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/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/wait.h> 31 #include <linux/sched/signal.h> 32 #include <linux/blockgroup_lock.h> 33 #include <linux/percpu_counter.h> 34 #include <linux/ratelimit.h> 35 #include <crypto/hash.h> 36 #include <linux/falloc.h> 37 #include <linux/percpu-rwsem.h> 38 #include <linux/fiemap.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #include <linux/fscrypt.h> 44 #include <linux/fsverity.h> 45 46 #include <linux/compiler.h> 47 48 /* 49 * The fourth extended filesystem constants/structures 50 */ 51 52 /* 53 * with AGGRESSIVE_CHECK allocator runs consistency checks over 54 * structures. these checks slow things down a lot 55 */ 56 #define AGGRESSIVE_CHECK__ 57 58 /* 59 * with DOUBLE_CHECK defined mballoc creates persistent in-core 60 * bitmaps, maintains and uses them to check for double allocations 61 */ 62 #define DOUBLE_CHECK__ 63 64 /* 65 * Define EXT4FS_DEBUG to produce debug messages 66 */ 67 #undef EXT4FS_DEBUG 68 69 /* 70 * Debug code 71 */ 72 #ifdef EXT4FS_DEBUG 73 #define ext4_debug(f, a...) \ 74 do { \ 75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 76 __FILE__, __LINE__, __func__); \ 77 printk(KERN_DEBUG f, ## a); \ 78 } while (0) 79 #else 80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 81 #endif 82 83 /* 84 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 85 */ 86 #define EXT_DEBUG__ 87 88 /* 89 * Dynamic printk for controlled extents debugging. 90 */ 91 #ifdef CONFIG_EXT4_DEBUG 92 #define ext_debug(ino, fmt, ...) \ 93 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 94 current->comm, task_pid_nr(current), \ 95 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 96 __func__, ##__VA_ARGS__) 97 #else 98 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 99 #endif 100 101 #define ASSERT(assert) \ 102 do { \ 103 if (unlikely(!(assert))) { \ 104 printk(KERN_EMERG \ 105 "Assertion failure in %s() at %s:%d: '%s'\n", \ 106 __func__, __FILE__, __LINE__, #assert); \ 107 BUG(); \ 108 } \ 109 } while (0) 110 111 /* data type for block offset of block group */ 112 typedef int ext4_grpblk_t; 113 114 /* data type for filesystem-wide blocks number */ 115 typedef unsigned long long ext4_fsblk_t; 116 117 /* data type for file logical block number */ 118 typedef __u32 ext4_lblk_t; 119 120 /* data type for block group number */ 121 typedef unsigned int ext4_group_t; 122 123 enum SHIFT_DIRECTION { 124 SHIFT_LEFT = 0, 125 SHIFT_RIGHT, 126 }; 127 128 /* 129 * Flags used in mballoc's allocation_context flags field. 130 * 131 * Also used to show what's going on for debugging purposes when the 132 * flag field is exported via the traceport interface 133 */ 134 135 /* prefer goal again. length */ 136 #define EXT4_MB_HINT_MERGE 0x0001 137 /* blocks already reserved */ 138 #define EXT4_MB_HINT_RESERVED 0x0002 139 /* metadata is being allocated */ 140 #define EXT4_MB_HINT_METADATA 0x0004 141 /* first blocks in the file */ 142 #define EXT4_MB_HINT_FIRST 0x0008 143 /* search for the best chunk */ 144 #define EXT4_MB_HINT_BEST 0x0010 145 /* data is being allocated */ 146 #define EXT4_MB_HINT_DATA 0x0020 147 /* don't preallocate (for tails) */ 148 #define EXT4_MB_HINT_NOPREALLOC 0x0040 149 /* allocate for locality group */ 150 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 151 /* allocate goal blocks or none */ 152 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 153 /* goal is meaningful */ 154 #define EXT4_MB_HINT_TRY_GOAL 0x0200 155 /* blocks already pre-reserved by delayed allocation */ 156 #define EXT4_MB_DELALLOC_RESERVED 0x0400 157 /* We are doing stream allocation */ 158 #define EXT4_MB_STREAM_ALLOC 0x0800 159 /* Use reserved root blocks if needed */ 160 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 161 /* Use blocks from reserved pool */ 162 #define EXT4_MB_USE_RESERVED 0x2000 163 /* Do strict check for free blocks while retrying block allocation */ 164 #define EXT4_MB_STRICT_CHECK 0x4000 165 166 struct ext4_allocation_request { 167 /* target inode for block we're allocating */ 168 struct inode *inode; 169 /* how many blocks we want to allocate */ 170 unsigned int len; 171 /* logical block in target inode */ 172 ext4_lblk_t logical; 173 /* the closest logical allocated block to the left */ 174 ext4_lblk_t lleft; 175 /* the closest logical allocated block to the right */ 176 ext4_lblk_t lright; 177 /* phys. target (a hint) */ 178 ext4_fsblk_t goal; 179 /* phys. block for the closest logical allocated block to the left */ 180 ext4_fsblk_t pleft; 181 /* phys. block for the closest logical allocated block to the right */ 182 ext4_fsblk_t pright; 183 /* flags. see above EXT4_MB_HINT_* */ 184 unsigned int flags; 185 }; 186 187 /* 188 * Logical to physical block mapping, used by ext4_map_blocks() 189 * 190 * This structure is used to pass requests into ext4_map_blocks() as 191 * well as to store the information returned by ext4_map_blocks(). It 192 * takes less room on the stack than a struct buffer_head. 193 */ 194 #define EXT4_MAP_NEW BIT(BH_New) 195 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 196 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 197 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 198 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 199 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 200 201 struct ext4_map_blocks { 202 ext4_fsblk_t m_pblk; 203 ext4_lblk_t m_lblk; 204 unsigned int m_len; 205 unsigned int m_flags; 206 }; 207 208 /* 209 * Block validity checking, system zone rbtree. 210 */ 211 struct ext4_system_blocks { 212 struct rb_root root; 213 struct rcu_head rcu; 214 }; 215 216 /* 217 * Flags for ext4_io_end->flags 218 */ 219 #define EXT4_IO_END_UNWRITTEN 0x0001 220 221 struct ext4_io_end_vec { 222 struct list_head list; /* list of io_end_vec */ 223 loff_t offset; /* offset in the file */ 224 ssize_t size; /* size of the extent */ 225 }; 226 227 /* 228 * For converting unwritten extents on a work queue. 'handle' is used for 229 * buffered writeback. 230 */ 231 typedef struct ext4_io_end { 232 struct list_head list; /* per-file finished IO list */ 233 handle_t *handle; /* handle reserved for extent 234 * conversion */ 235 struct inode *inode; /* file being written to */ 236 struct bio *bio; /* Linked list of completed 237 * bios covering the extent */ 238 unsigned int flag; /* unwritten or not */ 239 atomic_t count; /* reference counter */ 240 struct list_head list_vec; /* list of ext4_io_end_vec */ 241 } ext4_io_end_t; 242 243 struct ext4_io_submit { 244 struct writeback_control *io_wbc; 245 struct bio *io_bio; 246 ext4_io_end_t *io_end; 247 sector_t io_next_block; 248 }; 249 250 /* 251 * Special inodes numbers 252 */ 253 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 254 #define EXT4_ROOT_INO 2 /* Root inode */ 255 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 256 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 257 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 258 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 259 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 260 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 261 262 /* First non-reserved inode for old ext4 filesystems */ 263 #define EXT4_GOOD_OLD_FIRST_INO 11 264 265 /* 266 * Maximal count of links to a file 267 */ 268 #define EXT4_LINK_MAX 65000 269 270 /* 271 * Macro-instructions used to manage several block sizes 272 */ 273 #define EXT4_MIN_BLOCK_SIZE 1024 274 #define EXT4_MAX_BLOCK_SIZE 65536 275 #define EXT4_MIN_BLOCK_LOG_SIZE 10 276 #define EXT4_MAX_BLOCK_LOG_SIZE 16 277 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 278 #ifdef __KERNEL__ 279 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 280 #else 281 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 282 #endif 283 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 284 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 285 EXT4_SB(s)->s_cluster_bits) 286 #ifdef __KERNEL__ 287 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 288 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 289 #else 290 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 291 #endif 292 #ifdef __KERNEL__ 293 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 294 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 295 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 296 #else 297 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 298 EXT4_GOOD_OLD_INODE_SIZE : \ 299 (s)->s_inode_size) 300 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 301 EXT4_GOOD_OLD_FIRST_INO : \ 302 (s)->s_first_ino) 303 #endif 304 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 305 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 306 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 307 blkbits)) 308 309 /* Translate a block number to a cluster number */ 310 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 311 /* Translate a cluster number to a block number */ 312 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 313 /* Translate # of blks to # of clusters */ 314 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 315 (sbi)->s_cluster_bits) 316 /* Mask out the low bits to get the starting block of the cluster */ 317 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 318 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 319 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 320 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 321 /* Fill in the low bits to get the last block of the cluster */ 322 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 323 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 324 /* Get the cluster offset */ 325 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 326 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 327 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 328 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 329 330 /* 331 * Structure of a blocks group descriptor 332 */ 333 struct ext4_group_desc 334 { 335 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 336 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 337 __le32 bg_inode_table_lo; /* Inodes table block */ 338 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 339 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 340 __le16 bg_used_dirs_count_lo; /* Directories count */ 341 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 342 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 343 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 344 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 345 __le16 bg_itable_unused_lo; /* Unused inodes count */ 346 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 347 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 348 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 349 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 350 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 351 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 352 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 353 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 354 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 355 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 356 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 357 __u32 bg_reserved; 358 }; 359 360 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 361 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 362 sizeof(__le16)) 363 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 364 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 365 sizeof(__le16)) 366 367 /* 368 * Structure of a flex block group info 369 */ 370 371 struct flex_groups { 372 atomic64_t free_clusters; 373 atomic_t free_inodes; 374 atomic_t used_dirs; 375 }; 376 377 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 378 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 379 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 380 381 /* 382 * Macro-instructions used to manage group descriptors 383 */ 384 #define EXT4_MIN_DESC_SIZE 32 385 #define EXT4_MIN_DESC_SIZE_64BIT 64 386 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 387 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 388 #ifdef __KERNEL__ 389 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 390 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 391 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 392 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 393 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 394 #else 395 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 396 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 397 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 398 #endif 399 400 /* 401 * Constants relative to the data blocks 402 */ 403 #define EXT4_NDIR_BLOCKS 12 404 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 405 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 406 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 407 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 408 409 /* 410 * Inode flags 411 */ 412 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 413 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 414 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 415 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 416 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 417 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 418 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 419 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 420 /* Reserved for compression usage... */ 421 #define EXT4_DIRTY_FL 0x00000100 422 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 423 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 424 /* nb: was previously EXT2_ECOMPR_FL */ 425 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 426 /* End compression flags --- maybe not all used */ 427 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 428 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 429 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 430 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 431 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 432 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 433 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 434 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 435 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 436 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 437 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 438 439 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 440 441 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 442 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 443 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 444 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 445 446 /* User modifiable flags */ 447 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 448 EXT4_UNRM_FL | \ 449 EXT4_COMPR_FL | \ 450 EXT4_SYNC_FL | \ 451 EXT4_IMMUTABLE_FL | \ 452 EXT4_APPEND_FL | \ 453 EXT4_NODUMP_FL | \ 454 EXT4_NOATIME_FL | \ 455 EXT4_JOURNAL_DATA_FL | \ 456 EXT4_NOTAIL_FL | \ 457 EXT4_DIRSYNC_FL | \ 458 EXT4_TOPDIR_FL | \ 459 EXT4_EXTENTS_FL | \ 460 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 461 EXT4_DAX_FL | \ 462 EXT4_PROJINHERIT_FL | \ 463 EXT4_CASEFOLD_FL) 464 465 /* User visible flags */ 466 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 467 EXT4_DIRTY_FL | \ 468 EXT4_COMPRBLK_FL | \ 469 EXT4_NOCOMPR_FL | \ 470 EXT4_ENCRYPT_FL | \ 471 EXT4_INDEX_FL | \ 472 EXT4_VERITY_FL | \ 473 EXT4_INLINE_DATA_FL) 474 475 /* Flags that should be inherited by new inodes from their parent. */ 476 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 477 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 478 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 479 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 480 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 481 EXT4_DAX_FL) 482 483 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 484 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 485 EXT4_PROJINHERIT_FL)) 486 487 /* Flags that are appropriate for non-directories/regular files. */ 488 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 489 490 /* The only flags that should be swapped */ 491 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 492 493 /* Flags which are mutually exclusive to DAX */ 494 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 495 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 496 497 /* Mask out flags that are inappropriate for the given type of inode. */ 498 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 499 { 500 if (S_ISDIR(mode)) 501 return flags; 502 else if (S_ISREG(mode)) 503 return flags & EXT4_REG_FLMASK; 504 else 505 return flags & EXT4_OTHER_FLMASK; 506 } 507 508 /* 509 * Inode flags used for atomic set/get 510 */ 511 enum { 512 EXT4_INODE_SECRM = 0, /* Secure deletion */ 513 EXT4_INODE_UNRM = 1, /* Undelete */ 514 EXT4_INODE_COMPR = 2, /* Compress file */ 515 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 516 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 517 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 518 EXT4_INODE_NODUMP = 6, /* do not dump file */ 519 EXT4_INODE_NOATIME = 7, /* do not update atime */ 520 /* Reserved for compression usage... */ 521 EXT4_INODE_DIRTY = 8, 522 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 523 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 524 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 525 /* End compression flags --- maybe not all used */ 526 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 527 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 528 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 529 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 530 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 531 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 532 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 533 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 534 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 535 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 536 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 537 EXT4_INODE_DAX = 25, /* Inode is DAX */ 538 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 539 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 540 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 541 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 542 }; 543 544 /* 545 * Since it's pretty easy to mix up bit numbers and hex values, we use a 546 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 547 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 548 * any extra space in the compiled kernel image, otherwise, the build will fail. 549 * It's important that these values are the same, since we are using 550 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 551 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 552 * values found in ext2, ext3 and ext4 filesystems, and of course the values 553 * defined in e2fsprogs. 554 * 555 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 556 */ 557 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 558 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 559 560 static inline void ext4_check_flag_values(void) 561 { 562 CHECK_FLAG_VALUE(SECRM); 563 CHECK_FLAG_VALUE(UNRM); 564 CHECK_FLAG_VALUE(COMPR); 565 CHECK_FLAG_VALUE(SYNC); 566 CHECK_FLAG_VALUE(IMMUTABLE); 567 CHECK_FLAG_VALUE(APPEND); 568 CHECK_FLAG_VALUE(NODUMP); 569 CHECK_FLAG_VALUE(NOATIME); 570 CHECK_FLAG_VALUE(DIRTY); 571 CHECK_FLAG_VALUE(COMPRBLK); 572 CHECK_FLAG_VALUE(NOCOMPR); 573 CHECK_FLAG_VALUE(ENCRYPT); 574 CHECK_FLAG_VALUE(INDEX); 575 CHECK_FLAG_VALUE(IMAGIC); 576 CHECK_FLAG_VALUE(JOURNAL_DATA); 577 CHECK_FLAG_VALUE(NOTAIL); 578 CHECK_FLAG_VALUE(DIRSYNC); 579 CHECK_FLAG_VALUE(TOPDIR); 580 CHECK_FLAG_VALUE(HUGE_FILE); 581 CHECK_FLAG_VALUE(EXTENTS); 582 CHECK_FLAG_VALUE(VERITY); 583 CHECK_FLAG_VALUE(EA_INODE); 584 CHECK_FLAG_VALUE(INLINE_DATA); 585 CHECK_FLAG_VALUE(PROJINHERIT); 586 CHECK_FLAG_VALUE(CASEFOLD); 587 CHECK_FLAG_VALUE(RESERVED); 588 } 589 590 /* Used to pass group descriptor data when online resize is done */ 591 struct ext4_new_group_input { 592 __u32 group; /* Group number for this data */ 593 __u64 block_bitmap; /* Absolute block number of block bitmap */ 594 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 595 __u64 inode_table; /* Absolute block number of inode table start */ 596 __u32 blocks_count; /* Total number of blocks in this group */ 597 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 598 __u16 unused; 599 }; 600 601 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 602 struct compat_ext4_new_group_input { 603 u32 group; 604 compat_u64 block_bitmap; 605 compat_u64 inode_bitmap; 606 compat_u64 inode_table; 607 u32 blocks_count; 608 u16 reserved_blocks; 609 u16 unused; 610 }; 611 #endif 612 613 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 614 struct ext4_new_group_data { 615 __u32 group; 616 __u64 block_bitmap; 617 __u64 inode_bitmap; 618 __u64 inode_table; 619 __u32 blocks_count; 620 __u16 reserved_blocks; 621 __u16 mdata_blocks; 622 __u32 free_clusters_count; 623 }; 624 625 /* Indexes used to index group tables in ext4_new_group_data */ 626 enum { 627 BLOCK_BITMAP = 0, /* block bitmap */ 628 INODE_BITMAP, /* inode bitmap */ 629 INODE_TABLE, /* inode tables */ 630 GROUP_TABLE_COUNT, 631 }; 632 633 /* 634 * Flags used by ext4_map_blocks() 635 */ 636 /* Allocate any needed blocks and/or convert an unwritten 637 extent to be an initialized ext4 */ 638 #define EXT4_GET_BLOCKS_CREATE 0x0001 639 /* Request the creation of an unwritten extent */ 640 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 641 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 642 EXT4_GET_BLOCKS_CREATE) 643 /* Caller is from the delayed allocation writeout path 644 * finally doing the actual allocation of delayed blocks */ 645 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 646 /* caller is from the direct IO path, request to creation of an 647 unwritten extents if not allocated, split the unwritten 648 extent if blocks has been preallocated already*/ 649 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 650 #define EXT4_GET_BLOCKS_CONVERT 0x0010 651 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 652 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 653 /* Convert extent to initialized after IO complete */ 654 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 655 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 656 /* Eventual metadata allocation (due to growing extent tree) 657 * should not fail, so try to use reserved blocks for that.*/ 658 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 659 /* Don't normalize allocation size (used for fallocate) */ 660 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 661 /* Convert written extents to unwritten */ 662 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 663 /* Write zeros to newly created written extents */ 664 #define EXT4_GET_BLOCKS_ZERO 0x0200 665 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 666 EXT4_GET_BLOCKS_ZERO) 667 /* Caller will submit data before dropping transaction handle. This 668 * allows jbd2 to avoid submitting data before commit. */ 669 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 670 671 /* 672 * The bit position of these flags must not overlap with any of the 673 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 674 * read_extent_tree_block(), ext4_split_extent_at(), 675 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 676 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 677 * caching the extents when reading from the extent tree while a 678 * truncate or punch hole operation is in progress. 679 */ 680 #define EXT4_EX_NOCACHE 0x40000000 681 #define EXT4_EX_FORCE_CACHE 0x20000000 682 #define EXT4_EX_NOFAIL 0x10000000 683 684 /* 685 * Flags used by ext4_free_blocks 686 */ 687 #define EXT4_FREE_BLOCKS_METADATA 0x0001 688 #define EXT4_FREE_BLOCKS_FORGET 0x0002 689 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 690 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 691 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 692 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 693 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 694 695 /* 696 * ioctl commands 697 */ 698 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 699 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 700 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 701 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 702 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 703 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 704 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 705 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 706 #define EXT4_IOC_MIGRATE _IO('f', 9) 707 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 708 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 709 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 710 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 711 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 712 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 713 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 714 /* ioctl codes 19--39 are reserved for fscrypt */ 715 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 716 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 717 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 718 719 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 720 721 /* 722 * Flags for going down operation 723 */ 724 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 725 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 726 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 727 728 /* 729 * Flags returned by EXT4_IOC_GETSTATE 730 * 731 * We only expose to userspace a subset of the state flags in 732 * i_state_flags 733 */ 734 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 735 #define EXT4_STATE_FLAG_NEW 0x00000002 736 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 737 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 738 739 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 740 /* 741 * ioctl commands in 32 bit emulation 742 */ 743 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 744 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 745 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 746 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 747 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 748 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 749 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 750 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 751 #endif 752 753 /* 754 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 755 * It indicates that the entry in extent status cache is for a hole. 756 */ 757 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 758 759 /* Max physical block we can address w/o extents */ 760 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 761 762 /* Max logical block we can support */ 763 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 764 765 /* 766 * Structure of an inode on the disk 767 */ 768 struct ext4_inode { 769 __le16 i_mode; /* File mode */ 770 __le16 i_uid; /* Low 16 bits of Owner Uid */ 771 __le32 i_size_lo; /* Size in bytes */ 772 __le32 i_atime; /* Access time */ 773 __le32 i_ctime; /* Inode Change time */ 774 __le32 i_mtime; /* Modification time */ 775 __le32 i_dtime; /* Deletion Time */ 776 __le16 i_gid; /* Low 16 bits of Group Id */ 777 __le16 i_links_count; /* Links count */ 778 __le32 i_blocks_lo; /* Blocks count */ 779 __le32 i_flags; /* File flags */ 780 union { 781 struct { 782 __le32 l_i_version; 783 } linux1; 784 struct { 785 __u32 h_i_translator; 786 } hurd1; 787 struct { 788 __u32 m_i_reserved1; 789 } masix1; 790 } osd1; /* OS dependent 1 */ 791 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 792 __le32 i_generation; /* File version (for NFS) */ 793 __le32 i_file_acl_lo; /* File ACL */ 794 __le32 i_size_high; 795 __le32 i_obso_faddr; /* Obsoleted fragment address */ 796 union { 797 struct { 798 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 799 __le16 l_i_file_acl_high; 800 __le16 l_i_uid_high; /* these 2 fields */ 801 __le16 l_i_gid_high; /* were reserved2[0] */ 802 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 803 __le16 l_i_reserved; 804 } linux2; 805 struct { 806 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 807 __u16 h_i_mode_high; 808 __u16 h_i_uid_high; 809 __u16 h_i_gid_high; 810 __u32 h_i_author; 811 } hurd2; 812 struct { 813 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 814 __le16 m_i_file_acl_high; 815 __u32 m_i_reserved2[2]; 816 } masix2; 817 } osd2; /* OS dependent 2 */ 818 __le16 i_extra_isize; 819 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 820 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 821 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 822 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 823 __le32 i_crtime; /* File Creation time */ 824 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 825 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 826 __le32 i_projid; /* Project ID */ 827 }; 828 829 struct move_extent { 830 __u32 reserved; /* should be zero */ 831 __u32 donor_fd; /* donor file descriptor */ 832 __u64 orig_start; /* logical start offset in block for orig */ 833 __u64 donor_start; /* logical start offset in block for donor */ 834 __u64 len; /* block length to be moved */ 835 __u64 moved_len; /* moved block length */ 836 }; 837 838 #define EXT4_EPOCH_BITS 2 839 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 840 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 841 842 /* 843 * Extended fields will fit into an inode if the filesystem was formatted 844 * with large inodes (-I 256 or larger) and there are not currently any EAs 845 * consuming all of the available space. For new inodes we always reserve 846 * enough space for the kernel's known extended fields, but for inodes 847 * created with an old kernel this might not have been the case. None of 848 * the extended inode fields is critical for correct filesystem operation. 849 * This macro checks if a certain field fits in the inode. Note that 850 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 851 */ 852 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 853 ((offsetof(typeof(*ext4_inode), field) + \ 854 sizeof((ext4_inode)->field)) \ 855 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 856 (einode)->i_extra_isize)) \ 857 858 /* 859 * We use an encoding that preserves the times for extra epoch "00": 860 * 861 * extra msb of adjust for signed 862 * epoch 32-bit 32-bit tv_sec to 863 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 864 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 865 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 866 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 867 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 868 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 869 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 870 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 871 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 872 * 873 * Note that previous versions of the kernel on 64-bit systems would 874 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 875 * 1970. e2fsck will correct this, assuming that it is run on the 876 * affected filesystem before 2242. 877 */ 878 879 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 880 { 881 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 882 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 883 } 884 885 static inline void ext4_decode_extra_time(struct timespec64 *time, 886 __le32 extra) 887 { 888 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 889 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 890 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 891 } 892 893 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 894 do { \ 895 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 896 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 897 (raw_inode)->xtime ## _extra = \ 898 ext4_encode_extra_time(&(inode)->xtime); \ 899 } \ 900 else \ 901 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 902 } while (0) 903 904 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 905 do { \ 906 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 907 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 908 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 909 (raw_inode)->xtime ## _extra = \ 910 ext4_encode_extra_time(&(einode)->xtime); \ 911 } while (0) 912 913 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 914 do { \ 915 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 916 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 917 ext4_decode_extra_time(&(inode)->xtime, \ 918 raw_inode->xtime ## _extra); \ 919 } \ 920 else \ 921 (inode)->xtime.tv_nsec = 0; \ 922 } while (0) 923 924 925 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 926 do { \ 927 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 928 (einode)->xtime.tv_sec = \ 929 (signed)le32_to_cpu((raw_inode)->xtime); \ 930 else \ 931 (einode)->xtime.tv_sec = 0; \ 932 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 933 ext4_decode_extra_time(&(einode)->xtime, \ 934 raw_inode->xtime ## _extra); \ 935 else \ 936 (einode)->xtime.tv_nsec = 0; \ 937 } while (0) 938 939 #define i_disk_version osd1.linux1.l_i_version 940 941 #if defined(__KERNEL__) || defined(__linux__) 942 #define i_reserved1 osd1.linux1.l_i_reserved1 943 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 944 #define i_blocks_high osd2.linux2.l_i_blocks_high 945 #define i_uid_low i_uid 946 #define i_gid_low i_gid 947 #define i_uid_high osd2.linux2.l_i_uid_high 948 #define i_gid_high osd2.linux2.l_i_gid_high 949 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 950 951 #elif defined(__GNU__) 952 953 #define i_translator osd1.hurd1.h_i_translator 954 #define i_uid_high osd2.hurd2.h_i_uid_high 955 #define i_gid_high osd2.hurd2.h_i_gid_high 956 #define i_author osd2.hurd2.h_i_author 957 958 #elif defined(__masix__) 959 960 #define i_reserved1 osd1.masix1.m_i_reserved1 961 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 962 #define i_reserved2 osd2.masix2.m_i_reserved2 963 964 #endif /* defined(__KERNEL__) || defined(__linux__) */ 965 966 #include "extents_status.h" 967 #include "fast_commit.h" 968 969 /* 970 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 971 * 972 * These are needed to avoid lockdep false positives when we need to 973 * allocate blocks to the quota inode during ext4_map_blocks(), while 974 * holding i_data_sem for a normal (non-quota) inode. Since we don't 975 * do quota tracking for the quota inode, this avoids deadlock (as 976 * well as infinite recursion, since it isn't turtles all the way 977 * down...) 978 * 979 * I_DATA_SEM_NORMAL - Used for most inodes 980 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 981 * where the second inode has larger inode number 982 * than the first 983 * I_DATA_SEM_QUOTA - Used for quota inodes only 984 */ 985 enum { 986 I_DATA_SEM_NORMAL = 0, 987 I_DATA_SEM_OTHER, 988 I_DATA_SEM_QUOTA, 989 }; 990 991 992 /* 993 * fourth extended file system inode data in memory 994 */ 995 struct ext4_inode_info { 996 __le32 i_data[15]; /* unconverted */ 997 __u32 i_dtime; 998 ext4_fsblk_t i_file_acl; 999 1000 /* 1001 * i_block_group is the number of the block group which contains 1002 * this file's inode. Constant across the lifetime of the inode, 1003 * it is used for making block allocation decisions - we try to 1004 * place a file's data blocks near its inode block, and new inodes 1005 * near to their parent directory's inode. 1006 */ 1007 ext4_group_t i_block_group; 1008 ext4_lblk_t i_dir_start_lookup; 1009 #if (BITS_PER_LONG < 64) 1010 unsigned long i_state_flags; /* Dynamic state flags */ 1011 #endif 1012 unsigned long i_flags; 1013 1014 /* 1015 * Extended attributes can be read independently of the main file 1016 * data. Taking i_mutex even when reading would cause contention 1017 * between readers of EAs and writers of regular file data, so 1018 * instead we synchronize on xattr_sem when reading or changing 1019 * EAs. 1020 */ 1021 struct rw_semaphore xattr_sem; 1022 1023 struct list_head i_orphan; /* unlinked but open inodes */ 1024 1025 /* Fast commit related info */ 1026 1027 struct list_head i_fc_list; /* 1028 * inodes that need fast commit 1029 * protected by sbi->s_fc_lock. 1030 */ 1031 1032 /* Start of lblk range that needs to be committed in this fast commit */ 1033 ext4_lblk_t i_fc_lblk_start; 1034 1035 /* End of lblk range that needs to be committed in this fast commit */ 1036 ext4_lblk_t i_fc_lblk_len; 1037 1038 /* Number of ongoing updates on this inode */ 1039 atomic_t i_fc_updates; 1040 1041 /* Fast commit wait queue for this inode */ 1042 wait_queue_head_t i_fc_wait; 1043 1044 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1045 struct mutex i_fc_lock; 1046 1047 /* 1048 * i_disksize keeps track of what the inode size is ON DISK, not 1049 * in memory. During truncate, i_size is set to the new size by 1050 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1051 * set i_disksize to 0 until the truncate is actually under way. 1052 * 1053 * The intent is that i_disksize always represents the blocks which 1054 * are used by this file. This allows recovery to restart truncate 1055 * on orphans if we crash during truncate. We actually write i_disksize 1056 * into the on-disk inode when writing inodes out, instead of i_size. 1057 * 1058 * The only time when i_disksize and i_size may be different is when 1059 * a truncate is in progress. The only things which change i_disksize 1060 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1061 */ 1062 loff_t i_disksize; 1063 1064 /* 1065 * i_data_sem is for serialising ext4_truncate() against 1066 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1067 * data tree are chopped off during truncate. We can't do that in 1068 * ext4 because whenever we perform intermediate commits during 1069 * truncate, the inode and all the metadata blocks *must* be in a 1070 * consistent state which allows truncation of the orphans to restart 1071 * during recovery. Hence we must fix the get_block-vs-truncate race 1072 * by other means, so we have i_data_sem. 1073 */ 1074 struct rw_semaphore i_data_sem; 1075 /* 1076 * i_mmap_sem is for serializing page faults with truncate / punch hole 1077 * operations. We have to make sure that new page cannot be faulted in 1078 * a section of the inode that is being punched. We cannot easily use 1079 * i_data_sem for this since we need protection for the whole punch 1080 * operation and i_data_sem ranks below transaction start so we have 1081 * to occasionally drop it. 1082 */ 1083 struct rw_semaphore i_mmap_sem; 1084 struct inode vfs_inode; 1085 struct jbd2_inode *jinode; 1086 1087 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1088 1089 /* 1090 * File creation time. Its function is same as that of 1091 * struct timespec64 i_{a,c,m}time in the generic inode. 1092 */ 1093 struct timespec64 i_crtime; 1094 1095 /* mballoc */ 1096 atomic_t i_prealloc_active; 1097 struct list_head i_prealloc_list; 1098 spinlock_t i_prealloc_lock; 1099 1100 /* extents status tree */ 1101 struct ext4_es_tree i_es_tree; 1102 rwlock_t i_es_lock; 1103 struct list_head i_es_list; 1104 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1105 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1106 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1107 extents to shrink. Protected by 1108 i_es_lock */ 1109 1110 /* ialloc */ 1111 ext4_group_t i_last_alloc_group; 1112 1113 /* allocation reservation info for delalloc */ 1114 /* In case of bigalloc, this refer to clusters rather than blocks */ 1115 unsigned int i_reserved_data_blocks; 1116 1117 /* pending cluster reservations for bigalloc file systems */ 1118 struct ext4_pending_tree i_pending_tree; 1119 1120 /* on-disk additional length */ 1121 __u16 i_extra_isize; 1122 1123 /* Indicate the inline data space. */ 1124 u16 i_inline_off; 1125 u16 i_inline_size; 1126 1127 #ifdef CONFIG_QUOTA 1128 /* quota space reservation, managed internally by quota code */ 1129 qsize_t i_reserved_quota; 1130 #endif 1131 1132 /* Lock protecting lists below */ 1133 spinlock_t i_completed_io_lock; 1134 /* 1135 * Completed IOs that need unwritten extents handling and have 1136 * transaction reserved 1137 */ 1138 struct list_head i_rsv_conversion_list; 1139 struct work_struct i_rsv_conversion_work; 1140 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1141 1142 spinlock_t i_block_reservation_lock; 1143 1144 /* 1145 * Transactions that contain inode's metadata needed to complete 1146 * fsync and fdatasync, respectively. 1147 */ 1148 tid_t i_sync_tid; 1149 tid_t i_datasync_tid; 1150 1151 #ifdef CONFIG_QUOTA 1152 struct dquot *i_dquot[MAXQUOTAS]; 1153 #endif 1154 1155 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1156 __u32 i_csum_seed; 1157 1158 kprojid_t i_projid; 1159 }; 1160 1161 /* 1162 * File system states 1163 */ 1164 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1165 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1166 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1167 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1168 1169 /* 1170 * Misc. filesystem flags 1171 */ 1172 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1173 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1174 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1175 1176 /* 1177 * Mount flags set via mount options or defaults 1178 */ 1179 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1180 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1181 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1182 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1183 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1184 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1185 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1186 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1187 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1188 #ifdef CONFIG_FS_DAX 1189 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1190 #else 1191 #define EXT4_MOUNT_DAX_ALWAYS 0 1192 #endif 1193 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1194 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1195 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1196 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1197 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1198 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1199 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1200 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1201 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1202 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1203 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1204 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1205 * enable enforcement for hidden 1206 * quota files */ 1207 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1208 * enforcement for hidden quota 1209 * files */ 1210 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1211 * enforcement */ 1212 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1213 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1214 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1215 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1216 #define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000 1217 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1218 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1219 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1220 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1221 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1222 1223 /* 1224 * Mount flags set either automatically (could not be set by mount option) 1225 * based on per file system feature or property or in special cases such as 1226 * distinguishing between explicit mount option definition and default. 1227 */ 1228 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1229 specified delalloc */ 1230 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1231 size of blocksize * 8 1232 blocks */ 1233 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1234 file systems */ 1235 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1236 specified journal checksum */ 1237 1238 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1239 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1240 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1241 1242 1243 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1244 ~EXT4_MOUNT_##opt 1245 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1246 EXT4_MOUNT_##opt 1247 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1248 EXT4_MOUNT_##opt) 1249 1250 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1251 ~EXT4_MOUNT2_##opt 1252 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1253 EXT4_MOUNT2_##opt 1254 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1255 EXT4_MOUNT2_##opt) 1256 1257 #define ext4_test_and_set_bit __test_and_set_bit_le 1258 #define ext4_set_bit __set_bit_le 1259 #define ext4_set_bit_atomic ext2_set_bit_atomic 1260 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1261 #define ext4_clear_bit __clear_bit_le 1262 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1263 #define ext4_test_bit test_bit_le 1264 #define ext4_find_next_zero_bit find_next_zero_bit_le 1265 #define ext4_find_next_bit find_next_bit_le 1266 1267 extern void ext4_set_bits(void *bm, int cur, int len); 1268 1269 /* 1270 * Maximal mount counts between two filesystem checks 1271 */ 1272 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1273 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1274 1275 /* 1276 * Behaviour when detecting errors 1277 */ 1278 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1279 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1280 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1281 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1282 1283 /* Metadata checksum algorithm codes */ 1284 #define EXT4_CRC32C_CHKSUM 1 1285 1286 /* 1287 * Structure of the super block 1288 */ 1289 struct ext4_super_block { 1290 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1291 __le32 s_blocks_count_lo; /* Blocks count */ 1292 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1293 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1294 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1295 __le32 s_first_data_block; /* First Data Block */ 1296 __le32 s_log_block_size; /* Block size */ 1297 __le32 s_log_cluster_size; /* Allocation cluster size */ 1298 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1299 __le32 s_clusters_per_group; /* # Clusters per group */ 1300 __le32 s_inodes_per_group; /* # Inodes per group */ 1301 __le32 s_mtime; /* Mount time */ 1302 /*30*/ __le32 s_wtime; /* Write time */ 1303 __le16 s_mnt_count; /* Mount count */ 1304 __le16 s_max_mnt_count; /* Maximal mount count */ 1305 __le16 s_magic; /* Magic signature */ 1306 __le16 s_state; /* File system state */ 1307 __le16 s_errors; /* Behaviour when detecting errors */ 1308 __le16 s_minor_rev_level; /* minor revision level */ 1309 /*40*/ __le32 s_lastcheck; /* time of last check */ 1310 __le32 s_checkinterval; /* max. time between checks */ 1311 __le32 s_creator_os; /* OS */ 1312 __le32 s_rev_level; /* Revision level */ 1313 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1314 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1315 /* 1316 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1317 * 1318 * Note: the difference between the compatible feature set and 1319 * the incompatible feature set is that if there is a bit set 1320 * in the incompatible feature set that the kernel doesn't 1321 * know about, it should refuse to mount the filesystem. 1322 * 1323 * e2fsck's requirements are more strict; if it doesn't know 1324 * about a feature in either the compatible or incompatible 1325 * feature set, it must abort and not try to meddle with 1326 * things it doesn't understand... 1327 */ 1328 __le32 s_first_ino; /* First non-reserved inode */ 1329 __le16 s_inode_size; /* size of inode structure */ 1330 __le16 s_block_group_nr; /* block group # of this superblock */ 1331 __le32 s_feature_compat; /* compatible feature set */ 1332 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1333 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1334 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1335 /*78*/ char s_volume_name[16]; /* volume name */ 1336 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1337 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1338 /* 1339 * Performance hints. Directory preallocation should only 1340 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1341 */ 1342 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1343 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1344 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1345 /* 1346 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1347 */ 1348 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1349 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1350 __le32 s_journal_dev; /* device number of journal file */ 1351 __le32 s_last_orphan; /* start of list of inodes to delete */ 1352 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1353 __u8 s_def_hash_version; /* Default hash version to use */ 1354 __u8 s_jnl_backup_type; 1355 __le16 s_desc_size; /* size of group descriptor */ 1356 /*100*/ __le32 s_default_mount_opts; 1357 __le32 s_first_meta_bg; /* First metablock block group */ 1358 __le32 s_mkfs_time; /* When the filesystem was created */ 1359 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1360 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1361 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1362 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1363 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1364 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1365 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1366 __le32 s_flags; /* Miscellaneous flags */ 1367 __le16 s_raid_stride; /* RAID stride */ 1368 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1369 __le64 s_mmp_block; /* Block for multi-mount protection */ 1370 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1371 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1372 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1373 __u8 s_encryption_level; /* versioning level for encryption */ 1374 __u8 s_reserved_pad; /* Padding to next 32bits */ 1375 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1376 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1377 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1378 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1379 snapshot's future use */ 1380 __le32 s_snapshot_list; /* inode number of the head of the 1381 on-disk snapshot list */ 1382 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1383 __le32 s_error_count; /* number of fs errors */ 1384 __le32 s_first_error_time; /* first time an error happened */ 1385 __le32 s_first_error_ino; /* inode involved in first error */ 1386 __le64 s_first_error_block; /* block involved of first error */ 1387 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1388 __le32 s_first_error_line; /* line number where error happened */ 1389 __le32 s_last_error_time; /* most recent time of an error */ 1390 __le32 s_last_error_ino; /* inode involved in last error */ 1391 __le32 s_last_error_line; /* line number where error happened */ 1392 __le64 s_last_error_block; /* block involved of last error */ 1393 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1394 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1395 __u8 s_mount_opts[64]; 1396 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1397 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1398 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1399 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1400 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1401 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1402 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1403 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1404 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1405 __u8 s_wtime_hi; 1406 __u8 s_mtime_hi; 1407 __u8 s_mkfs_time_hi; 1408 __u8 s_lastcheck_hi; 1409 __u8 s_first_error_time_hi; 1410 __u8 s_last_error_time_hi; 1411 __u8 s_first_error_errcode; 1412 __u8 s_last_error_errcode; 1413 __le16 s_encoding; /* Filename charset encoding */ 1414 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1415 __le32 s_reserved[95]; /* Padding to the end of the block */ 1416 __le32 s_checksum; /* crc32c(superblock) */ 1417 }; 1418 1419 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1420 1421 #ifdef __KERNEL__ 1422 1423 #ifdef CONFIG_FS_ENCRYPTION 1424 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL) 1425 #else 1426 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1427 #endif 1428 1429 /* Number of quota types we support */ 1430 #define EXT4_MAXQUOTAS 3 1431 1432 #define EXT4_ENC_UTF8_12_1 1 1433 1434 /* 1435 * fourth extended-fs super-block data in memory 1436 */ 1437 struct ext4_sb_info { 1438 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1439 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1440 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1441 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1442 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1443 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1444 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1445 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1446 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1447 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1448 unsigned long s_overhead; /* # of fs overhead clusters */ 1449 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1450 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1451 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1452 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1453 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1454 struct buffer_head * __rcu *s_group_desc; 1455 unsigned int s_mount_opt; 1456 unsigned int s_mount_opt2; 1457 unsigned long s_mount_flags; 1458 unsigned int s_def_mount_opt; 1459 ext4_fsblk_t s_sb_block; 1460 atomic64_t s_resv_clusters; 1461 kuid_t s_resuid; 1462 kgid_t s_resgid; 1463 unsigned short s_mount_state; 1464 unsigned short s_pad; 1465 int s_addr_per_block_bits; 1466 int s_desc_per_block_bits; 1467 int s_inode_size; 1468 int s_first_ino; 1469 unsigned int s_inode_readahead_blks; 1470 unsigned int s_inode_goal; 1471 u32 s_hash_seed[4]; 1472 int s_def_hash_version; 1473 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1474 struct percpu_counter s_freeclusters_counter; 1475 struct percpu_counter s_freeinodes_counter; 1476 struct percpu_counter s_dirs_counter; 1477 struct percpu_counter s_dirtyclusters_counter; 1478 struct percpu_counter s_sra_exceeded_retry_limit; 1479 struct blockgroup_lock *s_blockgroup_lock; 1480 struct proc_dir_entry *s_proc; 1481 struct kobject s_kobj; 1482 struct completion s_kobj_unregister; 1483 struct super_block *s_sb; 1484 1485 /* Journaling */ 1486 struct journal_s *s_journal; 1487 struct list_head s_orphan; 1488 struct mutex s_orphan_lock; 1489 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1490 unsigned long s_commit_interval; 1491 u32 s_max_batch_time; 1492 u32 s_min_batch_time; 1493 struct block_device *s_journal_bdev; 1494 #ifdef CONFIG_QUOTA 1495 /* Names of quota files with journalled quota */ 1496 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1497 int s_jquota_fmt; /* Format of quota to use */ 1498 #endif 1499 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1500 struct ext4_system_blocks __rcu *s_system_blks; 1501 1502 #ifdef EXTENTS_STATS 1503 /* ext4 extents stats */ 1504 unsigned long s_ext_min; 1505 unsigned long s_ext_max; 1506 unsigned long s_depth_max; 1507 spinlock_t s_ext_stats_lock; 1508 unsigned long s_ext_blocks; 1509 unsigned long s_ext_extents; 1510 #endif 1511 1512 /* for buddy allocator */ 1513 struct ext4_group_info ** __rcu *s_group_info; 1514 struct inode *s_buddy_cache; 1515 spinlock_t s_md_lock; 1516 unsigned short *s_mb_offsets; 1517 unsigned int *s_mb_maxs; 1518 unsigned int s_group_info_size; 1519 unsigned int s_mb_free_pending; 1520 struct list_head s_freed_data_list; /* List of blocks to be freed 1521 after commit completed */ 1522 1523 /* tunables */ 1524 unsigned long s_stripe; 1525 unsigned int s_mb_stream_request; 1526 unsigned int s_mb_max_to_scan; 1527 unsigned int s_mb_min_to_scan; 1528 unsigned int s_mb_stats; 1529 unsigned int s_mb_order2_reqs; 1530 unsigned int s_mb_group_prealloc; 1531 unsigned int s_mb_max_inode_prealloc; 1532 unsigned int s_max_dir_size_kb; 1533 /* where last allocation was done - for stream allocation */ 1534 unsigned long s_mb_last_group; 1535 unsigned long s_mb_last_start; 1536 unsigned int s_mb_prefetch; 1537 unsigned int s_mb_prefetch_limit; 1538 1539 /* stats for buddy allocator */ 1540 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1541 atomic_t s_bal_success; /* we found long enough chunks */ 1542 atomic_t s_bal_allocated; /* in blocks */ 1543 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1544 atomic_t s_bal_goals; /* goal hits */ 1545 atomic_t s_bal_breaks; /* too long searches */ 1546 atomic_t s_bal_2orders; /* 2^order hits */ 1547 spinlock_t s_bal_lock; 1548 unsigned long s_mb_buddies_generated; 1549 unsigned long long s_mb_generation_time; 1550 atomic_t s_mb_lost_chunks; 1551 atomic_t s_mb_preallocated; 1552 atomic_t s_mb_discarded; 1553 atomic_t s_lock_busy; 1554 1555 /* locality groups */ 1556 struct ext4_locality_group __percpu *s_locality_groups; 1557 1558 /* for write statistics */ 1559 unsigned long s_sectors_written_start; 1560 u64 s_kbytes_written; 1561 1562 /* the size of zero-out chunk */ 1563 unsigned int s_extent_max_zeroout_kb; 1564 1565 unsigned int s_log_groups_per_flex; 1566 struct flex_groups * __rcu *s_flex_groups; 1567 ext4_group_t s_flex_groups_allocated; 1568 1569 /* workqueue for reserved extent conversions (buffered io) */ 1570 struct workqueue_struct *rsv_conversion_wq; 1571 1572 /* timer for periodic error stats printing */ 1573 struct timer_list s_err_report; 1574 1575 /* Lazy inode table initialization info */ 1576 struct ext4_li_request *s_li_request; 1577 /* Wait multiplier for lazy initialization thread */ 1578 unsigned int s_li_wait_mult; 1579 1580 /* Kernel thread for multiple mount protection */ 1581 struct task_struct *s_mmp_tsk; 1582 1583 /* record the last minlen when FITRIM is called. */ 1584 atomic_t s_last_trim_minblks; 1585 1586 /* Reference to checksum algorithm driver via cryptoapi */ 1587 struct crypto_shash *s_chksum_driver; 1588 1589 /* Precomputed FS UUID checksum for seeding other checksums */ 1590 __u32 s_csum_seed; 1591 1592 /* Reclaim extents from extent status tree */ 1593 struct shrinker s_es_shrinker; 1594 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1595 long s_es_nr_inode; 1596 struct ext4_es_stats s_es_stats; 1597 struct mb_cache *s_ea_block_cache; 1598 struct mb_cache *s_ea_inode_cache; 1599 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1600 1601 /* Ratelimit ext4 messages. */ 1602 struct ratelimit_state s_err_ratelimit_state; 1603 struct ratelimit_state s_warning_ratelimit_state; 1604 struct ratelimit_state s_msg_ratelimit_state; 1605 atomic_t s_warning_count; 1606 atomic_t s_msg_count; 1607 1608 /* Encryption policy for '-o test_dummy_encryption' */ 1609 struct fscrypt_dummy_policy s_dummy_enc_policy; 1610 1611 /* 1612 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1613 * or EXTENTS flag. 1614 */ 1615 struct percpu_rw_semaphore s_writepages_rwsem; 1616 struct dax_device *s_daxdev; 1617 #ifdef CONFIG_EXT4_DEBUG 1618 unsigned long s_simulate_fail; 1619 #endif 1620 /* Record the errseq of the backing block device */ 1621 errseq_t s_bdev_wb_err; 1622 spinlock_t s_bdev_wb_lock; 1623 1624 /* Information about errors that happened during this mount */ 1625 spinlock_t s_error_lock; 1626 int s_add_error_count; 1627 int s_first_error_code; 1628 __u32 s_first_error_line; 1629 __u32 s_first_error_ino; 1630 __u64 s_first_error_block; 1631 const char *s_first_error_func; 1632 time64_t s_first_error_time; 1633 int s_last_error_code; 1634 __u32 s_last_error_line; 1635 __u32 s_last_error_ino; 1636 __u64 s_last_error_block; 1637 const char *s_last_error_func; 1638 time64_t s_last_error_time; 1639 /* 1640 * If we are in a context where we cannot update error information in 1641 * the on-disk superblock, we queue this work to do it. 1642 */ 1643 struct work_struct s_error_work; 1644 1645 /* Ext4 fast commit stuff */ 1646 atomic_t s_fc_subtid; 1647 atomic_t s_fc_ineligible_updates; 1648 /* 1649 * After commit starts, the main queue gets locked, and the further 1650 * updates get added in the staging queue. 1651 */ 1652 #define FC_Q_MAIN 0 1653 #define FC_Q_STAGING 1 1654 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1655 * that have data changes in them. 1656 */ 1657 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1658 unsigned int s_fc_bytes; 1659 /* 1660 * Main fast commit lock. This lock protects accesses to the 1661 * following fields: 1662 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1663 */ 1664 spinlock_t s_fc_lock; 1665 struct buffer_head *s_fc_bh; 1666 struct ext4_fc_stats s_fc_stats; 1667 u64 s_fc_avg_commit_time; 1668 #ifdef CONFIG_EXT4_DEBUG 1669 int s_fc_debug_max_replay; 1670 #endif 1671 struct ext4_fc_replay_state s_fc_replay_state; 1672 }; 1673 1674 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1675 { 1676 return sb->s_fs_info; 1677 } 1678 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1679 { 1680 return container_of(inode, struct ext4_inode_info, vfs_inode); 1681 } 1682 1683 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1684 { 1685 return ino == EXT4_ROOT_INO || 1686 (ino >= EXT4_FIRST_INO(sb) && 1687 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1688 } 1689 1690 /* 1691 * Returns: sbi->field[index] 1692 * Used to access an array element from the following sbi fields which require 1693 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1694 * - s_group_desc 1695 * - s_group_info 1696 * - s_flex_group 1697 */ 1698 #define sbi_array_rcu_deref(sbi, field, index) \ 1699 ({ \ 1700 typeof(*((sbi)->field)) _v; \ 1701 rcu_read_lock(); \ 1702 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1703 rcu_read_unlock(); \ 1704 _v; \ 1705 }) 1706 1707 /* 1708 * run-time mount flags 1709 */ 1710 enum { 1711 EXT4_MF_MNTDIR_SAMPLED, 1712 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1713 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1714 EXT4_MF_FC_COMMITTING /* File system underoing a fast 1715 * commit. 1716 */ 1717 }; 1718 1719 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1720 { 1721 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1722 } 1723 1724 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1725 { 1726 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1727 } 1728 1729 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1730 { 1731 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1732 } 1733 1734 1735 /* 1736 * Simulate_fail codes 1737 */ 1738 #define EXT4_SIM_BBITMAP_EIO 1 1739 #define EXT4_SIM_BBITMAP_CRC 2 1740 #define EXT4_SIM_IBITMAP_EIO 3 1741 #define EXT4_SIM_IBITMAP_CRC 4 1742 #define EXT4_SIM_INODE_EIO 5 1743 #define EXT4_SIM_INODE_CRC 6 1744 #define EXT4_SIM_DIRBLOCK_EIO 7 1745 #define EXT4_SIM_DIRBLOCK_CRC 8 1746 1747 static inline bool ext4_simulate_fail(struct super_block *sb, 1748 unsigned long code) 1749 { 1750 #ifdef CONFIG_EXT4_DEBUG 1751 struct ext4_sb_info *sbi = EXT4_SB(sb); 1752 1753 if (unlikely(sbi->s_simulate_fail == code)) { 1754 sbi->s_simulate_fail = 0; 1755 return true; 1756 } 1757 #endif 1758 return false; 1759 } 1760 1761 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1762 struct buffer_head *bh, 1763 unsigned long code) 1764 { 1765 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1766 clear_buffer_uptodate(bh); 1767 } 1768 1769 /* 1770 * Error number codes for s_{first,last}_error_errno 1771 * 1772 * Linux errno numbers are architecture specific, so we need to translate 1773 * them into something which is architecture independent. We don't define 1774 * codes for all errno's; just the ones which are most likely to be the cause 1775 * of an ext4_error() call. 1776 */ 1777 #define EXT4_ERR_UNKNOWN 1 1778 #define EXT4_ERR_EIO 2 1779 #define EXT4_ERR_ENOMEM 3 1780 #define EXT4_ERR_EFSBADCRC 4 1781 #define EXT4_ERR_EFSCORRUPTED 5 1782 #define EXT4_ERR_ENOSPC 6 1783 #define EXT4_ERR_ENOKEY 7 1784 #define EXT4_ERR_EROFS 8 1785 #define EXT4_ERR_EFBIG 9 1786 #define EXT4_ERR_EEXIST 10 1787 #define EXT4_ERR_ERANGE 11 1788 #define EXT4_ERR_EOVERFLOW 12 1789 #define EXT4_ERR_EBUSY 13 1790 #define EXT4_ERR_ENOTDIR 14 1791 #define EXT4_ERR_ENOTEMPTY 15 1792 #define EXT4_ERR_ESHUTDOWN 16 1793 #define EXT4_ERR_EFAULT 17 1794 1795 /* 1796 * Inode dynamic state flags 1797 */ 1798 enum { 1799 EXT4_STATE_JDATA, /* journaled data exists */ 1800 EXT4_STATE_NEW, /* inode is newly created */ 1801 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1802 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1803 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1804 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1805 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1806 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1807 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1808 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1809 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1810 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1811 }; 1812 1813 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1814 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1815 { \ 1816 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1817 } \ 1818 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1819 { \ 1820 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1821 } \ 1822 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1823 { \ 1824 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1825 } 1826 1827 /* Add these declarations here only so that these functions can be 1828 * found by name. Otherwise, they are very hard to locate. */ 1829 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1830 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1831 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1832 EXT4_INODE_BIT_FNS(flag, flags, 0) 1833 1834 /* Add these declarations here only so that these functions can be 1835 * found by name. Otherwise, they are very hard to locate. */ 1836 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1837 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1838 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1839 #if (BITS_PER_LONG < 64) 1840 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1841 1842 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1843 { 1844 (ei)->i_state_flags = 0; 1845 } 1846 #else 1847 EXT4_INODE_BIT_FNS(state, flags, 32) 1848 1849 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1850 { 1851 /* We depend on the fact that callers will set i_flags */ 1852 } 1853 #endif 1854 #else 1855 /* Assume that user mode programs are passing in an ext4fs superblock, not 1856 * a kernel struct super_block. This will allow us to call the feature-test 1857 * macros from user land. */ 1858 #define EXT4_SB(sb) (sb) 1859 #endif 1860 1861 static inline bool ext4_verity_in_progress(struct inode *inode) 1862 { 1863 return IS_ENABLED(CONFIG_FS_VERITY) && 1864 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1865 } 1866 1867 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1868 1869 /* 1870 * Codes for operating systems 1871 */ 1872 #define EXT4_OS_LINUX 0 1873 #define EXT4_OS_HURD 1 1874 #define EXT4_OS_MASIX 2 1875 #define EXT4_OS_FREEBSD 3 1876 #define EXT4_OS_LITES 4 1877 1878 /* 1879 * Revision levels 1880 */ 1881 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1882 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1883 1884 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1885 1886 #define EXT4_GOOD_OLD_INODE_SIZE 128 1887 1888 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1889 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1890 #define EXT4_TIMESTAMP_MIN S32_MIN 1891 1892 /* 1893 * Feature set definitions 1894 */ 1895 1896 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1897 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1898 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1899 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1900 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1901 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1902 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1903 /* 1904 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1905 * incompatible only if fast commit blocks are present in the FS. Since we 1906 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1907 * incompatible. We also have a JBD2 incompat feature, which gets set when 1908 * there are fast commit blocks present in the journal. 1909 */ 1910 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1911 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1912 1913 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1914 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1915 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1916 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1917 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1918 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1919 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1920 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1921 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1922 /* 1923 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1924 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1925 * all other data structures' checksums. However, the METADATA_CSUM and 1926 * GDT_CSUM bits are mutually exclusive. 1927 */ 1928 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1929 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1930 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1931 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1932 1933 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1934 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1935 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1936 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1937 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1938 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1939 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1940 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1941 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1942 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1943 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1944 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1945 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1946 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1947 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1948 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1949 1950 extern void ext4_update_dynamic_rev(struct super_block *sb); 1951 1952 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1953 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1954 { \ 1955 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1956 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1957 } \ 1958 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1959 { \ 1960 ext4_update_dynamic_rev(sb); \ 1961 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1962 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1963 } \ 1964 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1965 { \ 1966 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1967 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1968 } 1969 1970 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1971 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1972 { \ 1973 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1974 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1975 } \ 1976 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1977 { \ 1978 ext4_update_dynamic_rev(sb); \ 1979 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1980 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1981 } \ 1982 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1983 { \ 1984 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1985 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1986 } 1987 1988 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1989 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1990 { \ 1991 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1992 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1993 } \ 1994 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1995 { \ 1996 ext4_update_dynamic_rev(sb); \ 1997 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1998 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1999 } \ 2000 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2001 { \ 2002 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2003 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2004 } 2005 2006 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2007 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2008 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2009 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2010 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2011 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2012 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2013 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2014 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2015 2016 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2017 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2018 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2019 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2020 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2021 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2022 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2023 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2024 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2025 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2026 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2027 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2028 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2029 2030 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2031 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2032 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2033 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2034 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2035 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2036 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2037 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2038 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2039 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2040 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2041 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2042 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2043 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2044 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2045 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2046 2047 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2048 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2049 EXT4_FEATURE_INCOMPAT_META_BG) 2050 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2051 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2052 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2053 2054 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2055 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2056 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2057 EXT4_FEATURE_INCOMPAT_META_BG) 2058 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2059 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2060 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2061 2062 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2063 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2064 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2065 EXT4_FEATURE_INCOMPAT_META_BG| \ 2066 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2067 EXT4_FEATURE_INCOMPAT_64BIT| \ 2068 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2069 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2070 EXT4_FEATURE_INCOMPAT_MMP | \ 2071 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2072 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2073 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2074 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2075 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2076 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2077 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2078 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2079 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2080 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2081 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2082 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2083 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2084 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2085 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2086 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2087 EXT4_FEATURE_RO_COMPAT_VERITY) 2088 2089 #define EXTN_FEATURE_FUNCS(ver) \ 2090 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2091 { \ 2092 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2093 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2094 } \ 2095 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2096 { \ 2097 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2098 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2099 } \ 2100 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2101 { \ 2102 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2103 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2104 } 2105 2106 EXTN_FEATURE_FUNCS(2) 2107 EXTN_FEATURE_FUNCS(3) 2108 EXTN_FEATURE_FUNCS(4) 2109 2110 static inline bool ext4_has_compat_features(struct super_block *sb) 2111 { 2112 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2113 } 2114 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2115 { 2116 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2117 } 2118 static inline bool ext4_has_incompat_features(struct super_block *sb) 2119 { 2120 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2121 } 2122 2123 /* 2124 * Superblock flags 2125 */ 2126 #define EXT4_FLAGS_RESIZING 0 2127 #define EXT4_FLAGS_SHUTDOWN 1 2128 #define EXT4_FLAGS_BDEV_IS_DAX 2 2129 2130 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2131 { 2132 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2133 } 2134 2135 2136 /* 2137 * Default values for user and/or group using reserved blocks 2138 */ 2139 #define EXT4_DEF_RESUID 0 2140 #define EXT4_DEF_RESGID 0 2141 2142 /* 2143 * Default project ID 2144 */ 2145 #define EXT4_DEF_PROJID 0 2146 2147 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2148 2149 /* 2150 * Default mount options 2151 */ 2152 #define EXT4_DEFM_DEBUG 0x0001 2153 #define EXT4_DEFM_BSDGROUPS 0x0002 2154 #define EXT4_DEFM_XATTR_USER 0x0004 2155 #define EXT4_DEFM_ACL 0x0008 2156 #define EXT4_DEFM_UID16 0x0010 2157 #define EXT4_DEFM_JMODE 0x0060 2158 #define EXT4_DEFM_JMODE_DATA 0x0020 2159 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2160 #define EXT4_DEFM_JMODE_WBACK 0x0060 2161 #define EXT4_DEFM_NOBARRIER 0x0100 2162 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2163 #define EXT4_DEFM_DISCARD 0x0400 2164 #define EXT4_DEFM_NODELALLOC 0x0800 2165 2166 /* 2167 * Default journal batch times 2168 */ 2169 #define EXT4_DEF_MIN_BATCH_TIME 0 2170 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2171 2172 /* 2173 * Minimum number of groups in a flexgroup before we separate out 2174 * directories into the first block group of a flexgroup 2175 */ 2176 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2177 2178 /* 2179 * Structure of a directory entry 2180 */ 2181 #define EXT4_NAME_LEN 255 2182 2183 struct ext4_dir_entry { 2184 __le32 inode; /* Inode number */ 2185 __le16 rec_len; /* Directory entry length */ 2186 __le16 name_len; /* Name length */ 2187 char name[EXT4_NAME_LEN]; /* File name */ 2188 }; 2189 2190 /* 2191 * The new version of the directory entry. Since EXT4 structures are 2192 * stored in intel byte order, and the name_len field could never be 2193 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2194 * file_type field. 2195 */ 2196 struct ext4_dir_entry_2 { 2197 __le32 inode; /* Inode number */ 2198 __le16 rec_len; /* Directory entry length */ 2199 __u8 name_len; /* Name length */ 2200 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2201 char name[EXT4_NAME_LEN]; /* File name */ 2202 }; 2203 2204 /* 2205 * This is a bogus directory entry at the end of each leaf block that 2206 * records checksums. 2207 */ 2208 struct ext4_dir_entry_tail { 2209 __le32 det_reserved_zero1; /* Pretend to be unused */ 2210 __le16 det_rec_len; /* 12 */ 2211 __u8 det_reserved_zero2; /* Zero name length */ 2212 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2213 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2214 }; 2215 2216 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2217 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2218 ((blocksize) - \ 2219 sizeof(struct ext4_dir_entry_tail)))) 2220 2221 /* 2222 * Ext4 directory file types. Only the low 3 bits are used. The 2223 * other bits are reserved for now. 2224 */ 2225 #define EXT4_FT_UNKNOWN 0 2226 #define EXT4_FT_REG_FILE 1 2227 #define EXT4_FT_DIR 2 2228 #define EXT4_FT_CHRDEV 3 2229 #define EXT4_FT_BLKDEV 4 2230 #define EXT4_FT_FIFO 5 2231 #define EXT4_FT_SOCK 6 2232 #define EXT4_FT_SYMLINK 7 2233 2234 #define EXT4_FT_MAX 8 2235 2236 #define EXT4_FT_DIR_CSUM 0xDE 2237 2238 /* 2239 * EXT4_DIR_PAD defines the directory entries boundaries 2240 * 2241 * NOTE: It must be a multiple of 4 2242 */ 2243 #define EXT4_DIR_PAD 4 2244 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2245 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 2246 ~EXT4_DIR_ROUND) 2247 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2248 2249 /* 2250 * If we ever get support for fs block sizes > page_size, we'll need 2251 * to remove the #if statements in the next two functions... 2252 */ 2253 static inline unsigned int 2254 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2255 { 2256 unsigned len = le16_to_cpu(dlen); 2257 2258 #if (PAGE_SIZE >= 65536) 2259 if (len == EXT4_MAX_REC_LEN || len == 0) 2260 return blocksize; 2261 return (len & 65532) | ((len & 3) << 16); 2262 #else 2263 return len; 2264 #endif 2265 } 2266 2267 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2268 { 2269 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2270 BUG(); 2271 #if (PAGE_SIZE >= 65536) 2272 if (len < 65536) 2273 return cpu_to_le16(len); 2274 if (len == blocksize) { 2275 if (blocksize == 65536) 2276 return cpu_to_le16(EXT4_MAX_REC_LEN); 2277 else 2278 return cpu_to_le16(0); 2279 } 2280 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2281 #else 2282 return cpu_to_le16(len); 2283 #endif 2284 } 2285 2286 /* 2287 * Hash Tree Directory indexing 2288 * (c) Daniel Phillips, 2001 2289 */ 2290 2291 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2292 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2293 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2294 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2295 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2296 2297 /* Legal values for the dx_root hash_version field: */ 2298 2299 #define DX_HASH_LEGACY 0 2300 #define DX_HASH_HALF_MD4 1 2301 #define DX_HASH_TEA 2 2302 #define DX_HASH_LEGACY_UNSIGNED 3 2303 #define DX_HASH_HALF_MD4_UNSIGNED 4 2304 #define DX_HASH_TEA_UNSIGNED 5 2305 2306 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2307 const void *address, unsigned int length) 2308 { 2309 struct { 2310 struct shash_desc shash; 2311 char ctx[4]; 2312 } desc; 2313 2314 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2315 2316 desc.shash.tfm = sbi->s_chksum_driver; 2317 *(u32 *)desc.ctx = crc; 2318 2319 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2320 2321 return *(u32 *)desc.ctx; 2322 } 2323 2324 #ifdef __KERNEL__ 2325 2326 /* hash info structure used by the directory hash */ 2327 struct dx_hash_info 2328 { 2329 u32 hash; 2330 u32 minor_hash; 2331 int hash_version; 2332 u32 *seed; 2333 }; 2334 2335 2336 /* 32 and 64 bit signed EOF for dx directories */ 2337 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2338 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2339 2340 2341 /* 2342 * Control parameters used by ext4_htree_next_block 2343 */ 2344 #define HASH_NB_ALWAYS 1 2345 2346 struct ext4_filename { 2347 const struct qstr *usr_fname; 2348 struct fscrypt_str disk_name; 2349 struct dx_hash_info hinfo; 2350 #ifdef CONFIG_FS_ENCRYPTION 2351 struct fscrypt_str crypto_buf; 2352 #endif 2353 #ifdef CONFIG_UNICODE 2354 struct fscrypt_str cf_name; 2355 #endif 2356 }; 2357 2358 #define fname_name(p) ((p)->disk_name.name) 2359 #define fname_len(p) ((p)->disk_name.len) 2360 2361 /* 2362 * Describe an inode's exact location on disk and in memory 2363 */ 2364 struct ext4_iloc 2365 { 2366 struct buffer_head *bh; 2367 unsigned long offset; 2368 ext4_group_t block_group; 2369 }; 2370 2371 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2372 { 2373 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2374 } 2375 2376 static inline bool ext4_is_quota_file(struct inode *inode) 2377 { 2378 return IS_NOQUOTA(inode) && 2379 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2380 } 2381 2382 /* 2383 * This structure is stuffed into the struct file's private_data field 2384 * for directories. It is where we put information so that we can do 2385 * readdir operations in hash tree order. 2386 */ 2387 struct dir_private_info { 2388 struct rb_root root; 2389 struct rb_node *curr_node; 2390 struct fname *extra_fname; 2391 loff_t last_pos; 2392 __u32 curr_hash; 2393 __u32 curr_minor_hash; 2394 __u32 next_hash; 2395 }; 2396 2397 /* calculate the first block number of the group */ 2398 static inline ext4_fsblk_t 2399 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2400 { 2401 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2402 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2403 } 2404 2405 /* 2406 * Special error return code only used by dx_probe() and its callers. 2407 */ 2408 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2409 2410 /* htree levels for ext4 */ 2411 #define EXT4_HTREE_LEVEL_COMPAT 2 2412 #define EXT4_HTREE_LEVEL 3 2413 2414 static inline int ext4_dir_htree_level(struct super_block *sb) 2415 { 2416 return ext4_has_feature_largedir(sb) ? 2417 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2418 } 2419 2420 /* 2421 * Timeout and state flag for lazy initialization inode thread. 2422 */ 2423 #define EXT4_DEF_LI_WAIT_MULT 10 2424 #define EXT4_DEF_LI_MAX_START_DELAY 5 2425 #define EXT4_LAZYINIT_QUIT 0x0001 2426 #define EXT4_LAZYINIT_RUNNING 0x0002 2427 2428 /* 2429 * Lazy inode table initialization info 2430 */ 2431 struct ext4_lazy_init { 2432 unsigned long li_state; 2433 struct list_head li_request_list; 2434 struct mutex li_list_mtx; 2435 }; 2436 2437 enum ext4_li_mode { 2438 EXT4_LI_MODE_PREFETCH_BBITMAP, 2439 EXT4_LI_MODE_ITABLE, 2440 }; 2441 2442 struct ext4_li_request { 2443 struct super_block *lr_super; 2444 enum ext4_li_mode lr_mode; 2445 ext4_group_t lr_first_not_zeroed; 2446 ext4_group_t lr_next_group; 2447 struct list_head lr_request; 2448 unsigned long lr_next_sched; 2449 unsigned long lr_timeout; 2450 }; 2451 2452 struct ext4_features { 2453 struct kobject f_kobj; 2454 struct completion f_kobj_unregister; 2455 }; 2456 2457 /* 2458 * This structure will be used for multiple mount protection. It will be 2459 * written into the block number saved in the s_mmp_block field in the 2460 * superblock. Programs that check MMP should assume that if 2461 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2462 * to use the filesystem, regardless of how old the timestamp is. 2463 */ 2464 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2465 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2466 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2467 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2468 2469 struct mmp_struct { 2470 __le32 mmp_magic; /* Magic number for MMP */ 2471 __le32 mmp_seq; /* Sequence no. updated periodically */ 2472 2473 /* 2474 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2475 * purposes and do not affect the correctness of the algorithm 2476 */ 2477 __le64 mmp_time; /* Time last updated */ 2478 char mmp_nodename[64]; /* Node which last updated MMP block */ 2479 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2480 2481 /* 2482 * mmp_check_interval is used to verify if the MMP block has been 2483 * updated on the block device. The value is updated based on the 2484 * maximum time to write the MMP block during an update cycle. 2485 */ 2486 __le16 mmp_check_interval; 2487 2488 __le16 mmp_pad1; 2489 __le32 mmp_pad2[226]; 2490 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2491 }; 2492 2493 /* arguments passed to the mmp thread */ 2494 struct mmpd_data { 2495 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2496 struct super_block *sb; /* super block of the fs */ 2497 }; 2498 2499 /* 2500 * Check interval multiplier 2501 * The MMP block is written every update interval and initially checked every 2502 * update interval x the multiplier (the value is then adapted based on the 2503 * write latency). The reason is that writes can be delayed under load and we 2504 * don't want readers to incorrectly assume that the filesystem is no longer 2505 * in use. 2506 */ 2507 #define EXT4_MMP_CHECK_MULT 2UL 2508 2509 /* 2510 * Minimum interval for MMP checking in seconds. 2511 */ 2512 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2513 2514 /* 2515 * Maximum interval for MMP checking in seconds. 2516 */ 2517 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2518 2519 /* 2520 * Function prototypes 2521 */ 2522 2523 /* 2524 * Ok, these declarations are also in <linux/kernel.h> but none of the 2525 * ext4 source programs needs to include it so they are duplicated here. 2526 */ 2527 # define NORET_TYPE /**/ 2528 # define ATTRIB_NORET __attribute__((noreturn)) 2529 # define NORET_AND noreturn, 2530 2531 /* bitmap.c */ 2532 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2533 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2534 struct ext4_group_desc *gdp, 2535 struct buffer_head *bh, int sz); 2536 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2537 struct ext4_group_desc *gdp, 2538 struct buffer_head *bh, int sz); 2539 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2540 struct ext4_group_desc *gdp, 2541 struct buffer_head *bh); 2542 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2543 struct ext4_group_desc *gdp, 2544 struct buffer_head *bh); 2545 2546 /* balloc.c */ 2547 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2548 ext4_fsblk_t blocknr, 2549 ext4_group_t *blockgrpp, 2550 ext4_grpblk_t *offsetp); 2551 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2552 ext4_fsblk_t block); 2553 2554 extern unsigned int ext4_block_group(struct super_block *sb, 2555 ext4_fsblk_t blocknr); 2556 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2557 ext4_fsblk_t blocknr); 2558 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2559 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2560 ext4_group_t group); 2561 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2562 ext4_fsblk_t goal, 2563 unsigned int flags, 2564 unsigned long *count, 2565 int *errp); 2566 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2567 s64 nclusters, unsigned int flags); 2568 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2569 extern void ext4_check_blocks_bitmap(struct super_block *); 2570 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2571 ext4_group_t block_group, 2572 struct buffer_head ** bh); 2573 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2574 2575 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2576 ext4_group_t block_group, 2577 bool ignore_locked); 2578 extern int ext4_wait_block_bitmap(struct super_block *sb, 2579 ext4_group_t block_group, 2580 struct buffer_head *bh); 2581 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2582 ext4_group_t block_group); 2583 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2584 ext4_group_t block_group, 2585 struct ext4_group_desc *gdp); 2586 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2587 2588 #ifdef CONFIG_UNICODE 2589 extern void ext4_fname_setup_ci_filename(struct inode *dir, 2590 const struct qstr *iname, 2591 struct fscrypt_str *fname); 2592 #endif 2593 2594 #ifdef CONFIG_FS_ENCRYPTION 2595 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst, 2596 const struct fscrypt_name *src) 2597 { 2598 memset(dst, 0, sizeof(*dst)); 2599 2600 dst->usr_fname = src->usr_fname; 2601 dst->disk_name = src->disk_name; 2602 dst->hinfo.hash = src->hash; 2603 dst->hinfo.minor_hash = src->minor_hash; 2604 dst->crypto_buf = src->crypto_buf; 2605 } 2606 2607 static inline int ext4_fname_setup_filename(struct inode *dir, 2608 const struct qstr *iname, 2609 int lookup, 2610 struct ext4_filename *fname) 2611 { 2612 struct fscrypt_name name; 2613 int err; 2614 2615 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2616 if (err) 2617 return err; 2618 2619 ext4_fname_from_fscrypt_name(fname, &name); 2620 2621 #ifdef CONFIG_UNICODE 2622 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2623 #endif 2624 return 0; 2625 } 2626 2627 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2628 struct dentry *dentry, 2629 struct ext4_filename *fname) 2630 { 2631 struct fscrypt_name name; 2632 int err; 2633 2634 err = fscrypt_prepare_lookup(dir, dentry, &name); 2635 if (err) 2636 return err; 2637 2638 ext4_fname_from_fscrypt_name(fname, &name); 2639 2640 #ifdef CONFIG_UNICODE 2641 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name); 2642 #endif 2643 return 0; 2644 } 2645 2646 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2647 { 2648 struct fscrypt_name name; 2649 2650 name.crypto_buf = fname->crypto_buf; 2651 fscrypt_free_filename(&name); 2652 2653 fname->crypto_buf.name = NULL; 2654 fname->usr_fname = NULL; 2655 fname->disk_name.name = NULL; 2656 2657 #ifdef CONFIG_UNICODE 2658 kfree(fname->cf_name.name); 2659 fname->cf_name.name = NULL; 2660 #endif 2661 } 2662 #else /* !CONFIG_FS_ENCRYPTION */ 2663 static inline int ext4_fname_setup_filename(struct inode *dir, 2664 const struct qstr *iname, 2665 int lookup, 2666 struct ext4_filename *fname) 2667 { 2668 fname->usr_fname = iname; 2669 fname->disk_name.name = (unsigned char *) iname->name; 2670 fname->disk_name.len = iname->len; 2671 2672 #ifdef CONFIG_UNICODE 2673 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2674 #endif 2675 2676 return 0; 2677 } 2678 2679 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2680 struct dentry *dentry, 2681 struct ext4_filename *fname) 2682 { 2683 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2684 } 2685 2686 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2687 { 2688 #ifdef CONFIG_UNICODE 2689 kfree(fname->cf_name.name); 2690 fname->cf_name.name = NULL; 2691 #endif 2692 } 2693 #endif /* !CONFIG_FS_ENCRYPTION */ 2694 2695 /* dir.c */ 2696 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2697 struct file *, 2698 struct ext4_dir_entry_2 *, 2699 struct buffer_head *, char *, int, 2700 unsigned int); 2701 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2702 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2703 (de), (bh), (buf), (size), (offset))) 2704 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2705 __u32 minor_hash, 2706 struct ext4_dir_entry_2 *dirent, 2707 struct fscrypt_str *ent_name); 2708 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2709 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2710 struct buffer_head *bh, 2711 void *buf, int buf_size, 2712 struct ext4_filename *fname, 2713 struct ext4_dir_entry_2 **dest_de); 2714 void ext4_insert_dentry(struct inode *inode, 2715 struct ext4_dir_entry_2 *de, 2716 int buf_size, 2717 struct ext4_filename *fname); 2718 static inline void ext4_update_dx_flag(struct inode *inode) 2719 { 2720 if (!ext4_has_feature_dir_index(inode->i_sb) && 2721 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2722 /* ext4_iget() should have caught this... */ 2723 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2724 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2725 } 2726 } 2727 static const unsigned char ext4_filetype_table[] = { 2728 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2729 }; 2730 2731 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2732 { 2733 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2734 return DT_UNKNOWN; 2735 2736 return ext4_filetype_table[filetype]; 2737 } 2738 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2739 void *buf, int buf_size); 2740 2741 /* fsync.c */ 2742 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2743 2744 /* hash.c */ 2745 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2746 struct dx_hash_info *hinfo); 2747 2748 /* ialloc.c */ 2749 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2750 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *, 2751 struct inode *, umode_t, 2752 const struct qstr *qstr, __u32 goal, 2753 uid_t *owner, __u32 i_flags, 2754 int handle_type, unsigned int line_no, 2755 int nblocks); 2756 2757 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2758 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \ 2759 (goal), (owner), i_flags, 0, 0, 0) 2760 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \ 2761 type, nblocks) \ 2762 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2763 0, (type), __LINE__, (nblocks)) 2764 2765 2766 extern void ext4_free_inode(handle_t *, struct inode *); 2767 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2768 extern unsigned long ext4_count_free_inodes(struct super_block *); 2769 extern unsigned long ext4_count_dirs(struct super_block *); 2770 extern void ext4_check_inodes_bitmap(struct super_block *); 2771 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2772 extern int ext4_init_inode_table(struct super_block *sb, 2773 ext4_group_t group, int barrier); 2774 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2775 2776 /* fast_commit.c */ 2777 int ext4_fc_info_show(struct seq_file *seq, void *v); 2778 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2779 void ext4_fc_init_inode(struct inode *inode); 2780 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2781 ext4_lblk_t end); 2782 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2783 struct dentry *dentry); 2784 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2785 struct dentry *dentry); 2786 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2787 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2788 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2789 struct dentry *dentry); 2790 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2791 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2792 void ext4_fc_mark_ineligible(struct super_block *sb, int reason); 2793 void ext4_fc_start_ineligible(struct super_block *sb, int reason); 2794 void ext4_fc_stop_ineligible(struct super_block *sb); 2795 void ext4_fc_start_update(struct inode *inode); 2796 void ext4_fc_stop_update(struct inode *inode); 2797 void ext4_fc_del(struct inode *inode); 2798 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2799 void ext4_fc_replay_cleanup(struct super_block *sb); 2800 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2801 int __init ext4_fc_init_dentry_cache(void); 2802 2803 /* mballoc.c */ 2804 extern const struct seq_operations ext4_mb_seq_groups_ops; 2805 extern long ext4_mb_stats; 2806 extern long ext4_mb_max_to_scan; 2807 extern int ext4_mb_init(struct super_block *); 2808 extern int ext4_mb_release(struct super_block *); 2809 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2810 struct ext4_allocation_request *, int *); 2811 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2812 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2813 extern int __init ext4_init_mballoc(void); 2814 extern void ext4_exit_mballoc(void); 2815 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2816 ext4_group_t group, 2817 unsigned int nr, int *cnt); 2818 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2819 unsigned int nr); 2820 2821 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2822 struct buffer_head *bh, ext4_fsblk_t block, 2823 unsigned long count, int flags); 2824 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2825 ext4_group_t ngroups); 2826 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2827 ext4_group_t i, struct ext4_group_desc *desc); 2828 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2829 ext4_fsblk_t block, unsigned long count); 2830 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2831 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2832 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2833 int len, int state); 2834 2835 /* inode.c */ 2836 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2837 struct ext4_inode_info *ei); 2838 int ext4_inode_is_fast_symlink(struct inode *inode); 2839 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2840 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2841 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2842 bool wait, struct buffer_head **bhs); 2843 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2844 struct buffer_head *bh_result, int create); 2845 int ext4_get_block(struct inode *inode, sector_t iblock, 2846 struct buffer_head *bh_result, int create); 2847 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2848 struct buffer_head *bh, int create); 2849 int ext4_walk_page_buffers(handle_t *handle, 2850 struct buffer_head *head, 2851 unsigned from, 2852 unsigned to, 2853 int *partial, 2854 int (*fn)(handle_t *handle, 2855 struct buffer_head *bh)); 2856 int do_journal_get_write_access(handle_t *handle, 2857 struct buffer_head *bh); 2858 #define FALL_BACK_TO_NONDELALLOC 1 2859 #define CONVERT_INLINE_DATA 2 2860 2861 typedef enum { 2862 EXT4_IGET_NORMAL = 0, 2863 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2864 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2865 } ext4_iget_flags; 2866 2867 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2868 ext4_iget_flags flags, const char *function, 2869 unsigned int line); 2870 2871 #define ext4_iget(sb, ino, flags) \ 2872 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2873 2874 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2875 extern int ext4_setattr(struct user_namespace *, struct dentry *, 2876 struct iattr *); 2877 extern int ext4_getattr(struct user_namespace *, const struct path *, 2878 struct kstat *, u32, unsigned int); 2879 extern void ext4_evict_inode(struct inode *); 2880 extern void ext4_clear_inode(struct inode *); 2881 extern int ext4_file_getattr(struct user_namespace *, const struct path *, 2882 struct kstat *, u32, unsigned int); 2883 extern int ext4_sync_inode(handle_t *, struct inode *); 2884 extern void ext4_dirty_inode(struct inode *, int); 2885 extern int ext4_change_inode_journal_flag(struct inode *, int); 2886 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2887 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2888 struct ext4_iloc *iloc); 2889 extern int ext4_inode_attach_jinode(struct inode *inode); 2890 extern int ext4_can_truncate(struct inode *inode); 2891 extern int ext4_truncate(struct inode *); 2892 extern int ext4_break_layouts(struct inode *); 2893 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2894 extern void ext4_set_inode_flags(struct inode *, bool init); 2895 extern int ext4_alloc_da_blocks(struct inode *inode); 2896 extern void ext4_set_aops(struct inode *inode); 2897 extern int ext4_writepage_trans_blocks(struct inode *); 2898 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2899 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2900 loff_t lstart, loff_t lend); 2901 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2902 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2903 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2904 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2905 extern void ext4_da_release_space(struct inode *inode, int to_free); 2906 extern void ext4_da_update_reserve_space(struct inode *inode, 2907 int used, int quota_claim); 2908 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2909 ext4_fsblk_t pblk, ext4_lblk_t len); 2910 2911 /* indirect.c */ 2912 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2913 struct ext4_map_blocks *map, int flags); 2914 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2915 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2916 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2917 ext4_lblk_t start, ext4_lblk_t end); 2918 2919 /* ioctl.c */ 2920 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2921 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2922 int ext4_fileattr_set(struct user_namespace *mnt_userns, 2923 struct dentry *dentry, struct fileattr *fa); 2924 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 2925 extern void ext4_reset_inode_seed(struct inode *inode); 2926 2927 /* migrate.c */ 2928 extern int ext4_ext_migrate(struct inode *); 2929 extern int ext4_ind_migrate(struct inode *inode); 2930 2931 /* namei.c */ 2932 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 2933 struct inode *inode); 2934 extern int ext4_dirblock_csum_verify(struct inode *inode, 2935 struct buffer_head *bh); 2936 extern int ext4_orphan_add(handle_t *, struct inode *); 2937 extern int ext4_orphan_del(handle_t *, struct inode *); 2938 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2939 __u32 start_minor_hash, __u32 *next_hash); 2940 extern int ext4_search_dir(struct buffer_head *bh, 2941 char *search_buf, 2942 int buf_size, 2943 struct inode *dir, 2944 struct ext4_filename *fname, 2945 unsigned int offset, 2946 struct ext4_dir_entry_2 **res_dir); 2947 extern int ext4_generic_delete_entry(struct inode *dir, 2948 struct ext4_dir_entry_2 *de_del, 2949 struct buffer_head *bh, 2950 void *entry_buf, 2951 int buf_size, 2952 int csum_size); 2953 extern bool ext4_empty_dir(struct inode *inode); 2954 2955 /* resize.c */ 2956 extern void ext4_kvfree_array_rcu(void *to_free); 2957 extern int ext4_group_add(struct super_block *sb, 2958 struct ext4_new_group_data *input); 2959 extern int ext4_group_extend(struct super_block *sb, 2960 struct ext4_super_block *es, 2961 ext4_fsblk_t n_blocks_count); 2962 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2963 2964 /* super.c */ 2965 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2966 sector_t block, int op_flags); 2967 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 2968 sector_t block); 2969 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags, 2970 bh_end_io_t *end_io); 2971 extern int ext4_read_bh(struct buffer_head *bh, int op_flags, 2972 bh_end_io_t *end_io); 2973 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait); 2974 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 2975 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2976 extern int ext4_calculate_overhead(struct super_block *sb); 2977 extern void ext4_superblock_csum_set(struct super_block *sb); 2978 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2979 ext4_group_t ngroup); 2980 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2981 char nbuf[16]); 2982 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2983 ext4_group_t block_group, 2984 unsigned int flags); 2985 2986 extern __printf(7, 8) 2987 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 2988 int, __u64, const char *, ...); 2989 extern __printf(6, 7) 2990 void __ext4_error_inode(struct inode *, const char *, unsigned int, 2991 ext4_fsblk_t, int, const char *, ...); 2992 extern __printf(5, 6) 2993 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2994 const char *, ...); 2995 extern void __ext4_std_error(struct super_block *, const char *, 2996 unsigned int, int); 2997 extern __printf(4, 5) 2998 void __ext4_warning(struct super_block *, const char *, unsigned int, 2999 const char *, ...); 3000 extern __printf(4, 5) 3001 void __ext4_warning_inode(const struct inode *inode, const char *function, 3002 unsigned int line, const char *fmt, ...); 3003 extern __printf(3, 4) 3004 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3005 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3006 const char *, unsigned int, const char *); 3007 extern __printf(7, 8) 3008 void __ext4_grp_locked_error(const char *, unsigned int, 3009 struct super_block *, ext4_group_t, 3010 unsigned long, ext4_fsblk_t, 3011 const char *, ...); 3012 3013 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3014 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3015 3016 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3017 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3018 3019 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3020 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3021 (fmt), ## a) 3022 3023 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3024 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3025 3026 #define ext4_abort(sb, err, fmt, a...) \ 3027 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3028 3029 #ifdef CONFIG_PRINTK 3030 3031 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3032 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3033 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3034 __ext4_error_inode((inode), (func), (line), (block), \ 3035 (err), (fmt), ##__VA_ARGS__) 3036 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3037 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3038 #define ext4_error(sb, fmt, ...) \ 3039 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3040 ##__VA_ARGS__) 3041 #define ext4_error_err(sb, err, fmt, ...) \ 3042 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3043 ##__VA_ARGS__) 3044 #define ext4_warning(sb, fmt, ...) \ 3045 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3046 #define ext4_warning_inode(inode, fmt, ...) \ 3047 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3048 #define ext4_msg(sb, level, fmt, ...) \ 3049 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3050 #define dump_mmp_msg(sb, mmp, msg) \ 3051 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3052 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3053 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3054 fmt, ##__VA_ARGS__) 3055 3056 #else 3057 3058 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3059 do { \ 3060 no_printk(fmt, ##__VA_ARGS__); \ 3061 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3062 } while (0) 3063 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3064 do { \ 3065 no_printk(fmt, ##__VA_ARGS__); \ 3066 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3067 } while (0) 3068 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3069 do { \ 3070 no_printk(fmt, ##__VA_ARGS__); \ 3071 __ext4_error_file(file, "", 0, block, " "); \ 3072 } while (0) 3073 #define ext4_error(sb, fmt, ...) \ 3074 do { \ 3075 no_printk(fmt, ##__VA_ARGS__); \ 3076 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3077 } while (0) 3078 #define ext4_error_err(sb, err, fmt, ...) \ 3079 do { \ 3080 no_printk(fmt, ##__VA_ARGS__); \ 3081 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3082 } while (0) 3083 #define ext4_warning(sb, fmt, ...) \ 3084 do { \ 3085 no_printk(fmt, ##__VA_ARGS__); \ 3086 __ext4_warning(sb, "", 0, " "); \ 3087 } while (0) 3088 #define ext4_warning_inode(inode, fmt, ...) \ 3089 do { \ 3090 no_printk(fmt, ##__VA_ARGS__); \ 3091 __ext4_warning_inode(inode, "", 0, " "); \ 3092 } while (0) 3093 #define ext4_msg(sb, level, fmt, ...) \ 3094 do { \ 3095 no_printk(fmt, ##__VA_ARGS__); \ 3096 __ext4_msg(sb, "", " "); \ 3097 } while (0) 3098 #define dump_mmp_msg(sb, mmp, msg) \ 3099 __dump_mmp_msg(sb, mmp, "", 0, "") 3100 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3101 do { \ 3102 no_printk(fmt, ##__VA_ARGS__); \ 3103 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3104 } while (0) 3105 3106 #endif 3107 3108 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3109 struct ext4_group_desc *bg); 3110 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3111 struct ext4_group_desc *bg); 3112 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3113 struct ext4_group_desc *bg); 3114 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3115 struct ext4_group_desc *bg); 3116 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3117 struct ext4_group_desc *bg); 3118 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3119 struct ext4_group_desc *bg); 3120 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3121 struct ext4_group_desc *bg); 3122 extern void ext4_block_bitmap_set(struct super_block *sb, 3123 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3124 extern void ext4_inode_bitmap_set(struct super_block *sb, 3125 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3126 extern void ext4_inode_table_set(struct super_block *sb, 3127 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3128 extern void ext4_free_group_clusters_set(struct super_block *sb, 3129 struct ext4_group_desc *bg, 3130 __u32 count); 3131 extern void ext4_free_inodes_set(struct super_block *sb, 3132 struct ext4_group_desc *bg, __u32 count); 3133 extern void ext4_used_dirs_set(struct super_block *sb, 3134 struct ext4_group_desc *bg, __u32 count); 3135 extern void ext4_itable_unused_set(struct super_block *sb, 3136 struct ext4_group_desc *bg, __u32 count); 3137 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3138 struct ext4_group_desc *gdp); 3139 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3140 struct ext4_group_desc *gdp); 3141 extern int ext4_register_li_request(struct super_block *sb, 3142 ext4_group_t first_not_zeroed); 3143 3144 static inline int ext4_has_metadata_csum(struct super_block *sb) 3145 { 3146 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3147 !EXT4_SB(sb)->s_chksum_driver); 3148 3149 return ext4_has_feature_metadata_csum(sb) && 3150 (EXT4_SB(sb)->s_chksum_driver != NULL); 3151 } 3152 3153 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3154 { 3155 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3156 } 3157 3158 #define ext4_read_incompat_64bit_val(es, name) \ 3159 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3160 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3161 le32_to_cpu(es->name##_lo)) 3162 3163 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3164 { 3165 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3166 } 3167 3168 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3169 { 3170 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3171 } 3172 3173 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3174 { 3175 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3176 } 3177 3178 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3179 ext4_fsblk_t blk) 3180 { 3181 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3182 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3183 } 3184 3185 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3186 ext4_fsblk_t blk) 3187 { 3188 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3189 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3190 } 3191 3192 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3193 ext4_fsblk_t blk) 3194 { 3195 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3196 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3197 } 3198 3199 static inline loff_t ext4_isize(struct super_block *sb, 3200 struct ext4_inode *raw_inode) 3201 { 3202 if (ext4_has_feature_largedir(sb) || 3203 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3204 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3205 le32_to_cpu(raw_inode->i_size_lo); 3206 3207 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3208 } 3209 3210 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3211 { 3212 raw_inode->i_size_lo = cpu_to_le32(i_size); 3213 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3214 } 3215 3216 static inline 3217 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3218 ext4_group_t group) 3219 { 3220 struct ext4_group_info **grp_info; 3221 long indexv, indexh; 3222 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3223 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3224 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3225 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3226 return grp_info[indexh]; 3227 } 3228 3229 /* 3230 * Reading s_groups_count requires using smp_rmb() afterwards. See 3231 * the locking protocol documented in the comments of ext4_group_add() 3232 * in resize.c 3233 */ 3234 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3235 { 3236 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3237 3238 smp_rmb(); 3239 return ngroups; 3240 } 3241 3242 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3243 ext4_group_t block_group) 3244 { 3245 return block_group >> sbi->s_log_groups_per_flex; 3246 } 3247 3248 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3249 { 3250 return 1 << sbi->s_log_groups_per_flex; 3251 } 3252 3253 #define ext4_std_error(sb, errno) \ 3254 do { \ 3255 if ((errno)) \ 3256 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3257 } while (0) 3258 3259 #ifdef CONFIG_SMP 3260 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3261 * counters. So we need to make sure we have free clusters more 3262 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3263 */ 3264 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3265 #else 3266 #define EXT4_FREECLUSTERS_WATERMARK 0 3267 #endif 3268 3269 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 3270 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3271 { 3272 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3273 !inode_is_locked(inode)); 3274 down_write(&EXT4_I(inode)->i_data_sem); 3275 if (newsize > EXT4_I(inode)->i_disksize) 3276 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3277 up_write(&EXT4_I(inode)->i_data_sem); 3278 } 3279 3280 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 3281 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3282 { 3283 int changed = 0; 3284 3285 if (newsize > inode->i_size) { 3286 i_size_write(inode, newsize); 3287 changed = 1; 3288 } 3289 if (newsize > EXT4_I(inode)->i_disksize) { 3290 ext4_update_i_disksize(inode, newsize); 3291 changed |= 2; 3292 } 3293 return changed; 3294 } 3295 3296 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3297 loff_t len); 3298 3299 struct ext4_group_info { 3300 unsigned long bb_state; 3301 #ifdef AGGRESSIVE_CHECK 3302 unsigned long bb_check_counter; 3303 #endif 3304 struct rb_root bb_free_root; 3305 ext4_grpblk_t bb_first_free; /* first free block */ 3306 ext4_grpblk_t bb_free; /* total free blocks */ 3307 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3308 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3309 struct list_head bb_prealloc_list; 3310 #ifdef DOUBLE_CHECK 3311 void *bb_bitmap; 3312 #endif 3313 struct rw_semaphore alloc_sem; 3314 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3315 * regions, index is order. 3316 * bb_counters[3] = 5 means 3317 * 5 free 8-block regions. */ 3318 }; 3319 3320 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3321 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3322 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3323 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3324 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3325 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3326 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3327 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3328 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3329 3330 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3331 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3332 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3333 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3334 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3335 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3336 3337 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3338 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3339 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3340 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3341 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3342 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3343 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3344 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3345 3346 #define EXT4_MAX_CONTENTION 8 3347 #define EXT4_CONTENTION_THRESHOLD 2 3348 3349 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3350 ext4_group_t group) 3351 { 3352 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3353 } 3354 3355 /* 3356 * Returns true if the filesystem is busy enough that attempts to 3357 * access the block group locks has run into contention. 3358 */ 3359 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3360 { 3361 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3362 } 3363 3364 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3365 { 3366 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3367 if (spin_trylock(lock)) 3368 /* 3369 * We're able to grab the lock right away, so drop the 3370 * lock contention counter. 3371 */ 3372 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3373 else { 3374 /* 3375 * The lock is busy, so bump the contention counter, 3376 * and then wait on the spin lock. 3377 */ 3378 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3379 EXT4_MAX_CONTENTION); 3380 spin_lock(lock); 3381 } 3382 } 3383 3384 static inline void ext4_unlock_group(struct super_block *sb, 3385 ext4_group_t group) 3386 { 3387 spin_unlock(ext4_group_lock_ptr(sb, group)); 3388 } 3389 3390 #ifdef CONFIG_QUOTA 3391 static inline bool ext4_quota_capable(struct super_block *sb) 3392 { 3393 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3394 } 3395 3396 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3397 { 3398 struct ext4_sb_info *sbi = EXT4_SB(sb); 3399 3400 return (ext4_has_feature_quota(sb) || 3401 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3402 } 3403 #endif 3404 3405 /* 3406 * Block validity checking 3407 */ 3408 #define ext4_check_indirect_blockref(inode, bh) \ 3409 ext4_check_blockref(__func__, __LINE__, inode, \ 3410 (__le32 *)(bh)->b_data, \ 3411 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3412 3413 #define ext4_ind_check_inode(inode) \ 3414 ext4_check_blockref(__func__, __LINE__, inode, \ 3415 EXT4_I(inode)->i_data, \ 3416 EXT4_NDIR_BLOCKS) 3417 3418 /* 3419 * Inodes and files operations 3420 */ 3421 3422 /* dir.c */ 3423 extern const struct file_operations ext4_dir_operations; 3424 3425 /* file.c */ 3426 extern const struct inode_operations ext4_file_inode_operations; 3427 extern const struct file_operations ext4_file_operations; 3428 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3429 3430 /* inline.c */ 3431 extern int ext4_get_max_inline_size(struct inode *inode); 3432 extern int ext4_find_inline_data_nolock(struct inode *inode); 3433 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3434 unsigned int len); 3435 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3436 3437 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3438 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3439 struct inode *inode, 3440 loff_t pos, unsigned len, 3441 unsigned flags, 3442 struct page **pagep); 3443 extern int ext4_write_inline_data_end(struct inode *inode, 3444 loff_t pos, unsigned len, 3445 unsigned copied, 3446 struct page *page); 3447 extern struct buffer_head * 3448 ext4_journalled_write_inline_data(struct inode *inode, 3449 unsigned len, 3450 struct page *page); 3451 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3452 struct inode *inode, 3453 loff_t pos, unsigned len, 3454 unsigned flags, 3455 struct page **pagep, 3456 void **fsdata); 3457 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3458 unsigned len, unsigned copied, 3459 struct page *page); 3460 extern int ext4_try_add_inline_entry(handle_t *handle, 3461 struct ext4_filename *fname, 3462 struct inode *dir, struct inode *inode); 3463 extern int ext4_try_create_inline_dir(handle_t *handle, 3464 struct inode *parent, 3465 struct inode *inode); 3466 extern int ext4_read_inline_dir(struct file *filp, 3467 struct dir_context *ctx, 3468 int *has_inline_data); 3469 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3470 struct inode *dir, ext4_lblk_t block, 3471 struct dx_hash_info *hinfo, 3472 __u32 start_hash, __u32 start_minor_hash, 3473 int *has_inline_data); 3474 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3475 struct ext4_filename *fname, 3476 struct ext4_dir_entry_2 **res_dir, 3477 int *has_inline_data); 3478 extern int ext4_delete_inline_entry(handle_t *handle, 3479 struct inode *dir, 3480 struct ext4_dir_entry_2 *de_del, 3481 struct buffer_head *bh, 3482 int *has_inline_data); 3483 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3484 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3485 struct ext4_dir_entry_2 **parent_de, 3486 int *retval); 3487 extern int ext4_inline_data_fiemap(struct inode *inode, 3488 struct fiemap_extent_info *fieinfo, 3489 int *has_inline, __u64 start, __u64 len); 3490 3491 struct iomap; 3492 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3493 3494 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3495 3496 extern int ext4_convert_inline_data(struct inode *inode); 3497 3498 static inline int ext4_has_inline_data(struct inode *inode) 3499 { 3500 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3501 EXT4_I(inode)->i_inline_off; 3502 } 3503 3504 /* namei.c */ 3505 extern const struct inode_operations ext4_dir_inode_operations; 3506 extern const struct inode_operations ext4_special_inode_operations; 3507 extern struct dentry *ext4_get_parent(struct dentry *child); 3508 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3509 struct ext4_dir_entry_2 *de, 3510 int blocksize, int csum_size, 3511 unsigned int parent_ino, int dotdot_real_len); 3512 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3513 unsigned int blocksize); 3514 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3515 struct buffer_head *bh); 3516 extern int ext4_ci_compare(const struct inode *parent, 3517 const struct qstr *fname, 3518 const struct qstr *entry, bool quick); 3519 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name, 3520 struct inode *inode); 3521 extern int __ext4_link(struct inode *dir, struct inode *inode, 3522 struct dentry *dentry); 3523 3524 #define S_SHIFT 12 3525 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3526 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3527 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3528 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3529 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3530 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3531 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3532 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3533 }; 3534 3535 static inline void ext4_set_de_type(struct super_block *sb, 3536 struct ext4_dir_entry_2 *de, 3537 umode_t mode) { 3538 if (ext4_has_feature_filetype(sb)) 3539 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3540 } 3541 3542 /* readpages.c */ 3543 extern int ext4_mpage_readpages(struct inode *inode, 3544 struct readahead_control *rac, struct page *page); 3545 extern int __init ext4_init_post_read_processing(void); 3546 extern void ext4_exit_post_read_processing(void); 3547 3548 /* symlink.c */ 3549 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3550 extern const struct inode_operations ext4_symlink_inode_operations; 3551 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3552 3553 /* sysfs.c */ 3554 extern int ext4_register_sysfs(struct super_block *sb); 3555 extern void ext4_unregister_sysfs(struct super_block *sb); 3556 extern int __init ext4_init_sysfs(void); 3557 extern void ext4_exit_sysfs(void); 3558 3559 /* block_validity */ 3560 extern void ext4_release_system_zone(struct super_block *sb); 3561 extern int ext4_setup_system_zone(struct super_block *sb); 3562 extern int __init ext4_init_system_zone(void); 3563 extern void ext4_exit_system_zone(void); 3564 extern int ext4_inode_block_valid(struct inode *inode, 3565 ext4_fsblk_t start_blk, 3566 unsigned int count); 3567 extern int ext4_check_blockref(const char *, unsigned int, 3568 struct inode *, __le32 *, unsigned int); 3569 3570 /* extents.c */ 3571 struct ext4_ext_path; 3572 struct ext4_extent; 3573 3574 /* 3575 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3576 * __le32. 3577 */ 3578 #define EXT_MAX_BLOCKS 0xffffffff 3579 3580 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3581 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3582 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3583 struct ext4_map_blocks *map, int flags); 3584 extern int ext4_ext_truncate(handle_t *, struct inode *); 3585 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3586 ext4_lblk_t end); 3587 extern void ext4_ext_init(struct super_block *); 3588 extern void ext4_ext_release(struct super_block *); 3589 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3590 loff_t len); 3591 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3592 loff_t offset, ssize_t len); 3593 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3594 ext4_io_end_t *io_end); 3595 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3596 struct ext4_map_blocks *map, int flags); 3597 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3598 int num, 3599 struct ext4_ext_path *path); 3600 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3601 struct ext4_ext_path **, 3602 struct ext4_extent *, int); 3603 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3604 struct ext4_ext_path **, 3605 int flags); 3606 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3607 extern int ext4_ext_check_inode(struct inode *inode); 3608 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3609 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3610 __u64 start, __u64 len); 3611 extern int ext4_get_es_cache(struct inode *inode, 3612 struct fiemap_extent_info *fieinfo, 3613 __u64 start, __u64 len); 3614 extern int ext4_ext_precache(struct inode *inode); 3615 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3616 struct inode *inode2, ext4_lblk_t lblk1, 3617 ext4_lblk_t lblk2, ext4_lblk_t count, 3618 int mark_unwritten,int *err); 3619 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3620 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3621 int check_cred, int restart_cred, 3622 int revoke_cred); 3623 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3624 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3625 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3626 int len, int unwritten, ext4_fsblk_t pblk); 3627 extern int ext4_ext_clear_bb(struct inode *inode); 3628 3629 3630 /* move_extent.c */ 3631 extern void ext4_double_down_write_data_sem(struct inode *first, 3632 struct inode *second); 3633 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3634 struct inode *donor_inode); 3635 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3636 __u64 start_orig, __u64 start_donor, 3637 __u64 len, __u64 *moved_len); 3638 3639 /* page-io.c */ 3640 extern int __init ext4_init_pageio(void); 3641 extern void ext4_exit_pageio(void); 3642 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3643 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3644 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3645 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3646 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3647 struct writeback_control *wbc); 3648 extern void ext4_end_io_rsv_work(struct work_struct *work); 3649 extern void ext4_io_submit(struct ext4_io_submit *io); 3650 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3651 struct page *page, 3652 int len, 3653 bool keep_towrite); 3654 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3655 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3656 3657 /* mmp.c */ 3658 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3659 3660 /* verity.c */ 3661 extern const struct fsverity_operations ext4_verityops; 3662 3663 /* 3664 * Add new method to test whether block and inode bitmaps are properly 3665 * initialized. With uninit_bg reading the block from disk is not enough 3666 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3667 */ 3668 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3669 3670 static inline int bitmap_uptodate(struct buffer_head *bh) 3671 { 3672 return (buffer_uptodate(bh) && 3673 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3674 } 3675 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3676 { 3677 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3678 } 3679 3680 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3681 3682 /* For ioend & aio unwritten conversion wait queues */ 3683 #define EXT4_WQ_HASH_SZ 37 3684 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3685 EXT4_WQ_HASH_SZ]) 3686 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3687 3688 extern int ext4_resize_begin(struct super_block *sb); 3689 extern void ext4_resize_end(struct super_block *sb); 3690 3691 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3692 struct ext4_io_end *io_end) 3693 { 3694 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3695 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3696 atomic_inc(&EXT4_I(inode)->i_unwritten); 3697 } 3698 } 3699 3700 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3701 { 3702 struct inode *inode = io_end->inode; 3703 3704 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3705 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3706 /* Wake up anyone waiting on unwritten extent conversion */ 3707 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3708 wake_up_all(ext4_ioend_wq(inode)); 3709 } 3710 } 3711 3712 extern const struct iomap_ops ext4_iomap_ops; 3713 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3714 extern const struct iomap_ops ext4_iomap_report_ops; 3715 3716 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3717 { 3718 /* 3719 * If the buffer has the write error flag, we have failed 3720 * to write out data in the block. In this case, we don't 3721 * have to read the block because we may read the old data 3722 * successfully. 3723 */ 3724 if (!buffer_uptodate(bh) && buffer_write_io_error(bh)) 3725 set_buffer_uptodate(bh); 3726 return buffer_uptodate(bh); 3727 } 3728 3729 #endif /* __KERNEL__ */ 3730 3731 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3732 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3733 3734 #endif /* _EXT4_H */ 3735