 2798 } 2799 #endif /* !CONFIG_FS_ENCRYPTION */ 2800 2801 /* dir.c */ 2802 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2803 struct file *, 2804 struct ext4_dir_entry_2 *, 2805 struct buffer_head *, char *, int, 2806 unsigned int); 2807 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2808 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2809 (de), (bh), (buf), (size), (offset))) 2810 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2811 __u32 minor_hash, 2812 struct ext4_dir_entry_2 *dirent, 2813 struct fscrypt_str *ent_name); 2814 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2815 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2816 struct buffer_head *bh, 2817 void *buf, int buf_size, 2818 struct ext4_filename *fname, 2819 struct ext4_dir_entry_2 **dest_de); 2820 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2821 struct ext4_dir_entry_2 *de, 2822 int buf_size, 2823 struct ext4_filename *fname); 2824 static inline void ext4_update_dx_flag(struct inode *inode) 2825 { 2826 if (!ext4_has_feature_dir_index(inode->i_sb) && 2827 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2828 /* ext4_iget() should have caught this... */ 2829 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2830 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2831 } 2832 } 2833 static const unsigned char ext4_filetype_table[] = { 2834 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2835 }; 2836 2837 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2838 { 2839 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2840 return DT_UNKNOWN; 2841 2842 return ext4_filetype_table[filetype]; 2843 } 2844 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2845 void *buf, int buf_size); 2846 2847 /* fsync.c */ 2848 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2849 2850 /* hash.c */ 2851 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2852 struct dx_hash_info *hinfo); 2853 2854 /* ialloc.c */ 2855 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2856 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2857 struct inode *, umode_t, 2858 const struct qstr *qstr, __u32 goal, 2859 uid_t *owner, __u32 i_flags, 2860 int handle_type, unsigned int line_no, 2861 int nblocks); 2862 2863 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2864 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2865 (goal), (owner), i_flags, 0, 0, 0) 2866 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2867 type, nblocks) \ 2868 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2869 0, (type), __LINE__, (nblocks)) 2870 2871 2872 extern void ext4_free_inode(handle_t *, struct inode *); 2873 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2874 extern unsigned long ext4_count_free_inodes(struct super_block *); 2875 extern unsigned long ext4_count_dirs(struct super_block *); 2876 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2877 extern int ext4_init_inode_table(struct super_block *sb, 2878 ext4_group_t group, int barrier); 2879 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2880 2881 /* fast_commit.c */ 2882 int ext4_fc_info_show(struct seq_file *seq, void *v); 2883 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2884 void ext4_fc_init_inode(struct inode *inode); 2885 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2886 ext4_lblk_t end); 2887 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2888 struct dentry *dentry); 2889 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2890 struct dentry *dentry); 2891 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2892 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2893 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2894 struct dentry *dentry); 2895 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2896 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2897 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2898 void ext4_fc_start_update(struct inode *inode); 2899 void ext4_fc_stop_update(struct inode *inode); 2900 void ext4_fc_del(struct inode *inode); 2901 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2902 void ext4_fc_replay_cleanup(struct super_block *sb); 2903 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2904 int __init ext4_fc_init_dentry_cache(void); 2905 void ext4_fc_destroy_dentry_cache(void); 2906 int ext4_fc_record_regions(struct super_block *sb, int ino, 2907 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2908 int len, int replay); 2909 2910 /* mballoc.c */ 2911 extern const struct seq_operations ext4_mb_seq_groups_ops; 2912 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2913 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2914 extern int ext4_mb_init(struct super_block *); 2915 extern int ext4_mb_release(struct super_block *); 2916 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2917 struct ext4_allocation_request *, int *); 2918 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2919 extern int __init ext4_init_mballoc(void); 2920 extern void ext4_exit_mballoc(void); 2921 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2922 ext4_group_t group, 2923 unsigned int nr, int *cnt); 2924 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2925 unsigned int nr); 2926 2927 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2928 struct buffer_head *bh, ext4_fsblk_t block, 2929 unsigned long count, int flags); 2930 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2931 ext4_group_t ngroups); 2932 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2933 ext4_group_t i, struct ext4_group_desc *desc); 2934 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2935 ext4_fsblk_t block, unsigned long count); 2936 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2937 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2938 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2939 int len, bool state); 2940 static inline bool ext4_mb_cr_expensive(enum criteria cr) 2941 { 2942 return cr >= CR_GOAL_LEN_SLOW; 2943 } 2944 2945 /* inode.c */ 2946 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2947 struct ext4_inode_info *ei); 2948 int ext4_inode_is_fast_symlink(struct inode *inode); 2949 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2950 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2951 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2952 bool wait, struct buffer_head **bhs); 2953 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2954 struct buffer_head *bh_result, int create); 2955 int ext4_get_block(struct inode *inode, sector_t iblock, 2956 struct buffer_head *bh_result, int create); 2957 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2958 struct buffer_head *bh, int create); 2959 int ext4_walk_page_buffers(handle_t *handle, 2960 struct inode *inode, 2961 struct buffer_head *head, 2962 unsigned from, 2963 unsigned to, 2964 int *partial, 2965 int (*fn)(handle_t *handle, struct inode *inode, 2966 struct buffer_head *bh)); 2967 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2968 struct buffer_head *bh); 2969 #define FALL_BACK_TO_NONDELALLOC 1 2970 #define CONVERT_INLINE_DATA 2 2971 2972 typedef enum { 2973 EXT4_IGET_NORMAL = 0, 2974 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2975 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2976 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 2977 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 2978 } ext4_iget_flags; 2979 2980 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2981 ext4_iget_flags flags, const char *function, 2982 unsigned int line); 2983 2984 #define ext4_iget(sb, ino, flags) \ 2985 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2986 2987 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2988 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2989 struct iattr *); 2990 extern u32 ext4_dio_alignment(struct inode *inode); 2991 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2992 struct kstat *, u32, unsigned int); 2993 extern void ext4_evict_inode(struct inode *); 2994 extern void ext4_clear_inode(struct inode *); 2995 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2996 struct kstat *, u32, unsigned int); 2997 extern void ext4_dirty_inode(struct inode *, int); 2998 extern int ext4_change_inode_journal_flag(struct inode *, int); 2999 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3000 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3001 struct ext4_iloc *iloc); 3002 extern int ext4_inode_attach_jinode(struct inode *inode); 3003 extern int ext4_can_truncate(struct inode *inode); 3004 extern int ext4_truncate(struct inode *); 3005 extern int ext4_break_layouts(struct inode *); 3006 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3007 extern void ext4_set_inode_flags(struct inode *, bool init); 3008 extern int ext4_alloc_da_blocks(struct inode *inode); 3009 extern void ext4_set_aops(struct inode *inode); 3010 extern int ext4_writepage_trans_blocks(struct inode *); 3011 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3012 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3013 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3014 loff_t lstart, loff_t lend); 3015 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3016 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3017 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3018 extern void ext4_da_release_space(struct inode *inode, int to_free); 3019 extern void ext4_da_update_reserve_space(struct inode *inode, 3020 int used, int quota_claim); 3021 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3022 ext4_fsblk_t pblk, ext4_lblk_t len); 3023 3024 /* indirect.c */ 3025 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3026 struct ext4_map_blocks *map, int flags); 3027 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3028 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3029 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3030 ext4_lblk_t start, ext4_lblk_t end); 3031 3032 /* ioctl.c */ 3033 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3034 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3035 int ext4_fileattr_set(struct mnt_idmap *idmap, 3036 struct dentry *dentry, struct fileattr *fa); 3037 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3038 extern void ext4_reset_inode_seed(struct inode *inode); 3039 int ext4_update_overhead(struct super_block *sb, bool force); 3040 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3041 3042 /* migrate.c */ 3043 extern int ext4_ext_migrate(struct inode *); 3044 extern int ext4_ind_migrate(struct inode *inode); 3045 3046 /* namei.c */ 3047 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3048 struct inode *inode); 3049 extern int ext4_dirblock_csum_verify(struct inode *inode, 3050 struct buffer_head *bh); 3051 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3052 __u32 start_minor_hash, __u32 *next_hash); 3053 extern int ext4_search_dir(struct buffer_head *bh, 3054 char *search_buf, 3055 int buf_size, 3056 struct inode *dir, 3057 struct ext4_filename *fname, 3058 unsigned int offset, 3059 struct ext4_dir_entry_2 **res_dir); 3060 extern int ext4_generic_delete_entry(struct inode *dir, 3061 struct ext4_dir_entry_2 *de_del, 3062 struct buffer_head *bh, 3063 void *entry_buf, 3064 int buf_size, 3065 int csum_size); 3066 extern bool ext4_empty_dir(struct inode *inode); 3067 3068 /* resize.c */ 3069 extern void ext4_kvfree_array_rcu(void *to_free); 3070 extern int ext4_group_add(struct super_block *sb, 3071 struct ext4_new_group_data *input); 3072 extern int ext4_group_extend(struct super_block *sb, 3073 struct ext4_super_block *es, 3074 ext4_fsblk_t n_blocks_count); 3075 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3076 extern unsigned int ext4_list_backups(struct super_block *sb, 3077 unsigned int *three, unsigned int *five, 3078 unsigned int *seven); 3079 3080 /* super.c */ 3081 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3082 sector_t block, blk_opf_t op_flags); 3083 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3084 sector_t block); 3085 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3086 bh_end_io_t *end_io); 3087 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3088 bh_end_io_t *end_io); 3089 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3090 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3091 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3092 extern int ext4_calculate_overhead(struct super_block *sb); 3093 extern __le32 ext4_superblock_csum(struct super_block *sb, 3094 struct ext4_super_block *es); 3095 extern void ext4_superblock_csum_set(struct super_block *sb); 3096 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3097 ext4_group_t ngroup); 3098 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3099 char nbuf[16]); 3100 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3101 ext4_group_t block_group, 3102 unsigned int flags); 3103 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3104 ext4_group_t block_group); 3105 3106 extern __printf(7, 8) 3107 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3108 int, __u64, const char *, ...); 3109 extern __printf(6, 7) 3110 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3111 ext4_fsblk_t, int, const char *, ...); 3112 extern __printf(5, 6) 3113 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3114 const char *, ...); 3115 extern void __ext4_std_error(struct super_block *, const char *, 3116 unsigned int, int); 3117 extern __printf(4, 5) 3118 void __ext4_warning(struct super_block *, const char *, unsigned int, 3119 const char *, ...); 3120 extern __printf(4, 5) 3121 void __ext4_warning_inode(const struct inode *inode, const char *function, 3122 unsigned int line, const char *fmt, ...); 3123 extern __printf(3, 4) 3124 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3125 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3126 const char *, unsigned int, const char *); 3127 extern __printf(7, 8) 3128 void __ext4_grp_locked_error(const char *, unsigned int, 3129 struct super_block *, ext4_group_t, 3130 unsigned long, ext4_fsblk_t, 3131 const char *, ...); 3132 3133 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3134 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3135 3136 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3137 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3138 3139 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3140 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3141 (fmt), ## a) 3142 3143 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3144 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3145 3146 #define ext4_abort(sb, err, fmt, a...) \ 3147 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3148 3149 #ifdef CONFIG_PRINTK 3150 3151 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3152 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3153 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3154 __ext4_error_inode((inode), (func), (line), (block), \ 3155 (err), (fmt), ##__VA_ARGS__) 3156 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3157 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3158 #define ext4_error(sb, fmt, ...) \ 3159 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3160 ##__VA_ARGS__) 3161 #define ext4_error_err(sb, err, fmt, ...) \ 3162 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3163 ##__VA_ARGS__) 3164 #define ext4_warning(sb, fmt, ...) \ 3165 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3166 #define ext4_warning_inode(inode, fmt, ...) \ 3167 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3168 #define ext4_msg(sb, level, fmt, ...) \ 3169 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3170 #define dump_mmp_msg(sb, mmp, msg) \ 3171 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3172 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3173 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3174 fmt, ##__VA_ARGS__) 3175 3176 #else 3177 3178 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3179 do { \ 3180 no_printk(fmt, ##__VA_ARGS__); \ 3181 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3182 } while (0) 3183 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3184 do { \ 3185 no_printk(fmt, ##__VA_ARGS__); \ 3186 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3187 } while (0) 3188 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3189 do { \ 3190 no_printk(fmt, ##__VA_ARGS__); \ 3191 __ext4_error_file(file, "", 0, block, " "); \ 3192 } while (0) 3193 #define ext4_error(sb, fmt, ...) \ 3194 do { \ 3195 no_printk(fmt, ##__VA_ARGS__); \ 3196 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3197 } while (0) 3198 #define ext4_error_err(sb, err, fmt, ...) \ 3199 do { \ 3200 no_printk(fmt, ##__VA_ARGS__); \ 3201 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3202 } while (0) 3203 #define ext4_warning(sb, fmt, ...) \ 3204 do { \ 3205 no_printk(fmt, ##__VA_ARGS__); \ 3206 __ext4_warning(sb, "", 0, " "); \ 3207 } while (0) 3208 #define ext4_warning_inode(inode, fmt, ...) \ 3209 do { \ 3210 no_printk(fmt, ##__VA_ARGS__); \ 3211 __ext4_warning_inode(inode, "", 0, " "); \ 3212 } while (0) 3213 #define ext4_msg(sb, level, fmt, ...) \ 3214 do { \ 3215 no_printk(fmt, ##__VA_ARGS__); \ 3216 __ext4_msg(sb, "", " "); \ 3217 } while (0) 3218 #define dump_mmp_msg(sb, mmp, msg) \ 3219 __dump_mmp_msg(sb, mmp, "", 0, "") 3220 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3221 do { \ 3222 no_printk(fmt, ##__VA_ARGS__); \ 3223 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3224 } while (0) 3225 3226 #endif 3227 3228 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3229 struct ext4_group_desc *bg); 3230 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3231 struct ext4_group_desc *bg); 3232 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3233 struct ext4_group_desc *bg); 3234 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3235 struct ext4_group_desc *bg); 3236 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3237 struct ext4_group_desc *bg); 3238 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3239 struct ext4_group_desc *bg); 3240 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3241 struct ext4_group_desc *bg); 3242 extern void ext4_block_bitmap_set(struct super_block *sb, 3243 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3244 extern void ext4_inode_bitmap_set(struct super_block *sb, 3245 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3246 extern void ext4_inode_table_set(struct super_block *sb, 3247 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3248 extern void ext4_free_group_clusters_set(struct super_block *sb, 3249 struct ext4_group_desc *bg, 3250 __u32 count); 3251 extern void ext4_free_inodes_set(struct super_block *sb, 3252 struct ext4_group_desc *bg, __u32 count); 3253 extern void ext4_used_dirs_set(struct super_block *sb, 3254 struct ext4_group_desc *bg, __u32 count); 3255 extern void ext4_itable_unused_set(struct super_block *sb, 3256 struct ext4_group_desc *bg, __u32 count); 3257 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3258 struct ext4_group_desc *gdp); 3259 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3260 struct ext4_group_desc *gdp); 3261 extern int ext4_register_li_request(struct super_block *sb, 3262 ext4_group_t first_not_zeroed); 3263 3264 static inline int ext4_has_metadata_csum(struct super_block *sb) 3265 { 3266 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3267 !EXT4_SB(sb)->s_chksum_driver); 3268 3269 return ext4_has_feature_metadata_csum(sb) && 3270 (EXT4_SB(sb)->s_chksum_driver != NULL); 3271 } 3272 3273 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3274 { 3275 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3276 } 3277 3278 #define ext4_read_incompat_64bit_val(es, name) \ 3279 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3280 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3281 le32_to_cpu(es->name##_lo)) 3282 3283 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3284 { 3285 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3286 } 3287 3288 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3289 { 3290 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3291 } 3292 3293 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3294 { 3295 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3296 } 3297 3298 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3299 ext4_fsblk_t blk) 3300 { 3301 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3302 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3303 } 3304 3305 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3306 ext4_fsblk_t blk) 3307 { 3308 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3309 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3310 } 3311 3312 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3313 ext4_fsblk_t blk) 3314 { 3315 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3316 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3317 } 3318 3319 static inline loff_t ext4_isize(struct super_block *sb, 3320 struct ext4_inode *raw_inode) 3321 { 3322 if (ext4_has_feature_largedir(sb) || 3323 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3324 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3325 le32_to_cpu(raw_inode->i_size_lo); 3326 3327 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3328 } 3329 3330 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3331 { 3332 raw_inode->i_size_lo = cpu_to_le32(i_size); 3333 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3334 } 3335 3336 /* 3337 * Reading s_groups_count requires using smp_rmb() afterwards. See 3338 * the locking protocol documented in the comments of ext4_group_add() 3339 * in resize.c 3340 */ 3341 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3342 { 3343 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3344 3345 smp_rmb(); 3346 return ngroups; 3347 } 3348 3349 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3350 ext4_group_t block_group) 3351 { 3352 return block_group >> sbi->s_log_groups_per_flex; 3353 } 3354 3355 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3356 { 3357 return 1 << sbi->s_log_groups_per_flex; 3358 } 3359 3360 #define ext4_std_error(sb, errno) \ 3361 do { \ 3362 if ((errno)) \ 3363 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3364 } while (0) 3365 3366 #ifdef CONFIG_SMP 3367 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3368 * counters. So we need to make sure we have free clusters more 3369 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3370 */ 3371 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3372 #else 3373 #define EXT4_FREECLUSTERS_WATERMARK 0 3374 #endif 3375 3376 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3377 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3378 { 3379 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3380 !inode_is_locked(inode)); 3381 down_write(&EXT4_I(inode)->i_data_sem); 3382 if (newsize > EXT4_I(inode)->i_disksize) 3383 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3384 up_write(&EXT4_I(inode)->i_data_sem); 3385 } 3386 3387 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3388 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3389 { 3390 int changed = 0; 3391 3392 if (newsize > inode->i_size) { 3393 i_size_write(inode, newsize); 3394 changed = 1; 3395 } 3396 if (newsize > EXT4_I(inode)->i_disksize) { 3397 ext4_update_i_disksize(inode, newsize); 3398 changed |= 2; 3399 } 3400 return changed; 3401 } 3402 3403 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3404 loff_t len); 3405 3406 struct ext4_group_info { 3407 unsigned long bb_state; 3408 #ifdef AGGRESSIVE_CHECK 3409 unsigned long bb_check_counter; 3410 #endif 3411 struct rb_root bb_free_root; 3412 ext4_grpblk_t bb_first_free; /* first free block */ 3413 ext4_grpblk_t bb_free; /* total free blocks */ 3414 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3415 int bb_avg_fragment_size_order; /* order of average 3416 fragment in BG */ 3417 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3418 ext4_group_t bb_group; /* Group number */ 3419 struct list_head bb_prealloc_list; 3420 #ifdef DOUBLE_CHECK 3421 void *bb_bitmap; 3422 #endif 3423 struct rw_semaphore alloc_sem; 3424 struct list_head bb_avg_fragment_size_node; 3425 struct list_head bb_largest_free_order_node; 3426 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3427 * regions, index is order. 3428 * bb_counters[3] = 5 means 3429 * 5 free 8-block regions. */ 3430 }; 3431 3432 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3433 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3434 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3435 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3436 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3437 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3438 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3439 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3440 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3441 3442 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3443 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3444 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3445 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3446 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3447 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3448 3449 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3450 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3451 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3452 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3453 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3454 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3455 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3456 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3457 3458 #define EXT4_MAX_CONTENTION 8 3459 #define EXT4_CONTENTION_THRESHOLD 2 3460 3461 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3462 ext4_group_t group) 3463 { 3464 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3465 } 3466 3467 /* 3468 * Returns true if the filesystem is busy enough that attempts to 3469 * access the block group locks has run into contention. 3470 */ 3471 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3472 { 3473 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3474 } 3475 3476 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3477 { 3478 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3479 if (spin_trylock(lock)) 3480 /* 3481 * We're able to grab the lock right away, so drop the 3482 * lock contention counter. 3483 */ 3484 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3485 else { 3486 /* 3487 * The lock is busy, so bump the contention counter, 3488 * and then wait on the spin lock. 3489 */ 3490 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3491 EXT4_MAX_CONTENTION); 3492 spin_lock(lock); 3493 } 3494 } 3495 3496 static inline void ext4_unlock_group(struct super_block *sb, 3497 ext4_group_t group) 3498 { 3499 spin_unlock(ext4_group_lock_ptr(sb, group)); 3500 } 3501 3502 #ifdef CONFIG_QUOTA 3503 static inline bool ext4_quota_capable(struct super_block *sb) 3504 { 3505 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3506 } 3507 3508 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3509 { 3510 struct ext4_sb_info *sbi = EXT4_SB(sb); 3511 3512 return (ext4_has_feature_quota(sb) || 3513 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3514 } 3515 int ext4_enable_quotas(struct super_block *sb); 3516 #endif 3517 3518 /* 3519 * Block validity checking 3520 */ 3521 #define ext4_check_indirect_blockref(inode, bh) \ 3522 ext4_check_blockref(__func__, __LINE__, inode, \ 3523 (__le32 *)(bh)->b_data, \ 3524 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3525 3526 #define ext4_ind_check_inode(inode) \ 3527 ext4_check_blockref(__func__, __LINE__, inode, \ 3528 EXT4_I(inode)->i_data, \ 3529 EXT4_NDIR_BLOCKS) 3530 3531 /* 3532 * Inodes and files operations 3533 */ 3534 3535 /* dir.c */ 3536 extern const struct file_operations ext4_dir_operations; 3537 3538 /* file.c */ 3539 extern const struct inode_operations ext4_file_inode_operations; 3540 extern const struct file_operations ext4_file_operations; 3541 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3542 3543 /* inline.c */ 3544 extern int ext4_get_max_inline_size(struct inode *inode); 3545 extern int ext4_find_inline_data_nolock(struct inode *inode); 3546 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3547 3548 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3549 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3550 struct inode *inode, 3551 loff_t pos, unsigned len, 3552 struct page **pagep); 3553 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3554 unsigned copied, struct folio *folio); 3555 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3556 struct inode *inode, 3557 loff_t pos, unsigned len, 3558 struct page **pagep, 3559 void **fsdata); 3560 extern int ext4_try_add_inline_entry(handle_t *handle, 3561 struct ext4_filename *fname, 3562 struct inode *dir, struct inode *inode); 3563 extern int ext4_try_create_inline_dir(handle_t *handle, 3564 struct inode *parent, 3565 struct inode *inode); 3566 extern int ext4_read_inline_dir(struct file *filp, 3567 struct dir_context *ctx, 3568 int *has_inline_data); 3569 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3570 struct inode *dir, ext4_lblk_t block, 3571 struct dx_hash_info *hinfo, 3572 __u32 start_hash, __u32 start_minor_hash, 3573 int *has_inline_data); 3574 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3575 struct ext4_filename *fname, 3576 struct ext4_dir_entry_2 **res_dir, 3577 int *has_inline_data); 3578 extern int ext4_delete_inline_entry(handle_t *handle, 3579 struct inode *dir, 3580 struct ext4_dir_entry_2 *de_del, 3581 struct buffer_head *bh, 3582 int *has_inline_data); 3583 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3584 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3585 struct ext4_dir_entry_2 **parent_de, 3586 int *retval); 3587 extern void *ext4_read_inline_link(struct inode *inode); 3588 3589 struct iomap; 3590 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3591 3592 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3593 3594 extern int ext4_convert_inline_data(struct inode *inode); 3595 3596 static inline int ext4_has_inline_data(struct inode *inode) 3597 { 3598 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3599 EXT4_I(inode)->i_inline_off; 3600 } 3601 3602 /* namei.c */ 3603 extern const struct inode_operations ext4_dir_inode_operations; 3604 extern const struct inode_operations ext4_special_inode_operations; 3605 extern struct dentry *ext4_get_parent(struct dentry *child); 3606 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3607 struct ext4_dir_entry_2 *de, 3608 int blocksize, int csum_size, 3609 unsigned int parent_ino, int dotdot_real_len); 3610 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3611 unsigned int blocksize); 3612 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3613 struct buffer_head *bh); 3614 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3615 struct inode *inode, struct dentry *dentry); 3616 extern int __ext4_link(struct inode *dir, struct inode *inode, 3617 struct dentry *dentry); 3618 3619 #define S_SHIFT 12 3620 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3621 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3622 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3623 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3624 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3625 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3626 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3627 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3628 }; 3629 3630 static inline void ext4_set_de_type(struct super_block *sb, 3631 struct ext4_dir_entry_2 *de, 3632 umode_t mode) { 3633 if (ext4_has_feature_filetype(sb)) 3634 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3635 } 3636 3637 /* readpages.c */ 3638 extern int ext4_mpage_readpages(struct inode *inode, 3639 struct readahead_control *rac, struct folio *folio); 3640 extern int __init ext4_init_post_read_processing(void); 3641 extern void ext4_exit_post_read_processing(void); 3642 3643 /* symlink.c */ 3644 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3645 extern const struct inode_operations ext4_symlink_inode_operations; 3646 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3647 3648 /* sysfs.c */ 3649 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3650 extern int ext4_register_sysfs(struct super_block *sb); 3651 extern void ext4_unregister_sysfs(struct super_block *sb); 3652 extern int __init ext4_init_sysfs(void); 3653 extern void ext4_exit_sysfs(void); 3654 3655 /* block_validity */ 3656 extern void ext4_release_system_zone(struct super_block *sb); 3657 extern int ext4_setup_system_zone(struct super_block *sb); 3658 extern int __init ext4_init_system_zone(void); 3659 extern void ext4_exit_system_zone(void); 3660 extern int ext4_inode_block_valid(struct inode *inode, 3661 ext4_fsblk_t start_blk, 3662 unsigned int count); 3663 extern int ext4_check_blockref(const char *, unsigned int, 3664 struct inode *, __le32 *, unsigned int); 3665 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3666 ext4_fsblk_t start_blk, unsigned int count); 3667 3668 3669 /* extents.c */ 3670 struct ext4_ext_path; 3671 struct ext4_extent; 3672 3673 /* 3674 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3675 * __le32. 3676 */ 3677 #define EXT_MAX_BLOCKS 0xffffffff 3678 3679 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3680 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3681 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3682 struct ext4_map_blocks *map, int flags); 3683 extern int ext4_ext_truncate(handle_t *, struct inode *); 3684 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3685 ext4_lblk_t end); 3686 extern void ext4_ext_init(struct super_block *); 3687 extern void ext4_ext_release(struct super_block *); 3688 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3689 loff_t len); 3690 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3691 loff_t offset, ssize_t len); 3692 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3693 ext4_io_end_t *io_end); 3694 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3695 struct ext4_map_blocks *map, int flags); 3696 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3697 int num, 3698 struct ext4_ext_path *path); 3699 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3700 struct ext4_ext_path **, 3701 struct ext4_extent *, int); 3702 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3703 struct ext4_ext_path **, 3704 int flags); 3705 extern void ext4_free_ext_path(struct ext4_ext_path *); 3706 extern int ext4_ext_check_inode(struct inode *inode); 3707 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3708 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3709 __u64 start, __u64 len); 3710 extern int ext4_get_es_cache(struct inode *inode, 3711 struct fiemap_extent_info *fieinfo, 3712 __u64 start, __u64 len); 3713 extern int ext4_ext_precache(struct inode *inode); 3714 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3715 struct inode *inode2, ext4_lblk_t lblk1, 3716 ext4_lblk_t lblk2, ext4_lblk_t count, 3717 int mark_unwritten,int *err); 3718 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3719 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3720 int check_cred, int restart_cred, 3721 int revoke_cred); 3722 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3723 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3724 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3725 int len, int unwritten, ext4_fsblk_t pblk); 3726 extern int ext4_ext_clear_bb(struct inode *inode); 3727 3728 3729 /* move_extent.c */ 3730 extern void ext4_double_down_write_data_sem(struct inode *first, 3731 struct inode *second); 3732 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3733 struct inode *donor_inode); 3734 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3735 __u64 start_orig, __u64 start_donor, 3736 __u64 len, __u64 *moved_len); 3737 3738 /* page-io.c */ 3739 extern int __init ext4_init_pageio(void); 3740 extern void ext4_exit_pageio(void); 3741 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3742 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3743 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3744 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3745 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3746 struct writeback_control *wbc); 3747 extern void ext4_end_io_rsv_work(struct work_struct *work); 3748 extern void ext4_io_submit(struct ext4_io_submit *io); 3749 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3750 size_t len); 3751 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3752 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3753 3754 /* mmp.c */ 3755 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3756 3757 /* mmp.c */ 3758 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3759 3760 /* verity.c */ 3761 extern const struct fsverity_operations ext4_verityops; 3762 3763 /* orphan.c */ 3764 extern int ext4_orphan_add(handle_t *, struct inode *); 3765 extern int ext4_orphan_del(handle_t *, struct inode *); 3766 extern void ext4_orphan_cleanup(struct super_block *sb, 3767 struct ext4_super_block *es); 3768 extern void ext4_release_orphan_info(struct super_block *sb); 3769 extern int ext4_init_orphan_info(struct super_block *sb); 3770 extern int ext4_orphan_file_empty(struct super_block *sb); 3771 extern void ext4_orphan_file_block_trigger( 3772 struct jbd2_buffer_trigger_type *triggers, 3773 struct buffer_head *bh, 3774 void *data, size_t size); 3775 3776 /* 3777 * Add new method to test whether block and inode bitmaps are properly 3778 * initialized. With uninit_bg reading the block from disk is not enough 3779 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3780 */ 3781 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3782 3783 static inline int bitmap_uptodate(struct buffer_head *bh) 3784 { 3785 return (buffer_uptodate(bh) && 3786 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3787 } 3788 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3789 { 3790 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3791 } 3792 3793 /* For ioend & aio unwritten conversion wait queues */ 3794 #define EXT4_WQ_HASH_SZ 37 3795 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3796 EXT4_WQ_HASH_SZ]) 3797 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3798 3799 extern int ext4_resize_begin(struct super_block *sb); 3800 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3801 3802 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3803 struct ext4_io_end *io_end) 3804 { 3805 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3806 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3807 atomic_inc(&EXT4_I(inode)->i_unwritten); 3808 } 3809 } 3810 3811 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3812 { 3813 struct inode *inode = io_end->inode; 3814 3815 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3816 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3817 /* Wake up anyone waiting on unwritten extent conversion */ 3818 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3819 wake_up_all(ext4_ioend_wq(inode)); 3820 } 3821 } 3822 3823 extern const struct iomap_ops ext4_iomap_ops; 3824 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3825 extern const struct iomap_ops ext4_iomap_report_ops; 3826 3827 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3828 { 3829 /* 3830 * If the buffer has the write error flag, we have failed 3831 * to write out data in the block. In this case, we don't 3832 * have to read the block because we may read the old data 3833 * successfully. 3834 */ 3835 if (buffer_write_io_error(bh)) 3836 set_buffer_uptodate(bh); 3837 return buffer_uptodate(bh); 3838 } 3839 3840 #endif /* __KERNEL__ */ 3841 3842 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3843 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3844 3845 #endif /* _EXT4_H */ 3846