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 /* criteria below which we use fast block scanning and avoid unnecessary IO */ 180 #define CR_FAST CR_GOAL_LEN_SLOW 181 182 /* 183 * Flags used in mballoc's allocation_context flags field. 184 * 185 * Also used to show what's going on for debugging purposes when the 186 * flag field is exported via the traceport interface 187 */ 188 189 /* prefer goal again. length */ 190 #define EXT4_MB_HINT_MERGE 0x0001 191 /* blocks already reserved */ 192 #define EXT4_MB_HINT_RESERVED 0x0002 193 /* metadata is being allocated */ 194 #define EXT4_MB_HINT_METADATA 0x0004 195 /* first blocks in the file */ 196 #define EXT4_MB_HINT_FIRST 0x0008 197 /* search for the best chunk */ 198 #define EXT4_MB_HINT_BEST 0x0010 199 /* data is being allocated */ 200 #define EXT4_MB_HINT_DATA 0x0020 201 /* don't preallocate (for tails) */ 202 #define EXT4_MB_HINT_NOPREALLOC 0x0040 203 /* allocate for locality group */ 204 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 205 /* allocate goal blocks or none */ 206 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 207 /* goal is meaningful */ 208 #define EXT4_MB_HINT_TRY_GOAL 0x0200 209 /* blocks already pre-reserved by delayed allocation */ 210 #define EXT4_MB_DELALLOC_RESERVED 0x0400 211 /* We are doing stream allocation */ 212 #define EXT4_MB_STREAM_ALLOC 0x0800 213 /* Use reserved root blocks if needed */ 214 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 215 /* Use blocks from reserved pool */ 216 #define EXT4_MB_USE_RESERVED 0x2000 217 /* Do strict check for free blocks while retrying block allocation */ 218 #define EXT4_MB_STRICT_CHECK 0x4000 219 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 220 #define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000 221 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 222 #define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000 223 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1.5 */ 224 #define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000 225 226 struct ext4_allocation_request { 227 /* target inode for block we're allocating */ 228 struct inode *inode; 229 /* how many blocks we want to allocate */ 230 unsigned int len; 231 /* logical block in target inode */ 232 ext4_lblk_t logical; 233 /* the closest logical allocated block to the left */ 234 ext4_lblk_t lleft; 235 /* the closest logical allocated block to the right */ 236 ext4_lblk_t lright; 237 /* phys. target (a hint) */ 238 ext4_fsblk_t goal; 239 /* phys. block for the closest logical allocated block to the left */ 240 ext4_fsblk_t pleft; 241 /* phys. block for the closest logical allocated block to the right */ 242 ext4_fsblk_t pright; 243 /* flags. see above EXT4_MB_HINT_* */ 244 unsigned int flags; 245 }; 246 247 /* 248 * Logical to physical block mapping, used by ext4_map_blocks() 249 * 250 * This structure is used to pass requests into ext4_map_blocks() as 251 * well as to store the information returned by ext4_map_blocks(). It 252 * takes less room on the stack than a struct buffer_head. 253 */ 254 #define EXT4_MAP_NEW BIT(BH_New) 255 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 256 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 257 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 258 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 259 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 260 261 struct ext4_map_blocks { 262 ext4_fsblk_t m_pblk; 263 ext4_lblk_t m_lblk; 264 unsigned int m_len; 265 unsigned int m_flags; 266 }; 267 268 /* 269 * Block validity checking, system zone rbtree. 270 */ 271 struct ext4_system_blocks { 272 struct rb_root root; 273 struct rcu_head rcu; 274 }; 275 276 /* 277 * Flags for ext4_io_end->flags 278 */ 279 #define EXT4_IO_END_UNWRITTEN 0x0001 280 281 struct ext4_io_end_vec { 282 struct list_head list; /* list of io_end_vec */ 283 loff_t offset; /* offset in the file */ 284 ssize_t size; /* size of the extent */ 285 }; 286 287 /* 288 * For converting unwritten extents on a work queue. 'handle' is used for 289 * buffered writeback. 290 */ 291 typedef struct ext4_io_end { 292 struct list_head list; /* per-file finished IO list */ 293 handle_t *handle; /* handle reserved for extent 294 * conversion */ 295 struct inode *inode; /* file being written to */ 296 struct bio *bio; /* Linked list of completed 297 * bios covering the extent */ 298 unsigned int flag; /* unwritten or not */ 299 refcount_t count; /* reference counter */ 300 struct list_head list_vec; /* list of ext4_io_end_vec */ 301 } ext4_io_end_t; 302 303 struct ext4_io_submit { 304 struct writeback_control *io_wbc; 305 struct bio *io_bio; 306 ext4_io_end_t *io_end; 307 sector_t io_next_block; 308 }; 309 310 /* 311 * Special inodes numbers 312 */ 313 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 314 #define EXT4_ROOT_INO 2 /* Root inode */ 315 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 316 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 317 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 318 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 319 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 320 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 321 322 /* First non-reserved inode for old ext4 filesystems */ 323 #define EXT4_GOOD_OLD_FIRST_INO 11 324 325 /* 326 * Maximal count of links to a file 327 */ 328 #define EXT4_LINK_MAX 65000 329 330 /* 331 * Macro-instructions used to manage several block sizes 332 */ 333 #define EXT4_MIN_BLOCK_SIZE 1024 334 #define EXT4_MAX_BLOCK_SIZE 65536 335 #define EXT4_MIN_BLOCK_LOG_SIZE 10 336 #define EXT4_MAX_BLOCK_LOG_SIZE 16 337 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 338 #ifdef __KERNEL__ 339 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 340 #else 341 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 342 #endif 343 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 344 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 345 EXT4_SB(s)->s_cluster_bits) 346 #ifdef __KERNEL__ 347 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 348 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 349 #else 350 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 351 #endif 352 #ifdef __KERNEL__ 353 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 354 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 355 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 356 #else 357 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 358 EXT4_GOOD_OLD_INODE_SIZE : \ 359 (s)->s_inode_size) 360 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 361 EXT4_GOOD_OLD_FIRST_INO : \ 362 (s)->s_first_ino) 363 #endif 364 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 365 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 366 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 367 blkbits)) 368 369 /* Translate a block number to a cluster number */ 370 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 371 /* Translate a cluster number to a block number */ 372 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 373 /* Translate # of blks to # of clusters */ 374 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 375 (sbi)->s_cluster_bits) 376 /* Mask out the low bits to get the starting block of the cluster */ 377 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 378 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 379 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 380 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 381 /* Fill in the low bits to get the last block of the cluster */ 382 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 383 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 384 /* Get the cluster offset */ 385 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 386 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 387 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 388 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 389 390 /* 391 * Structure of a blocks group descriptor 392 */ 393 struct ext4_group_desc 394 { 395 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 396 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 397 __le32 bg_inode_table_lo; /* Inodes table block */ 398 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 399 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 400 __le16 bg_used_dirs_count_lo; /* Directories count */ 401 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 402 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 403 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 404 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 405 __le16 bg_itable_unused_lo; /* Unused inodes count */ 406 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 407 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 408 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 409 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 410 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 411 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 412 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 413 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 414 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 415 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 416 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 417 __u32 bg_reserved; 418 }; 419 420 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 421 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 422 sizeof(__le16)) 423 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 424 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 425 sizeof(__le16)) 426 427 /* 428 * Structure of a flex block group info 429 */ 430 431 struct flex_groups { 432 atomic64_t free_clusters; 433 atomic_t free_inodes; 434 atomic_t used_dirs; 435 }; 436 437 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 438 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 439 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 440 441 /* 442 * Macro-instructions used to manage group descriptors 443 */ 444 #define EXT4_MIN_DESC_SIZE 32 445 #define EXT4_MIN_DESC_SIZE_64BIT 64 446 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 447 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 448 #ifdef __KERNEL__ 449 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 450 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 451 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 452 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 453 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 454 #else 455 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 456 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 457 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 458 #endif 459 460 /* 461 * Constants relative to the data blocks 462 */ 463 #define EXT4_NDIR_BLOCKS 12 464 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 465 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 466 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 467 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 468 469 /* 470 * Inode flags 471 */ 472 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 473 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 474 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 475 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 476 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 477 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 478 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 479 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 480 /* Reserved for compression usage... */ 481 #define EXT4_DIRTY_FL 0x00000100 482 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 483 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 484 /* nb: was previously EXT2_ECOMPR_FL */ 485 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 486 /* End compression flags --- maybe not all used */ 487 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 488 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 489 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 490 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 491 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 492 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 493 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 494 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 495 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 496 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 497 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 498 499 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 500 501 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 502 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 503 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 504 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 505 506 /* User modifiable flags */ 507 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 508 EXT4_UNRM_FL | \ 509 EXT4_COMPR_FL | \ 510 EXT4_SYNC_FL | \ 511 EXT4_IMMUTABLE_FL | \ 512 EXT4_APPEND_FL | \ 513 EXT4_NODUMP_FL | \ 514 EXT4_NOATIME_FL | \ 515 EXT4_JOURNAL_DATA_FL | \ 516 EXT4_NOTAIL_FL | \ 517 EXT4_DIRSYNC_FL | \ 518 EXT4_TOPDIR_FL | \ 519 EXT4_EXTENTS_FL | \ 520 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 521 EXT4_DAX_FL | \ 522 EXT4_PROJINHERIT_FL | \ 523 EXT4_CASEFOLD_FL) 524 525 /* User visible flags */ 526 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 527 EXT4_DIRTY_FL | \ 528 EXT4_COMPRBLK_FL | \ 529 EXT4_NOCOMPR_FL | \ 530 EXT4_ENCRYPT_FL | \ 531 EXT4_INDEX_FL | \ 532 EXT4_VERITY_FL | \ 533 EXT4_INLINE_DATA_FL) 534 535 /* Flags that should be inherited by new inodes from their parent. */ 536 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 537 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 538 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 539 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 540 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 541 EXT4_DAX_FL) 542 543 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 544 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 545 EXT4_PROJINHERIT_FL)) 546 547 /* Flags that are appropriate for non-directories/regular files. */ 548 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 549 550 /* The only flags that should be swapped */ 551 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 552 553 /* Flags which are mutually exclusive to DAX */ 554 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 555 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 556 557 /* Mask out flags that are inappropriate for the given type of inode. */ 558 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 559 { 560 if (S_ISDIR(mode)) 561 return flags; 562 else if (S_ISREG(mode)) 563 return flags & EXT4_REG_FLMASK; 564 else 565 return flags & EXT4_OTHER_FLMASK; 566 } 567 568 /* 569 * Inode flags used for atomic set/get 570 */ 571 enum { 572 EXT4_INODE_SECRM = 0, /* Secure deletion */ 573 EXT4_INODE_UNRM = 1, /* Undelete */ 574 EXT4_INODE_COMPR = 2, /* Compress file */ 575 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 576 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 577 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 578 EXT4_INODE_NODUMP = 6, /* do not dump file */ 579 EXT4_INODE_NOATIME = 7, /* do not update atime */ 580 /* Reserved for compression usage... */ 581 EXT4_INODE_DIRTY = 8, 582 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 583 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 584 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 585 /* End compression flags --- maybe not all used */ 586 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 587 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 588 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 589 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 590 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 591 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 592 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 593 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 594 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 595 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 596 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 597 EXT4_INODE_DAX = 25, /* Inode is DAX */ 598 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 599 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 600 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 601 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 602 }; 603 604 /* 605 * Since it's pretty easy to mix up bit numbers and hex values, we use a 606 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 607 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 608 * any extra space in the compiled kernel image, otherwise, the build will fail. 609 * It's important that these values are the same, since we are using 610 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 611 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 612 * values found in ext2, ext3 and ext4 filesystems, and of course the values 613 * defined in e2fsprogs. 614 * 615 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 616 */ 617 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 618 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 619 620 static inline void ext4_check_flag_values(void) 621 { 622 CHECK_FLAG_VALUE(SECRM); 623 CHECK_FLAG_VALUE(UNRM); 624 CHECK_FLAG_VALUE(COMPR); 625 CHECK_FLAG_VALUE(SYNC); 626 CHECK_FLAG_VALUE(IMMUTABLE); 627 CHECK_FLAG_VALUE(APPEND); 628 CHECK_FLAG_VALUE(NODUMP); 629 CHECK_FLAG_VALUE(NOATIME); 630 CHECK_FLAG_VALUE(DIRTY); 631 CHECK_FLAG_VALUE(COMPRBLK); 632 CHECK_FLAG_VALUE(NOCOMPR); 633 CHECK_FLAG_VALUE(ENCRYPT); 634 CHECK_FLAG_VALUE(INDEX); 635 CHECK_FLAG_VALUE(IMAGIC); 636 CHECK_FLAG_VALUE(JOURNAL_DATA); 637 CHECK_FLAG_VALUE(NOTAIL); 638 CHECK_FLAG_VALUE(DIRSYNC); 639 CHECK_FLAG_VALUE(TOPDIR); 640 CHECK_FLAG_VALUE(HUGE_FILE); 641 CHECK_FLAG_VALUE(EXTENTS); 642 CHECK_FLAG_VALUE(VERITY); 643 CHECK_FLAG_VALUE(EA_INODE); 644 CHECK_FLAG_VALUE(INLINE_DATA); 645 CHECK_FLAG_VALUE(PROJINHERIT); 646 CHECK_FLAG_VALUE(CASEFOLD); 647 CHECK_FLAG_VALUE(RESERVED); 648 } 649 650 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 651 struct compat_ext4_new_group_input { 652 u32 group; 653 compat_u64 block_bitmap; 654 compat_u64 inode_bitmap; 655 compat_u64 inode_table; 656 u32 blocks_count; 657 u16 reserved_blocks; 658 u16 unused; 659 }; 660 #endif 661 662 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 663 struct ext4_new_group_data { 664 __u32 group; 665 __u64 block_bitmap; 666 __u64 inode_bitmap; 667 __u64 inode_table; 668 __u32 blocks_count; 669 __u16 reserved_blocks; 670 __u16 mdata_blocks; 671 __u32 free_clusters_count; 672 }; 673 674 /* Indexes used to index group tables in ext4_new_group_data */ 675 enum { 676 BLOCK_BITMAP = 0, /* block bitmap */ 677 INODE_BITMAP, /* inode bitmap */ 678 INODE_TABLE, /* inode tables */ 679 GROUP_TABLE_COUNT, 680 }; 681 682 /* 683 * Flags used by ext4_map_blocks() 684 */ 685 /* Allocate any needed blocks and/or convert an unwritten 686 extent to be an initialized ext4 */ 687 #define EXT4_GET_BLOCKS_CREATE 0x0001 688 /* Request the creation of an unwritten extent */ 689 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 690 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 691 EXT4_GET_BLOCKS_CREATE) 692 /* Caller is from the delayed allocation writeout path 693 * finally doing the actual allocation of delayed blocks */ 694 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 695 /* caller is from the direct IO path, request to creation of an 696 unwritten extents if not allocated, split the unwritten 697 extent if blocks has been preallocated already*/ 698 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 699 #define EXT4_GET_BLOCKS_CONVERT 0x0010 700 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 701 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 702 /* Convert extent to initialized after IO complete */ 703 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 704 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 705 /* Eventual metadata allocation (due to growing extent tree) 706 * should not fail, so try to use reserved blocks for that.*/ 707 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 708 /* Don't normalize allocation size (used for fallocate) */ 709 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 710 /* Convert written extents to unwritten */ 711 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 712 /* Write zeros to newly created written extents */ 713 #define EXT4_GET_BLOCKS_ZERO 0x0200 714 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 715 EXT4_GET_BLOCKS_ZERO) 716 /* Caller will submit data before dropping transaction handle. This 717 * allows jbd2 to avoid submitting data before commit. */ 718 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 719 /* Caller is in the atomic contex, find extent if it has been cached */ 720 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 721 722 /* 723 * The bit position of these flags must not overlap with any of the 724 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 725 * read_extent_tree_block(), ext4_split_extent_at(), 726 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 727 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 728 * caching the extents when reading from the extent tree while a 729 * truncate or punch hole operation is in progress. 730 */ 731 #define EXT4_EX_NOCACHE 0x40000000 732 #define EXT4_EX_FORCE_CACHE 0x20000000 733 #define EXT4_EX_NOFAIL 0x10000000 734 735 /* 736 * Flags used by ext4_free_blocks 737 */ 738 #define EXT4_FREE_BLOCKS_METADATA 0x0001 739 #define EXT4_FREE_BLOCKS_FORGET 0x0002 740 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 741 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 742 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 743 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 744 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 745 746 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 747 /* 748 * ioctl commands in 32 bit emulation 749 */ 750 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 751 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 752 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 753 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 754 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 755 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 756 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 757 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 758 #endif 759 760 /* Max physical block we can address w/o extents */ 761 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 762 763 /* Max logical block we can support */ 764 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 765 766 /* 767 * Structure of an inode on the disk 768 */ 769 struct ext4_inode { 770 __le16 i_mode; /* File mode */ 771 __le16 i_uid; /* Low 16 bits of Owner Uid */ 772 __le32 i_size_lo; /* Size in bytes */ 773 __le32 i_atime; /* Access time */ 774 __le32 i_ctime; /* Inode Change time */ 775 __le32 i_mtime; /* Modification time */ 776 __le32 i_dtime; /* Deletion Time */ 777 __le16 i_gid; /* Low 16 bits of Group Id */ 778 __le16 i_links_count; /* Links count */ 779 __le32 i_blocks_lo; /* Blocks count */ 780 __le32 i_flags; /* File flags */ 781 union { 782 struct { 783 __le32 l_i_version; 784 } linux1; 785 struct { 786 __u32 h_i_translator; 787 } hurd1; 788 struct { 789 __u32 m_i_reserved1; 790 } masix1; 791 } osd1; /* OS dependent 1 */ 792 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 793 __le32 i_generation; /* File version (for NFS) */ 794 __le32 i_file_acl_lo; /* File ACL */ 795 __le32 i_size_high; 796 __le32 i_obso_faddr; /* Obsoleted fragment address */ 797 union { 798 struct { 799 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 800 __le16 l_i_file_acl_high; 801 __le16 l_i_uid_high; /* these 2 fields */ 802 __le16 l_i_gid_high; /* were reserved2[0] */ 803 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 804 __le16 l_i_reserved; 805 } linux2; 806 struct { 807 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 808 __u16 h_i_mode_high; 809 __u16 h_i_uid_high; 810 __u16 h_i_gid_high; 811 __u32 h_i_author; 812 } hurd2; 813 struct { 814 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 815 __le16 m_i_file_acl_high; 816 __u32 m_i_reserved2[2]; 817 } masix2; 818 } osd2; /* OS dependent 2 */ 819 __le16 i_extra_isize; 820 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 821 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 822 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 823 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 824 __le32 i_crtime; /* File Creation time */ 825 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 826 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 827 __le32 i_projid; /* Project ID */ 828 }; 829 830 #define EXT4_EPOCH_BITS 2 831 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 832 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 833 834 /* 835 * Extended fields will fit into an inode if the filesystem was formatted 836 * with large inodes (-I 256 or larger) and there are not currently any EAs 837 * consuming all of the available space. For new inodes we always reserve 838 * enough space for the kernel's known extended fields, but for inodes 839 * created with an old kernel this might not have been the case. None of 840 * the extended inode fields is critical for correct filesystem operation. 841 * This macro checks if a certain field fits in the inode. Note that 842 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 843 */ 844 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 845 ((offsetof(typeof(*ext4_inode), field) + \ 846 sizeof((ext4_inode)->field)) \ 847 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 848 (einode)->i_extra_isize)) \ 849 850 /* 851 * We use an encoding that preserves the times for extra epoch "00": 852 * 853 * extra msb of adjust for signed 854 * epoch 32-bit 32-bit tv_sec to 855 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 856 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 857 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 858 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 859 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 860 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 861 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 862 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 863 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 864 * 865 * Note that previous versions of the kernel on 64-bit systems would 866 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 867 * 1970. e2fsck will correct this, assuming that it is run on the 868 * affected filesystem before 2242. 869 */ 870 871 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 872 { 873 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 874 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 875 } 876 877 static inline void ext4_decode_extra_time(struct timespec64 *time, 878 __le32 extra) 879 { 880 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 881 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 882 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 883 } 884 885 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 886 do { \ 887 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 888 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 889 (raw_inode)->xtime ## _extra = \ 890 ext4_encode_extra_time(&(inode)->xtime); \ 891 } \ 892 else \ 893 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 894 } while (0) 895 896 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 897 do { \ 898 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 899 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 900 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 901 (raw_inode)->xtime ## _extra = \ 902 ext4_encode_extra_time(&(einode)->xtime); \ 903 } while (0) 904 905 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 906 do { \ 907 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 908 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 909 ext4_decode_extra_time(&(inode)->xtime, \ 910 raw_inode->xtime ## _extra); \ 911 } \ 912 else \ 913 (inode)->xtime.tv_nsec = 0; \ 914 } while (0) 915 916 917 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 918 do { \ 919 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 920 (einode)->xtime.tv_sec = \ 921 (signed)le32_to_cpu((raw_inode)->xtime); \ 922 else \ 923 (einode)->xtime.tv_sec = 0; \ 924 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 925 ext4_decode_extra_time(&(einode)->xtime, \ 926 raw_inode->xtime ## _extra); \ 927 else \ 928 (einode)->xtime.tv_nsec = 0; \ 929 } while (0) 930 931 #define i_disk_version osd1.linux1.l_i_version 932 933 #if defined(__KERNEL__) || defined(__linux__) 934 #define i_reserved1 osd1.linux1.l_i_reserved1 935 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 936 #define i_blocks_high osd2.linux2.l_i_blocks_high 937 #define i_uid_low i_uid 938 #define i_gid_low i_gid 939 #define i_uid_high osd2.linux2.l_i_uid_high 940 #define i_gid_high osd2.linux2.l_i_gid_high 941 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 942 943 #elif defined(__GNU__) 944 945 #define i_translator osd1.hurd1.h_i_translator 946 #define i_uid_high osd2.hurd2.h_i_uid_high 947 #define i_gid_high osd2.hurd2.h_i_gid_high 948 #define i_author osd2.hurd2.h_i_author 949 950 #elif defined(__masix__) 951 952 #define i_reserved1 osd1.masix1.m_i_reserved1 953 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 954 #define i_reserved2 osd2.masix2.m_i_reserved2 955 956 #endif /* defined(__KERNEL__) || defined(__linux__) */ 957 958 #include "extents_status.h" 959 #include "fast_commit.h" 960 961 /* 962 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 963 * 964 * These are needed to avoid lockdep false positives when we need to 965 * allocate blocks to the quota inode during ext4_map_blocks(), while 966 * holding i_data_sem for a normal (non-quota) inode. Since we don't 967 * do quota tracking for the quota inode, this avoids deadlock (as 968 * well as infinite recursion, since it isn't turtles all the way 969 * down...) 970 * 971 * I_DATA_SEM_NORMAL - Used for most inodes 972 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 973 * where the second inode has larger inode number 974 * than the first 975 * I_DATA_SEM_QUOTA - Used for quota inodes only 976 * I_DATA_SEM_EA - Used for ea_inodes only 977 */ 978 enum { 979 I_DATA_SEM_NORMAL = 0, 980 I_DATA_SEM_OTHER, 981 I_DATA_SEM_QUOTA, 982 I_DATA_SEM_EA 983 }; 984 985 986 /* 987 * fourth extended file system inode data in memory 988 */ 989 struct ext4_inode_info { 990 __le32 i_data[15]; /* unconverted */ 991 __u32 i_dtime; 992 ext4_fsblk_t i_file_acl; 993 994 /* 995 * i_block_group is the number of the block group which contains 996 * this file's inode. Constant across the lifetime of the inode, 997 * it is used for making block allocation decisions - we try to 998 * place a file's data blocks near its inode block, and new inodes 999 * near to their parent directory's inode. 1000 */ 1001 ext4_group_t i_block_group; 1002 ext4_lblk_t i_dir_start_lookup; 1003 #if (BITS_PER_LONG < 64) 1004 unsigned long i_state_flags; /* Dynamic state flags */ 1005 #endif 1006 unsigned long i_flags; 1007 1008 /* 1009 * Extended attributes can be read independently of the main file 1010 * data. Taking i_rwsem even when reading would cause contention 1011 * between readers of EAs and writers of regular file data, so 1012 * instead we synchronize on xattr_sem when reading or changing 1013 * EAs. 1014 */ 1015 struct rw_semaphore xattr_sem; 1016 1017 /* 1018 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1019 * i_orphan is used. 1020 */ 1021 union { 1022 struct list_head i_orphan; /* unlinked but open inodes */ 1023 unsigned int i_orphan_idx; /* Index in orphan file */ 1024 }; 1025 1026 /* Fast commit related info */ 1027 1028 /* For tracking dentry create updates */ 1029 struct list_head i_fc_dilist; 1030 struct list_head i_fc_list; /* 1031 * inodes that need fast commit 1032 * protected by sbi->s_fc_lock. 1033 */ 1034 1035 /* Start of lblk range that needs to be committed in this fast commit */ 1036 ext4_lblk_t i_fc_lblk_start; 1037 1038 /* End of lblk range that needs to be committed in this fast commit */ 1039 ext4_lblk_t i_fc_lblk_len; 1040 1041 /* Number of ongoing updates on this inode */ 1042 atomic_t i_fc_updates; 1043 1044 /* Fast commit wait queue for this inode */ 1045 wait_queue_head_t i_fc_wait; 1046 1047 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1048 struct mutex i_fc_lock; 1049 1050 /* 1051 * i_disksize keeps track of what the inode size is ON DISK, not 1052 * in memory. During truncate, i_size is set to the new size by 1053 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1054 * set i_disksize to 0 until the truncate is actually under way. 1055 * 1056 * The intent is that i_disksize always represents the blocks which 1057 * are used by this file. This allows recovery to restart truncate 1058 * on orphans if we crash during truncate. We actually write i_disksize 1059 * into the on-disk inode when writing inodes out, instead of i_size. 1060 * 1061 * The only time when i_disksize and i_size may be different is when 1062 * a truncate is in progress. The only things which change i_disksize 1063 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1064 */ 1065 loff_t i_disksize; 1066 1067 /* 1068 * i_data_sem is for serialising ext4_truncate() against 1069 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1070 * data tree are chopped off during truncate. We can't do that in 1071 * ext4 because whenever we perform intermediate commits during 1072 * truncate, the inode and all the metadata blocks *must* be in a 1073 * consistent state which allows truncation of the orphans to restart 1074 * during recovery. Hence we must fix the get_block-vs-truncate race 1075 * by other means, so we have i_data_sem. 1076 */ 1077 struct rw_semaphore i_data_sem; 1078 struct inode vfs_inode; 1079 struct jbd2_inode *jinode; 1080 1081 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1082 1083 /* 1084 * File creation time. Its function is same as that of 1085 * struct timespec64 i_{a,c,m}time in the generic inode. 1086 */ 1087 struct timespec64 i_crtime; 1088 1089 /* mballoc */ 1090 atomic_t i_prealloc_active; 1091 struct rb_root i_prealloc_node; 1092 rwlock_t i_prealloc_lock; 1093 1094 /* extents status tree */ 1095 struct ext4_es_tree i_es_tree; 1096 rwlock_t i_es_lock; 1097 struct list_head i_es_list; 1098 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1099 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1100 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1101 extents to shrink. Protected by 1102 i_es_lock */ 1103 1104 /* ialloc */ 1105 ext4_group_t i_last_alloc_group; 1106 1107 /* allocation reservation info for delalloc */ 1108 /* In case of bigalloc, this refer to clusters rather than blocks */ 1109 unsigned int i_reserved_data_blocks; 1110 1111 /* pending cluster reservations for bigalloc file systems */ 1112 struct ext4_pending_tree i_pending_tree; 1113 1114 /* on-disk additional length */ 1115 __u16 i_extra_isize; 1116 1117 /* Indicate the inline data space. */ 1118 u16 i_inline_off; 1119 u16 i_inline_size; 1120 1121 #ifdef CONFIG_QUOTA 1122 /* quota space reservation, managed internally by quota code */ 1123 qsize_t i_reserved_quota; 1124 #endif 1125 1126 /* Lock protecting lists below */ 1127 spinlock_t i_completed_io_lock; 1128 /* 1129 * Completed IOs that need unwritten extents handling and have 1130 * transaction reserved 1131 */ 1132 struct list_head i_rsv_conversion_list; 1133 struct work_struct i_rsv_conversion_work; 1134 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1135 1136 spinlock_t i_block_reservation_lock; 1137 1138 /* 1139 * Transactions that contain inode's metadata needed to complete 1140 * fsync and fdatasync, respectively. 1141 */ 1142 tid_t i_sync_tid; 1143 tid_t i_datasync_tid; 1144 1145 #ifdef CONFIG_QUOTA 1146 struct dquot *i_dquot[MAXQUOTAS]; 1147 #endif 1148 1149 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1150 __u32 i_csum_seed; 1151 1152 kprojid_t i_projid; 1153 }; 1154 1155 /* 1156 * File system states 1157 */ 1158 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1159 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1160 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1161 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1162 1163 /* 1164 * Misc. filesystem flags 1165 */ 1166 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1167 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1168 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1169 1170 /* 1171 * Mount flags set via mount options or defaults 1172 */ 1173 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1174 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1175 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1176 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1177 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1178 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1179 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1180 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1181 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1182 #ifdef CONFIG_FS_DAX 1183 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1184 #else 1185 #define EXT4_MOUNT_DAX_ALWAYS 0 1186 #endif 1187 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1188 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1189 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1190 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1191 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1192 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1193 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1194 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1195 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1196 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1197 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1198 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1199 * enable enforcement for hidden 1200 * quota files */ 1201 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1202 * enforcement for hidden quota 1203 * files */ 1204 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1205 * enforcement */ 1206 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1207 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1208 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1209 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1210 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1211 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1212 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1213 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1214 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1215 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1216 1217 /* 1218 * Mount flags set either automatically (could not be set by mount option) 1219 * based on per file system feature or property or in special cases such as 1220 * distinguishing between explicit mount option definition and default. 1221 */ 1222 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1223 specified delalloc */ 1224 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1225 size of blocksize * 8 1226 blocks */ 1227 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1228 file systems */ 1229 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1230 specified journal checksum */ 1231 1232 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1233 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1234 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1235 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1236 * scanning in mballoc 1237 */ 1238 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1239 1240 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1241 ~EXT4_MOUNT_##opt 1242 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1243 EXT4_MOUNT_##opt 1244 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1245 EXT4_MOUNT_##opt) 1246 1247 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1248 ~EXT4_MOUNT2_##opt 1249 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1250 EXT4_MOUNT2_##opt 1251 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1252 EXT4_MOUNT2_##opt) 1253 1254 #define ext4_test_and_set_bit __test_and_set_bit_le 1255 #define ext4_set_bit __set_bit_le 1256 #define ext4_set_bit_atomic ext2_set_bit_atomic 1257 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1258 #define ext4_clear_bit __clear_bit_le 1259 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1260 #define ext4_test_bit test_bit_le 1261 #define ext4_find_next_zero_bit find_next_zero_bit_le 1262 #define ext4_find_next_bit find_next_bit_le 1263 1264 extern void mb_set_bits(void *bm, int cur, int len); 1265 1266 /* 1267 * Maximal mount counts between two filesystem checks 1268 */ 1269 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1270 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1271 1272 /* 1273 * Behaviour when detecting errors 1274 */ 1275 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1276 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1277 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1278 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1279 1280 /* Metadata checksum algorithm codes */ 1281 #define EXT4_CRC32C_CHKSUM 1 1282 1283 #define EXT4_LABEL_MAX 16 1284 1285 /* 1286 * Structure of the super block 1287 */ 1288 struct ext4_super_block { 1289 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1290 __le32 s_blocks_count_lo; /* Blocks count */ 1291 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1292 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1293 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1294 __le32 s_first_data_block; /* First Data Block */ 1295 __le32 s_log_block_size; /* Block size */ 1296 __le32 s_log_cluster_size; /* Allocation cluster size */ 1297 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1298 __le32 s_clusters_per_group; /* # Clusters per group */ 1299 __le32 s_inodes_per_group; /* # Inodes per group */ 1300 __le32 s_mtime; /* Mount time */ 1301 /*30*/ __le32 s_wtime; /* Write time */ 1302 __le16 s_mnt_count; /* Mount count */ 1303 __le16 s_max_mnt_count; /* Maximal mount count */ 1304 __le16 s_magic; /* Magic signature */ 1305 __le16 s_state; /* File system state */ 1306 __le16 s_errors; /* Behaviour when detecting errors */ 1307 __le16 s_minor_rev_level; /* minor revision level */ 1308 /*40*/ __le32 s_lastcheck; /* time of last check */ 1309 __le32 s_checkinterval; /* max. time between checks */ 1310 __le32 s_creator_os; /* OS */ 1311 __le32 s_rev_level; /* Revision level */ 1312 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1313 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1314 /* 1315 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1316 * 1317 * Note: the difference between the compatible feature set and 1318 * the incompatible feature set is that if there is a bit set 1319 * in the incompatible feature set that the kernel doesn't 1320 * know about, it should refuse to mount the filesystem. 1321 * 1322 * e2fsck's requirements are more strict; if it doesn't know 1323 * about a feature in either the compatible or incompatible 1324 * feature set, it must abort and not try to meddle with 1325 * things it doesn't understand... 1326 */ 1327 __le32 s_first_ino; /* First non-reserved inode */ 1328 __le16 s_inode_size; /* size of inode structure */ 1329 __le16 s_block_group_nr; /* block group # of this superblock */ 1330 __le32 s_feature_compat; /* compatible feature set */ 1331 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1332 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1333 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1334 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1335 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1336 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1337 /* 1338 * Performance hints. Directory preallocation should only 1339 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1340 */ 1341 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1342 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1343 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1344 /* 1345 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1346 */ 1347 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1348 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1349 __le32 s_journal_dev; /* device number of journal file */ 1350 __le32 s_last_orphan; /* start of list of inodes to delete */ 1351 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1352 __u8 s_def_hash_version; /* Default hash version to use */ 1353 __u8 s_jnl_backup_type; 1354 __le16 s_desc_size; /* size of group descriptor */ 1355 /*100*/ __le32 s_default_mount_opts; 1356 __le32 s_first_meta_bg; /* First metablock block group */ 1357 __le32 s_mkfs_time; /* When the filesystem was created */ 1358 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1359 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1360 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1361 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1362 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1363 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1364 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1365 __le32 s_flags; /* Miscellaneous flags */ 1366 __le16 s_raid_stride; /* RAID stride */ 1367 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1368 __le64 s_mmp_block; /* Block for multi-mount protection */ 1369 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1370 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1371 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1372 __u8 s_encryption_level; /* versioning level for encryption */ 1373 __u8 s_reserved_pad; /* Padding to next 32bits */ 1374 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1375 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1376 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1377 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1378 snapshot's future use */ 1379 __le32 s_snapshot_list; /* inode number of the head of the 1380 on-disk snapshot list */ 1381 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1382 __le32 s_error_count; /* number of fs errors */ 1383 __le32 s_first_error_time; /* first time an error happened */ 1384 __le32 s_first_error_ino; /* inode involved in first error */ 1385 __le64 s_first_error_block; /* block involved of first error */ 1386 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1387 __le32 s_first_error_line; /* line number where error happened */ 1388 __le32 s_last_error_time; /* most recent time of an error */ 1389 __le32 s_last_error_ino; /* inode involved in last error */ 1390 __le32 s_last_error_line; /* line number where error happened */ 1391 __le64 s_last_error_block; /* block involved of last error */ 1392 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1393 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1394 __u8 s_mount_opts[64]; 1395 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1396 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1397 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1398 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1399 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1400 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1401 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1402 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1403 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1404 __u8 s_wtime_hi; 1405 __u8 s_mtime_hi; 1406 __u8 s_mkfs_time_hi; 1407 __u8 s_lastcheck_hi; 1408 __u8 s_first_error_time_hi; 1409 __u8 s_last_error_time_hi; 1410 __u8 s_first_error_errcode; 1411 __u8 s_last_error_errcode; 1412 __le16 s_encoding; /* Filename charset encoding */ 1413 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1414 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1415 __le32 s_reserved[94]; /* Padding to the end of the block */ 1416 __le32 s_checksum; /* crc32c(superblock) */ 1417 }; 1418 1419 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1420 1421 #ifdef __KERNEL__ 1422 1423 /* Number of quota types we support */ 1424 #define EXT4_MAXQUOTAS 3 1425 1426 #define EXT4_ENC_UTF8_12_1 1 1427 1428 /* Types of ext4 journal triggers */ 1429 enum ext4_journal_trigger_type { 1430 EXT4_JTR_ORPHAN_FILE, 1431 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1432 }; 1433 1434 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1435 1436 struct ext4_journal_trigger { 1437 struct jbd2_buffer_trigger_type tr_triggers; 1438 struct super_block *sb; 1439 }; 1440 1441 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1442 struct jbd2_buffer_trigger_type *trigger) 1443 { 1444 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1445 } 1446 1447 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1448 1449 /* Structure at the tail of orphan block */ 1450 struct ext4_orphan_block_tail { 1451 __le32 ob_magic; 1452 __le32 ob_checksum; 1453 }; 1454 1455 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1456 { 1457 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1458 sizeof(u32); 1459 } 1460 1461 struct ext4_orphan_block { 1462 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1463 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1464 }; 1465 1466 /* 1467 * Info about orphan file. 1468 */ 1469 struct ext4_orphan_info { 1470 int of_blocks; /* Number of orphan blocks in a file */ 1471 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1472 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1473 * file blocks */ 1474 }; 1475 1476 /* 1477 * fourth extended-fs super-block data in memory 1478 */ 1479 struct ext4_sb_info { 1480 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1481 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1482 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1483 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1484 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1485 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1486 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1487 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1488 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1489 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1490 unsigned long s_overhead; /* # of fs overhead clusters */ 1491 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1492 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1493 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1494 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1495 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1496 struct buffer_head * __rcu *s_group_desc; 1497 unsigned int s_mount_opt; 1498 unsigned int s_mount_opt2; 1499 unsigned long s_mount_flags; 1500 unsigned int s_def_mount_opt; 1501 unsigned int s_def_mount_opt2; 1502 ext4_fsblk_t s_sb_block; 1503 atomic64_t s_resv_clusters; 1504 kuid_t s_resuid; 1505 kgid_t s_resgid; 1506 unsigned short s_mount_state; 1507 unsigned short s_pad; 1508 int s_addr_per_block_bits; 1509 int s_desc_per_block_bits; 1510 int s_inode_size; 1511 int s_first_ino; 1512 unsigned int s_inode_readahead_blks; 1513 unsigned int s_inode_goal; 1514 u32 s_hash_seed[4]; 1515 int s_def_hash_version; 1516 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1517 struct percpu_counter s_freeclusters_counter; 1518 struct percpu_counter s_freeinodes_counter; 1519 struct percpu_counter s_dirs_counter; 1520 struct percpu_counter s_dirtyclusters_counter; 1521 struct percpu_counter s_sra_exceeded_retry_limit; 1522 struct blockgroup_lock *s_blockgroup_lock; 1523 struct proc_dir_entry *s_proc; 1524 struct kobject s_kobj; 1525 struct completion s_kobj_unregister; 1526 struct super_block *s_sb; 1527 struct buffer_head *s_mmp_bh; 1528 1529 /* Journaling */ 1530 struct journal_s *s_journal; 1531 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1532 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1533 struct list_head s_orphan; /* List of orphaned inodes in on disk 1534 list */ 1535 struct ext4_orphan_info s_orphan_info; 1536 unsigned long s_commit_interval; 1537 u32 s_max_batch_time; 1538 u32 s_min_batch_time; 1539 struct block_device *s_journal_bdev; 1540 #ifdef CONFIG_QUOTA 1541 /* Names of quota files with journalled quota */ 1542 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1543 int s_jquota_fmt; /* Format of quota to use */ 1544 #endif 1545 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1546 struct ext4_system_blocks __rcu *s_system_blks; 1547 1548 #ifdef EXTENTS_STATS 1549 /* ext4 extents stats */ 1550 unsigned long s_ext_min; 1551 unsigned long s_ext_max; 1552 unsigned long s_depth_max; 1553 spinlock_t s_ext_stats_lock; 1554 unsigned long s_ext_blocks; 1555 unsigned long s_ext_extents; 1556 #endif 1557 1558 /* for buddy allocator */ 1559 struct ext4_group_info ** __rcu *s_group_info; 1560 struct inode *s_buddy_cache; 1561 spinlock_t s_md_lock; 1562 unsigned short *s_mb_offsets; 1563 unsigned int *s_mb_maxs; 1564 unsigned int s_group_info_size; 1565 unsigned int s_mb_free_pending; 1566 struct list_head s_freed_data_list; /* List of blocks to be freed 1567 after commit completed */ 1568 struct list_head s_discard_list; 1569 struct work_struct s_discard_work; 1570 atomic_t s_retry_alloc_pending; 1571 struct list_head *s_mb_avg_fragment_size; 1572 rwlock_t *s_mb_avg_fragment_size_locks; 1573 struct list_head *s_mb_largest_free_orders; 1574 rwlock_t *s_mb_largest_free_orders_locks; 1575 1576 /* tunables */ 1577 unsigned long s_stripe; 1578 unsigned int s_mb_max_linear_groups; 1579 unsigned int s_mb_stream_request; 1580 unsigned int s_mb_max_to_scan; 1581 unsigned int s_mb_min_to_scan; 1582 unsigned int s_mb_stats; 1583 unsigned int s_mb_order2_reqs; 1584 unsigned int s_mb_group_prealloc; 1585 unsigned int s_max_dir_size_kb; 1586 /* where last allocation was done - for stream allocation */ 1587 unsigned long s_mb_last_group; 1588 unsigned long s_mb_last_start; 1589 unsigned int s_mb_prefetch; 1590 unsigned int s_mb_prefetch_limit; 1591 unsigned int s_mb_best_avail_max_trim_order; 1592 1593 /* stats for buddy allocator */ 1594 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1595 atomic_t s_bal_success; /* we found long enough chunks */ 1596 atomic_t s_bal_allocated; /* in blocks */ 1597 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1598 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1599 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1600 atomic_t s_bal_goals; /* goal hits */ 1601 atomic_t s_bal_len_goals; /* len goal hits */ 1602 atomic_t s_bal_breaks; /* too long searches */ 1603 atomic_t s_bal_2orders; /* 2^order hits */ 1604 atomic_t s_bal_p2_aligned_bad_suggestions; 1605 atomic_t s_bal_goal_fast_bad_suggestions; 1606 atomic_t s_bal_best_avail_bad_suggestions; 1607 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1608 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1609 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1610 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1611 atomic64_t s_mb_generation_time; 1612 atomic_t s_mb_lost_chunks; 1613 atomic_t s_mb_preallocated; 1614 atomic_t s_mb_discarded; 1615 atomic_t s_lock_busy; 1616 1617 /* locality groups */ 1618 struct ext4_locality_group __percpu *s_locality_groups; 1619 1620 /* for write statistics */ 1621 unsigned long s_sectors_written_start; 1622 u64 s_kbytes_written; 1623 1624 /* the size of zero-out chunk */ 1625 unsigned int s_extent_max_zeroout_kb; 1626 1627 unsigned int s_log_groups_per_flex; 1628 struct flex_groups * __rcu *s_flex_groups; 1629 ext4_group_t s_flex_groups_allocated; 1630 1631 /* workqueue for reserved extent conversions (buffered io) */ 1632 struct workqueue_struct *rsv_conversion_wq; 1633 1634 /* timer for periodic error stats printing */ 1635 struct timer_list s_err_report; 1636 1637 /* Lazy inode table initialization info */ 1638 struct ext4_li_request *s_li_request; 1639 /* Wait multiplier for lazy initialization thread */ 1640 unsigned int s_li_wait_mult; 1641 1642 /* Kernel thread for multiple mount protection */ 1643 struct task_struct *s_mmp_tsk; 1644 1645 /* record the last minlen when FITRIM is called. */ 1646 unsigned long s_last_trim_minblks; 1647 1648 /* Reference to checksum algorithm driver via cryptoapi */ 1649 struct crypto_shash *s_chksum_driver; 1650 1651 /* Precomputed FS UUID checksum for seeding other checksums */ 1652 __u32 s_csum_seed; 1653 1654 /* Reclaim extents from extent status tree */ 1655 struct shrinker s_es_shrinker; 1656 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1657 long s_es_nr_inode; 1658 struct ext4_es_stats s_es_stats; 1659 struct mb_cache *s_ea_block_cache; 1660 struct mb_cache *s_ea_inode_cache; 1661 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1662 1663 /* Journal triggers for checksum computation */ 1664 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1665 1666 /* Ratelimit ext4 messages. */ 1667 struct ratelimit_state s_err_ratelimit_state; 1668 struct ratelimit_state s_warning_ratelimit_state; 1669 struct ratelimit_state s_msg_ratelimit_state; 1670 atomic_t s_warning_count; 1671 atomic_t s_msg_count; 1672 1673 /* Encryption policy for '-o test_dummy_encryption' */ 1674 struct fscrypt_dummy_policy s_dummy_enc_policy; 1675 1676 /* 1677 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1678 * or EXTENTS flag. 1679 */ 1680 struct percpu_rw_semaphore s_writepages_rwsem; 1681 struct dax_device *s_daxdev; 1682 u64 s_dax_part_off; 1683 #ifdef CONFIG_EXT4_DEBUG 1684 unsigned long s_simulate_fail; 1685 #endif 1686 /* Record the errseq of the backing block device */ 1687 errseq_t s_bdev_wb_err; 1688 spinlock_t s_bdev_wb_lock; 1689 1690 /* Information about errors that happened during this mount */ 1691 spinlock_t s_error_lock; 1692 int s_add_error_count; 1693 int s_first_error_code; 1694 __u32 s_first_error_line; 1695 __u32 s_first_error_ino; 1696 __u64 s_first_error_block; 1697 const char *s_first_error_func; 1698 time64_t s_first_error_time; 1699 int s_last_error_code; 1700 __u32 s_last_error_line; 1701 __u32 s_last_error_ino; 1702 __u64 s_last_error_block; 1703 const char *s_last_error_func; 1704 time64_t s_last_error_time; 1705 /* 1706 * If we are in a context where we cannot update error information in 1707 * the on-disk superblock, we queue this work to do it. 1708 */ 1709 struct work_struct s_error_work; 1710 1711 /* Ext4 fast commit sub transaction ID */ 1712 atomic_t s_fc_subtid; 1713 1714 /* 1715 * After commit starts, the main queue gets locked, and the further 1716 * updates get added in the staging queue. 1717 */ 1718 #define FC_Q_MAIN 0 1719 #define FC_Q_STAGING 1 1720 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1721 * that have data changes in them. 1722 */ 1723 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1724 unsigned int s_fc_bytes; 1725 /* 1726 * Main fast commit lock. This lock protects accesses to the 1727 * following fields: 1728 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1729 */ 1730 spinlock_t s_fc_lock; 1731 struct buffer_head *s_fc_bh; 1732 struct ext4_fc_stats s_fc_stats; 1733 tid_t s_fc_ineligible_tid; 1734 #ifdef CONFIG_EXT4_DEBUG 1735 int s_fc_debug_max_replay; 1736 #endif 1737 struct ext4_fc_replay_state s_fc_replay_state; 1738 }; 1739 1740 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1741 { 1742 return sb->s_fs_info; 1743 } 1744 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1745 { 1746 return container_of(inode, struct ext4_inode_info, vfs_inode); 1747 } 1748 1749 static inline int ext4_writepages_down_read(struct super_block *sb) 1750 { 1751 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1752 return memalloc_nofs_save(); 1753 } 1754 1755 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1756 { 1757 memalloc_nofs_restore(ctx); 1758 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1759 } 1760 1761 static inline int ext4_writepages_down_write(struct super_block *sb) 1762 { 1763 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1764 return memalloc_nofs_save(); 1765 } 1766 1767 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1768 { 1769 memalloc_nofs_restore(ctx); 1770 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1771 } 1772 1773 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1774 { 1775 return ino == EXT4_ROOT_INO || 1776 (ino >= EXT4_FIRST_INO(sb) && 1777 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1778 } 1779 1780 /* 1781 * Returns: sbi->field[index] 1782 * Used to access an array element from the following sbi fields which require 1783 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1784 * - s_group_desc 1785 * - s_group_info 1786 * - s_flex_group 1787 */ 1788 #define sbi_array_rcu_deref(sbi, field, index) \ 1789 ({ \ 1790 typeof(*((sbi)->field)) _v; \ 1791 rcu_read_lock(); \ 1792 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1793 rcu_read_unlock(); \ 1794 _v; \ 1795 }) 1796 1797 /* 1798 * run-time mount flags 1799 */ 1800 enum { 1801 EXT4_MF_MNTDIR_SAMPLED, 1802 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1803 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1804 }; 1805 1806 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1807 { 1808 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1809 } 1810 1811 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1812 { 1813 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1814 } 1815 1816 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1817 { 1818 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1819 } 1820 1821 1822 /* 1823 * Simulate_fail codes 1824 */ 1825 #define EXT4_SIM_BBITMAP_EIO 1 1826 #define EXT4_SIM_BBITMAP_CRC 2 1827 #define EXT4_SIM_IBITMAP_EIO 3 1828 #define EXT4_SIM_IBITMAP_CRC 4 1829 #define EXT4_SIM_INODE_EIO 5 1830 #define EXT4_SIM_INODE_CRC 6 1831 #define EXT4_SIM_DIRBLOCK_EIO 7 1832 #define EXT4_SIM_DIRBLOCK_CRC 8 1833 1834 static inline bool ext4_simulate_fail(struct super_block *sb, 1835 unsigned long code) 1836 { 1837 #ifdef CONFIG_EXT4_DEBUG 1838 struct ext4_sb_info *sbi = EXT4_SB(sb); 1839 1840 if (unlikely(sbi->s_simulate_fail == code)) { 1841 sbi->s_simulate_fail = 0; 1842 return true; 1843 } 1844 #endif 1845 return false; 1846 } 1847 1848 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1849 struct buffer_head *bh, 1850 unsigned long code) 1851 { 1852 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1853 clear_buffer_uptodate(bh); 1854 } 1855 1856 /* 1857 * Error number codes for s_{first,last}_error_errno 1858 * 1859 * Linux errno numbers are architecture specific, so we need to translate 1860 * them into something which is architecture independent. We don't define 1861 * codes for all errno's; just the ones which are most likely to be the cause 1862 * of an ext4_error() call. 1863 */ 1864 #define EXT4_ERR_UNKNOWN 1 1865 #define EXT4_ERR_EIO 2 1866 #define EXT4_ERR_ENOMEM 3 1867 #define EXT4_ERR_EFSBADCRC 4 1868 #define EXT4_ERR_EFSCORRUPTED 5 1869 #define EXT4_ERR_ENOSPC 6 1870 #define EXT4_ERR_ENOKEY 7 1871 #define EXT4_ERR_EROFS 8 1872 #define EXT4_ERR_EFBIG 9 1873 #define EXT4_ERR_EEXIST 10 1874 #define EXT4_ERR_ERANGE 11 1875 #define EXT4_ERR_EOVERFLOW 12 1876 #define EXT4_ERR_EBUSY 13 1877 #define EXT4_ERR_ENOTDIR 14 1878 #define EXT4_ERR_ENOTEMPTY 15 1879 #define EXT4_ERR_ESHUTDOWN 16 1880 #define EXT4_ERR_EFAULT 17 1881 1882 /* 1883 * Inode dynamic state flags 1884 */ 1885 enum { 1886 EXT4_STATE_NEW, /* inode is newly created */ 1887 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1888 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1889 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1890 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1891 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1892 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1893 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1894 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1895 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1896 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1897 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1898 }; 1899 1900 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1901 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1902 { \ 1903 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1904 } \ 1905 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1906 { \ 1907 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1908 } \ 1909 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1910 { \ 1911 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1912 } 1913 1914 /* Add these declarations here only so that these functions can be 1915 * found by name. Otherwise, they are very hard to locate. */ 1916 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1917 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1918 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1919 EXT4_INODE_BIT_FNS(flag, flags, 0) 1920 1921 /* Add these declarations here only so that these functions can be 1922 * found by name. Otherwise, they are very hard to locate. */ 1923 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1924 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1925 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1926 #if (BITS_PER_LONG < 64) 1927 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1928 1929 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1930 { 1931 (ei)->i_state_flags = 0; 1932 } 1933 #else 1934 EXT4_INODE_BIT_FNS(state, flags, 32) 1935 1936 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1937 { 1938 /* We depend on the fact that callers will set i_flags */ 1939 } 1940 #endif 1941 #else 1942 /* Assume that user mode programs are passing in an ext4fs superblock, not 1943 * a kernel struct super_block. This will allow us to call the feature-test 1944 * macros from user land. */ 1945 #define EXT4_SB(sb) (sb) 1946 #endif 1947 1948 static inline bool ext4_verity_in_progress(struct inode *inode) 1949 { 1950 return IS_ENABLED(CONFIG_FS_VERITY) && 1951 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1952 } 1953 1954 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1955 1956 /* 1957 * Codes for operating systems 1958 */ 1959 #define EXT4_OS_LINUX 0 1960 #define EXT4_OS_HURD 1 1961 #define EXT4_OS_MASIX 2 1962 #define EXT4_OS_FREEBSD 3 1963 #define EXT4_OS_LITES 4 1964 1965 /* 1966 * Revision levels 1967 */ 1968 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1969 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1970 1971 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1972 1973 #define EXT4_GOOD_OLD_INODE_SIZE 128 1974 1975 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1976 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1977 #define EXT4_TIMESTAMP_MIN S32_MIN 1978 1979 /* 1980 * Feature set definitions 1981 */ 1982 1983 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1984 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1985 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1986 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1987 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1988 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1989 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1990 /* 1991 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1992 * incompatible only if fast commit blocks are present in the FS. Since we 1993 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1994 * incompatible. We also have a JBD2 incompat feature, which gets set when 1995 * there are fast commit blocks present in the journal. 1996 */ 1997 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1998 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1999 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2000 2001 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2002 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2003 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2004 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2005 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2006 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2007 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2008 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2009 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2010 /* 2011 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2012 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2013 * all other data structures' checksums. However, the METADATA_CSUM and 2014 * GDT_CSUM bits are mutually exclusive. 2015 */ 2016 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2017 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2018 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2019 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2020 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2021 non-empty */ 2022 2023 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2024 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2025 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2026 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2027 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2028 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2029 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2030 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2031 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2032 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2033 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2034 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2035 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2036 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2037 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2038 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2039 2040 extern void ext4_update_dynamic_rev(struct super_block *sb); 2041 2042 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2043 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2044 { \ 2045 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2046 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2047 } \ 2048 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2049 { \ 2050 ext4_update_dynamic_rev(sb); \ 2051 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2052 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2053 } \ 2054 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2055 { \ 2056 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2057 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2058 } 2059 2060 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2061 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2062 { \ 2063 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2064 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2065 } \ 2066 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2067 { \ 2068 ext4_update_dynamic_rev(sb); \ 2069 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2070 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2071 } \ 2072 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2073 { \ 2074 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2075 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2076 } 2077 2078 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2079 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2080 { \ 2081 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2082 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2083 } \ 2084 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2085 { \ 2086 ext4_update_dynamic_rev(sb); \ 2087 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2088 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2089 } \ 2090 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2091 { \ 2092 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2093 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2094 } 2095 2096 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2097 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2098 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2099 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2100 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2101 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2102 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2103 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2104 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2105 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2106 2107 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2108 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2109 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2110 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2111 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2112 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2113 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2114 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2115 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2116 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2117 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2118 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2119 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2120 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2121 2122 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2123 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2124 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2125 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2126 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2127 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2128 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2129 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2130 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2131 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2132 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2133 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2134 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2135 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2136 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2137 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2138 2139 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2140 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2141 EXT4_FEATURE_INCOMPAT_META_BG) 2142 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2143 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2144 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2145 2146 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2147 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2148 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2149 EXT4_FEATURE_INCOMPAT_META_BG) 2150 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2151 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2152 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2153 2154 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2155 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2156 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2157 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2158 EXT4_FEATURE_INCOMPAT_META_BG| \ 2159 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2160 EXT4_FEATURE_INCOMPAT_64BIT| \ 2161 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2162 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2163 EXT4_FEATURE_INCOMPAT_MMP | \ 2164 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2165 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2166 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2167 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2168 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2169 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2170 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2171 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2172 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2173 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2174 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2175 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2176 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2177 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2178 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2179 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2180 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2181 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2182 2183 #define EXTN_FEATURE_FUNCS(ver) \ 2184 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2185 { \ 2186 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2187 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2188 } \ 2189 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2190 { \ 2191 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2192 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2193 } \ 2194 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2195 { \ 2196 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2197 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2198 } 2199 2200 EXTN_FEATURE_FUNCS(2) 2201 EXTN_FEATURE_FUNCS(3) 2202 EXTN_FEATURE_FUNCS(4) 2203 2204 static inline bool ext4_has_compat_features(struct super_block *sb) 2205 { 2206 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2207 } 2208 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2209 { 2210 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2211 } 2212 static inline bool ext4_has_incompat_features(struct super_block *sb) 2213 { 2214 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2215 } 2216 2217 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2218 2219 /* 2220 * Superblock flags 2221 */ 2222 #define EXT4_FLAGS_RESIZING 0 2223 #define EXT4_FLAGS_SHUTDOWN 1 2224 #define EXT4_FLAGS_BDEV_IS_DAX 2 2225 2226 static inline int ext4_forced_shutdown(struct super_block *sb) 2227 { 2228 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2229 } 2230 2231 /* 2232 * Default values for user and/or group using reserved blocks 2233 */ 2234 #define EXT4_DEF_RESUID 0 2235 #define EXT4_DEF_RESGID 0 2236 2237 /* 2238 * Default project ID 2239 */ 2240 #define EXT4_DEF_PROJID 0 2241 2242 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2243 2244 /* 2245 * Default mount options 2246 */ 2247 #define EXT4_DEFM_DEBUG 0x0001 2248 #define EXT4_DEFM_BSDGROUPS 0x0002 2249 #define EXT4_DEFM_XATTR_USER 0x0004 2250 #define EXT4_DEFM_ACL 0x0008 2251 #define EXT4_DEFM_UID16 0x0010 2252 #define EXT4_DEFM_JMODE 0x0060 2253 #define EXT4_DEFM_JMODE_DATA 0x0020 2254 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2255 #define EXT4_DEFM_JMODE_WBACK 0x0060 2256 #define EXT4_DEFM_NOBARRIER 0x0100 2257 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2258 #define EXT4_DEFM_DISCARD 0x0400 2259 #define EXT4_DEFM_NODELALLOC 0x0800 2260 2261 /* 2262 * Default journal batch times 2263 */ 2264 #define EXT4_DEF_MIN_BATCH_TIME 0 2265 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2266 2267 /* 2268 * Minimum number of groups in a flexgroup before we separate out 2269 * directories into the first block group of a flexgroup 2270 */ 2271 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2272 2273 /* 2274 * Structure of a directory entry 2275 */ 2276 #define EXT4_NAME_LEN 255 2277 /* 2278 * Base length of the ext4 directory entry excluding the name length 2279 */ 2280 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2281 2282 struct ext4_dir_entry { 2283 __le32 inode; /* Inode number */ 2284 __le16 rec_len; /* Directory entry length */ 2285 __le16 name_len; /* Name length */ 2286 char name[EXT4_NAME_LEN]; /* File name */ 2287 }; 2288 2289 2290 /* 2291 * Encrypted Casefolded entries require saving the hash on disk. This structure 2292 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2293 * boundary. 2294 */ 2295 struct ext4_dir_entry_hash { 2296 __le32 hash; 2297 __le32 minor_hash; 2298 }; 2299 2300 /* 2301 * The new version of the directory entry. Since EXT4 structures are 2302 * stored in intel byte order, and the name_len field could never be 2303 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2304 * file_type field. 2305 */ 2306 struct ext4_dir_entry_2 { 2307 __le32 inode; /* Inode number */ 2308 __le16 rec_len; /* Directory entry length */ 2309 __u8 name_len; /* Name length */ 2310 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2311 char name[EXT4_NAME_LEN]; /* File name */ 2312 }; 2313 2314 /* 2315 * Access the hashes at the end of ext4_dir_entry_2 2316 */ 2317 #define EXT4_DIRENT_HASHES(entry) \ 2318 ((struct ext4_dir_entry_hash *) \ 2319 (((void *)(entry)) + \ 2320 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2321 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2322 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2323 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2324 2325 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2326 { 2327 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2328 } 2329 2330 /* 2331 * This is a bogus directory entry at the end of each leaf block that 2332 * records checksums. 2333 */ 2334 struct ext4_dir_entry_tail { 2335 __le32 det_reserved_zero1; /* Pretend to be unused */ 2336 __le16 det_rec_len; /* 12 */ 2337 __u8 det_reserved_zero2; /* Zero name length */ 2338 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2339 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2340 }; 2341 2342 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2343 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2344 ((blocksize) - \ 2345 sizeof(struct ext4_dir_entry_tail)))) 2346 2347 /* 2348 * Ext4 directory file types. Only the low 3 bits are used. The 2349 * other bits are reserved for now. 2350 */ 2351 #define EXT4_FT_UNKNOWN 0 2352 #define EXT4_FT_REG_FILE 1 2353 #define EXT4_FT_DIR 2 2354 #define EXT4_FT_CHRDEV 3 2355 #define EXT4_FT_BLKDEV 4 2356 #define EXT4_FT_FIFO 5 2357 #define EXT4_FT_SOCK 6 2358 #define EXT4_FT_SYMLINK 7 2359 2360 #define EXT4_FT_MAX 8 2361 2362 #define EXT4_FT_DIR_CSUM 0xDE 2363 2364 /* 2365 * EXT4_DIR_PAD defines the directory entries boundaries 2366 * 2367 * NOTE: It must be a multiple of 4 2368 */ 2369 #define EXT4_DIR_PAD 4 2370 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2371 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2372 2373 /* 2374 * The rec_len is dependent on the type of directory. Directories that are 2375 * casefolded and encrypted need to store the hash as well, so we add room for 2376 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2377 * pass NULL for dir, as those entries do not use the extra fields. 2378 */ 2379 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2380 const struct inode *dir) 2381 { 2382 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2383 2384 if (dir && ext4_hash_in_dirent(dir)) 2385 rec_len += sizeof(struct ext4_dir_entry_hash); 2386 return (rec_len & ~EXT4_DIR_ROUND); 2387 } 2388 2389 /* 2390 * If we ever get support for fs block sizes > page_size, we'll need 2391 * to remove the #if statements in the next two functions... 2392 */ 2393 static inline unsigned int 2394 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2395 { 2396 unsigned len = le16_to_cpu(dlen); 2397 2398 #if (PAGE_SIZE >= 65536) 2399 if (len == EXT4_MAX_REC_LEN || len == 0) 2400 return blocksize; 2401 return (len & 65532) | ((len & 3) << 16); 2402 #else 2403 return len; 2404 #endif 2405 } 2406 2407 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2408 { 2409 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2410 #if (PAGE_SIZE >= 65536) 2411 if (len < 65536) 2412 return cpu_to_le16(len); 2413 if (len == blocksize) { 2414 if (blocksize == 65536) 2415 return cpu_to_le16(EXT4_MAX_REC_LEN); 2416 else 2417 return cpu_to_le16(0); 2418 } 2419 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2420 #else 2421 return cpu_to_le16(len); 2422 #endif 2423 } 2424 2425 /* 2426 * Hash Tree Directory indexing 2427 * (c) Daniel Phillips, 2001 2428 */ 2429 2430 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2431 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2432 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2433 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2434 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2435 2436 /* Legal values for the dx_root hash_version field: */ 2437 2438 #define DX_HASH_LEGACY 0 2439 #define DX_HASH_HALF_MD4 1 2440 #define DX_HASH_TEA 2 2441 #define DX_HASH_LEGACY_UNSIGNED 3 2442 #define DX_HASH_HALF_MD4_UNSIGNED 4 2443 #define DX_HASH_TEA_UNSIGNED 5 2444 #define DX_HASH_SIPHASH 6 2445 2446 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2447 const void *address, unsigned int length) 2448 { 2449 struct { 2450 struct shash_desc shash; 2451 char ctx[4]; 2452 } desc; 2453 2454 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2455 2456 desc.shash.tfm = sbi->s_chksum_driver; 2457 *(u32 *)desc.ctx = crc; 2458 2459 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2460 2461 return *(u32 *)desc.ctx; 2462 } 2463 2464 #ifdef __KERNEL__ 2465 2466 /* hash info structure used by the directory hash */ 2467 struct dx_hash_info 2468 { 2469 u32 hash; 2470 u32 minor_hash; 2471 int hash_version; 2472 u32 *seed; 2473 }; 2474 2475 2476 /* 32 and 64 bit signed EOF for dx directories */ 2477 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2478 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2479 2480 2481 /* 2482 * Control parameters used by ext4_htree_next_block 2483 */ 2484 #define HASH_NB_ALWAYS 1 2485 2486 struct ext4_filename { 2487 const struct qstr *usr_fname; 2488 struct fscrypt_str disk_name; 2489 struct dx_hash_info hinfo; 2490 #ifdef CONFIG_FS_ENCRYPTION 2491 struct fscrypt_str crypto_buf; 2492 #endif 2493 #if IS_ENABLED(CONFIG_UNICODE) 2494 struct fscrypt_str cf_name; 2495 #endif 2496 }; 2497 2498 #define fname_name(p) ((p)->disk_name.name) 2499 #define fname_usr_name(p) ((p)->usr_fname->name) 2500 #define fname_len(p) ((p)->disk_name.len) 2501 2502 /* 2503 * Describe an inode's exact location on disk and in memory 2504 */ 2505 struct ext4_iloc 2506 { 2507 struct buffer_head *bh; 2508 unsigned long offset; 2509 ext4_group_t block_group; 2510 }; 2511 2512 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2513 { 2514 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2515 } 2516 2517 static inline bool ext4_is_quota_file(struct inode *inode) 2518 { 2519 return IS_NOQUOTA(inode) && 2520 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2521 } 2522 2523 /* 2524 * This structure is stuffed into the struct file's private_data field 2525 * for directories. It is where we put information so that we can do 2526 * readdir operations in hash tree order. 2527 */ 2528 struct dir_private_info { 2529 struct rb_root root; 2530 struct rb_node *curr_node; 2531 struct fname *extra_fname; 2532 loff_t last_pos; 2533 __u32 curr_hash; 2534 __u32 curr_minor_hash; 2535 __u32 next_hash; 2536 }; 2537 2538 /* calculate the first block number of the group */ 2539 static inline ext4_fsblk_t 2540 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2541 { 2542 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2543 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2544 } 2545 2546 /* 2547 * Special error return code only used by dx_probe() and its callers. 2548 */ 2549 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2550 2551 /* htree levels for ext4 */ 2552 #define EXT4_HTREE_LEVEL_COMPAT 2 2553 #define EXT4_HTREE_LEVEL 3 2554 2555 static inline int ext4_dir_htree_level(struct super_block *sb) 2556 { 2557 return ext4_has_feature_largedir(sb) ? 2558 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2559 } 2560 2561 /* 2562 * Timeout and state flag for lazy initialization inode thread. 2563 */ 2564 #define EXT4_DEF_LI_WAIT_MULT 10 2565 #define EXT4_DEF_LI_MAX_START_DELAY 5 2566 #define EXT4_LAZYINIT_QUIT 0x0001 2567 #define EXT4_LAZYINIT_RUNNING 0x0002 2568 2569 /* 2570 * Lazy inode table initialization info 2571 */ 2572 struct ext4_lazy_init { 2573 unsigned long li_state; 2574 struct list_head li_request_list; 2575 struct mutex li_list_mtx; 2576 }; 2577 2578 enum ext4_li_mode { 2579 EXT4_LI_MODE_PREFETCH_BBITMAP, 2580 EXT4_LI_MODE_ITABLE, 2581 }; 2582 2583 struct ext4_li_request { 2584 struct super_block *lr_super; 2585 enum ext4_li_mode lr_mode; 2586 ext4_group_t lr_first_not_zeroed; 2587 ext4_group_t lr_next_group; 2588 struct list_head lr_request; 2589 unsigned long lr_next_sched; 2590 unsigned long lr_timeout; 2591 }; 2592 2593 struct ext4_features { 2594 struct kobject f_kobj; 2595 struct completion f_kobj_unregister; 2596 }; 2597 2598 /* 2599 * This structure will be used for multiple mount protection. It will be 2600 * written into the block number saved in the s_mmp_block field in the 2601 * superblock. Programs that check MMP should assume that if 2602 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2603 * to use the filesystem, regardless of how old the timestamp is. 2604 */ 2605 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2606 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2607 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2608 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2609 2610 struct mmp_struct { 2611 __le32 mmp_magic; /* Magic number for MMP */ 2612 __le32 mmp_seq; /* Sequence no. updated periodically */ 2613 2614 /* 2615 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2616 * purposes and do not affect the correctness of the algorithm 2617 */ 2618 __le64 mmp_time; /* Time last updated */ 2619 char mmp_nodename[64]; /* Node which last updated MMP block */ 2620 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2621 2622 /* 2623 * mmp_check_interval is used to verify if the MMP block has been 2624 * updated on the block device. The value is updated based on the 2625 * maximum time to write the MMP block during an update cycle. 2626 */ 2627 __le16 mmp_check_interval; 2628 2629 __le16 mmp_pad1; 2630 __le32 mmp_pad2[226]; 2631 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2632 }; 2633 2634 /* arguments passed to the mmp thread */ 2635 struct mmpd_data { 2636 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2637 struct super_block *sb; /* super block of the fs */ 2638 }; 2639 2640 /* 2641 * Check interval multiplier 2642 * The MMP block is written every update interval and initially checked every 2643 * update interval x the multiplier (the value is then adapted based on the 2644 * write latency). The reason is that writes can be delayed under load and we 2645 * don't want readers to incorrectly assume that the filesystem is no longer 2646 * in use. 2647 */ 2648 #define EXT4_MMP_CHECK_MULT 2UL 2649 2650 /* 2651 * Minimum interval for MMP checking in seconds. 2652 */ 2653 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2654 2655 /* 2656 * Maximum interval for MMP checking in seconds. 2657 */ 2658 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2659 2660 /* 2661 * Function prototypes 2662 */ 2663 2664 /* 2665 * Ok, these declarations are also in <linux/kernel.h> but none of the 2666 * ext4 source programs needs to include it so they are duplicated here. 2667 */ 2668 # define NORET_TYPE /**/ 2669 # define ATTRIB_NORET __attribute__((noreturn)) 2670 # define NORET_AND noreturn, 2671 2672 /* bitmap.c */ 2673 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2674 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2675 struct ext4_group_desc *gdp, 2676 struct buffer_head *bh, int sz); 2677 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2678 struct ext4_group_desc *gdp, 2679 struct buffer_head *bh, int sz); 2680 void ext4_block_bitmap_csum_set(struct super_block *sb, 2681 struct ext4_group_desc *gdp, 2682 struct buffer_head *bh); 2683 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2684 struct ext4_group_desc *gdp, 2685 struct buffer_head *bh); 2686 2687 /* balloc.c */ 2688 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2689 ext4_fsblk_t blocknr, 2690 ext4_group_t *blockgrpp, 2691 ext4_grpblk_t *offsetp); 2692 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2693 ext4_fsblk_t block); 2694 2695 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2696 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2697 ext4_group_t group); 2698 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2699 ext4_fsblk_t goal, 2700 unsigned int flags, 2701 unsigned long *count, 2702 int *errp); 2703 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2704 s64 nclusters, unsigned int flags); 2705 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2706 extern void ext4_check_blocks_bitmap(struct super_block *); 2707 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2708 ext4_group_t block_group, 2709 struct buffer_head ** bh); 2710 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2711 ext4_group_t group); 2712 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2713 2714 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2715 ext4_group_t block_group, 2716 bool ignore_locked); 2717 extern int ext4_wait_block_bitmap(struct super_block *sb, 2718 ext4_group_t block_group, 2719 struct buffer_head *bh); 2720 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2721 ext4_group_t block_group); 2722 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2723 ext4_group_t block_group, 2724 struct ext4_group_desc *gdp); 2725 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2726 2727 #if IS_ENABLED(CONFIG_UNICODE) 2728 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2729 const struct qstr *iname, 2730 struct ext4_filename *fname); 2731 #endif 2732 2733 /* ext4 encryption related stuff goes here crypto.c */ 2734 #ifdef CONFIG_FS_ENCRYPTION 2735 extern const struct fscrypt_operations ext4_cryptops; 2736 2737 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2738 int lookup, struct ext4_filename *fname); 2739 2740 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2741 struct ext4_filename *fname); 2742 2743 void ext4_fname_free_filename(struct ext4_filename *fname); 2744 2745 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2746 2747 #else /* !CONFIG_FS_ENCRYPTION */ 2748 static inline int ext4_fname_setup_filename(struct inode *dir, 2749 const struct qstr *iname, 2750 int lookup, 2751 struct ext4_filename *fname) 2752 { 2753 int err = 0; 2754 fname->usr_fname = iname; 2755 fname->disk_name.name = (unsigned char *) iname->name; 2756 fname->disk_name.len = iname->len; 2757 2758 #if IS_ENABLED(CONFIG_UNICODE) 2759 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2760 #endif 2761 2762 return err; 2763 } 2764 2765 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2766 struct dentry *dentry, 2767 struct ext4_filename *fname) 2768 { 2769 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2770 } 2771 2772 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2773 { 2774 #if IS_ENABLED(CONFIG_UNICODE) 2775 kfree(fname->cf_name.name); 2776 fname->cf_name.name = NULL; 2777 #endif 2778 } 2779 2780 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2781 void __user *arg) 2782 { 2783 return -EOPNOTSUPP; 2784 } 2785 #endif /* !CONFIG_FS_ENCRYPTION */ 2786 2787 /* dir.c */ 2788 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2789 struct file *, 2790 struct ext4_dir_entry_2 *, 2791 struct buffer_head *, char *, int, 2792 unsigned int); 2793 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2794 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2795 (de), (bh), (buf), (size), (offset))) 2796 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2797 __u32 minor_hash, 2798 struct ext4_dir_entry_2 *dirent, 2799 struct fscrypt_str *ent_name); 2800 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2801 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2802 struct buffer_head *bh, 2803 void *buf, int buf_size, 2804 struct ext4_filename *fname, 2805 struct ext4_dir_entry_2 **dest_de); 2806 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2807 struct ext4_dir_entry_2 *de, 2808 int buf_size, 2809 struct ext4_filename *fname); 2810 static inline void ext4_update_dx_flag(struct inode *inode) 2811 { 2812 if (!ext4_has_feature_dir_index(inode->i_sb) && 2813 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2814 /* ext4_iget() should have caught this... */ 2815 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2816 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2817 } 2818 } 2819 static const unsigned char ext4_filetype_table[] = { 2820 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2821 }; 2822 2823 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2824 { 2825 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2826 return DT_UNKNOWN; 2827 2828 return ext4_filetype_table[filetype]; 2829 } 2830 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2831 void *buf, int buf_size); 2832 2833 /* fsync.c */ 2834 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2835 2836 /* hash.c */ 2837 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2838 struct dx_hash_info *hinfo); 2839 2840 /* ialloc.c */ 2841 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2842 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2843 struct inode *, umode_t, 2844 const struct qstr *qstr, __u32 goal, 2845 uid_t *owner, __u32 i_flags, 2846 int handle_type, unsigned int line_no, 2847 int nblocks); 2848 2849 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2850 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2851 (goal), (owner), i_flags, 0, 0, 0) 2852 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2853 type, nblocks) \ 2854 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2855 0, (type), __LINE__, (nblocks)) 2856 2857 2858 extern void ext4_free_inode(handle_t *, struct inode *); 2859 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2860 extern unsigned long ext4_count_free_inodes(struct super_block *); 2861 extern unsigned long ext4_count_dirs(struct super_block *); 2862 extern void ext4_check_inodes_bitmap(struct super_block *); 2863 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2864 extern int ext4_init_inode_table(struct super_block *sb, 2865 ext4_group_t group, int barrier); 2866 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2867 2868 /* fast_commit.c */ 2869 int ext4_fc_info_show(struct seq_file *seq, void *v); 2870 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2871 void ext4_fc_init_inode(struct inode *inode); 2872 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2873 ext4_lblk_t end); 2874 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2875 struct dentry *dentry); 2876 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2877 struct dentry *dentry); 2878 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2879 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2880 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2881 struct dentry *dentry); 2882 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2883 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2884 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2885 void ext4_fc_start_update(struct inode *inode); 2886 void ext4_fc_stop_update(struct inode *inode); 2887 void ext4_fc_del(struct inode *inode); 2888 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2889 void ext4_fc_replay_cleanup(struct super_block *sb); 2890 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2891 int __init ext4_fc_init_dentry_cache(void); 2892 void ext4_fc_destroy_dentry_cache(void); 2893 int ext4_fc_record_regions(struct super_block *sb, int ino, 2894 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2895 int len, int replay); 2896 2897 /* mballoc.c */ 2898 extern const struct seq_operations ext4_mb_seq_groups_ops; 2899 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2900 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2901 extern int ext4_mb_init(struct super_block *); 2902 extern int ext4_mb_release(struct super_block *); 2903 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2904 struct ext4_allocation_request *, int *); 2905 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2906 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2907 extern int __init ext4_init_mballoc(void); 2908 extern void ext4_exit_mballoc(void); 2909 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2910 ext4_group_t group, 2911 unsigned int nr, int *cnt); 2912 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2913 unsigned int nr); 2914 2915 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2916 struct buffer_head *bh, ext4_fsblk_t block, 2917 unsigned long count, int flags); 2918 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2919 ext4_group_t ngroups); 2920 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2921 ext4_group_t i, struct ext4_group_desc *desc); 2922 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2923 ext4_fsblk_t block, unsigned long count); 2924 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2925 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2926 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2927 int len, int state); 2928 2929 /* inode.c */ 2930 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2931 struct ext4_inode_info *ei); 2932 int ext4_inode_is_fast_symlink(struct inode *inode); 2933 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2934 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2935 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2936 bool wait, struct buffer_head **bhs); 2937 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2938 struct buffer_head *bh_result, int create); 2939 int ext4_get_block(struct inode *inode, sector_t iblock, 2940 struct buffer_head *bh_result, int create); 2941 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2942 struct buffer_head *bh, int create); 2943 int ext4_walk_page_buffers(handle_t *handle, 2944 struct inode *inode, 2945 struct buffer_head *head, 2946 unsigned from, 2947 unsigned to, 2948 int *partial, 2949 int (*fn)(handle_t *handle, struct inode *inode, 2950 struct buffer_head *bh)); 2951 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2952 struct buffer_head *bh); 2953 #define FALL_BACK_TO_NONDELALLOC 1 2954 #define CONVERT_INLINE_DATA 2 2955 2956 typedef enum { 2957 EXT4_IGET_NORMAL = 0, 2958 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2959 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2960 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 2961 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 2962 } ext4_iget_flags; 2963 2964 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2965 ext4_iget_flags flags, const char *function, 2966 unsigned int line); 2967 2968 #define ext4_iget(sb, ino, flags) \ 2969 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2970 2971 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2972 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2973 struct iattr *); 2974 extern u32 ext4_dio_alignment(struct inode *inode); 2975 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2976 struct kstat *, u32, unsigned int); 2977 extern void ext4_evict_inode(struct inode *); 2978 extern void ext4_clear_inode(struct inode *); 2979 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2980 struct kstat *, u32, unsigned int); 2981 extern int ext4_sync_inode(handle_t *, struct inode *); 2982 extern void ext4_dirty_inode(struct inode *, int); 2983 extern int ext4_change_inode_journal_flag(struct inode *, int); 2984 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2985 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2986 struct ext4_iloc *iloc); 2987 extern int ext4_inode_attach_jinode(struct inode *inode); 2988 extern int ext4_can_truncate(struct inode *inode); 2989 extern int ext4_truncate(struct inode *); 2990 extern int ext4_break_layouts(struct inode *); 2991 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2992 extern void ext4_set_inode_flags(struct inode *, bool init); 2993 extern int ext4_alloc_da_blocks(struct inode *inode); 2994 extern void ext4_set_aops(struct inode *inode); 2995 extern int ext4_writepage_trans_blocks(struct inode *); 2996 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 2997 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2998 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2999 loff_t lstart, loff_t lend); 3000 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3001 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3002 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3003 extern void ext4_da_release_space(struct inode *inode, int to_free); 3004 extern void ext4_da_update_reserve_space(struct inode *inode, 3005 int used, int quota_claim); 3006 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3007 ext4_fsblk_t pblk, ext4_lblk_t len); 3008 3009 /* indirect.c */ 3010 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3011 struct ext4_map_blocks *map, int flags); 3012 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3013 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3014 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3015 ext4_lblk_t start, ext4_lblk_t end); 3016 3017 /* ioctl.c */ 3018 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3019 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3020 int ext4_fileattr_set(struct mnt_idmap *idmap, 3021 struct dentry *dentry, struct fileattr *fa); 3022 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3023 extern void ext4_reset_inode_seed(struct inode *inode); 3024 int ext4_update_overhead(struct super_block *sb, bool force); 3025 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3026 3027 /* migrate.c */ 3028 extern int ext4_ext_migrate(struct inode *); 3029 extern int ext4_ind_migrate(struct inode *inode); 3030 3031 /* namei.c */ 3032 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3033 struct inode *inode); 3034 extern int ext4_dirblock_csum_verify(struct inode *inode, 3035 struct buffer_head *bh); 3036 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3037 __u32 start_minor_hash, __u32 *next_hash); 3038 extern int ext4_search_dir(struct buffer_head *bh, 3039 char *search_buf, 3040 int buf_size, 3041 struct inode *dir, 3042 struct ext4_filename *fname, 3043 unsigned int offset, 3044 struct ext4_dir_entry_2 **res_dir); 3045 extern int ext4_generic_delete_entry(struct inode *dir, 3046 struct ext4_dir_entry_2 *de_del, 3047 struct buffer_head *bh, 3048 void *entry_buf, 3049 int buf_size, 3050 int csum_size); 3051 extern bool ext4_empty_dir(struct inode *inode); 3052 3053 /* resize.c */ 3054 extern void ext4_kvfree_array_rcu(void *to_free); 3055 extern int ext4_group_add(struct super_block *sb, 3056 struct ext4_new_group_data *input); 3057 extern int ext4_group_extend(struct super_block *sb, 3058 struct ext4_super_block *es, 3059 ext4_fsblk_t n_blocks_count); 3060 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3061 extern unsigned int ext4_list_backups(struct super_block *sb, 3062 unsigned int *three, unsigned int *five, 3063 unsigned int *seven); 3064 3065 /* super.c */ 3066 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3067 sector_t block, blk_opf_t op_flags); 3068 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3069 sector_t block); 3070 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3071 bh_end_io_t *end_io); 3072 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3073 bh_end_io_t *end_io); 3074 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3075 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3076 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3077 extern int ext4_calculate_overhead(struct super_block *sb); 3078 extern __le32 ext4_superblock_csum(struct super_block *sb, 3079 struct ext4_super_block *es); 3080 extern void ext4_superblock_csum_set(struct super_block *sb); 3081 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3082 ext4_group_t ngroup); 3083 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3084 char nbuf[16]); 3085 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3086 ext4_group_t block_group, 3087 unsigned int flags); 3088 3089 extern __printf(7, 8) 3090 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3091 int, __u64, const char *, ...); 3092 extern __printf(6, 7) 3093 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3094 ext4_fsblk_t, int, const char *, ...); 3095 extern __printf(5, 6) 3096 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3097 const char *, ...); 3098 extern void __ext4_std_error(struct super_block *, const char *, 3099 unsigned int, int); 3100 extern __printf(4, 5) 3101 void __ext4_warning(struct super_block *, const char *, unsigned int, 3102 const char *, ...); 3103 extern __printf(4, 5) 3104 void __ext4_warning_inode(const struct inode *inode, const char *function, 3105 unsigned int line, const char *fmt, ...); 3106 extern __printf(3, 4) 3107 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3108 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3109 const char *, unsigned int, const char *); 3110 extern __printf(7, 8) 3111 void __ext4_grp_locked_error(const char *, unsigned int, 3112 struct super_block *, ext4_group_t, 3113 unsigned long, ext4_fsblk_t, 3114 const char *, ...); 3115 3116 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3117 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3118 3119 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3120 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3121 3122 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3123 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3124 (fmt), ## a) 3125 3126 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3127 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3128 3129 #define ext4_abort(sb, err, fmt, a...) \ 3130 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3131 3132 #ifdef CONFIG_PRINTK 3133 3134 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3135 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3136 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3137 __ext4_error_inode((inode), (func), (line), (block), \ 3138 (err), (fmt), ##__VA_ARGS__) 3139 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3140 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3141 #define ext4_error(sb, fmt, ...) \ 3142 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3143 ##__VA_ARGS__) 3144 #define ext4_error_err(sb, err, fmt, ...) \ 3145 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3146 ##__VA_ARGS__) 3147 #define ext4_warning(sb, fmt, ...) \ 3148 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3149 #define ext4_warning_inode(inode, fmt, ...) \ 3150 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3151 #define ext4_msg(sb, level, fmt, ...) \ 3152 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3153 #define dump_mmp_msg(sb, mmp, msg) \ 3154 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3155 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3156 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3157 fmt, ##__VA_ARGS__) 3158 3159 #else 3160 3161 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3162 do { \ 3163 no_printk(fmt, ##__VA_ARGS__); \ 3164 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3165 } while (0) 3166 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3167 do { \ 3168 no_printk(fmt, ##__VA_ARGS__); \ 3169 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3170 } while (0) 3171 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3172 do { \ 3173 no_printk(fmt, ##__VA_ARGS__); \ 3174 __ext4_error_file(file, "", 0, block, " "); \ 3175 } while (0) 3176 #define ext4_error(sb, fmt, ...) \ 3177 do { \ 3178 no_printk(fmt, ##__VA_ARGS__); \ 3179 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3180 } while (0) 3181 #define ext4_error_err(sb, err, fmt, ...) \ 3182 do { \ 3183 no_printk(fmt, ##__VA_ARGS__); \ 3184 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3185 } while (0) 3186 #define ext4_warning(sb, fmt, ...) \ 3187 do { \ 3188 no_printk(fmt, ##__VA_ARGS__); \ 3189 __ext4_warning(sb, "", 0, " "); \ 3190 } while (0) 3191 #define ext4_warning_inode(inode, fmt, ...) \ 3192 do { \ 3193 no_printk(fmt, ##__VA_ARGS__); \ 3194 __ext4_warning_inode(inode, "", 0, " "); \ 3195 } while (0) 3196 #define ext4_msg(sb, level, fmt, ...) \ 3197 do { \ 3198 no_printk(fmt, ##__VA_ARGS__); \ 3199 __ext4_msg(sb, "", " "); \ 3200 } while (0) 3201 #define dump_mmp_msg(sb, mmp, msg) \ 3202 __dump_mmp_msg(sb, mmp, "", 0, "") 3203 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3204 do { \ 3205 no_printk(fmt, ##__VA_ARGS__); \ 3206 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3207 } while (0) 3208 3209 #endif 3210 3211 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3212 struct ext4_group_desc *bg); 3213 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3214 struct ext4_group_desc *bg); 3215 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3216 struct ext4_group_desc *bg); 3217 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3218 struct ext4_group_desc *bg); 3219 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3220 struct ext4_group_desc *bg); 3221 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3222 struct ext4_group_desc *bg); 3223 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3224 struct ext4_group_desc *bg); 3225 extern void ext4_block_bitmap_set(struct super_block *sb, 3226 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3227 extern void ext4_inode_bitmap_set(struct super_block *sb, 3228 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3229 extern void ext4_inode_table_set(struct super_block *sb, 3230 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3231 extern void ext4_free_group_clusters_set(struct super_block *sb, 3232 struct ext4_group_desc *bg, 3233 __u32 count); 3234 extern void ext4_free_inodes_set(struct super_block *sb, 3235 struct ext4_group_desc *bg, __u32 count); 3236 extern void ext4_used_dirs_set(struct super_block *sb, 3237 struct ext4_group_desc *bg, __u32 count); 3238 extern void ext4_itable_unused_set(struct super_block *sb, 3239 struct ext4_group_desc *bg, __u32 count); 3240 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3241 struct ext4_group_desc *gdp); 3242 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3243 struct ext4_group_desc *gdp); 3244 extern int ext4_register_li_request(struct super_block *sb, 3245 ext4_group_t first_not_zeroed); 3246 3247 static inline int ext4_has_metadata_csum(struct super_block *sb) 3248 { 3249 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3250 !EXT4_SB(sb)->s_chksum_driver); 3251 3252 return ext4_has_feature_metadata_csum(sb) && 3253 (EXT4_SB(sb)->s_chksum_driver != NULL); 3254 } 3255 3256 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3257 { 3258 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3259 } 3260 3261 #define ext4_read_incompat_64bit_val(es, name) \ 3262 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3263 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3264 le32_to_cpu(es->name##_lo)) 3265 3266 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3267 { 3268 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3269 } 3270 3271 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3272 { 3273 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3274 } 3275 3276 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3277 { 3278 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3279 } 3280 3281 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3282 ext4_fsblk_t blk) 3283 { 3284 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3285 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3286 } 3287 3288 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3289 ext4_fsblk_t blk) 3290 { 3291 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3292 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3293 } 3294 3295 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3296 ext4_fsblk_t blk) 3297 { 3298 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3299 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3300 } 3301 3302 static inline loff_t ext4_isize(struct super_block *sb, 3303 struct ext4_inode *raw_inode) 3304 { 3305 if (ext4_has_feature_largedir(sb) || 3306 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3307 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3308 le32_to_cpu(raw_inode->i_size_lo); 3309 3310 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3311 } 3312 3313 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3314 { 3315 raw_inode->i_size_lo = cpu_to_le32(i_size); 3316 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3317 } 3318 3319 /* 3320 * Reading s_groups_count requires using smp_rmb() afterwards. See 3321 * the locking protocol documented in the comments of ext4_group_add() 3322 * in resize.c 3323 */ 3324 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3325 { 3326 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3327 3328 smp_rmb(); 3329 return ngroups; 3330 } 3331 3332 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3333 ext4_group_t block_group) 3334 { 3335 return block_group >> sbi->s_log_groups_per_flex; 3336 } 3337 3338 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3339 { 3340 return 1 << sbi->s_log_groups_per_flex; 3341 } 3342 3343 #define ext4_std_error(sb, errno) \ 3344 do { \ 3345 if ((errno)) \ 3346 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3347 } while (0) 3348 3349 #ifdef CONFIG_SMP 3350 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3351 * counters. So we need to make sure we have free clusters more 3352 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3353 */ 3354 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3355 #else 3356 #define EXT4_FREECLUSTERS_WATERMARK 0 3357 #endif 3358 3359 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3360 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3361 { 3362 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3363 !inode_is_locked(inode)); 3364 down_write(&EXT4_I(inode)->i_data_sem); 3365 if (newsize > EXT4_I(inode)->i_disksize) 3366 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3367 up_write(&EXT4_I(inode)->i_data_sem); 3368 } 3369 3370 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3371 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3372 { 3373 int changed = 0; 3374 3375 if (newsize > inode->i_size) { 3376 i_size_write(inode, newsize); 3377 changed = 1; 3378 } 3379 if (newsize > EXT4_I(inode)->i_disksize) { 3380 ext4_update_i_disksize(inode, newsize); 3381 changed |= 2; 3382 } 3383 return changed; 3384 } 3385 3386 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3387 loff_t len); 3388 3389 struct ext4_group_info { 3390 unsigned long bb_state; 3391 #ifdef AGGRESSIVE_CHECK 3392 unsigned long bb_check_counter; 3393 #endif 3394 struct rb_root bb_free_root; 3395 ext4_grpblk_t bb_first_free; /* first free block */ 3396 ext4_grpblk_t bb_free; /* total free blocks */ 3397 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3398 int bb_avg_fragment_size_order; /* order of average 3399 fragment in BG */ 3400 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3401 ext4_group_t bb_group; /* Group number */ 3402 struct list_head bb_prealloc_list; 3403 #ifdef DOUBLE_CHECK 3404 void *bb_bitmap; 3405 #endif 3406 struct rw_semaphore alloc_sem; 3407 struct list_head bb_avg_fragment_size_node; 3408 struct list_head bb_largest_free_order_node; 3409 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3410 * regions, index is order. 3411 * bb_counters[3] = 5 means 3412 * 5 free 8-block regions. */ 3413 }; 3414 3415 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3416 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3417 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3418 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3419 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3420 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3421 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3422 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3423 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3424 3425 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3426 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3427 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3428 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3429 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3430 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3431 3432 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3433 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3434 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3435 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3436 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3437 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3438 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3439 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3440 3441 #define EXT4_MAX_CONTENTION 8 3442 #define EXT4_CONTENTION_THRESHOLD 2 3443 3444 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3445 ext4_group_t group) 3446 { 3447 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3448 } 3449 3450 /* 3451 * Returns true if the filesystem is busy enough that attempts to 3452 * access the block group locks has run into contention. 3453 */ 3454 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3455 { 3456 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3457 } 3458 3459 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3460 { 3461 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3462 if (spin_trylock(lock)) 3463 /* 3464 * We're able to grab the lock right away, so drop the 3465 * lock contention counter. 3466 */ 3467 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3468 else { 3469 /* 3470 * The lock is busy, so bump the contention counter, 3471 * and then wait on the spin lock. 3472 */ 3473 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3474 EXT4_MAX_CONTENTION); 3475 spin_lock(lock); 3476 } 3477 } 3478 3479 static inline void ext4_unlock_group(struct super_block *sb, 3480 ext4_group_t group) 3481 { 3482 spin_unlock(ext4_group_lock_ptr(sb, group)); 3483 } 3484 3485 #ifdef CONFIG_QUOTA 3486 static inline bool ext4_quota_capable(struct super_block *sb) 3487 { 3488 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3489 } 3490 3491 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3492 { 3493 struct ext4_sb_info *sbi = EXT4_SB(sb); 3494 3495 return (ext4_has_feature_quota(sb) || 3496 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3497 } 3498 int ext4_enable_quotas(struct super_block *sb); 3499 #endif 3500 3501 /* 3502 * Block validity checking 3503 */ 3504 #define ext4_check_indirect_blockref(inode, bh) \ 3505 ext4_check_blockref(__func__, __LINE__, inode, \ 3506 (__le32 *)(bh)->b_data, \ 3507 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3508 3509 #define ext4_ind_check_inode(inode) \ 3510 ext4_check_blockref(__func__, __LINE__, inode, \ 3511 EXT4_I(inode)->i_data, \ 3512 EXT4_NDIR_BLOCKS) 3513 3514 /* 3515 * Inodes and files operations 3516 */ 3517 3518 /* dir.c */ 3519 extern const struct file_operations ext4_dir_operations; 3520 3521 /* file.c */ 3522 extern const struct inode_operations ext4_file_inode_operations; 3523 extern const struct file_operations ext4_file_operations; 3524 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3525 3526 /* inline.c */ 3527 extern int ext4_get_max_inline_size(struct inode *inode); 3528 extern int ext4_find_inline_data_nolock(struct inode *inode); 3529 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3530 unsigned int len); 3531 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3532 3533 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3534 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3535 struct inode *inode, 3536 loff_t pos, unsigned len, 3537 struct page **pagep); 3538 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3539 unsigned copied, struct folio *folio); 3540 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3541 struct inode *inode, 3542 loff_t pos, unsigned len, 3543 struct page **pagep, 3544 void **fsdata); 3545 extern int ext4_try_add_inline_entry(handle_t *handle, 3546 struct ext4_filename *fname, 3547 struct inode *dir, struct inode *inode); 3548 extern int ext4_try_create_inline_dir(handle_t *handle, 3549 struct inode *parent, 3550 struct inode *inode); 3551 extern int ext4_read_inline_dir(struct file *filp, 3552 struct dir_context *ctx, 3553 int *has_inline_data); 3554 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3555 struct inode *dir, ext4_lblk_t block, 3556 struct dx_hash_info *hinfo, 3557 __u32 start_hash, __u32 start_minor_hash, 3558 int *has_inline_data); 3559 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3560 struct ext4_filename *fname, 3561 struct ext4_dir_entry_2 **res_dir, 3562 int *has_inline_data); 3563 extern int ext4_delete_inline_entry(handle_t *handle, 3564 struct inode *dir, 3565 struct ext4_dir_entry_2 *de_del, 3566 struct buffer_head *bh, 3567 int *has_inline_data); 3568 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3569 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3570 struct ext4_dir_entry_2 **parent_de, 3571 int *retval); 3572 extern void *ext4_read_inline_link(struct inode *inode); 3573 3574 struct iomap; 3575 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3576 3577 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3578 3579 extern int ext4_convert_inline_data(struct inode *inode); 3580 3581 static inline int ext4_has_inline_data(struct inode *inode) 3582 { 3583 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3584 EXT4_I(inode)->i_inline_off; 3585 } 3586 3587 /* namei.c */ 3588 extern const struct inode_operations ext4_dir_inode_operations; 3589 extern const struct inode_operations ext4_special_inode_operations; 3590 extern struct dentry *ext4_get_parent(struct dentry *child); 3591 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3592 struct ext4_dir_entry_2 *de, 3593 int blocksize, int csum_size, 3594 unsigned int parent_ino, int dotdot_real_len); 3595 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3596 unsigned int blocksize); 3597 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3598 struct buffer_head *bh); 3599 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3600 struct inode *inode, struct dentry *dentry); 3601 extern int __ext4_link(struct inode *dir, struct inode *inode, 3602 struct dentry *dentry); 3603 3604 #define S_SHIFT 12 3605 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3606 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3607 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3608 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3609 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3610 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3611 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3612 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3613 }; 3614 3615 static inline void ext4_set_de_type(struct super_block *sb, 3616 struct ext4_dir_entry_2 *de, 3617 umode_t mode) { 3618 if (ext4_has_feature_filetype(sb)) 3619 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3620 } 3621 3622 /* readpages.c */ 3623 extern int ext4_mpage_readpages(struct inode *inode, 3624 struct readahead_control *rac, struct folio *folio); 3625 extern int __init ext4_init_post_read_processing(void); 3626 extern void ext4_exit_post_read_processing(void); 3627 3628 /* symlink.c */ 3629 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3630 extern const struct inode_operations ext4_symlink_inode_operations; 3631 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3632 3633 /* sysfs.c */ 3634 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3635 extern int ext4_register_sysfs(struct super_block *sb); 3636 extern void ext4_unregister_sysfs(struct super_block *sb); 3637 extern int __init ext4_init_sysfs(void); 3638 extern void ext4_exit_sysfs(void); 3639 3640 /* block_validity */ 3641 extern void ext4_release_system_zone(struct super_block *sb); 3642 extern int ext4_setup_system_zone(struct super_block *sb); 3643 extern int __init ext4_init_system_zone(void); 3644 extern void ext4_exit_system_zone(void); 3645 extern int ext4_inode_block_valid(struct inode *inode, 3646 ext4_fsblk_t start_blk, 3647 unsigned int count); 3648 extern int ext4_check_blockref(const char *, unsigned int, 3649 struct inode *, __le32 *, unsigned int); 3650 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3651 ext4_fsblk_t start_blk, unsigned int count); 3652 3653 3654 /* extents.c */ 3655 struct ext4_ext_path; 3656 struct ext4_extent; 3657 3658 /* 3659 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3660 * __le32. 3661 */ 3662 #define EXT_MAX_BLOCKS 0xffffffff 3663 3664 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3665 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3666 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3667 struct ext4_map_blocks *map, int flags); 3668 extern int ext4_ext_truncate(handle_t *, struct inode *); 3669 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3670 ext4_lblk_t end); 3671 extern void ext4_ext_init(struct super_block *); 3672 extern void ext4_ext_release(struct super_block *); 3673 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3674 loff_t len); 3675 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3676 loff_t offset, ssize_t len); 3677 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3678 ext4_io_end_t *io_end); 3679 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3680 struct ext4_map_blocks *map, int flags); 3681 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3682 int num, 3683 struct ext4_ext_path *path); 3684 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3685 struct ext4_ext_path **, 3686 struct ext4_extent *, int); 3687 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3688 struct ext4_ext_path **, 3689 int flags); 3690 extern void ext4_free_ext_path(struct ext4_ext_path *); 3691 extern int ext4_ext_check_inode(struct inode *inode); 3692 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3693 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3694 __u64 start, __u64 len); 3695 extern int ext4_get_es_cache(struct inode *inode, 3696 struct fiemap_extent_info *fieinfo, 3697 __u64 start, __u64 len); 3698 extern int ext4_ext_precache(struct inode *inode); 3699 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3700 struct inode *inode2, ext4_lblk_t lblk1, 3701 ext4_lblk_t lblk2, ext4_lblk_t count, 3702 int mark_unwritten,int *err); 3703 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3704 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3705 int check_cred, int restart_cred, 3706 int revoke_cred); 3707 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3708 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3709 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3710 int len, int unwritten, ext4_fsblk_t pblk); 3711 extern int ext4_ext_clear_bb(struct inode *inode); 3712 3713 3714 /* move_extent.c */ 3715 extern void ext4_double_down_write_data_sem(struct inode *first, 3716 struct inode *second); 3717 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3718 struct inode *donor_inode); 3719 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3720 __u64 start_orig, __u64 start_donor, 3721 __u64 len, __u64 *moved_len); 3722 3723 /* page-io.c */ 3724 extern int __init ext4_init_pageio(void); 3725 extern void ext4_exit_pageio(void); 3726 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3727 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3728 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3729 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3730 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3731 struct writeback_control *wbc); 3732 extern void ext4_end_io_rsv_work(struct work_struct *work); 3733 extern void ext4_io_submit(struct ext4_io_submit *io); 3734 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3735 size_t len); 3736 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3737 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3738 3739 /* mmp.c */ 3740 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3741 3742 /* mmp.c */ 3743 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3744 3745 /* verity.c */ 3746 extern const struct fsverity_operations ext4_verityops; 3747 3748 /* orphan.c */ 3749 extern int ext4_orphan_add(handle_t *, struct inode *); 3750 extern int ext4_orphan_del(handle_t *, struct inode *); 3751 extern void ext4_orphan_cleanup(struct super_block *sb, 3752 struct ext4_super_block *es); 3753 extern void ext4_release_orphan_info(struct super_block *sb); 3754 extern int ext4_init_orphan_info(struct super_block *sb); 3755 extern int ext4_orphan_file_empty(struct super_block *sb); 3756 extern void ext4_orphan_file_block_trigger( 3757 struct jbd2_buffer_trigger_type *triggers, 3758 struct buffer_head *bh, 3759 void *data, size_t size); 3760 3761 /* 3762 * Add new method to test whether block and inode bitmaps are properly 3763 * initialized. With uninit_bg reading the block from disk is not enough 3764 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3765 */ 3766 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3767 3768 static inline int bitmap_uptodate(struct buffer_head *bh) 3769 { 3770 return (buffer_uptodate(bh) && 3771 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3772 } 3773 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3774 { 3775 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3776 } 3777 3778 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3779 3780 /* For ioend & aio unwritten conversion wait queues */ 3781 #define EXT4_WQ_HASH_SZ 37 3782 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3783 EXT4_WQ_HASH_SZ]) 3784 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3785 3786 extern int ext4_resize_begin(struct super_block *sb); 3787 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3788 3789 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3790 struct ext4_io_end *io_end) 3791 { 3792 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3793 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3794 atomic_inc(&EXT4_I(inode)->i_unwritten); 3795 } 3796 } 3797 3798 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3799 { 3800 struct inode *inode = io_end->inode; 3801 3802 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3803 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3804 /* Wake up anyone waiting on unwritten extent conversion */ 3805 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3806 wake_up_all(ext4_ioend_wq(inode)); 3807 } 3808 } 3809 3810 extern const struct iomap_ops ext4_iomap_ops; 3811 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3812 extern const struct iomap_ops ext4_iomap_report_ops; 3813 3814 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3815 { 3816 /* 3817 * If the buffer has the write error flag, we have failed 3818 * to write out data in the block. In this case, we don't 3819 * have to read the block because we may read the old data 3820 * successfully. 3821 */ 3822 if (buffer_write_io_error(bh)) 3823 set_buffer_uptodate(bh); 3824 return buffer_uptodate(bh); 3825 } 3826 3827 #endif /* __KERNEL__ */ 3828 3829 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3830 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3831 3832 #endif /* _EXT4_H */ 3833