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 * Number of criterias defined. For each criteria, mballoc has slightly 132 * different way of finding the required blocks nad usually, higher the 133 * criteria the slower the allocation. We start at lower criterias and keep 134 * falling back to higher ones if we are not able to find any blocks. 135 */ 136 #define EXT4_MB_NUM_CRS 5 137 /* 138 * All possible allocation criterias for mballoc 139 */ 140 enum criteria { 141 CR0, 142 CR1, 143 CR1_5, 144 CR2, 145 CR3, 146 }; 147 148 /* 149 * Flags used in mballoc's allocation_context flags field. 150 * 151 * Also used to show what's going on for debugging purposes when the 152 * flag field is exported via the traceport interface 153 */ 154 155 /* prefer goal again. length */ 156 #define EXT4_MB_HINT_MERGE 0x0001 157 /* blocks already reserved */ 158 #define EXT4_MB_HINT_RESERVED 0x0002 159 /* metadata is being allocated */ 160 #define EXT4_MB_HINT_METADATA 0x0004 161 /* first blocks in the file */ 162 #define EXT4_MB_HINT_FIRST 0x0008 163 /* search for the best chunk */ 164 #define EXT4_MB_HINT_BEST 0x0010 165 /* data is being allocated */ 166 #define EXT4_MB_HINT_DATA 0x0020 167 /* don't preallocate (for tails) */ 168 #define EXT4_MB_HINT_NOPREALLOC 0x0040 169 /* allocate for locality group */ 170 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 171 /* allocate goal blocks or none */ 172 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 173 /* goal is meaningful */ 174 #define EXT4_MB_HINT_TRY_GOAL 0x0200 175 /* blocks already pre-reserved by delayed allocation */ 176 #define EXT4_MB_DELALLOC_RESERVED 0x0400 177 /* We are doing stream allocation */ 178 #define EXT4_MB_STREAM_ALLOC 0x0800 179 /* Use reserved root blocks if needed */ 180 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 181 /* Use blocks from reserved pool */ 182 #define EXT4_MB_USE_RESERVED 0x2000 183 /* Do strict check for free blocks while retrying block allocation */ 184 #define EXT4_MB_STRICT_CHECK 0x4000 185 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 186 #define EXT4_MB_CR0_OPTIMIZED 0x8000 187 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 188 #define EXT4_MB_CR1_OPTIMIZED 0x00010000 189 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1.5 */ 190 #define EXT4_MB_CR1_5_OPTIMIZED 0x00020000 191 192 struct ext4_allocation_request { 193 /* target inode for block we're allocating */ 194 struct inode *inode; 195 /* how many blocks we want to allocate */ 196 unsigned int len; 197 /* logical block in target inode */ 198 ext4_lblk_t logical; 199 /* the closest logical allocated block to the left */ 200 ext4_lblk_t lleft; 201 /* the closest logical allocated block to the right */ 202 ext4_lblk_t lright; 203 /* phys. target (a hint) */ 204 ext4_fsblk_t goal; 205 /* phys. block for the closest logical allocated block to the left */ 206 ext4_fsblk_t pleft; 207 /* phys. block for the closest logical allocated block to the right */ 208 ext4_fsblk_t pright; 209 /* flags. see above EXT4_MB_HINT_* */ 210 unsigned int flags; 211 }; 212 213 /* 214 * Logical to physical block mapping, used by ext4_map_blocks() 215 * 216 * This structure is used to pass requests into ext4_map_blocks() as 217 * well as to store the information returned by ext4_map_blocks(). It 218 * takes less room on the stack than a struct buffer_head. 219 */ 220 #define EXT4_MAP_NEW BIT(BH_New) 221 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 222 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 223 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 224 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 225 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 226 227 struct ext4_map_blocks { 228 ext4_fsblk_t m_pblk; 229 ext4_lblk_t m_lblk; 230 unsigned int m_len; 231 unsigned int m_flags; 232 }; 233 234 /* 235 * Block validity checking, system zone rbtree. 236 */ 237 struct ext4_system_blocks { 238 struct rb_root root; 239 struct rcu_head rcu; 240 }; 241 242 /* 243 * Flags for ext4_io_end->flags 244 */ 245 #define EXT4_IO_END_UNWRITTEN 0x0001 246 247 struct ext4_io_end_vec { 248 struct list_head list; /* list of io_end_vec */ 249 loff_t offset; /* offset in the file */ 250 ssize_t size; /* size of the extent */ 251 }; 252 253 /* 254 * For converting unwritten extents on a work queue. 'handle' is used for 255 * buffered writeback. 256 */ 257 typedef struct ext4_io_end { 258 struct list_head list; /* per-file finished IO list */ 259 handle_t *handle; /* handle reserved for extent 260 * conversion */ 261 struct inode *inode; /* file being written to */ 262 struct bio *bio; /* Linked list of completed 263 * bios covering the extent */ 264 unsigned int flag; /* unwritten or not */ 265 refcount_t count; /* reference counter */ 266 struct list_head list_vec; /* list of ext4_io_end_vec */ 267 } ext4_io_end_t; 268 269 struct ext4_io_submit { 270 struct writeback_control *io_wbc; 271 struct bio *io_bio; 272 ext4_io_end_t *io_end; 273 sector_t io_next_block; 274 }; 275 276 /* 277 * Special inodes numbers 278 */ 279 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 280 #define EXT4_ROOT_INO 2 /* Root inode */ 281 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 282 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 283 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 284 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 285 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 286 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 287 288 /* First non-reserved inode for old ext4 filesystems */ 289 #define EXT4_GOOD_OLD_FIRST_INO 11 290 291 /* 292 * Maximal count of links to a file 293 */ 294 #define EXT4_LINK_MAX 65000 295 296 /* 297 * Macro-instructions used to manage several block sizes 298 */ 299 #define EXT4_MIN_BLOCK_SIZE 1024 300 #define EXT4_MAX_BLOCK_SIZE 65536 301 #define EXT4_MIN_BLOCK_LOG_SIZE 10 302 #define EXT4_MAX_BLOCK_LOG_SIZE 16 303 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 304 #ifdef __KERNEL__ 305 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 306 #else 307 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 308 #endif 309 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 310 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 311 EXT4_SB(s)->s_cluster_bits) 312 #ifdef __KERNEL__ 313 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 314 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 315 #else 316 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 317 #endif 318 #ifdef __KERNEL__ 319 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 320 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 321 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 322 #else 323 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 324 EXT4_GOOD_OLD_INODE_SIZE : \ 325 (s)->s_inode_size) 326 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 327 EXT4_GOOD_OLD_FIRST_INO : \ 328 (s)->s_first_ino) 329 #endif 330 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 331 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 332 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 333 blkbits)) 334 335 /* Translate a block number to a cluster number */ 336 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 337 /* Translate a cluster number to a block number */ 338 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 339 /* Translate # of blks to # of clusters */ 340 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 341 (sbi)->s_cluster_bits) 342 /* Mask out the low bits to get the starting block of the cluster */ 343 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 344 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 345 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 346 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 347 /* Fill in the low bits to get the last block of the cluster */ 348 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 349 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 350 /* Get the cluster offset */ 351 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 352 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 353 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 354 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 355 356 /* 357 * Structure of a blocks group descriptor 358 */ 359 struct ext4_group_desc 360 { 361 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 362 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 363 __le32 bg_inode_table_lo; /* Inodes table block */ 364 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 365 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 366 __le16 bg_used_dirs_count_lo; /* Directories count */ 367 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 368 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 369 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 370 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 371 __le16 bg_itable_unused_lo; /* Unused inodes count */ 372 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 373 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 374 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 375 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 376 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 377 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 378 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 379 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 380 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 381 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 382 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 383 __u32 bg_reserved; 384 }; 385 386 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 387 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 388 sizeof(__le16)) 389 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 390 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 391 sizeof(__le16)) 392 393 /* 394 * Structure of a flex block group info 395 */ 396 397 struct flex_groups { 398 atomic64_t free_clusters; 399 atomic_t free_inodes; 400 atomic_t used_dirs; 401 }; 402 403 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 404 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 405 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 406 407 /* 408 * Macro-instructions used to manage group descriptors 409 */ 410 #define EXT4_MIN_DESC_SIZE 32 411 #define EXT4_MIN_DESC_SIZE_64BIT 64 412 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 413 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 414 #ifdef __KERNEL__ 415 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 416 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 417 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 418 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 419 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 420 #else 421 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 422 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 423 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 424 #endif 425 426 /* 427 * Constants relative to the data blocks 428 */ 429 #define EXT4_NDIR_BLOCKS 12 430 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 431 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 432 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 433 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 434 435 /* 436 * Inode flags 437 */ 438 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 439 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 440 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 441 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 442 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 443 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 444 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 445 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 446 /* Reserved for compression usage... */ 447 #define EXT4_DIRTY_FL 0x00000100 448 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 449 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 450 /* nb: was previously EXT2_ECOMPR_FL */ 451 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 452 /* End compression flags --- maybe not all used */ 453 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 454 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 455 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 456 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 457 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 458 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 459 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 460 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 461 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 462 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 463 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 464 465 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 466 467 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 468 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 469 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 470 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 471 472 /* User modifiable flags */ 473 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 474 EXT4_UNRM_FL | \ 475 EXT4_COMPR_FL | \ 476 EXT4_SYNC_FL | \ 477 EXT4_IMMUTABLE_FL | \ 478 EXT4_APPEND_FL | \ 479 EXT4_NODUMP_FL | \ 480 EXT4_NOATIME_FL | \ 481 EXT4_JOURNAL_DATA_FL | \ 482 EXT4_NOTAIL_FL | \ 483 EXT4_DIRSYNC_FL | \ 484 EXT4_TOPDIR_FL | \ 485 EXT4_EXTENTS_FL | \ 486 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 487 EXT4_DAX_FL | \ 488 EXT4_PROJINHERIT_FL | \ 489 EXT4_CASEFOLD_FL) 490 491 /* User visible flags */ 492 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 493 EXT4_DIRTY_FL | \ 494 EXT4_COMPRBLK_FL | \ 495 EXT4_NOCOMPR_FL | \ 496 EXT4_ENCRYPT_FL | \ 497 EXT4_INDEX_FL | \ 498 EXT4_VERITY_FL | \ 499 EXT4_INLINE_DATA_FL) 500 501 /* Flags that should be inherited by new inodes from their parent. */ 502 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 503 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 504 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 505 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 506 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 507 EXT4_DAX_FL) 508 509 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 510 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 511 EXT4_PROJINHERIT_FL)) 512 513 /* Flags that are appropriate for non-directories/regular files. */ 514 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 515 516 /* The only flags that should be swapped */ 517 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 518 519 /* Flags which are mutually exclusive to DAX */ 520 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 521 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 522 523 /* Mask out flags that are inappropriate for the given type of inode. */ 524 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 525 { 526 if (S_ISDIR(mode)) 527 return flags; 528 else if (S_ISREG(mode)) 529 return flags & EXT4_REG_FLMASK; 530 else 531 return flags & EXT4_OTHER_FLMASK; 532 } 533 534 /* 535 * Inode flags used for atomic set/get 536 */ 537 enum { 538 EXT4_INODE_SECRM = 0, /* Secure deletion */ 539 EXT4_INODE_UNRM = 1, /* Undelete */ 540 EXT4_INODE_COMPR = 2, /* Compress file */ 541 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 542 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 543 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 544 EXT4_INODE_NODUMP = 6, /* do not dump file */ 545 EXT4_INODE_NOATIME = 7, /* do not update atime */ 546 /* Reserved for compression usage... */ 547 EXT4_INODE_DIRTY = 8, 548 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 549 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 550 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 551 /* End compression flags --- maybe not all used */ 552 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 553 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 554 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 555 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 556 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 557 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 558 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 559 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 560 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 561 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 562 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 563 EXT4_INODE_DAX = 25, /* Inode is DAX */ 564 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 565 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 566 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 567 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 568 }; 569 570 /* 571 * Since it's pretty easy to mix up bit numbers and hex values, we use a 572 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 573 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 574 * any extra space in the compiled kernel image, otherwise, the build will fail. 575 * It's important that these values are the same, since we are using 576 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 577 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 578 * values found in ext2, ext3 and ext4 filesystems, and of course the values 579 * defined in e2fsprogs. 580 * 581 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 582 */ 583 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 584 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 585 586 static inline void ext4_check_flag_values(void) 587 { 588 CHECK_FLAG_VALUE(SECRM); 589 CHECK_FLAG_VALUE(UNRM); 590 CHECK_FLAG_VALUE(COMPR); 591 CHECK_FLAG_VALUE(SYNC); 592 CHECK_FLAG_VALUE(IMMUTABLE); 593 CHECK_FLAG_VALUE(APPEND); 594 CHECK_FLAG_VALUE(NODUMP); 595 CHECK_FLAG_VALUE(NOATIME); 596 CHECK_FLAG_VALUE(DIRTY); 597 CHECK_FLAG_VALUE(COMPRBLK); 598 CHECK_FLAG_VALUE(NOCOMPR); 599 CHECK_FLAG_VALUE(ENCRYPT); 600 CHECK_FLAG_VALUE(INDEX); 601 CHECK_FLAG_VALUE(IMAGIC); 602 CHECK_FLAG_VALUE(JOURNAL_DATA); 603 CHECK_FLAG_VALUE(NOTAIL); 604 CHECK_FLAG_VALUE(DIRSYNC); 605 CHECK_FLAG_VALUE(TOPDIR); 606 CHECK_FLAG_VALUE(HUGE_FILE); 607 CHECK_FLAG_VALUE(EXTENTS); 608 CHECK_FLAG_VALUE(VERITY); 609 CHECK_FLAG_VALUE(EA_INODE); 610 CHECK_FLAG_VALUE(INLINE_DATA); 611 CHECK_FLAG_VALUE(PROJINHERIT); 612 CHECK_FLAG_VALUE(CASEFOLD); 613 CHECK_FLAG_VALUE(RESERVED); 614 } 615 616 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 617 struct compat_ext4_new_group_input { 618 u32 group; 619 compat_u64 block_bitmap; 620 compat_u64 inode_bitmap; 621 compat_u64 inode_table; 622 u32 blocks_count; 623 u16 reserved_blocks; 624 u16 unused; 625 }; 626 #endif 627 628 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 629 struct ext4_new_group_data { 630 __u32 group; 631 __u64 block_bitmap; 632 __u64 inode_bitmap; 633 __u64 inode_table; 634 __u32 blocks_count; 635 __u16 reserved_blocks; 636 __u16 mdata_blocks; 637 __u32 free_clusters_count; 638 }; 639 640 /* Indexes used to index group tables in ext4_new_group_data */ 641 enum { 642 BLOCK_BITMAP = 0, /* block bitmap */ 643 INODE_BITMAP, /* inode bitmap */ 644 INODE_TABLE, /* inode tables */ 645 GROUP_TABLE_COUNT, 646 }; 647 648 /* 649 * Flags used by ext4_map_blocks() 650 */ 651 /* Allocate any needed blocks and/or convert an unwritten 652 extent to be an initialized ext4 */ 653 #define EXT4_GET_BLOCKS_CREATE 0x0001 654 /* Request the creation of an unwritten extent */ 655 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 656 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 657 EXT4_GET_BLOCKS_CREATE) 658 /* Caller is from the delayed allocation writeout path 659 * finally doing the actual allocation of delayed blocks */ 660 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 661 /* caller is from the direct IO path, request to creation of an 662 unwritten extents if not allocated, split the unwritten 663 extent if blocks has been preallocated already*/ 664 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 665 #define EXT4_GET_BLOCKS_CONVERT 0x0010 666 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 667 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 668 /* Convert extent to initialized after IO complete */ 669 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 670 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 671 /* Eventual metadata allocation (due to growing extent tree) 672 * should not fail, so try to use reserved blocks for that.*/ 673 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 674 /* Don't normalize allocation size (used for fallocate) */ 675 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 676 /* Convert written extents to unwritten */ 677 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 678 /* Write zeros to newly created written extents */ 679 #define EXT4_GET_BLOCKS_ZERO 0x0200 680 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 681 EXT4_GET_BLOCKS_ZERO) 682 /* Caller will submit data before dropping transaction handle. This 683 * allows jbd2 to avoid submitting data before commit. */ 684 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 685 /* Caller is in the atomic contex, find extent if it has been cached */ 686 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 687 688 /* 689 * The bit position of these flags must not overlap with any of the 690 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 691 * read_extent_tree_block(), ext4_split_extent_at(), 692 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 693 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 694 * caching the extents when reading from the extent tree while a 695 * truncate or punch hole operation is in progress. 696 */ 697 #define EXT4_EX_NOCACHE 0x40000000 698 #define EXT4_EX_FORCE_CACHE 0x20000000 699 #define EXT4_EX_NOFAIL 0x10000000 700 701 /* 702 * Flags used by ext4_free_blocks 703 */ 704 #define EXT4_FREE_BLOCKS_METADATA 0x0001 705 #define EXT4_FREE_BLOCKS_FORGET 0x0002 706 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 707 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 708 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 709 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 710 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 711 712 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 713 /* 714 * ioctl commands in 32 bit emulation 715 */ 716 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 717 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 718 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 719 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 720 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 721 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 722 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 723 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 724 #endif 725 726 /* Max physical block we can address w/o extents */ 727 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 728 729 /* Max logical block we can support */ 730 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 731 732 /* 733 * Structure of an inode on the disk 734 */ 735 struct ext4_inode { 736 __le16 i_mode; /* File mode */ 737 __le16 i_uid; /* Low 16 bits of Owner Uid */ 738 __le32 i_size_lo; /* Size in bytes */ 739 __le32 i_atime; /* Access time */ 740 __le32 i_ctime; /* Inode Change time */ 741 __le32 i_mtime; /* Modification time */ 742 __le32 i_dtime; /* Deletion Time */ 743 __le16 i_gid; /* Low 16 bits of Group Id */ 744 __le16 i_links_count; /* Links count */ 745 __le32 i_blocks_lo; /* Blocks count */ 746 __le32 i_flags; /* File flags */ 747 union { 748 struct { 749 __le32 l_i_version; 750 } linux1; 751 struct { 752 __u32 h_i_translator; 753 } hurd1; 754 struct { 755 __u32 m_i_reserved1; 756 } masix1; 757 } osd1; /* OS dependent 1 */ 758 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 759 __le32 i_generation; /* File version (for NFS) */ 760 __le32 i_file_acl_lo; /* File ACL */ 761 __le32 i_size_high; 762 __le32 i_obso_faddr; /* Obsoleted fragment address */ 763 union { 764 struct { 765 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 766 __le16 l_i_file_acl_high; 767 __le16 l_i_uid_high; /* these 2 fields */ 768 __le16 l_i_gid_high; /* were reserved2[0] */ 769 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 770 __le16 l_i_reserved; 771 } linux2; 772 struct { 773 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 774 __u16 h_i_mode_high; 775 __u16 h_i_uid_high; 776 __u16 h_i_gid_high; 777 __u32 h_i_author; 778 } hurd2; 779 struct { 780 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 781 __le16 m_i_file_acl_high; 782 __u32 m_i_reserved2[2]; 783 } masix2; 784 } osd2; /* OS dependent 2 */ 785 __le16 i_extra_isize; 786 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 787 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 788 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 789 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 790 __le32 i_crtime; /* File Creation time */ 791 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 792 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 793 __le32 i_projid; /* Project ID */ 794 }; 795 796 #define EXT4_EPOCH_BITS 2 797 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 798 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 799 800 /* 801 * Extended fields will fit into an inode if the filesystem was formatted 802 * with large inodes (-I 256 or larger) and there are not currently any EAs 803 * consuming all of the available space. For new inodes we always reserve 804 * enough space for the kernel's known extended fields, but for inodes 805 * created with an old kernel this might not have been the case. None of 806 * the extended inode fields is critical for correct filesystem operation. 807 * This macro checks if a certain field fits in the inode. Note that 808 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 809 */ 810 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 811 ((offsetof(typeof(*ext4_inode), field) + \ 812 sizeof((ext4_inode)->field)) \ 813 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 814 (einode)->i_extra_isize)) \ 815 816 /* 817 * We use an encoding that preserves the times for extra epoch "00": 818 * 819 * extra msb of adjust for signed 820 * epoch 32-bit 32-bit tv_sec to 821 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 822 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 823 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 824 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 825 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 826 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 827 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 828 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 829 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 830 * 831 * Note that previous versions of the kernel on 64-bit systems would 832 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 833 * 1970. e2fsck will correct this, assuming that it is run on the 834 * affected filesystem before 2242. 835 */ 836 837 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 838 { 839 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 840 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 841 } 842 843 static inline void ext4_decode_extra_time(struct timespec64 *time, 844 __le32 extra) 845 { 846 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 847 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 848 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 849 } 850 851 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 852 do { \ 853 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 854 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 855 (raw_inode)->xtime ## _extra = \ 856 ext4_encode_extra_time(&(inode)->xtime); \ 857 } \ 858 else \ 859 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 860 } while (0) 861 862 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 863 do { \ 864 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 865 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 866 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 867 (raw_inode)->xtime ## _extra = \ 868 ext4_encode_extra_time(&(einode)->xtime); \ 869 } while (0) 870 871 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 872 do { \ 873 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 874 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 875 ext4_decode_extra_time(&(inode)->xtime, \ 876 raw_inode->xtime ## _extra); \ 877 } \ 878 else \ 879 (inode)->xtime.tv_nsec = 0; \ 880 } while (0) 881 882 883 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 884 do { \ 885 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 886 (einode)->xtime.tv_sec = \ 887 (signed)le32_to_cpu((raw_inode)->xtime); \ 888 else \ 889 (einode)->xtime.tv_sec = 0; \ 890 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 891 ext4_decode_extra_time(&(einode)->xtime, \ 892 raw_inode->xtime ## _extra); \ 893 else \ 894 (einode)->xtime.tv_nsec = 0; \ 895 } while (0) 896 897 #define i_disk_version osd1.linux1.l_i_version 898 899 #if defined(__KERNEL__) || defined(__linux__) 900 #define i_reserved1 osd1.linux1.l_i_reserved1 901 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 902 #define i_blocks_high osd2.linux2.l_i_blocks_high 903 #define i_uid_low i_uid 904 #define i_gid_low i_gid 905 #define i_uid_high osd2.linux2.l_i_uid_high 906 #define i_gid_high osd2.linux2.l_i_gid_high 907 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 908 909 #elif defined(__GNU__) 910 911 #define i_translator osd1.hurd1.h_i_translator 912 #define i_uid_high osd2.hurd2.h_i_uid_high 913 #define i_gid_high osd2.hurd2.h_i_gid_high 914 #define i_author osd2.hurd2.h_i_author 915 916 #elif defined(__masix__) 917 918 #define i_reserved1 osd1.masix1.m_i_reserved1 919 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 920 #define i_reserved2 osd2.masix2.m_i_reserved2 921 922 #endif /* defined(__KERNEL__) || defined(__linux__) */ 923 924 #include "extents_status.h" 925 #include "fast_commit.h" 926 927 /* 928 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 929 * 930 * These are needed to avoid lockdep false positives when we need to 931 * allocate blocks to the quota inode during ext4_map_blocks(), while 932 * holding i_data_sem for a normal (non-quota) inode. Since we don't 933 * do quota tracking for the quota inode, this avoids deadlock (as 934 * well as infinite recursion, since it isn't turtles all the way 935 * down...) 936 * 937 * I_DATA_SEM_NORMAL - Used for most inodes 938 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 939 * where the second inode has larger inode number 940 * than the first 941 * I_DATA_SEM_QUOTA - Used for quota inodes only 942 * I_DATA_SEM_EA - Used for ea_inodes only 943 */ 944 enum { 945 I_DATA_SEM_NORMAL = 0, 946 I_DATA_SEM_OTHER, 947 I_DATA_SEM_QUOTA, 948 I_DATA_SEM_EA 949 }; 950 951 952 /* 953 * fourth extended file system inode data in memory 954 */ 955 struct ext4_inode_info { 956 __le32 i_data[15]; /* unconverted */ 957 __u32 i_dtime; 958 ext4_fsblk_t i_file_acl; 959 960 /* 961 * i_block_group is the number of the block group which contains 962 * this file's inode. Constant across the lifetime of the inode, 963 * it is used for making block allocation decisions - we try to 964 * place a file's data blocks near its inode block, and new inodes 965 * near to their parent directory's inode. 966 */ 967 ext4_group_t i_block_group; 968 ext4_lblk_t i_dir_start_lookup; 969 #if (BITS_PER_LONG < 64) 970 unsigned long i_state_flags; /* Dynamic state flags */ 971 #endif 972 unsigned long i_flags; 973 974 /* 975 * Extended attributes can be read independently of the main file 976 * data. Taking i_rwsem even when reading would cause contention 977 * between readers of EAs and writers of regular file data, so 978 * instead we synchronize on xattr_sem when reading or changing 979 * EAs. 980 */ 981 struct rw_semaphore xattr_sem; 982 983 /* 984 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 985 * i_orphan is used. 986 */ 987 union { 988 struct list_head i_orphan; /* unlinked but open inodes */ 989 unsigned int i_orphan_idx; /* Index in orphan file */ 990 }; 991 992 /* Fast commit related info */ 993 994 /* For tracking dentry create updates */ 995 struct list_head i_fc_dilist; 996 struct list_head i_fc_list; /* 997 * inodes that need fast commit 998 * protected by sbi->s_fc_lock. 999 */ 1000 1001 /* Start of lblk range that needs to be committed in this fast commit */ 1002 ext4_lblk_t i_fc_lblk_start; 1003 1004 /* End of lblk range that needs to be committed in this fast commit */ 1005 ext4_lblk_t i_fc_lblk_len; 1006 1007 /* Number of ongoing updates on this inode */ 1008 atomic_t i_fc_updates; 1009 1010 /* Fast commit wait queue for this inode */ 1011 wait_queue_head_t i_fc_wait; 1012 1013 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1014 struct mutex i_fc_lock; 1015 1016 /* 1017 * i_disksize keeps track of what the inode size is ON DISK, not 1018 * in memory. During truncate, i_size is set to the new size by 1019 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1020 * set i_disksize to 0 until the truncate is actually under way. 1021 * 1022 * The intent is that i_disksize always represents the blocks which 1023 * are used by this file. This allows recovery to restart truncate 1024 * on orphans if we crash during truncate. We actually write i_disksize 1025 * into the on-disk inode when writing inodes out, instead of i_size. 1026 * 1027 * The only time when i_disksize and i_size may be different is when 1028 * a truncate is in progress. The only things which change i_disksize 1029 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1030 */ 1031 loff_t i_disksize; 1032 1033 /* 1034 * i_data_sem is for serialising ext4_truncate() against 1035 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1036 * data tree are chopped off during truncate. We can't do that in 1037 * ext4 because whenever we perform intermediate commits during 1038 * truncate, the inode and all the metadata blocks *must* be in a 1039 * consistent state which allows truncation of the orphans to restart 1040 * during recovery. Hence we must fix the get_block-vs-truncate race 1041 * by other means, so we have i_data_sem. 1042 */ 1043 struct rw_semaphore i_data_sem; 1044 struct inode vfs_inode; 1045 struct jbd2_inode *jinode; 1046 1047 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1048 1049 /* 1050 * File creation time. Its function is same as that of 1051 * struct timespec64 i_{a,c,m}time in the generic inode. 1052 */ 1053 struct timespec64 i_crtime; 1054 1055 /* mballoc */ 1056 atomic_t i_prealloc_active; 1057 struct rb_root i_prealloc_node; 1058 rwlock_t i_prealloc_lock; 1059 1060 /* extents status tree */ 1061 struct ext4_es_tree i_es_tree; 1062 rwlock_t i_es_lock; 1063 struct list_head i_es_list; 1064 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1065 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1066 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1067 extents to shrink. Protected by 1068 i_es_lock */ 1069 1070 /* ialloc */ 1071 ext4_group_t i_last_alloc_group; 1072 1073 /* allocation reservation info for delalloc */ 1074 /* In case of bigalloc, this refer to clusters rather than blocks */ 1075 unsigned int i_reserved_data_blocks; 1076 1077 /* pending cluster reservations for bigalloc file systems */ 1078 struct ext4_pending_tree i_pending_tree; 1079 1080 /* on-disk additional length */ 1081 __u16 i_extra_isize; 1082 1083 /* Indicate the inline data space. */ 1084 u16 i_inline_off; 1085 u16 i_inline_size; 1086 1087 #ifdef CONFIG_QUOTA 1088 /* quota space reservation, managed internally by quota code */ 1089 qsize_t i_reserved_quota; 1090 #endif 1091 1092 /* Lock protecting lists below */ 1093 spinlock_t i_completed_io_lock; 1094 /* 1095 * Completed IOs that need unwritten extents handling and have 1096 * transaction reserved 1097 */ 1098 struct list_head i_rsv_conversion_list; 1099 struct work_struct i_rsv_conversion_work; 1100 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1101 1102 spinlock_t i_block_reservation_lock; 1103 1104 /* 1105 * Transactions that contain inode's metadata needed to complete 1106 * fsync and fdatasync, respectively. 1107 */ 1108 tid_t i_sync_tid; 1109 tid_t i_datasync_tid; 1110 1111 #ifdef CONFIG_QUOTA 1112 struct dquot *i_dquot[MAXQUOTAS]; 1113 #endif 1114 1115 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1116 __u32 i_csum_seed; 1117 1118 kprojid_t i_projid; 1119 }; 1120 1121 /* 1122 * File system states 1123 */ 1124 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1125 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1126 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1127 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1128 1129 /* 1130 * Misc. filesystem flags 1131 */ 1132 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1133 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1134 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1135 1136 /* 1137 * Mount flags set via mount options or defaults 1138 */ 1139 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1140 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1141 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1142 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1143 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1144 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1145 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1146 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1147 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1148 #ifdef CONFIG_FS_DAX 1149 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1150 #else 1151 #define EXT4_MOUNT_DAX_ALWAYS 0 1152 #endif 1153 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1154 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1155 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1156 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1157 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1158 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1159 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1160 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1161 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1162 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1163 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1164 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1165 * enable enforcement for hidden 1166 * quota files */ 1167 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1168 * enforcement for hidden quota 1169 * files */ 1170 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1171 * enforcement */ 1172 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1173 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1174 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1175 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1176 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1177 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1178 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1179 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1180 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1181 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1182 1183 /* 1184 * Mount flags set either automatically (could not be set by mount option) 1185 * based on per file system feature or property or in special cases such as 1186 * distinguishing between explicit mount option definition and default. 1187 */ 1188 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1189 specified delalloc */ 1190 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1191 size of blocksize * 8 1192 blocks */ 1193 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1194 file systems */ 1195 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1196 specified journal checksum */ 1197 1198 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1199 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1200 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1201 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1202 * scanning in mballoc 1203 */ 1204 1205 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1206 ~EXT4_MOUNT_##opt 1207 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1208 EXT4_MOUNT_##opt 1209 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1210 EXT4_MOUNT_##opt) 1211 1212 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1213 ~EXT4_MOUNT2_##opt 1214 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1215 EXT4_MOUNT2_##opt 1216 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1217 EXT4_MOUNT2_##opt) 1218 1219 #define ext4_test_and_set_bit __test_and_set_bit_le 1220 #define ext4_set_bit __set_bit_le 1221 #define ext4_set_bit_atomic ext2_set_bit_atomic 1222 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1223 #define ext4_clear_bit __clear_bit_le 1224 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1225 #define ext4_test_bit test_bit_le 1226 #define ext4_find_next_zero_bit find_next_zero_bit_le 1227 #define ext4_find_next_bit find_next_bit_le 1228 1229 extern void mb_set_bits(void *bm, int cur, int len); 1230 1231 /* 1232 * Maximal mount counts between two filesystem checks 1233 */ 1234 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1235 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1236 1237 /* 1238 * Behaviour when detecting errors 1239 */ 1240 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1241 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1242 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1243 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1244 1245 /* Metadata checksum algorithm codes */ 1246 #define EXT4_CRC32C_CHKSUM 1 1247 1248 #define EXT4_LABEL_MAX 16 1249 1250 /* 1251 * Structure of the super block 1252 */ 1253 struct ext4_super_block { 1254 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1255 __le32 s_blocks_count_lo; /* Blocks count */ 1256 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1257 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1258 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1259 __le32 s_first_data_block; /* First Data Block */ 1260 __le32 s_log_block_size; /* Block size */ 1261 __le32 s_log_cluster_size; /* Allocation cluster size */ 1262 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1263 __le32 s_clusters_per_group; /* # Clusters per group */ 1264 __le32 s_inodes_per_group; /* # Inodes per group */ 1265 __le32 s_mtime; /* Mount time */ 1266 /*30*/ __le32 s_wtime; /* Write time */ 1267 __le16 s_mnt_count; /* Mount count */ 1268 __le16 s_max_mnt_count; /* Maximal mount count */ 1269 __le16 s_magic; /* Magic signature */ 1270 __le16 s_state; /* File system state */ 1271 __le16 s_errors; /* Behaviour when detecting errors */ 1272 __le16 s_minor_rev_level; /* minor revision level */ 1273 /*40*/ __le32 s_lastcheck; /* time of last check */ 1274 __le32 s_checkinterval; /* max. time between checks */ 1275 __le32 s_creator_os; /* OS */ 1276 __le32 s_rev_level; /* Revision level */ 1277 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1278 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1279 /* 1280 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1281 * 1282 * Note: the difference between the compatible feature set and 1283 * the incompatible feature set is that if there is a bit set 1284 * in the incompatible feature set that the kernel doesn't 1285 * know about, it should refuse to mount the filesystem. 1286 * 1287 * e2fsck's requirements are more strict; if it doesn't know 1288 * about a feature in either the compatible or incompatible 1289 * feature set, it must abort and not try to meddle with 1290 * things it doesn't understand... 1291 */ 1292 __le32 s_first_ino; /* First non-reserved inode */ 1293 __le16 s_inode_size; /* size of inode structure */ 1294 __le16 s_block_group_nr; /* block group # of this superblock */ 1295 __le32 s_feature_compat; /* compatible feature set */ 1296 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1297 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1298 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1299 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1300 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1301 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1302 /* 1303 * Performance hints. Directory preallocation should only 1304 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1305 */ 1306 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1307 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1308 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1309 /* 1310 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1311 */ 1312 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1313 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1314 __le32 s_journal_dev; /* device number of journal file */ 1315 __le32 s_last_orphan; /* start of list of inodes to delete */ 1316 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1317 __u8 s_def_hash_version; /* Default hash version to use */ 1318 __u8 s_jnl_backup_type; 1319 __le16 s_desc_size; /* size of group descriptor */ 1320 /*100*/ __le32 s_default_mount_opts; 1321 __le32 s_first_meta_bg; /* First metablock block group */ 1322 __le32 s_mkfs_time; /* When the filesystem was created */ 1323 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1324 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1325 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1326 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1327 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1328 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1329 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1330 __le32 s_flags; /* Miscellaneous flags */ 1331 __le16 s_raid_stride; /* RAID stride */ 1332 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1333 __le64 s_mmp_block; /* Block for multi-mount protection */ 1334 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1335 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1336 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1337 __u8 s_encryption_level; /* versioning level for encryption */ 1338 __u8 s_reserved_pad; /* Padding to next 32bits */ 1339 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1340 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1341 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1342 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1343 snapshot's future use */ 1344 __le32 s_snapshot_list; /* inode number of the head of the 1345 on-disk snapshot list */ 1346 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1347 __le32 s_error_count; /* number of fs errors */ 1348 __le32 s_first_error_time; /* first time an error happened */ 1349 __le32 s_first_error_ino; /* inode involved in first error */ 1350 __le64 s_first_error_block; /* block involved of first error */ 1351 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1352 __le32 s_first_error_line; /* line number where error happened */ 1353 __le32 s_last_error_time; /* most recent time of an error */ 1354 __le32 s_last_error_ino; /* inode involved in last error */ 1355 __le32 s_last_error_line; /* line number where error happened */ 1356 __le64 s_last_error_block; /* block involved of last error */ 1357 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1358 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1359 __u8 s_mount_opts[64]; 1360 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1361 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1362 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1363 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1364 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1365 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1366 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1367 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1368 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1369 __u8 s_wtime_hi; 1370 __u8 s_mtime_hi; 1371 __u8 s_mkfs_time_hi; 1372 __u8 s_lastcheck_hi; 1373 __u8 s_first_error_time_hi; 1374 __u8 s_last_error_time_hi; 1375 __u8 s_first_error_errcode; 1376 __u8 s_last_error_errcode; 1377 __le16 s_encoding; /* Filename charset encoding */ 1378 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1379 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1380 __le32 s_reserved[94]; /* Padding to the end of the block */ 1381 __le32 s_checksum; /* crc32c(superblock) */ 1382 }; 1383 1384 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1385 1386 #ifdef __KERNEL__ 1387 1388 /* Number of quota types we support */ 1389 #define EXT4_MAXQUOTAS 3 1390 1391 #define EXT4_ENC_UTF8_12_1 1 1392 1393 /* Types of ext4 journal triggers */ 1394 enum ext4_journal_trigger_type { 1395 EXT4_JTR_ORPHAN_FILE, 1396 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1397 }; 1398 1399 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1400 1401 struct ext4_journal_trigger { 1402 struct jbd2_buffer_trigger_type tr_triggers; 1403 struct super_block *sb; 1404 }; 1405 1406 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1407 struct jbd2_buffer_trigger_type *trigger) 1408 { 1409 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1410 } 1411 1412 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1413 1414 /* Structure at the tail of orphan block */ 1415 struct ext4_orphan_block_tail { 1416 __le32 ob_magic; 1417 __le32 ob_checksum; 1418 }; 1419 1420 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1421 { 1422 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1423 sizeof(u32); 1424 } 1425 1426 struct ext4_orphan_block { 1427 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1428 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1429 }; 1430 1431 /* 1432 * Info about orphan file. 1433 */ 1434 struct ext4_orphan_info { 1435 int of_blocks; /* Number of orphan blocks in a file */ 1436 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1437 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1438 * file blocks */ 1439 }; 1440 1441 /* 1442 * fourth extended-fs super-block data in memory 1443 */ 1444 struct ext4_sb_info { 1445 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1446 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1447 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1448 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1449 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1450 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1451 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1452 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1453 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1454 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1455 unsigned long s_overhead; /* # of fs overhead clusters */ 1456 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1457 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1458 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1459 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1460 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1461 struct buffer_head * __rcu *s_group_desc; 1462 unsigned int s_mount_opt; 1463 unsigned int s_mount_opt2; 1464 unsigned long s_mount_flags; 1465 unsigned int s_def_mount_opt; 1466 unsigned int s_def_mount_opt2; 1467 ext4_fsblk_t s_sb_block; 1468 atomic64_t s_resv_clusters; 1469 kuid_t s_resuid; 1470 kgid_t s_resgid; 1471 unsigned short s_mount_state; 1472 unsigned short s_pad; 1473 int s_addr_per_block_bits; 1474 int s_desc_per_block_bits; 1475 int s_inode_size; 1476 int s_first_ino; 1477 unsigned int s_inode_readahead_blks; 1478 unsigned int s_inode_goal; 1479 u32 s_hash_seed[4]; 1480 int s_def_hash_version; 1481 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1482 struct percpu_counter s_freeclusters_counter; 1483 struct percpu_counter s_freeinodes_counter; 1484 struct percpu_counter s_dirs_counter; 1485 struct percpu_counter s_dirtyclusters_counter; 1486 struct percpu_counter s_sra_exceeded_retry_limit; 1487 struct blockgroup_lock *s_blockgroup_lock; 1488 struct proc_dir_entry *s_proc; 1489 struct kobject s_kobj; 1490 struct completion s_kobj_unregister; 1491 struct super_block *s_sb; 1492 struct buffer_head *s_mmp_bh; 1493 1494 /* Journaling */ 1495 struct journal_s *s_journal; 1496 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1497 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1498 struct list_head s_orphan; /* List of orphaned inodes in on disk 1499 list */ 1500 struct ext4_orphan_info s_orphan_info; 1501 unsigned long s_commit_interval; 1502 u32 s_max_batch_time; 1503 u32 s_min_batch_time; 1504 struct block_device *s_journal_bdev; 1505 #ifdef CONFIG_QUOTA 1506 /* Names of quota files with journalled quota */ 1507 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1508 int s_jquota_fmt; /* Format of quota to use */ 1509 #endif 1510 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1511 struct ext4_system_blocks __rcu *s_system_blks; 1512 1513 #ifdef EXTENTS_STATS 1514 /* ext4 extents stats */ 1515 unsigned long s_ext_min; 1516 unsigned long s_ext_max; 1517 unsigned long s_depth_max; 1518 spinlock_t s_ext_stats_lock; 1519 unsigned long s_ext_blocks; 1520 unsigned long s_ext_extents; 1521 #endif 1522 1523 /* for buddy allocator */ 1524 struct ext4_group_info ** __rcu *s_group_info; 1525 struct inode *s_buddy_cache; 1526 spinlock_t s_md_lock; 1527 unsigned short *s_mb_offsets; 1528 unsigned int *s_mb_maxs; 1529 unsigned int s_group_info_size; 1530 unsigned int s_mb_free_pending; 1531 struct list_head s_freed_data_list; /* List of blocks to be freed 1532 after commit completed */ 1533 struct list_head s_discard_list; 1534 struct work_struct s_discard_work; 1535 atomic_t s_retry_alloc_pending; 1536 struct list_head *s_mb_avg_fragment_size; 1537 rwlock_t *s_mb_avg_fragment_size_locks; 1538 struct list_head *s_mb_largest_free_orders; 1539 rwlock_t *s_mb_largest_free_orders_locks; 1540 1541 /* tunables */ 1542 unsigned long s_stripe; 1543 unsigned int s_mb_max_linear_groups; 1544 unsigned int s_mb_stream_request; 1545 unsigned int s_mb_max_to_scan; 1546 unsigned int s_mb_min_to_scan; 1547 unsigned int s_mb_stats; 1548 unsigned int s_mb_order2_reqs; 1549 unsigned int s_mb_group_prealloc; 1550 unsigned int s_max_dir_size_kb; 1551 /* where last allocation was done - for stream allocation */ 1552 unsigned long s_mb_last_group; 1553 unsigned long s_mb_last_start; 1554 unsigned int s_mb_prefetch; 1555 unsigned int s_mb_prefetch_limit; 1556 unsigned int s_mb_cr1_5_max_trim_order; 1557 1558 /* stats for buddy allocator */ 1559 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1560 atomic_t s_bal_success; /* we found long enough chunks */ 1561 atomic_t s_bal_allocated; /* in blocks */ 1562 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1563 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1564 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1565 atomic_t s_bal_goals; /* goal hits */ 1566 atomic_t s_bal_len_goals; /* len goal hits */ 1567 atomic_t s_bal_breaks; /* too long searches */ 1568 atomic_t s_bal_2orders; /* 2^order hits */ 1569 atomic_t s_bal_cr0_bad_suggestions; 1570 atomic_t s_bal_cr1_bad_suggestions; 1571 atomic_t s_bal_cr1_5_bad_suggestions; 1572 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1573 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1574 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1575 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1576 atomic64_t s_mb_generation_time; 1577 atomic_t s_mb_lost_chunks; 1578 atomic_t s_mb_preallocated; 1579 atomic_t s_mb_discarded; 1580 atomic_t s_lock_busy; 1581 1582 /* locality groups */ 1583 struct ext4_locality_group __percpu *s_locality_groups; 1584 1585 /* for write statistics */ 1586 unsigned long s_sectors_written_start; 1587 u64 s_kbytes_written; 1588 1589 /* the size of zero-out chunk */ 1590 unsigned int s_extent_max_zeroout_kb; 1591 1592 unsigned int s_log_groups_per_flex; 1593 struct flex_groups * __rcu *s_flex_groups; 1594 ext4_group_t s_flex_groups_allocated; 1595 1596 /* workqueue for reserved extent conversions (buffered io) */ 1597 struct workqueue_struct *rsv_conversion_wq; 1598 1599 /* timer for periodic error stats printing */ 1600 struct timer_list s_err_report; 1601 1602 /* Lazy inode table initialization info */ 1603 struct ext4_li_request *s_li_request; 1604 /* Wait multiplier for lazy initialization thread */ 1605 unsigned int s_li_wait_mult; 1606 1607 /* Kernel thread for multiple mount protection */ 1608 struct task_struct *s_mmp_tsk; 1609 1610 /* record the last minlen when FITRIM is called. */ 1611 unsigned long s_last_trim_minblks; 1612 1613 /* Reference to checksum algorithm driver via cryptoapi */ 1614 struct crypto_shash *s_chksum_driver; 1615 1616 /* Precomputed FS UUID checksum for seeding other checksums */ 1617 __u32 s_csum_seed; 1618 1619 /* Reclaim extents from extent status tree */ 1620 struct shrinker s_es_shrinker; 1621 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1622 long s_es_nr_inode; 1623 struct ext4_es_stats s_es_stats; 1624 struct mb_cache *s_ea_block_cache; 1625 struct mb_cache *s_ea_inode_cache; 1626 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1627 1628 /* Journal triggers for checksum computation */ 1629 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1630 1631 /* Ratelimit ext4 messages. */ 1632 struct ratelimit_state s_err_ratelimit_state; 1633 struct ratelimit_state s_warning_ratelimit_state; 1634 struct ratelimit_state s_msg_ratelimit_state; 1635 atomic_t s_warning_count; 1636 atomic_t s_msg_count; 1637 1638 /* Encryption policy for '-o test_dummy_encryption' */ 1639 struct fscrypt_dummy_policy s_dummy_enc_policy; 1640 1641 /* 1642 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1643 * or EXTENTS flag. 1644 */ 1645 struct percpu_rw_semaphore s_writepages_rwsem; 1646 struct dax_device *s_daxdev; 1647 u64 s_dax_part_off; 1648 #ifdef CONFIG_EXT4_DEBUG 1649 unsigned long s_simulate_fail; 1650 #endif 1651 /* Record the errseq of the backing block device */ 1652 errseq_t s_bdev_wb_err; 1653 spinlock_t s_bdev_wb_lock; 1654 1655 /* Information about errors that happened during this mount */ 1656 spinlock_t s_error_lock; 1657 int s_add_error_count; 1658 int s_first_error_code; 1659 __u32 s_first_error_line; 1660 __u32 s_first_error_ino; 1661 __u64 s_first_error_block; 1662 const char *s_first_error_func; 1663 time64_t s_first_error_time; 1664 int s_last_error_code; 1665 __u32 s_last_error_line; 1666 __u32 s_last_error_ino; 1667 __u64 s_last_error_block; 1668 const char *s_last_error_func; 1669 time64_t s_last_error_time; 1670 /* 1671 * If we are in a context where we cannot update error information in 1672 * the on-disk superblock, we queue this work to do it. 1673 */ 1674 struct work_struct s_error_work; 1675 1676 /* Ext4 fast commit sub transaction ID */ 1677 atomic_t s_fc_subtid; 1678 1679 /* 1680 * After commit starts, the main queue gets locked, and the further 1681 * updates get added in the staging queue. 1682 */ 1683 #define FC_Q_MAIN 0 1684 #define FC_Q_STAGING 1 1685 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1686 * that have data changes in them. 1687 */ 1688 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1689 unsigned int s_fc_bytes; 1690 /* 1691 * Main fast commit lock. This lock protects accesses to the 1692 * following fields: 1693 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1694 */ 1695 spinlock_t s_fc_lock; 1696 struct buffer_head *s_fc_bh; 1697 struct ext4_fc_stats s_fc_stats; 1698 tid_t s_fc_ineligible_tid; 1699 #ifdef CONFIG_EXT4_DEBUG 1700 int s_fc_debug_max_replay; 1701 #endif 1702 struct ext4_fc_replay_state s_fc_replay_state; 1703 }; 1704 1705 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1706 { 1707 return sb->s_fs_info; 1708 } 1709 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1710 { 1711 return container_of(inode, struct ext4_inode_info, vfs_inode); 1712 } 1713 1714 static inline int ext4_writepages_down_read(struct super_block *sb) 1715 { 1716 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1717 return memalloc_nofs_save(); 1718 } 1719 1720 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1721 { 1722 memalloc_nofs_restore(ctx); 1723 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1724 } 1725 1726 static inline int ext4_writepages_down_write(struct super_block *sb) 1727 { 1728 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1729 return memalloc_nofs_save(); 1730 } 1731 1732 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1733 { 1734 memalloc_nofs_restore(ctx); 1735 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1736 } 1737 1738 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1739 { 1740 return ino == EXT4_ROOT_INO || 1741 (ino >= EXT4_FIRST_INO(sb) && 1742 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1743 } 1744 1745 /* 1746 * Returns: sbi->field[index] 1747 * Used to access an array element from the following sbi fields which require 1748 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1749 * - s_group_desc 1750 * - s_group_info 1751 * - s_flex_group 1752 */ 1753 #define sbi_array_rcu_deref(sbi, field, index) \ 1754 ({ \ 1755 typeof(*((sbi)->field)) _v; \ 1756 rcu_read_lock(); \ 1757 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1758 rcu_read_unlock(); \ 1759 _v; \ 1760 }) 1761 1762 /* 1763 * run-time mount flags 1764 */ 1765 enum { 1766 EXT4_MF_MNTDIR_SAMPLED, 1767 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1768 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1769 }; 1770 1771 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1772 { 1773 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1774 } 1775 1776 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1777 { 1778 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1779 } 1780 1781 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1782 { 1783 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1784 } 1785 1786 1787 /* 1788 * Simulate_fail codes 1789 */ 1790 #define EXT4_SIM_BBITMAP_EIO 1 1791 #define EXT4_SIM_BBITMAP_CRC 2 1792 #define EXT4_SIM_IBITMAP_EIO 3 1793 #define EXT4_SIM_IBITMAP_CRC 4 1794 #define EXT4_SIM_INODE_EIO 5 1795 #define EXT4_SIM_INODE_CRC 6 1796 #define EXT4_SIM_DIRBLOCK_EIO 7 1797 #define EXT4_SIM_DIRBLOCK_CRC 8 1798 1799 static inline bool ext4_simulate_fail(struct super_block *sb, 1800 unsigned long code) 1801 { 1802 #ifdef CONFIG_EXT4_DEBUG 1803 struct ext4_sb_info *sbi = EXT4_SB(sb); 1804 1805 if (unlikely(sbi->s_simulate_fail == code)) { 1806 sbi->s_simulate_fail = 0; 1807 return true; 1808 } 1809 #endif 1810 return false; 1811 } 1812 1813 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1814 struct buffer_head *bh, 1815 unsigned long code) 1816 { 1817 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1818 clear_buffer_uptodate(bh); 1819 } 1820 1821 /* 1822 * Error number codes for s_{first,last}_error_errno 1823 * 1824 * Linux errno numbers are architecture specific, so we need to translate 1825 * them into something which is architecture independent. We don't define 1826 * codes for all errno's; just the ones which are most likely to be the cause 1827 * of an ext4_error() call. 1828 */ 1829 #define EXT4_ERR_UNKNOWN 1 1830 #define EXT4_ERR_EIO 2 1831 #define EXT4_ERR_ENOMEM 3 1832 #define EXT4_ERR_EFSBADCRC 4 1833 #define EXT4_ERR_EFSCORRUPTED 5 1834 #define EXT4_ERR_ENOSPC 6 1835 #define EXT4_ERR_ENOKEY 7 1836 #define EXT4_ERR_EROFS 8 1837 #define EXT4_ERR_EFBIG 9 1838 #define EXT4_ERR_EEXIST 10 1839 #define EXT4_ERR_ERANGE 11 1840 #define EXT4_ERR_EOVERFLOW 12 1841 #define EXT4_ERR_EBUSY 13 1842 #define EXT4_ERR_ENOTDIR 14 1843 #define EXT4_ERR_ENOTEMPTY 15 1844 #define EXT4_ERR_ESHUTDOWN 16 1845 #define EXT4_ERR_EFAULT 17 1846 1847 /* 1848 * Inode dynamic state flags 1849 */ 1850 enum { 1851 EXT4_STATE_NEW, /* inode is newly created */ 1852 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1853 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1854 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1855 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1856 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1857 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1858 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1859 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1860 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1861 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1862 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1863 }; 1864 1865 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1866 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1867 { \ 1868 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1869 } \ 1870 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1871 { \ 1872 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1873 } \ 1874 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1875 { \ 1876 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1877 } 1878 1879 /* Add these declarations here only so that these functions can be 1880 * found by name. Otherwise, they are very hard to locate. */ 1881 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1882 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1883 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1884 EXT4_INODE_BIT_FNS(flag, flags, 0) 1885 1886 /* Add these declarations here only so that these functions can be 1887 * found by name. Otherwise, they are very hard to locate. */ 1888 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1889 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1890 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1891 #if (BITS_PER_LONG < 64) 1892 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1893 1894 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1895 { 1896 (ei)->i_state_flags = 0; 1897 } 1898 #else 1899 EXT4_INODE_BIT_FNS(state, flags, 32) 1900 1901 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1902 { 1903 /* We depend on the fact that callers will set i_flags */ 1904 } 1905 #endif 1906 #else 1907 /* Assume that user mode programs are passing in an ext4fs superblock, not 1908 * a kernel struct super_block. This will allow us to call the feature-test 1909 * macros from user land. */ 1910 #define EXT4_SB(sb) (sb) 1911 #endif 1912 1913 static inline bool ext4_verity_in_progress(struct inode *inode) 1914 { 1915 return IS_ENABLED(CONFIG_FS_VERITY) && 1916 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1917 } 1918 1919 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1920 1921 /* 1922 * Codes for operating systems 1923 */ 1924 #define EXT4_OS_LINUX 0 1925 #define EXT4_OS_HURD 1 1926 #define EXT4_OS_MASIX 2 1927 #define EXT4_OS_FREEBSD 3 1928 #define EXT4_OS_LITES 4 1929 1930 /* 1931 * Revision levels 1932 */ 1933 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1934 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1935 1936 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1937 1938 #define EXT4_GOOD_OLD_INODE_SIZE 128 1939 1940 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1941 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1942 #define EXT4_TIMESTAMP_MIN S32_MIN 1943 1944 /* 1945 * Feature set definitions 1946 */ 1947 1948 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1949 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1950 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1951 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1952 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1953 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1954 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1955 /* 1956 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1957 * incompatible only if fast commit blocks are present in the FS. Since we 1958 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1959 * incompatible. We also have a JBD2 incompat feature, which gets set when 1960 * there are fast commit blocks present in the journal. 1961 */ 1962 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1963 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1964 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 1965 1966 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1967 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1968 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1969 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1970 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1971 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1972 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1973 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1974 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1975 /* 1976 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1977 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1978 * all other data structures' checksums. However, the METADATA_CSUM and 1979 * GDT_CSUM bits are mutually exclusive. 1980 */ 1981 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1982 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1983 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1984 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1985 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 1986 non-empty */ 1987 1988 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1989 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1990 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1991 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1992 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1993 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1994 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1995 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1996 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1997 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1998 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1999 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2000 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2001 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2002 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2003 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2004 2005 extern void ext4_update_dynamic_rev(struct super_block *sb); 2006 2007 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2008 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2009 { \ 2010 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2011 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2012 } \ 2013 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2014 { \ 2015 ext4_update_dynamic_rev(sb); \ 2016 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2017 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2018 } \ 2019 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2020 { \ 2021 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2022 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2023 } 2024 2025 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2026 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2027 { \ 2028 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2029 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2030 } \ 2031 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2032 { \ 2033 ext4_update_dynamic_rev(sb); \ 2034 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2035 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2036 } \ 2037 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2038 { \ 2039 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2040 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2041 } 2042 2043 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2044 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2045 { \ 2046 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2047 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2048 } \ 2049 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2050 { \ 2051 ext4_update_dynamic_rev(sb); \ 2052 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2053 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2054 } \ 2055 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2056 { \ 2057 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2058 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2059 } 2060 2061 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2062 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2063 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2064 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2065 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2066 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2067 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2068 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2069 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2070 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2071 2072 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2073 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2074 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2075 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2076 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2077 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2078 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2079 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2080 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2081 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2082 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2083 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2084 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2085 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2086 2087 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2088 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2089 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2090 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2091 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2092 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2093 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2094 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2095 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2096 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2097 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2098 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2099 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2100 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2101 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2102 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2103 2104 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2105 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2106 EXT4_FEATURE_INCOMPAT_META_BG) 2107 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2108 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2109 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2110 2111 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2112 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2113 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2114 EXT4_FEATURE_INCOMPAT_META_BG) 2115 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2116 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2117 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2118 2119 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2120 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2121 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2122 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2123 EXT4_FEATURE_INCOMPAT_META_BG| \ 2124 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2125 EXT4_FEATURE_INCOMPAT_64BIT| \ 2126 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2127 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2128 EXT4_FEATURE_INCOMPAT_MMP | \ 2129 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2130 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2131 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2132 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2133 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2134 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2135 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2136 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2137 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2138 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2139 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2140 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2141 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2142 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2143 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2144 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2145 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2146 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2147 2148 #define EXTN_FEATURE_FUNCS(ver) \ 2149 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2150 { \ 2151 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2152 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2153 } \ 2154 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2155 { \ 2156 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2157 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2158 } \ 2159 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2160 { \ 2161 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2162 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2163 } 2164 2165 EXTN_FEATURE_FUNCS(2) 2166 EXTN_FEATURE_FUNCS(3) 2167 EXTN_FEATURE_FUNCS(4) 2168 2169 static inline bool ext4_has_compat_features(struct super_block *sb) 2170 { 2171 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2172 } 2173 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2174 { 2175 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2176 } 2177 static inline bool ext4_has_incompat_features(struct super_block *sb) 2178 { 2179 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2180 } 2181 2182 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2183 2184 /* 2185 * Superblock flags 2186 */ 2187 #define EXT4_FLAGS_RESIZING 0 2188 #define EXT4_FLAGS_SHUTDOWN 1 2189 #define EXT4_FLAGS_BDEV_IS_DAX 2 2190 2191 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2192 { 2193 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2194 } 2195 2196 /* 2197 * Default values for user and/or group using reserved blocks 2198 */ 2199 #define EXT4_DEF_RESUID 0 2200 #define EXT4_DEF_RESGID 0 2201 2202 /* 2203 * Default project ID 2204 */ 2205 #define EXT4_DEF_PROJID 0 2206 2207 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2208 2209 /* 2210 * Default mount options 2211 */ 2212 #define EXT4_DEFM_DEBUG 0x0001 2213 #define EXT4_DEFM_BSDGROUPS 0x0002 2214 #define EXT4_DEFM_XATTR_USER 0x0004 2215 #define EXT4_DEFM_ACL 0x0008 2216 #define EXT4_DEFM_UID16 0x0010 2217 #define EXT4_DEFM_JMODE 0x0060 2218 #define EXT4_DEFM_JMODE_DATA 0x0020 2219 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2220 #define EXT4_DEFM_JMODE_WBACK 0x0060 2221 #define EXT4_DEFM_NOBARRIER 0x0100 2222 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2223 #define EXT4_DEFM_DISCARD 0x0400 2224 #define EXT4_DEFM_NODELALLOC 0x0800 2225 2226 /* 2227 * Default journal batch times 2228 */ 2229 #define EXT4_DEF_MIN_BATCH_TIME 0 2230 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2231 2232 /* 2233 * Minimum number of groups in a flexgroup before we separate out 2234 * directories into the first block group of a flexgroup 2235 */ 2236 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2237 2238 /* 2239 * Structure of a directory entry 2240 */ 2241 #define EXT4_NAME_LEN 255 2242 /* 2243 * Base length of the ext4 directory entry excluding the name length 2244 */ 2245 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2246 2247 struct ext4_dir_entry { 2248 __le32 inode; /* Inode number */ 2249 __le16 rec_len; /* Directory entry length */ 2250 __le16 name_len; /* Name length */ 2251 char name[EXT4_NAME_LEN]; /* File name */ 2252 }; 2253 2254 2255 /* 2256 * Encrypted Casefolded entries require saving the hash on disk. This structure 2257 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2258 * boundary. 2259 */ 2260 struct ext4_dir_entry_hash { 2261 __le32 hash; 2262 __le32 minor_hash; 2263 }; 2264 2265 /* 2266 * The new version of the directory entry. Since EXT4 structures are 2267 * stored in intel byte order, and the name_len field could never be 2268 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2269 * file_type field. 2270 */ 2271 struct ext4_dir_entry_2 { 2272 __le32 inode; /* Inode number */ 2273 __le16 rec_len; /* Directory entry length */ 2274 __u8 name_len; /* Name length */ 2275 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2276 char name[EXT4_NAME_LEN]; /* File name */ 2277 }; 2278 2279 /* 2280 * Access the hashes at the end of ext4_dir_entry_2 2281 */ 2282 #define EXT4_DIRENT_HASHES(entry) \ 2283 ((struct ext4_dir_entry_hash *) \ 2284 (((void *)(entry)) + \ 2285 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2286 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2287 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2288 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2289 2290 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2291 { 2292 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2293 } 2294 2295 /* 2296 * This is a bogus directory entry at the end of each leaf block that 2297 * records checksums. 2298 */ 2299 struct ext4_dir_entry_tail { 2300 __le32 det_reserved_zero1; /* Pretend to be unused */ 2301 __le16 det_rec_len; /* 12 */ 2302 __u8 det_reserved_zero2; /* Zero name length */ 2303 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2304 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2305 }; 2306 2307 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2308 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2309 ((blocksize) - \ 2310 sizeof(struct ext4_dir_entry_tail)))) 2311 2312 /* 2313 * Ext4 directory file types. Only the low 3 bits are used. The 2314 * other bits are reserved for now. 2315 */ 2316 #define EXT4_FT_UNKNOWN 0 2317 #define EXT4_FT_REG_FILE 1 2318 #define EXT4_FT_DIR 2 2319 #define EXT4_FT_CHRDEV 3 2320 #define EXT4_FT_BLKDEV 4 2321 #define EXT4_FT_FIFO 5 2322 #define EXT4_FT_SOCK 6 2323 #define EXT4_FT_SYMLINK 7 2324 2325 #define EXT4_FT_MAX 8 2326 2327 #define EXT4_FT_DIR_CSUM 0xDE 2328 2329 /* 2330 * EXT4_DIR_PAD defines the directory entries boundaries 2331 * 2332 * NOTE: It must be a multiple of 4 2333 */ 2334 #define EXT4_DIR_PAD 4 2335 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2336 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2337 2338 /* 2339 * The rec_len is dependent on the type of directory. Directories that are 2340 * casefolded and encrypted need to store the hash as well, so we add room for 2341 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2342 * pass NULL for dir, as those entries do not use the extra fields. 2343 */ 2344 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2345 const struct inode *dir) 2346 { 2347 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2348 2349 if (dir && ext4_hash_in_dirent(dir)) 2350 rec_len += sizeof(struct ext4_dir_entry_hash); 2351 return (rec_len & ~EXT4_DIR_ROUND); 2352 } 2353 2354 /* 2355 * If we ever get support for fs block sizes > page_size, we'll need 2356 * to remove the #if statements in the next two functions... 2357 */ 2358 static inline unsigned int 2359 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2360 { 2361 unsigned len = le16_to_cpu(dlen); 2362 2363 #if (PAGE_SIZE >= 65536) 2364 if (len == EXT4_MAX_REC_LEN || len == 0) 2365 return blocksize; 2366 return (len & 65532) | ((len & 3) << 16); 2367 #else 2368 return len; 2369 #endif 2370 } 2371 2372 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2373 { 2374 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2375 #if (PAGE_SIZE >= 65536) 2376 if (len < 65536) 2377 return cpu_to_le16(len); 2378 if (len == blocksize) { 2379 if (blocksize == 65536) 2380 return cpu_to_le16(EXT4_MAX_REC_LEN); 2381 else 2382 return cpu_to_le16(0); 2383 } 2384 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2385 #else 2386 return cpu_to_le16(len); 2387 #endif 2388 } 2389 2390 /* 2391 * Hash Tree Directory indexing 2392 * (c) Daniel Phillips, 2001 2393 */ 2394 2395 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2396 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2397 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2398 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2399 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2400 2401 /* Legal values for the dx_root hash_version field: */ 2402 2403 #define DX_HASH_LEGACY 0 2404 #define DX_HASH_HALF_MD4 1 2405 #define DX_HASH_TEA 2 2406 #define DX_HASH_LEGACY_UNSIGNED 3 2407 #define DX_HASH_HALF_MD4_UNSIGNED 4 2408 #define DX_HASH_TEA_UNSIGNED 5 2409 #define DX_HASH_SIPHASH 6 2410 2411 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2412 const void *address, unsigned int length) 2413 { 2414 struct { 2415 struct shash_desc shash; 2416 char ctx[4]; 2417 } desc; 2418 2419 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2420 2421 desc.shash.tfm = sbi->s_chksum_driver; 2422 *(u32 *)desc.ctx = crc; 2423 2424 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2425 2426 return *(u32 *)desc.ctx; 2427 } 2428 2429 #ifdef __KERNEL__ 2430 2431 /* hash info structure used by the directory hash */ 2432 struct dx_hash_info 2433 { 2434 u32 hash; 2435 u32 minor_hash; 2436 int hash_version; 2437 u32 *seed; 2438 }; 2439 2440 2441 /* 32 and 64 bit signed EOF for dx directories */ 2442 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2443 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2444 2445 2446 /* 2447 * Control parameters used by ext4_htree_next_block 2448 */ 2449 #define HASH_NB_ALWAYS 1 2450 2451 struct ext4_filename { 2452 const struct qstr *usr_fname; 2453 struct fscrypt_str disk_name; 2454 struct dx_hash_info hinfo; 2455 #ifdef CONFIG_FS_ENCRYPTION 2456 struct fscrypt_str crypto_buf; 2457 #endif 2458 #if IS_ENABLED(CONFIG_UNICODE) 2459 struct fscrypt_str cf_name; 2460 #endif 2461 }; 2462 2463 #define fname_name(p) ((p)->disk_name.name) 2464 #define fname_usr_name(p) ((p)->usr_fname->name) 2465 #define fname_len(p) ((p)->disk_name.len) 2466 2467 /* 2468 * Describe an inode's exact location on disk and in memory 2469 */ 2470 struct ext4_iloc 2471 { 2472 struct buffer_head *bh; 2473 unsigned long offset; 2474 ext4_group_t block_group; 2475 }; 2476 2477 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2478 { 2479 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2480 } 2481 2482 static inline bool ext4_is_quota_file(struct inode *inode) 2483 { 2484 return IS_NOQUOTA(inode) && 2485 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2486 } 2487 2488 /* 2489 * This structure is stuffed into the struct file's private_data field 2490 * for directories. It is where we put information so that we can do 2491 * readdir operations in hash tree order. 2492 */ 2493 struct dir_private_info { 2494 struct rb_root root; 2495 struct rb_node *curr_node; 2496 struct fname *extra_fname; 2497 loff_t last_pos; 2498 __u32 curr_hash; 2499 __u32 curr_minor_hash; 2500 __u32 next_hash; 2501 }; 2502 2503 /* calculate the first block number of the group */ 2504 static inline ext4_fsblk_t 2505 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2506 { 2507 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2508 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2509 } 2510 2511 /* 2512 * Special error return code only used by dx_probe() and its callers. 2513 */ 2514 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2515 2516 /* htree levels for ext4 */ 2517 #define EXT4_HTREE_LEVEL_COMPAT 2 2518 #define EXT4_HTREE_LEVEL 3 2519 2520 static inline int ext4_dir_htree_level(struct super_block *sb) 2521 { 2522 return ext4_has_feature_largedir(sb) ? 2523 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2524 } 2525 2526 /* 2527 * Timeout and state flag for lazy initialization inode thread. 2528 */ 2529 #define EXT4_DEF_LI_WAIT_MULT 10 2530 #define EXT4_DEF_LI_MAX_START_DELAY 5 2531 #define EXT4_LAZYINIT_QUIT 0x0001 2532 #define EXT4_LAZYINIT_RUNNING 0x0002 2533 2534 /* 2535 * Lazy inode table initialization info 2536 */ 2537 struct ext4_lazy_init { 2538 unsigned long li_state; 2539 struct list_head li_request_list; 2540 struct mutex li_list_mtx; 2541 }; 2542 2543 enum ext4_li_mode { 2544 EXT4_LI_MODE_PREFETCH_BBITMAP, 2545 EXT4_LI_MODE_ITABLE, 2546 }; 2547 2548 struct ext4_li_request { 2549 struct super_block *lr_super; 2550 enum ext4_li_mode lr_mode; 2551 ext4_group_t lr_first_not_zeroed; 2552 ext4_group_t lr_next_group; 2553 struct list_head lr_request; 2554 unsigned long lr_next_sched; 2555 unsigned long lr_timeout; 2556 }; 2557 2558 struct ext4_features { 2559 struct kobject f_kobj; 2560 struct completion f_kobj_unregister; 2561 }; 2562 2563 /* 2564 * This structure will be used for multiple mount protection. It will be 2565 * written into the block number saved in the s_mmp_block field in the 2566 * superblock. Programs that check MMP should assume that if 2567 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2568 * to use the filesystem, regardless of how old the timestamp is. 2569 */ 2570 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2571 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2572 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2573 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2574 2575 struct mmp_struct { 2576 __le32 mmp_magic; /* Magic number for MMP */ 2577 __le32 mmp_seq; /* Sequence no. updated periodically */ 2578 2579 /* 2580 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2581 * purposes and do not affect the correctness of the algorithm 2582 */ 2583 __le64 mmp_time; /* Time last updated */ 2584 char mmp_nodename[64]; /* Node which last updated MMP block */ 2585 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2586 2587 /* 2588 * mmp_check_interval is used to verify if the MMP block has been 2589 * updated on the block device. The value is updated based on the 2590 * maximum time to write the MMP block during an update cycle. 2591 */ 2592 __le16 mmp_check_interval; 2593 2594 __le16 mmp_pad1; 2595 __le32 mmp_pad2[226]; 2596 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2597 }; 2598 2599 /* arguments passed to the mmp thread */ 2600 struct mmpd_data { 2601 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2602 struct super_block *sb; /* super block of the fs */ 2603 }; 2604 2605 /* 2606 * Check interval multiplier 2607 * The MMP block is written every update interval and initially checked every 2608 * update interval x the multiplier (the value is then adapted based on the 2609 * write latency). The reason is that writes can be delayed under load and we 2610 * don't want readers to incorrectly assume that the filesystem is no longer 2611 * in use. 2612 */ 2613 #define EXT4_MMP_CHECK_MULT 2UL 2614 2615 /* 2616 * Minimum interval for MMP checking in seconds. 2617 */ 2618 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2619 2620 /* 2621 * Maximum interval for MMP checking in seconds. 2622 */ 2623 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2624 2625 /* 2626 * Function prototypes 2627 */ 2628 2629 /* 2630 * Ok, these declarations are also in <linux/kernel.h> but none of the 2631 * ext4 source programs needs to include it so they are duplicated here. 2632 */ 2633 # define NORET_TYPE /**/ 2634 # define ATTRIB_NORET __attribute__((noreturn)) 2635 # define NORET_AND noreturn, 2636 2637 /* bitmap.c */ 2638 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2639 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2640 struct ext4_group_desc *gdp, 2641 struct buffer_head *bh, int sz); 2642 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2643 struct ext4_group_desc *gdp, 2644 struct buffer_head *bh, int sz); 2645 void ext4_block_bitmap_csum_set(struct super_block *sb, 2646 struct ext4_group_desc *gdp, 2647 struct buffer_head *bh); 2648 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2649 struct ext4_group_desc *gdp, 2650 struct buffer_head *bh); 2651 2652 /* balloc.c */ 2653 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2654 ext4_fsblk_t blocknr, 2655 ext4_group_t *blockgrpp, 2656 ext4_grpblk_t *offsetp); 2657 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2658 ext4_fsblk_t block); 2659 2660 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2661 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2662 ext4_group_t group); 2663 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2664 ext4_fsblk_t goal, 2665 unsigned int flags, 2666 unsigned long *count, 2667 int *errp); 2668 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2669 s64 nclusters, unsigned int flags); 2670 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2671 extern void ext4_check_blocks_bitmap(struct super_block *); 2672 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2673 ext4_group_t block_group, 2674 struct buffer_head ** bh); 2675 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2676 ext4_group_t group); 2677 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2678 2679 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2680 ext4_group_t block_group, 2681 bool ignore_locked); 2682 extern int ext4_wait_block_bitmap(struct super_block *sb, 2683 ext4_group_t block_group, 2684 struct buffer_head *bh); 2685 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2686 ext4_group_t block_group); 2687 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2688 ext4_group_t block_group, 2689 struct ext4_group_desc *gdp); 2690 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2691 2692 #if IS_ENABLED(CONFIG_UNICODE) 2693 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2694 const struct qstr *iname, 2695 struct ext4_filename *fname); 2696 #endif 2697 2698 /* ext4 encryption related stuff goes here crypto.c */ 2699 #ifdef CONFIG_FS_ENCRYPTION 2700 extern const struct fscrypt_operations ext4_cryptops; 2701 2702 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2703 int lookup, struct ext4_filename *fname); 2704 2705 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2706 struct ext4_filename *fname); 2707 2708 void ext4_fname_free_filename(struct ext4_filename *fname); 2709 2710 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2711 2712 #else /* !CONFIG_FS_ENCRYPTION */ 2713 static inline int ext4_fname_setup_filename(struct inode *dir, 2714 const struct qstr *iname, 2715 int lookup, 2716 struct ext4_filename *fname) 2717 { 2718 int err = 0; 2719 fname->usr_fname = iname; 2720 fname->disk_name.name = (unsigned char *) iname->name; 2721 fname->disk_name.len = iname->len; 2722 2723 #if IS_ENABLED(CONFIG_UNICODE) 2724 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2725 #endif 2726 2727 return err; 2728 } 2729 2730 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2731 struct dentry *dentry, 2732 struct ext4_filename *fname) 2733 { 2734 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2735 } 2736 2737 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2738 { 2739 #if IS_ENABLED(CONFIG_UNICODE) 2740 kfree(fname->cf_name.name); 2741 fname->cf_name.name = NULL; 2742 #endif 2743 } 2744 2745 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2746 void __user *arg) 2747 { 2748 return -EOPNOTSUPP; 2749 } 2750 #endif /* !CONFIG_FS_ENCRYPTION */ 2751 2752 /* dir.c */ 2753 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2754 struct file *, 2755 struct ext4_dir_entry_2 *, 2756 struct buffer_head *, char *, int, 2757 unsigned int); 2758 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2759 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2760 (de), (bh), (buf), (size), (offset))) 2761 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2762 __u32 minor_hash, 2763 struct ext4_dir_entry_2 *dirent, 2764 struct fscrypt_str *ent_name); 2765 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2766 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2767 struct buffer_head *bh, 2768 void *buf, int buf_size, 2769 struct ext4_filename *fname, 2770 struct ext4_dir_entry_2 **dest_de); 2771 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2772 struct ext4_dir_entry_2 *de, 2773 int buf_size, 2774 struct ext4_filename *fname); 2775 static inline void ext4_update_dx_flag(struct inode *inode) 2776 { 2777 if (!ext4_has_feature_dir_index(inode->i_sb) && 2778 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2779 /* ext4_iget() should have caught this... */ 2780 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2781 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2782 } 2783 } 2784 static const unsigned char ext4_filetype_table[] = { 2785 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2786 }; 2787 2788 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2789 { 2790 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2791 return DT_UNKNOWN; 2792 2793 return ext4_filetype_table[filetype]; 2794 } 2795 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2796 void *buf, int buf_size); 2797 2798 /* fsync.c */ 2799 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2800 2801 /* hash.c */ 2802 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2803 struct dx_hash_info *hinfo); 2804 2805 /* ialloc.c */ 2806 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2807 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2808 struct inode *, umode_t, 2809 const struct qstr *qstr, __u32 goal, 2810 uid_t *owner, __u32 i_flags, 2811 int handle_type, unsigned int line_no, 2812 int nblocks); 2813 2814 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2815 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2816 (goal), (owner), i_flags, 0, 0, 0) 2817 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2818 type, nblocks) \ 2819 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2820 0, (type), __LINE__, (nblocks)) 2821 2822 2823 extern void ext4_free_inode(handle_t *, struct inode *); 2824 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2825 extern unsigned long ext4_count_free_inodes(struct super_block *); 2826 extern unsigned long ext4_count_dirs(struct super_block *); 2827 extern void ext4_check_inodes_bitmap(struct super_block *); 2828 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2829 extern int ext4_init_inode_table(struct super_block *sb, 2830 ext4_group_t group, int barrier); 2831 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2832 2833 /* fast_commit.c */ 2834 int ext4_fc_info_show(struct seq_file *seq, void *v); 2835 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2836 void ext4_fc_init_inode(struct inode *inode); 2837 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2838 ext4_lblk_t end); 2839 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2840 struct dentry *dentry); 2841 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2842 struct dentry *dentry); 2843 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2844 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2845 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2846 struct dentry *dentry); 2847 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2848 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2849 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2850 void ext4_fc_start_update(struct inode *inode); 2851 void ext4_fc_stop_update(struct inode *inode); 2852 void ext4_fc_del(struct inode *inode); 2853 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2854 void ext4_fc_replay_cleanup(struct super_block *sb); 2855 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2856 int __init ext4_fc_init_dentry_cache(void); 2857 void ext4_fc_destroy_dentry_cache(void); 2858 int ext4_fc_record_regions(struct super_block *sb, int ino, 2859 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2860 int len, int replay); 2861 2862 /* mballoc.c */ 2863 extern const struct seq_operations ext4_mb_seq_groups_ops; 2864 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2865 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2866 extern int ext4_mb_init(struct super_block *); 2867 extern int ext4_mb_release(struct super_block *); 2868 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2869 struct ext4_allocation_request *, int *); 2870 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2871 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2872 extern int __init ext4_init_mballoc(void); 2873 extern void ext4_exit_mballoc(void); 2874 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2875 ext4_group_t group, 2876 unsigned int nr, int *cnt); 2877 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2878 unsigned int nr); 2879 2880 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2881 struct buffer_head *bh, ext4_fsblk_t block, 2882 unsigned long count, int flags); 2883 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2884 ext4_group_t ngroups); 2885 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2886 ext4_group_t i, struct ext4_group_desc *desc); 2887 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2888 ext4_fsblk_t block, unsigned long count); 2889 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2890 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2891 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2892 int len, int state); 2893 2894 /* inode.c */ 2895 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2896 struct ext4_inode_info *ei); 2897 int ext4_inode_is_fast_symlink(struct inode *inode); 2898 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2899 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2900 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2901 bool wait, struct buffer_head **bhs); 2902 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2903 struct buffer_head *bh_result, int create); 2904 int ext4_get_block(struct inode *inode, sector_t iblock, 2905 struct buffer_head *bh_result, int create); 2906 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2907 struct buffer_head *bh, int create); 2908 int ext4_walk_page_buffers(handle_t *handle, 2909 struct inode *inode, 2910 struct buffer_head *head, 2911 unsigned from, 2912 unsigned to, 2913 int *partial, 2914 int (*fn)(handle_t *handle, struct inode *inode, 2915 struct buffer_head *bh)); 2916 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2917 struct buffer_head *bh); 2918 #define FALL_BACK_TO_NONDELALLOC 1 2919 #define CONVERT_INLINE_DATA 2 2920 2921 typedef enum { 2922 EXT4_IGET_NORMAL = 0, 2923 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2924 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2925 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 2926 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 2927 } ext4_iget_flags; 2928 2929 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2930 ext4_iget_flags flags, const char *function, 2931 unsigned int line); 2932 2933 #define ext4_iget(sb, ino, flags) \ 2934 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2935 2936 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2937 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2938 struct iattr *); 2939 extern u32 ext4_dio_alignment(struct inode *inode); 2940 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2941 struct kstat *, u32, unsigned int); 2942 extern void ext4_evict_inode(struct inode *); 2943 extern void ext4_clear_inode(struct inode *); 2944 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2945 struct kstat *, u32, unsigned int); 2946 extern int ext4_sync_inode(handle_t *, struct inode *); 2947 extern void ext4_dirty_inode(struct inode *, int); 2948 extern int ext4_change_inode_journal_flag(struct inode *, int); 2949 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2950 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2951 struct ext4_iloc *iloc); 2952 extern int ext4_inode_attach_jinode(struct inode *inode); 2953 extern int ext4_can_truncate(struct inode *inode); 2954 extern int ext4_truncate(struct inode *); 2955 extern int ext4_break_layouts(struct inode *); 2956 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2957 extern void ext4_set_inode_flags(struct inode *, bool init); 2958 extern int ext4_alloc_da_blocks(struct inode *inode); 2959 extern void ext4_set_aops(struct inode *inode); 2960 extern int ext4_writepage_trans_blocks(struct inode *); 2961 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 2962 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2963 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2964 loff_t lstart, loff_t lend); 2965 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2966 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2967 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2968 extern void ext4_da_release_space(struct inode *inode, int to_free); 2969 extern void ext4_da_update_reserve_space(struct inode *inode, 2970 int used, int quota_claim); 2971 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2972 ext4_fsblk_t pblk, ext4_lblk_t len); 2973 2974 /* indirect.c */ 2975 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2976 struct ext4_map_blocks *map, int flags); 2977 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2978 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2979 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2980 ext4_lblk_t start, ext4_lblk_t end); 2981 2982 /* ioctl.c */ 2983 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2984 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2985 int ext4_fileattr_set(struct mnt_idmap *idmap, 2986 struct dentry *dentry, struct fileattr *fa); 2987 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 2988 extern void ext4_reset_inode_seed(struct inode *inode); 2989 int ext4_update_overhead(struct super_block *sb, bool force); 2990 2991 /* migrate.c */ 2992 extern int ext4_ext_migrate(struct inode *); 2993 extern int ext4_ind_migrate(struct inode *inode); 2994 2995 /* namei.c */ 2996 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 2997 struct inode *inode); 2998 extern int ext4_dirblock_csum_verify(struct inode *inode, 2999 struct buffer_head *bh); 3000 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3001 __u32 start_minor_hash, __u32 *next_hash); 3002 extern int ext4_search_dir(struct buffer_head *bh, 3003 char *search_buf, 3004 int buf_size, 3005 struct inode *dir, 3006 struct ext4_filename *fname, 3007 unsigned int offset, 3008 struct ext4_dir_entry_2 **res_dir); 3009 extern int ext4_generic_delete_entry(struct inode *dir, 3010 struct ext4_dir_entry_2 *de_del, 3011 struct buffer_head *bh, 3012 void *entry_buf, 3013 int buf_size, 3014 int csum_size); 3015 extern bool ext4_empty_dir(struct inode *inode); 3016 3017 /* resize.c */ 3018 extern void ext4_kvfree_array_rcu(void *to_free); 3019 extern int ext4_group_add(struct super_block *sb, 3020 struct ext4_new_group_data *input); 3021 extern int ext4_group_extend(struct super_block *sb, 3022 struct ext4_super_block *es, 3023 ext4_fsblk_t n_blocks_count); 3024 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3025 extern unsigned int ext4_list_backups(struct super_block *sb, 3026 unsigned int *three, unsigned int *five, 3027 unsigned int *seven); 3028 3029 /* super.c */ 3030 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3031 sector_t block, blk_opf_t op_flags); 3032 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3033 sector_t block); 3034 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3035 bh_end_io_t *end_io); 3036 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3037 bh_end_io_t *end_io); 3038 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3039 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3040 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3041 extern int ext4_calculate_overhead(struct super_block *sb); 3042 extern __le32 ext4_superblock_csum(struct super_block *sb, 3043 struct ext4_super_block *es); 3044 extern void ext4_superblock_csum_set(struct super_block *sb); 3045 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3046 ext4_group_t ngroup); 3047 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3048 char nbuf[16]); 3049 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3050 ext4_group_t block_group, 3051 unsigned int flags); 3052 3053 extern __printf(7, 8) 3054 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3055 int, __u64, const char *, ...); 3056 extern __printf(6, 7) 3057 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3058 ext4_fsblk_t, int, const char *, ...); 3059 extern __printf(5, 6) 3060 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3061 const char *, ...); 3062 extern void __ext4_std_error(struct super_block *, const char *, 3063 unsigned int, int); 3064 extern __printf(4, 5) 3065 void __ext4_warning(struct super_block *, const char *, unsigned int, 3066 const char *, ...); 3067 extern __printf(4, 5) 3068 void __ext4_warning_inode(const struct inode *inode, const char *function, 3069 unsigned int line, const char *fmt, ...); 3070 extern __printf(3, 4) 3071 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3072 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3073 const char *, unsigned int, const char *); 3074 extern __printf(7, 8) 3075 void __ext4_grp_locked_error(const char *, unsigned int, 3076 struct super_block *, ext4_group_t, 3077 unsigned long, ext4_fsblk_t, 3078 const char *, ...); 3079 3080 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3081 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3082 3083 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3084 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3085 3086 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3087 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3088 (fmt), ## a) 3089 3090 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3091 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3092 3093 #define ext4_abort(sb, err, fmt, a...) \ 3094 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3095 3096 #ifdef CONFIG_PRINTK 3097 3098 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3099 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3100 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3101 __ext4_error_inode((inode), (func), (line), (block), \ 3102 (err), (fmt), ##__VA_ARGS__) 3103 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3104 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3105 #define ext4_error(sb, fmt, ...) \ 3106 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3107 ##__VA_ARGS__) 3108 #define ext4_error_err(sb, err, fmt, ...) \ 3109 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3110 ##__VA_ARGS__) 3111 #define ext4_warning(sb, fmt, ...) \ 3112 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3113 #define ext4_warning_inode(inode, fmt, ...) \ 3114 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3115 #define ext4_msg(sb, level, fmt, ...) \ 3116 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3117 #define dump_mmp_msg(sb, mmp, msg) \ 3118 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3119 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3120 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3121 fmt, ##__VA_ARGS__) 3122 3123 #else 3124 3125 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3126 do { \ 3127 no_printk(fmt, ##__VA_ARGS__); \ 3128 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3129 } while (0) 3130 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3131 do { \ 3132 no_printk(fmt, ##__VA_ARGS__); \ 3133 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3134 } while (0) 3135 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3136 do { \ 3137 no_printk(fmt, ##__VA_ARGS__); \ 3138 __ext4_error_file(file, "", 0, block, " "); \ 3139 } while (0) 3140 #define ext4_error(sb, fmt, ...) \ 3141 do { \ 3142 no_printk(fmt, ##__VA_ARGS__); \ 3143 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3144 } while (0) 3145 #define ext4_error_err(sb, err, fmt, ...) \ 3146 do { \ 3147 no_printk(fmt, ##__VA_ARGS__); \ 3148 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3149 } while (0) 3150 #define ext4_warning(sb, fmt, ...) \ 3151 do { \ 3152 no_printk(fmt, ##__VA_ARGS__); \ 3153 __ext4_warning(sb, "", 0, " "); \ 3154 } while (0) 3155 #define ext4_warning_inode(inode, fmt, ...) \ 3156 do { \ 3157 no_printk(fmt, ##__VA_ARGS__); \ 3158 __ext4_warning_inode(inode, "", 0, " "); \ 3159 } while (0) 3160 #define ext4_msg(sb, level, fmt, ...) \ 3161 do { \ 3162 no_printk(fmt, ##__VA_ARGS__); \ 3163 __ext4_msg(sb, "", " "); \ 3164 } while (0) 3165 #define dump_mmp_msg(sb, mmp, msg) \ 3166 __dump_mmp_msg(sb, mmp, "", 0, "") 3167 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3168 do { \ 3169 no_printk(fmt, ##__VA_ARGS__); \ 3170 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3171 } while (0) 3172 3173 #endif 3174 3175 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3176 struct ext4_group_desc *bg); 3177 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3178 struct ext4_group_desc *bg); 3179 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3180 struct ext4_group_desc *bg); 3181 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3182 struct ext4_group_desc *bg); 3183 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3184 struct ext4_group_desc *bg); 3185 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3186 struct ext4_group_desc *bg); 3187 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3188 struct ext4_group_desc *bg); 3189 extern void ext4_block_bitmap_set(struct super_block *sb, 3190 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3191 extern void ext4_inode_bitmap_set(struct super_block *sb, 3192 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3193 extern void ext4_inode_table_set(struct super_block *sb, 3194 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3195 extern void ext4_free_group_clusters_set(struct super_block *sb, 3196 struct ext4_group_desc *bg, 3197 __u32 count); 3198 extern void ext4_free_inodes_set(struct super_block *sb, 3199 struct ext4_group_desc *bg, __u32 count); 3200 extern void ext4_used_dirs_set(struct super_block *sb, 3201 struct ext4_group_desc *bg, __u32 count); 3202 extern void ext4_itable_unused_set(struct super_block *sb, 3203 struct ext4_group_desc *bg, __u32 count); 3204 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3205 struct ext4_group_desc *gdp); 3206 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3207 struct ext4_group_desc *gdp); 3208 extern int ext4_register_li_request(struct super_block *sb, 3209 ext4_group_t first_not_zeroed); 3210 3211 static inline int ext4_has_metadata_csum(struct super_block *sb) 3212 { 3213 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3214 !EXT4_SB(sb)->s_chksum_driver); 3215 3216 return ext4_has_feature_metadata_csum(sb) && 3217 (EXT4_SB(sb)->s_chksum_driver != NULL); 3218 } 3219 3220 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3221 { 3222 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3223 } 3224 3225 #define ext4_read_incompat_64bit_val(es, name) \ 3226 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3227 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3228 le32_to_cpu(es->name##_lo)) 3229 3230 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3231 { 3232 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3233 } 3234 3235 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3236 { 3237 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3238 } 3239 3240 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3241 { 3242 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3243 } 3244 3245 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3246 ext4_fsblk_t blk) 3247 { 3248 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3249 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3250 } 3251 3252 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3253 ext4_fsblk_t blk) 3254 { 3255 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3256 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3257 } 3258 3259 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3260 ext4_fsblk_t blk) 3261 { 3262 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3263 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3264 } 3265 3266 static inline loff_t ext4_isize(struct super_block *sb, 3267 struct ext4_inode *raw_inode) 3268 { 3269 if (ext4_has_feature_largedir(sb) || 3270 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3271 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3272 le32_to_cpu(raw_inode->i_size_lo); 3273 3274 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3275 } 3276 3277 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3278 { 3279 raw_inode->i_size_lo = cpu_to_le32(i_size); 3280 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3281 } 3282 3283 /* 3284 * Reading s_groups_count requires using smp_rmb() afterwards. See 3285 * the locking protocol documented in the comments of ext4_group_add() 3286 * in resize.c 3287 */ 3288 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3289 { 3290 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3291 3292 smp_rmb(); 3293 return ngroups; 3294 } 3295 3296 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3297 ext4_group_t block_group) 3298 { 3299 return block_group >> sbi->s_log_groups_per_flex; 3300 } 3301 3302 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3303 { 3304 return 1 << sbi->s_log_groups_per_flex; 3305 } 3306 3307 #define ext4_std_error(sb, errno) \ 3308 do { \ 3309 if ((errno)) \ 3310 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3311 } while (0) 3312 3313 #ifdef CONFIG_SMP 3314 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3315 * counters. So we need to make sure we have free clusters more 3316 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3317 */ 3318 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3319 #else 3320 #define EXT4_FREECLUSTERS_WATERMARK 0 3321 #endif 3322 3323 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3324 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3325 { 3326 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3327 !inode_is_locked(inode)); 3328 down_write(&EXT4_I(inode)->i_data_sem); 3329 if (newsize > EXT4_I(inode)->i_disksize) 3330 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3331 up_write(&EXT4_I(inode)->i_data_sem); 3332 } 3333 3334 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3335 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3336 { 3337 int changed = 0; 3338 3339 if (newsize > inode->i_size) { 3340 i_size_write(inode, newsize); 3341 changed = 1; 3342 } 3343 if (newsize > EXT4_I(inode)->i_disksize) { 3344 ext4_update_i_disksize(inode, newsize); 3345 changed |= 2; 3346 } 3347 return changed; 3348 } 3349 3350 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3351 loff_t len); 3352 3353 struct ext4_group_info { 3354 unsigned long bb_state; 3355 #ifdef AGGRESSIVE_CHECK 3356 unsigned long bb_check_counter; 3357 #endif 3358 struct rb_root bb_free_root; 3359 ext4_grpblk_t bb_first_free; /* first free block */ 3360 ext4_grpblk_t bb_free; /* total free blocks */ 3361 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3362 int bb_avg_fragment_size_order; /* order of average 3363 fragment in BG */ 3364 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3365 ext4_group_t bb_group; /* Group number */ 3366 struct list_head bb_prealloc_list; 3367 #ifdef DOUBLE_CHECK 3368 void *bb_bitmap; 3369 #endif 3370 struct rw_semaphore alloc_sem; 3371 struct list_head bb_avg_fragment_size_node; 3372 struct list_head bb_largest_free_order_node; 3373 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3374 * regions, index is order. 3375 * bb_counters[3] = 5 means 3376 * 5 free 8-block regions. */ 3377 }; 3378 3379 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3380 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3381 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3382 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3383 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3384 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3385 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3386 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3387 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3388 3389 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3390 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3391 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3392 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3393 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3394 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3395 3396 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3397 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3398 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3399 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3400 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3401 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3402 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3403 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3404 3405 #define EXT4_MAX_CONTENTION 8 3406 #define EXT4_CONTENTION_THRESHOLD 2 3407 3408 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3409 ext4_group_t group) 3410 { 3411 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3412 } 3413 3414 /* 3415 * Returns true if the filesystem is busy enough that attempts to 3416 * access the block group locks has run into contention. 3417 */ 3418 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3419 { 3420 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3421 } 3422 3423 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3424 { 3425 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3426 if (spin_trylock(lock)) 3427 /* 3428 * We're able to grab the lock right away, so drop the 3429 * lock contention counter. 3430 */ 3431 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3432 else { 3433 /* 3434 * The lock is busy, so bump the contention counter, 3435 * and then wait on the spin lock. 3436 */ 3437 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3438 EXT4_MAX_CONTENTION); 3439 spin_lock(lock); 3440 } 3441 } 3442 3443 static inline void ext4_unlock_group(struct super_block *sb, 3444 ext4_group_t group) 3445 { 3446 spin_unlock(ext4_group_lock_ptr(sb, group)); 3447 } 3448 3449 #ifdef CONFIG_QUOTA 3450 static inline bool ext4_quota_capable(struct super_block *sb) 3451 { 3452 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3453 } 3454 3455 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3456 { 3457 struct ext4_sb_info *sbi = EXT4_SB(sb); 3458 3459 return (ext4_has_feature_quota(sb) || 3460 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3461 } 3462 int ext4_enable_quotas(struct super_block *sb); 3463 #endif 3464 3465 /* 3466 * Block validity checking 3467 */ 3468 #define ext4_check_indirect_blockref(inode, bh) \ 3469 ext4_check_blockref(__func__, __LINE__, inode, \ 3470 (__le32 *)(bh)->b_data, \ 3471 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3472 3473 #define ext4_ind_check_inode(inode) \ 3474 ext4_check_blockref(__func__, __LINE__, inode, \ 3475 EXT4_I(inode)->i_data, \ 3476 EXT4_NDIR_BLOCKS) 3477 3478 /* 3479 * Inodes and files operations 3480 */ 3481 3482 /* dir.c */ 3483 extern const struct file_operations ext4_dir_operations; 3484 3485 /* file.c */ 3486 extern const struct inode_operations ext4_file_inode_operations; 3487 extern const struct file_operations ext4_file_operations; 3488 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3489 3490 /* inline.c */ 3491 extern int ext4_get_max_inline_size(struct inode *inode); 3492 extern int ext4_find_inline_data_nolock(struct inode *inode); 3493 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3494 unsigned int len); 3495 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3496 3497 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3498 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3499 struct inode *inode, 3500 loff_t pos, unsigned len, 3501 struct page **pagep); 3502 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3503 unsigned copied, struct folio *folio); 3504 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3505 struct inode *inode, 3506 loff_t pos, unsigned len, 3507 struct page **pagep, 3508 void **fsdata); 3509 extern int ext4_try_add_inline_entry(handle_t *handle, 3510 struct ext4_filename *fname, 3511 struct inode *dir, struct inode *inode); 3512 extern int ext4_try_create_inline_dir(handle_t *handle, 3513 struct inode *parent, 3514 struct inode *inode); 3515 extern int ext4_read_inline_dir(struct file *filp, 3516 struct dir_context *ctx, 3517 int *has_inline_data); 3518 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3519 struct inode *dir, ext4_lblk_t block, 3520 struct dx_hash_info *hinfo, 3521 __u32 start_hash, __u32 start_minor_hash, 3522 int *has_inline_data); 3523 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3524 struct ext4_filename *fname, 3525 struct ext4_dir_entry_2 **res_dir, 3526 int *has_inline_data); 3527 extern int ext4_delete_inline_entry(handle_t *handle, 3528 struct inode *dir, 3529 struct ext4_dir_entry_2 *de_del, 3530 struct buffer_head *bh, 3531 int *has_inline_data); 3532 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3533 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3534 struct ext4_dir_entry_2 **parent_de, 3535 int *retval); 3536 extern void *ext4_read_inline_link(struct inode *inode); 3537 3538 struct iomap; 3539 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3540 3541 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3542 3543 extern int ext4_convert_inline_data(struct inode *inode); 3544 3545 static inline int ext4_has_inline_data(struct inode *inode) 3546 { 3547 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3548 EXT4_I(inode)->i_inline_off; 3549 } 3550 3551 /* namei.c */ 3552 extern const struct inode_operations ext4_dir_inode_operations; 3553 extern const struct inode_operations ext4_special_inode_operations; 3554 extern struct dentry *ext4_get_parent(struct dentry *child); 3555 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3556 struct ext4_dir_entry_2 *de, 3557 int blocksize, int csum_size, 3558 unsigned int parent_ino, int dotdot_real_len); 3559 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3560 unsigned int blocksize); 3561 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3562 struct buffer_head *bh); 3563 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3564 struct inode *inode, struct dentry *dentry); 3565 extern int __ext4_link(struct inode *dir, struct inode *inode, 3566 struct dentry *dentry); 3567 3568 #define S_SHIFT 12 3569 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3570 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3571 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3572 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3573 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3574 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3575 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3576 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3577 }; 3578 3579 static inline void ext4_set_de_type(struct super_block *sb, 3580 struct ext4_dir_entry_2 *de, 3581 umode_t mode) { 3582 if (ext4_has_feature_filetype(sb)) 3583 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3584 } 3585 3586 /* readpages.c */ 3587 extern int ext4_mpage_readpages(struct inode *inode, 3588 struct readahead_control *rac, struct folio *folio); 3589 extern int __init ext4_init_post_read_processing(void); 3590 extern void ext4_exit_post_read_processing(void); 3591 3592 /* symlink.c */ 3593 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3594 extern const struct inode_operations ext4_symlink_inode_operations; 3595 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3596 3597 /* sysfs.c */ 3598 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3599 extern int ext4_register_sysfs(struct super_block *sb); 3600 extern void ext4_unregister_sysfs(struct super_block *sb); 3601 extern int __init ext4_init_sysfs(void); 3602 extern void ext4_exit_sysfs(void); 3603 3604 /* block_validity */ 3605 extern void ext4_release_system_zone(struct super_block *sb); 3606 extern int ext4_setup_system_zone(struct super_block *sb); 3607 extern int __init ext4_init_system_zone(void); 3608 extern void ext4_exit_system_zone(void); 3609 extern int ext4_inode_block_valid(struct inode *inode, 3610 ext4_fsblk_t start_blk, 3611 unsigned int count); 3612 extern int ext4_check_blockref(const char *, unsigned int, 3613 struct inode *, __le32 *, unsigned int); 3614 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3615 ext4_fsblk_t start_blk, unsigned int count); 3616 3617 3618 /* extents.c */ 3619 struct ext4_ext_path; 3620 struct ext4_extent; 3621 3622 /* 3623 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3624 * __le32. 3625 */ 3626 #define EXT_MAX_BLOCKS 0xffffffff 3627 3628 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3629 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3630 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3631 struct ext4_map_blocks *map, int flags); 3632 extern int ext4_ext_truncate(handle_t *, struct inode *); 3633 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3634 ext4_lblk_t end); 3635 extern void ext4_ext_init(struct super_block *); 3636 extern void ext4_ext_release(struct super_block *); 3637 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3638 loff_t len); 3639 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3640 loff_t offset, ssize_t len); 3641 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3642 ext4_io_end_t *io_end); 3643 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3644 struct ext4_map_blocks *map, int flags); 3645 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3646 int num, 3647 struct ext4_ext_path *path); 3648 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3649 struct ext4_ext_path **, 3650 struct ext4_extent *, int); 3651 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3652 struct ext4_ext_path **, 3653 int flags); 3654 extern void ext4_free_ext_path(struct ext4_ext_path *); 3655 extern int ext4_ext_check_inode(struct inode *inode); 3656 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3657 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3658 __u64 start, __u64 len); 3659 extern int ext4_get_es_cache(struct inode *inode, 3660 struct fiemap_extent_info *fieinfo, 3661 __u64 start, __u64 len); 3662 extern int ext4_ext_precache(struct inode *inode); 3663 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3664 struct inode *inode2, ext4_lblk_t lblk1, 3665 ext4_lblk_t lblk2, ext4_lblk_t count, 3666 int mark_unwritten,int *err); 3667 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3668 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3669 int check_cred, int restart_cred, 3670 int revoke_cred); 3671 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3672 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3673 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3674 int len, int unwritten, ext4_fsblk_t pblk); 3675 extern int ext4_ext_clear_bb(struct inode *inode); 3676 3677 3678 /* move_extent.c */ 3679 extern void ext4_double_down_write_data_sem(struct inode *first, 3680 struct inode *second); 3681 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3682 struct inode *donor_inode); 3683 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3684 __u64 start_orig, __u64 start_donor, 3685 __u64 len, __u64 *moved_len); 3686 3687 /* page-io.c */ 3688 extern int __init ext4_init_pageio(void); 3689 extern void ext4_exit_pageio(void); 3690 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3691 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3692 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3693 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3694 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3695 struct writeback_control *wbc); 3696 extern void ext4_end_io_rsv_work(struct work_struct *work); 3697 extern void ext4_io_submit(struct ext4_io_submit *io); 3698 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3699 size_t len); 3700 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3701 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3702 3703 /* mmp.c */ 3704 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3705 3706 /* mmp.c */ 3707 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3708 3709 /* verity.c */ 3710 extern const struct fsverity_operations ext4_verityops; 3711 3712 /* orphan.c */ 3713 extern int ext4_orphan_add(handle_t *, struct inode *); 3714 extern int ext4_orphan_del(handle_t *, struct inode *); 3715 extern void ext4_orphan_cleanup(struct super_block *sb, 3716 struct ext4_super_block *es); 3717 extern void ext4_release_orphan_info(struct super_block *sb); 3718 extern int ext4_init_orphan_info(struct super_block *sb); 3719 extern int ext4_orphan_file_empty(struct super_block *sb); 3720 extern void ext4_orphan_file_block_trigger( 3721 struct jbd2_buffer_trigger_type *triggers, 3722 struct buffer_head *bh, 3723 void *data, size_t size); 3724 3725 /* 3726 * Add new method to test whether block and inode bitmaps are properly 3727 * initialized. With uninit_bg reading the block from disk is not enough 3728 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3729 */ 3730 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3731 3732 static inline int bitmap_uptodate(struct buffer_head *bh) 3733 { 3734 return (buffer_uptodate(bh) && 3735 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3736 } 3737 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3738 { 3739 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3740 } 3741 3742 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3743 3744 /* For ioend & aio unwritten conversion wait queues */ 3745 #define EXT4_WQ_HASH_SZ 37 3746 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3747 EXT4_WQ_HASH_SZ]) 3748 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3749 3750 extern int ext4_resize_begin(struct super_block *sb); 3751 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3752 3753 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3754 struct ext4_io_end *io_end) 3755 { 3756 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3757 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3758 atomic_inc(&EXT4_I(inode)->i_unwritten); 3759 } 3760 } 3761 3762 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3763 { 3764 struct inode *inode = io_end->inode; 3765 3766 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3767 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3768 /* Wake up anyone waiting on unwritten extent conversion */ 3769 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3770 wake_up_all(ext4_ioend_wq(inode)); 3771 } 3772 } 3773 3774 extern const struct iomap_ops ext4_iomap_ops; 3775 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3776 extern const struct iomap_ops ext4_iomap_report_ops; 3777 3778 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3779 { 3780 /* 3781 * If the buffer has the write error flag, we have failed 3782 * to write out data in the block. In this case, we don't 3783 * have to read the block because we may read the old data 3784 * successfully. 3785 */ 3786 if (buffer_write_io_error(bh)) 3787 set_buffer_uptodate(bh); 3788 return buffer_uptodate(bh); 3789 } 3790 3791 #endif /* __KERNEL__ */ 3792 3793 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3794 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3795 3796 #endif /* _EXT4_H */ 3797