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 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 166 #define EXT4_MB_CR0_OPTIMIZED 0x8000 167 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 168 #define EXT4_MB_CR1_OPTIMIZED 0x00010000 169 /* Perform linear traversal for one group */ 170 #define EXT4_MB_SEARCH_NEXT_LINEAR 0x00020000 171 struct ext4_allocation_request { 172 /* target inode for block we're allocating */ 173 struct inode *inode; 174 /* how many blocks we want to allocate */ 175 unsigned int len; 176 /* logical block in target inode */ 177 ext4_lblk_t logical; 178 /* the closest logical allocated block to the left */ 179 ext4_lblk_t lleft; 180 /* the closest logical allocated block to the right */ 181 ext4_lblk_t lright; 182 /* phys. target (a hint) */ 183 ext4_fsblk_t goal; 184 /* phys. block for the closest logical allocated block to the left */ 185 ext4_fsblk_t pleft; 186 /* phys. block for the closest logical allocated block to the right */ 187 ext4_fsblk_t pright; 188 /* flags. see above EXT4_MB_HINT_* */ 189 unsigned int flags; 190 }; 191 192 /* 193 * Logical to physical block mapping, used by ext4_map_blocks() 194 * 195 * This structure is used to pass requests into ext4_map_blocks() as 196 * well as to store the information returned by ext4_map_blocks(). It 197 * takes less room on the stack than a struct buffer_head. 198 */ 199 #define EXT4_MAP_NEW BIT(BH_New) 200 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 201 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 202 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 203 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 204 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 205 206 struct ext4_map_blocks { 207 ext4_fsblk_t m_pblk; 208 ext4_lblk_t m_lblk; 209 unsigned int m_len; 210 unsigned int m_flags; 211 }; 212 213 /* 214 * Block validity checking, system zone rbtree. 215 */ 216 struct ext4_system_blocks { 217 struct rb_root root; 218 struct rcu_head rcu; 219 }; 220 221 /* 222 * Flags for ext4_io_end->flags 223 */ 224 #define EXT4_IO_END_UNWRITTEN 0x0001 225 226 struct ext4_io_end_vec { 227 struct list_head list; /* list of io_end_vec */ 228 loff_t offset; /* offset in the file */ 229 ssize_t size; /* size of the extent */ 230 }; 231 232 /* 233 * For converting unwritten extents on a work queue. 'handle' is used for 234 * buffered writeback. 235 */ 236 typedef struct ext4_io_end { 237 struct list_head list; /* per-file finished IO list */ 238 handle_t *handle; /* handle reserved for extent 239 * conversion */ 240 struct inode *inode; /* file being written to */ 241 struct bio *bio; /* Linked list of completed 242 * bios covering the extent */ 243 unsigned int flag; /* unwritten or not */ 244 atomic_t count; /* reference counter */ 245 struct list_head list_vec; /* list of ext4_io_end_vec */ 246 } ext4_io_end_t; 247 248 struct ext4_io_submit { 249 struct writeback_control *io_wbc; 250 struct bio *io_bio; 251 ext4_io_end_t *io_end; 252 sector_t io_next_block; 253 }; 254 255 /* 256 * Special inodes numbers 257 */ 258 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 259 #define EXT4_ROOT_INO 2 /* Root inode */ 260 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 261 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 262 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 263 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 264 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 265 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 266 267 /* First non-reserved inode for old ext4 filesystems */ 268 #define EXT4_GOOD_OLD_FIRST_INO 11 269 270 /* 271 * Maximal count of links to a file 272 */ 273 #define EXT4_LINK_MAX 65000 274 275 /* 276 * Macro-instructions used to manage several block sizes 277 */ 278 #define EXT4_MIN_BLOCK_SIZE 1024 279 #define EXT4_MAX_BLOCK_SIZE 65536 280 #define EXT4_MIN_BLOCK_LOG_SIZE 10 281 #define EXT4_MAX_BLOCK_LOG_SIZE 16 282 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 283 #ifdef __KERNEL__ 284 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 285 #else 286 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 287 #endif 288 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 289 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 290 EXT4_SB(s)->s_cluster_bits) 291 #ifdef __KERNEL__ 292 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 293 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 294 #else 295 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 296 #endif 297 #ifdef __KERNEL__ 298 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 299 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 300 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 301 #else 302 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 303 EXT4_GOOD_OLD_INODE_SIZE : \ 304 (s)->s_inode_size) 305 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 306 EXT4_GOOD_OLD_FIRST_INO : \ 307 (s)->s_first_ino) 308 #endif 309 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 310 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 311 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 312 blkbits)) 313 314 /* Translate a block number to a cluster number */ 315 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 316 /* Translate a cluster number to a block number */ 317 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 318 /* Translate # of blks to # of clusters */ 319 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 320 (sbi)->s_cluster_bits) 321 /* Mask out the low bits to get the starting block of the cluster */ 322 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 323 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 324 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 325 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 326 /* Fill in the low bits to get the last block of the cluster */ 327 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 328 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 329 /* Get the cluster offset */ 330 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 331 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 332 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 333 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 334 335 /* 336 * Structure of a blocks group descriptor 337 */ 338 struct ext4_group_desc 339 { 340 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 341 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 342 __le32 bg_inode_table_lo; /* Inodes table block */ 343 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 344 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 345 __le16 bg_used_dirs_count_lo; /* Directories count */ 346 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 347 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 348 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 349 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 350 __le16 bg_itable_unused_lo; /* Unused inodes count */ 351 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 352 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 353 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 354 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 355 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 356 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 357 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 358 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 359 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 360 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 361 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 362 __u32 bg_reserved; 363 }; 364 365 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 366 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 367 sizeof(__le16)) 368 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 369 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 370 sizeof(__le16)) 371 372 /* 373 * Structure of a flex block group info 374 */ 375 376 struct flex_groups { 377 atomic64_t free_clusters; 378 atomic_t free_inodes; 379 atomic_t used_dirs; 380 }; 381 382 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 383 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 384 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 385 386 /* 387 * Macro-instructions used to manage group descriptors 388 */ 389 #define EXT4_MIN_DESC_SIZE 32 390 #define EXT4_MIN_DESC_SIZE_64BIT 64 391 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 392 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 393 #ifdef __KERNEL__ 394 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 395 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 396 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 397 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 398 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 399 #else 400 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 401 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 402 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 403 #endif 404 405 /* 406 * Constants relative to the data blocks 407 */ 408 #define EXT4_NDIR_BLOCKS 12 409 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 410 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 411 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 412 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 413 414 /* 415 * Inode flags 416 */ 417 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 418 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 419 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 420 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 421 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 422 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 423 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 424 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 425 /* Reserved for compression usage... */ 426 #define EXT4_DIRTY_FL 0x00000100 427 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 428 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 429 /* nb: was previously EXT2_ECOMPR_FL */ 430 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 431 /* End compression flags --- maybe not all used */ 432 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 433 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 434 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 435 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 436 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 437 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 438 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 439 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 440 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 441 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 442 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 443 444 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 445 446 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 447 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 448 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 449 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 450 451 /* User modifiable flags */ 452 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 453 EXT4_UNRM_FL | \ 454 EXT4_COMPR_FL | \ 455 EXT4_SYNC_FL | \ 456 EXT4_IMMUTABLE_FL | \ 457 EXT4_APPEND_FL | \ 458 EXT4_NODUMP_FL | \ 459 EXT4_NOATIME_FL | \ 460 EXT4_JOURNAL_DATA_FL | \ 461 EXT4_NOTAIL_FL | \ 462 EXT4_DIRSYNC_FL | \ 463 EXT4_TOPDIR_FL | \ 464 EXT4_EXTENTS_FL | \ 465 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 466 EXT4_DAX_FL | \ 467 EXT4_PROJINHERIT_FL | \ 468 EXT4_CASEFOLD_FL) 469 470 /* User visible flags */ 471 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 472 EXT4_DIRTY_FL | \ 473 EXT4_COMPRBLK_FL | \ 474 EXT4_NOCOMPR_FL | \ 475 EXT4_ENCRYPT_FL | \ 476 EXT4_INDEX_FL | \ 477 EXT4_VERITY_FL | \ 478 EXT4_INLINE_DATA_FL) 479 480 /* Flags that should be inherited by new inodes from their parent. */ 481 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 482 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 483 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 484 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 485 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 486 EXT4_DAX_FL) 487 488 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 489 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 490 EXT4_PROJINHERIT_FL)) 491 492 /* Flags that are appropriate for non-directories/regular files. */ 493 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 494 495 /* The only flags that should be swapped */ 496 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 497 498 /* Flags which are mutually exclusive to DAX */ 499 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 500 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 501 502 /* Mask out flags that are inappropriate for the given type of inode. */ 503 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 504 { 505 if (S_ISDIR(mode)) 506 return flags; 507 else if (S_ISREG(mode)) 508 return flags & EXT4_REG_FLMASK; 509 else 510 return flags & EXT4_OTHER_FLMASK; 511 } 512 513 /* 514 * Inode flags used for atomic set/get 515 */ 516 enum { 517 EXT4_INODE_SECRM = 0, /* Secure deletion */ 518 EXT4_INODE_UNRM = 1, /* Undelete */ 519 EXT4_INODE_COMPR = 2, /* Compress file */ 520 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 521 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 522 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 523 EXT4_INODE_NODUMP = 6, /* do not dump file */ 524 EXT4_INODE_NOATIME = 7, /* do not update atime */ 525 /* Reserved for compression usage... */ 526 EXT4_INODE_DIRTY = 8, 527 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 528 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 529 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 530 /* End compression flags --- maybe not all used */ 531 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 532 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 533 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 534 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 535 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 536 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 537 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 538 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 539 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 540 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 541 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 542 EXT4_INODE_DAX = 25, /* Inode is DAX */ 543 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 544 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 545 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 546 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 547 }; 548 549 /* 550 * Since it's pretty easy to mix up bit numbers and hex values, we use a 551 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 552 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 553 * any extra space in the compiled kernel image, otherwise, the build will fail. 554 * It's important that these values are the same, since we are using 555 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 556 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 557 * values found in ext2, ext3 and ext4 filesystems, and of course the values 558 * defined in e2fsprogs. 559 * 560 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 561 */ 562 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 563 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 564 565 static inline void ext4_check_flag_values(void) 566 { 567 CHECK_FLAG_VALUE(SECRM); 568 CHECK_FLAG_VALUE(UNRM); 569 CHECK_FLAG_VALUE(COMPR); 570 CHECK_FLAG_VALUE(SYNC); 571 CHECK_FLAG_VALUE(IMMUTABLE); 572 CHECK_FLAG_VALUE(APPEND); 573 CHECK_FLAG_VALUE(NODUMP); 574 CHECK_FLAG_VALUE(NOATIME); 575 CHECK_FLAG_VALUE(DIRTY); 576 CHECK_FLAG_VALUE(COMPRBLK); 577 CHECK_FLAG_VALUE(NOCOMPR); 578 CHECK_FLAG_VALUE(ENCRYPT); 579 CHECK_FLAG_VALUE(INDEX); 580 CHECK_FLAG_VALUE(IMAGIC); 581 CHECK_FLAG_VALUE(JOURNAL_DATA); 582 CHECK_FLAG_VALUE(NOTAIL); 583 CHECK_FLAG_VALUE(DIRSYNC); 584 CHECK_FLAG_VALUE(TOPDIR); 585 CHECK_FLAG_VALUE(HUGE_FILE); 586 CHECK_FLAG_VALUE(EXTENTS); 587 CHECK_FLAG_VALUE(VERITY); 588 CHECK_FLAG_VALUE(EA_INODE); 589 CHECK_FLAG_VALUE(INLINE_DATA); 590 CHECK_FLAG_VALUE(PROJINHERIT); 591 CHECK_FLAG_VALUE(CASEFOLD); 592 CHECK_FLAG_VALUE(RESERVED); 593 } 594 595 /* Used to pass group descriptor data when online resize is done */ 596 struct ext4_new_group_input { 597 __u32 group; /* Group number for this data */ 598 __u64 block_bitmap; /* Absolute block number of block bitmap */ 599 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 600 __u64 inode_table; /* Absolute block number of inode table start */ 601 __u32 blocks_count; /* Total number of blocks in this group */ 602 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 603 __u16 unused; 604 }; 605 606 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 607 struct compat_ext4_new_group_input { 608 u32 group; 609 compat_u64 block_bitmap; 610 compat_u64 inode_bitmap; 611 compat_u64 inode_table; 612 u32 blocks_count; 613 u16 reserved_blocks; 614 u16 unused; 615 }; 616 #endif 617 618 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 619 struct ext4_new_group_data { 620 __u32 group; 621 __u64 block_bitmap; 622 __u64 inode_bitmap; 623 __u64 inode_table; 624 __u32 blocks_count; 625 __u16 reserved_blocks; 626 __u16 mdata_blocks; 627 __u32 free_clusters_count; 628 }; 629 630 /* Indexes used to index group tables in ext4_new_group_data */ 631 enum { 632 BLOCK_BITMAP = 0, /* block bitmap */ 633 INODE_BITMAP, /* inode bitmap */ 634 INODE_TABLE, /* inode tables */ 635 GROUP_TABLE_COUNT, 636 }; 637 638 /* 639 * Flags used by ext4_map_blocks() 640 */ 641 /* Allocate any needed blocks and/or convert an unwritten 642 extent to be an initialized ext4 */ 643 #define EXT4_GET_BLOCKS_CREATE 0x0001 644 /* Request the creation of an unwritten extent */ 645 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 646 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 647 EXT4_GET_BLOCKS_CREATE) 648 /* Caller is from the delayed allocation writeout path 649 * finally doing the actual allocation of delayed blocks */ 650 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 651 /* caller is from the direct IO path, request to creation of an 652 unwritten extents if not allocated, split the unwritten 653 extent if blocks has been preallocated already*/ 654 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 655 #define EXT4_GET_BLOCKS_CONVERT 0x0010 656 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 657 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 658 /* Convert extent to initialized after IO complete */ 659 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 660 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 661 /* Eventual metadata allocation (due to growing extent tree) 662 * should not fail, so try to use reserved blocks for that.*/ 663 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 664 /* Don't normalize allocation size (used for fallocate) */ 665 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 666 /* Convert written extents to unwritten */ 667 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 668 /* Write zeros to newly created written extents */ 669 #define EXT4_GET_BLOCKS_ZERO 0x0200 670 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 671 EXT4_GET_BLOCKS_ZERO) 672 /* Caller will submit data before dropping transaction handle. This 673 * allows jbd2 to avoid submitting data before commit. */ 674 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 675 676 /* 677 * The bit position of these flags must not overlap with any of the 678 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 679 * read_extent_tree_block(), ext4_split_extent_at(), 680 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 681 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 682 * caching the extents when reading from the extent tree while a 683 * truncate or punch hole operation is in progress. 684 */ 685 #define EXT4_EX_NOCACHE 0x40000000 686 #define EXT4_EX_FORCE_CACHE 0x20000000 687 #define EXT4_EX_NOFAIL 0x10000000 688 689 /* 690 * Flags used by ext4_free_blocks 691 */ 692 #define EXT4_FREE_BLOCKS_METADATA 0x0001 693 #define EXT4_FREE_BLOCKS_FORGET 0x0002 694 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 695 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 696 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 697 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 698 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 699 700 /* 701 * ioctl commands 702 */ 703 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 704 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 705 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 706 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 707 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 708 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 709 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 710 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 711 #define EXT4_IOC_MIGRATE _IO('f', 9) 712 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 713 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 714 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 715 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 716 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 717 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 718 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 719 /* ioctl codes 19--39 are reserved for fscrypt */ 720 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 721 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 722 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 723 724 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 725 726 /* 727 * Flags for going down operation 728 */ 729 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 730 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 731 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 732 733 /* 734 * Flags returned by EXT4_IOC_GETSTATE 735 * 736 * We only expose to userspace a subset of the state flags in 737 * i_state_flags 738 */ 739 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 740 #define EXT4_STATE_FLAG_NEW 0x00000002 741 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 742 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 743 744 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 745 /* 746 * ioctl commands in 32 bit emulation 747 */ 748 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 749 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 750 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 751 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 752 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 753 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 754 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 755 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 756 #endif 757 758 /* 759 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 760 * It indicates that the entry in extent status cache is for a hole. 761 */ 762 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 763 764 /* Max physical block we can address w/o extents */ 765 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 766 767 /* Max logical block we can support */ 768 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 769 770 /* 771 * Structure of an inode on the disk 772 */ 773 struct ext4_inode { 774 __le16 i_mode; /* File mode */ 775 __le16 i_uid; /* Low 16 bits of Owner Uid */ 776 __le32 i_size_lo; /* Size in bytes */ 777 __le32 i_atime; /* Access time */ 778 __le32 i_ctime; /* Inode Change time */ 779 __le32 i_mtime; /* Modification time */ 780 __le32 i_dtime; /* Deletion Time */ 781 __le16 i_gid; /* Low 16 bits of Group Id */ 782 __le16 i_links_count; /* Links count */ 783 __le32 i_blocks_lo; /* Blocks count */ 784 __le32 i_flags; /* File flags */ 785 union { 786 struct { 787 __le32 l_i_version; 788 } linux1; 789 struct { 790 __u32 h_i_translator; 791 } hurd1; 792 struct { 793 __u32 m_i_reserved1; 794 } masix1; 795 } osd1; /* OS dependent 1 */ 796 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 797 __le32 i_generation; /* File version (for NFS) */ 798 __le32 i_file_acl_lo; /* File ACL */ 799 __le32 i_size_high; 800 __le32 i_obso_faddr; /* Obsoleted fragment address */ 801 union { 802 struct { 803 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 804 __le16 l_i_file_acl_high; 805 __le16 l_i_uid_high; /* these 2 fields */ 806 __le16 l_i_gid_high; /* were reserved2[0] */ 807 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 808 __le16 l_i_reserved; 809 } linux2; 810 struct { 811 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 812 __u16 h_i_mode_high; 813 __u16 h_i_uid_high; 814 __u16 h_i_gid_high; 815 __u32 h_i_author; 816 } hurd2; 817 struct { 818 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 819 __le16 m_i_file_acl_high; 820 __u32 m_i_reserved2[2]; 821 } masix2; 822 } osd2; /* OS dependent 2 */ 823 __le16 i_extra_isize; 824 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 825 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 826 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 827 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 828 __le32 i_crtime; /* File Creation time */ 829 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 830 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 831 __le32 i_projid; /* Project ID */ 832 }; 833 834 struct move_extent { 835 __u32 reserved; /* should be zero */ 836 __u32 donor_fd; /* donor file descriptor */ 837 __u64 orig_start; /* logical start offset in block for orig */ 838 __u64 donor_start; /* logical start offset in block for donor */ 839 __u64 len; /* block length to be moved */ 840 __u64 moved_len; /* moved block length */ 841 }; 842 843 #define EXT4_EPOCH_BITS 2 844 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 845 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 846 847 /* 848 * Extended fields will fit into an inode if the filesystem was formatted 849 * with large inodes (-I 256 or larger) and there are not currently any EAs 850 * consuming all of the available space. For new inodes we always reserve 851 * enough space for the kernel's known extended fields, but for inodes 852 * created with an old kernel this might not have been the case. None of 853 * the extended inode fields is critical for correct filesystem operation. 854 * This macro checks if a certain field fits in the inode. Note that 855 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 856 */ 857 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 858 ((offsetof(typeof(*ext4_inode), field) + \ 859 sizeof((ext4_inode)->field)) \ 860 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 861 (einode)->i_extra_isize)) \ 862 863 /* 864 * We use an encoding that preserves the times for extra epoch "00": 865 * 866 * extra msb of adjust for signed 867 * epoch 32-bit 32-bit tv_sec to 868 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 869 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 870 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 871 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 872 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 873 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 874 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 875 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 876 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 877 * 878 * Note that previous versions of the kernel on 64-bit systems would 879 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 880 * 1970. e2fsck will correct this, assuming that it is run on the 881 * affected filesystem before 2242. 882 */ 883 884 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 885 { 886 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 887 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 888 } 889 890 static inline void ext4_decode_extra_time(struct timespec64 *time, 891 __le32 extra) 892 { 893 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 894 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 895 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 896 } 897 898 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 899 do { \ 900 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 901 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 902 (raw_inode)->xtime ## _extra = \ 903 ext4_encode_extra_time(&(inode)->xtime); \ 904 } \ 905 else \ 906 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 907 } while (0) 908 909 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 910 do { \ 911 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 912 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 913 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 914 (raw_inode)->xtime ## _extra = \ 915 ext4_encode_extra_time(&(einode)->xtime); \ 916 } while (0) 917 918 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 919 do { \ 920 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 921 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 922 ext4_decode_extra_time(&(inode)->xtime, \ 923 raw_inode->xtime ## _extra); \ 924 } \ 925 else \ 926 (inode)->xtime.tv_nsec = 0; \ 927 } while (0) 928 929 930 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 931 do { \ 932 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 933 (einode)->xtime.tv_sec = \ 934 (signed)le32_to_cpu((raw_inode)->xtime); \ 935 else \ 936 (einode)->xtime.tv_sec = 0; \ 937 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 938 ext4_decode_extra_time(&(einode)->xtime, \ 939 raw_inode->xtime ## _extra); \ 940 else \ 941 (einode)->xtime.tv_nsec = 0; \ 942 } while (0) 943 944 #define i_disk_version osd1.linux1.l_i_version 945 946 #if defined(__KERNEL__) || defined(__linux__) 947 #define i_reserved1 osd1.linux1.l_i_reserved1 948 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 949 #define i_blocks_high osd2.linux2.l_i_blocks_high 950 #define i_uid_low i_uid 951 #define i_gid_low i_gid 952 #define i_uid_high osd2.linux2.l_i_uid_high 953 #define i_gid_high osd2.linux2.l_i_gid_high 954 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 955 956 #elif defined(__GNU__) 957 958 #define i_translator osd1.hurd1.h_i_translator 959 #define i_uid_high osd2.hurd2.h_i_uid_high 960 #define i_gid_high osd2.hurd2.h_i_gid_high 961 #define i_author osd2.hurd2.h_i_author 962 963 #elif defined(__masix__) 964 965 #define i_reserved1 osd1.masix1.m_i_reserved1 966 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 967 #define i_reserved2 osd2.masix2.m_i_reserved2 968 969 #endif /* defined(__KERNEL__) || defined(__linux__) */ 970 971 #include "extents_status.h" 972 #include "fast_commit.h" 973 974 /* 975 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 976 * 977 * These are needed to avoid lockdep false positives when we need to 978 * allocate blocks to the quota inode during ext4_map_blocks(), while 979 * holding i_data_sem for a normal (non-quota) inode. Since we don't 980 * do quota tracking for the quota inode, this avoids deadlock (as 981 * well as infinite recursion, since it isn't turtles all the way 982 * down...) 983 * 984 * I_DATA_SEM_NORMAL - Used for most inodes 985 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 986 * where the second inode has larger inode number 987 * than the first 988 * I_DATA_SEM_QUOTA - Used for quota inodes only 989 */ 990 enum { 991 I_DATA_SEM_NORMAL = 0, 992 I_DATA_SEM_OTHER, 993 I_DATA_SEM_QUOTA, 994 }; 995 996 997 /* 998 * fourth extended file system inode data in memory 999 */ 1000 struct ext4_inode_info { 1001 __le32 i_data[15]; /* unconverted */ 1002 __u32 i_dtime; 1003 ext4_fsblk_t i_file_acl; 1004 1005 /* 1006 * i_block_group is the number of the block group which contains 1007 * this file's inode. Constant across the lifetime of the inode, 1008 * it is used for making block allocation decisions - we try to 1009 * place a file's data blocks near its inode block, and new inodes 1010 * near to their parent directory's inode. 1011 */ 1012 ext4_group_t i_block_group; 1013 ext4_lblk_t i_dir_start_lookup; 1014 #if (BITS_PER_LONG < 64) 1015 unsigned long i_state_flags; /* Dynamic state flags */ 1016 #endif 1017 unsigned long i_flags; 1018 1019 /* 1020 * Extended attributes can be read independently of the main file 1021 * data. Taking i_mutex even when reading would cause contention 1022 * between readers of EAs and writers of regular file data, so 1023 * instead we synchronize on xattr_sem when reading or changing 1024 * EAs. 1025 */ 1026 struct rw_semaphore xattr_sem; 1027 1028 struct list_head i_orphan; /* unlinked but open inodes */ 1029 1030 /* Fast commit related info */ 1031 1032 struct list_head i_fc_list; /* 1033 * inodes that need fast commit 1034 * protected by sbi->s_fc_lock. 1035 */ 1036 1037 /* Start of lblk range that needs to be committed in this fast commit */ 1038 ext4_lblk_t i_fc_lblk_start; 1039 1040 /* End of lblk range that needs to be committed in this fast commit */ 1041 ext4_lblk_t i_fc_lblk_len; 1042 1043 /* Number of ongoing updates on this inode */ 1044 atomic_t i_fc_updates; 1045 1046 /* Fast commit wait queue for this inode */ 1047 wait_queue_head_t i_fc_wait; 1048 1049 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1050 struct mutex i_fc_lock; 1051 1052 /* 1053 * i_disksize keeps track of what the inode size is ON DISK, not 1054 * in memory. During truncate, i_size is set to the new size by 1055 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1056 * set i_disksize to 0 until the truncate is actually under way. 1057 * 1058 * The intent is that i_disksize always represents the blocks which 1059 * are used by this file. This allows recovery to restart truncate 1060 * on orphans if we crash during truncate. We actually write i_disksize 1061 * into the on-disk inode when writing inodes out, instead of i_size. 1062 * 1063 * The only time when i_disksize and i_size may be different is when 1064 * a truncate is in progress. The only things which change i_disksize 1065 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1066 */ 1067 loff_t i_disksize; 1068 1069 /* 1070 * i_data_sem is for serialising ext4_truncate() against 1071 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1072 * data tree are chopped off during truncate. We can't do that in 1073 * ext4 because whenever we perform intermediate commits during 1074 * truncate, the inode and all the metadata blocks *must* be in a 1075 * consistent state which allows truncation of the orphans to restart 1076 * during recovery. Hence we must fix the get_block-vs-truncate race 1077 * by other means, so we have i_data_sem. 1078 */ 1079 struct rw_semaphore i_data_sem; 1080 /* 1081 * i_mmap_sem is for serializing page faults with truncate / punch hole 1082 * operations. We have to make sure that new page cannot be faulted in 1083 * a section of the inode that is being punched. We cannot easily use 1084 * i_data_sem for this since we need protection for the whole punch 1085 * operation and i_data_sem ranks below transaction start so we have 1086 * to occasionally drop it. 1087 */ 1088 struct rw_semaphore i_mmap_sem; 1089 struct inode vfs_inode; 1090 struct jbd2_inode *jinode; 1091 1092 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1093 1094 /* 1095 * File creation time. Its function is same as that of 1096 * struct timespec64 i_{a,c,m}time in the generic inode. 1097 */ 1098 struct timespec64 i_crtime; 1099 1100 /* mballoc */ 1101 atomic_t i_prealloc_active; 1102 struct list_head i_prealloc_list; 1103 spinlock_t i_prealloc_lock; 1104 1105 /* extents status tree */ 1106 struct ext4_es_tree i_es_tree; 1107 rwlock_t i_es_lock; 1108 struct list_head i_es_list; 1109 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1110 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1111 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1112 extents to shrink. Protected by 1113 i_es_lock */ 1114 1115 /* ialloc */ 1116 ext4_group_t i_last_alloc_group; 1117 1118 /* allocation reservation info for delalloc */ 1119 /* In case of bigalloc, this refer to clusters rather than blocks */ 1120 unsigned int i_reserved_data_blocks; 1121 1122 /* pending cluster reservations for bigalloc file systems */ 1123 struct ext4_pending_tree i_pending_tree; 1124 1125 /* on-disk additional length */ 1126 __u16 i_extra_isize; 1127 1128 /* Indicate the inline data space. */ 1129 u16 i_inline_off; 1130 u16 i_inline_size; 1131 1132 #ifdef CONFIG_QUOTA 1133 /* quota space reservation, managed internally by quota code */ 1134 qsize_t i_reserved_quota; 1135 #endif 1136 1137 /* Lock protecting lists below */ 1138 spinlock_t i_completed_io_lock; 1139 /* 1140 * Completed IOs that need unwritten extents handling and have 1141 * transaction reserved 1142 */ 1143 struct list_head i_rsv_conversion_list; 1144 struct work_struct i_rsv_conversion_work; 1145 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1146 1147 spinlock_t i_block_reservation_lock; 1148 1149 /* 1150 * Transactions that contain inode's metadata needed to complete 1151 * fsync and fdatasync, respectively. 1152 */ 1153 tid_t i_sync_tid; 1154 tid_t i_datasync_tid; 1155 1156 #ifdef CONFIG_QUOTA 1157 struct dquot *i_dquot[MAXQUOTAS]; 1158 #endif 1159 1160 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1161 __u32 i_csum_seed; 1162 1163 kprojid_t i_projid; 1164 }; 1165 1166 /* 1167 * File system states 1168 */ 1169 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1170 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1171 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1172 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1173 1174 /* 1175 * Misc. filesystem flags 1176 */ 1177 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1178 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1179 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1180 1181 /* 1182 * Mount flags set via mount options or defaults 1183 */ 1184 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1185 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1186 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1187 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1188 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1189 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1190 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1191 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1192 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1193 #ifdef CONFIG_FS_DAX 1194 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1195 #else 1196 #define EXT4_MOUNT_DAX_ALWAYS 0 1197 #endif 1198 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1199 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1200 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1201 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1202 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1203 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1204 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1205 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1206 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1207 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1208 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1209 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1210 * enable enforcement for hidden 1211 * quota files */ 1212 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1213 * enforcement for hidden quota 1214 * files */ 1215 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1216 * enforcement */ 1217 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1218 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1219 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1220 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1221 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1222 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1223 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1224 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1225 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1226 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1227 1228 /* 1229 * Mount flags set either automatically (could not be set by mount option) 1230 * based on per file system feature or property or in special cases such as 1231 * distinguishing between explicit mount option definition and default. 1232 */ 1233 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1234 specified delalloc */ 1235 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1236 size of blocksize * 8 1237 blocks */ 1238 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1239 file systems */ 1240 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1241 specified journal checksum */ 1242 1243 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1244 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1245 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1246 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1247 * scanning in mballoc 1248 */ 1249 1250 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1251 ~EXT4_MOUNT_##opt 1252 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1253 EXT4_MOUNT_##opt 1254 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1255 EXT4_MOUNT_##opt) 1256 1257 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1258 ~EXT4_MOUNT2_##opt 1259 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1260 EXT4_MOUNT2_##opt 1261 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1262 EXT4_MOUNT2_##opt) 1263 1264 #define ext4_test_and_set_bit __test_and_set_bit_le 1265 #define ext4_set_bit __set_bit_le 1266 #define ext4_set_bit_atomic ext2_set_bit_atomic 1267 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1268 #define ext4_clear_bit __clear_bit_le 1269 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1270 #define ext4_test_bit test_bit_le 1271 #define ext4_find_next_zero_bit find_next_zero_bit_le 1272 #define ext4_find_next_bit find_next_bit_le 1273 1274 extern void ext4_set_bits(void *bm, int cur, int len); 1275 1276 /* 1277 * Maximal mount counts between two filesystem checks 1278 */ 1279 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1280 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1281 1282 /* 1283 * Behaviour when detecting errors 1284 */ 1285 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1286 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1287 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1288 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1289 1290 /* Metadata checksum algorithm codes */ 1291 #define EXT4_CRC32C_CHKSUM 1 1292 1293 /* 1294 * Structure of the super block 1295 */ 1296 struct ext4_super_block { 1297 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1298 __le32 s_blocks_count_lo; /* Blocks count */ 1299 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1300 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1301 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1302 __le32 s_first_data_block; /* First Data Block */ 1303 __le32 s_log_block_size; /* Block size */ 1304 __le32 s_log_cluster_size; /* Allocation cluster size */ 1305 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1306 __le32 s_clusters_per_group; /* # Clusters per group */ 1307 __le32 s_inodes_per_group; /* # Inodes per group */ 1308 __le32 s_mtime; /* Mount time */ 1309 /*30*/ __le32 s_wtime; /* Write time */ 1310 __le16 s_mnt_count; /* Mount count */ 1311 __le16 s_max_mnt_count; /* Maximal mount count */ 1312 __le16 s_magic; /* Magic signature */ 1313 __le16 s_state; /* File system state */ 1314 __le16 s_errors; /* Behaviour when detecting errors */ 1315 __le16 s_minor_rev_level; /* minor revision level */ 1316 /*40*/ __le32 s_lastcheck; /* time of last check */ 1317 __le32 s_checkinterval; /* max. time between checks */ 1318 __le32 s_creator_os; /* OS */ 1319 __le32 s_rev_level; /* Revision level */ 1320 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1321 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1322 /* 1323 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1324 * 1325 * Note: the difference between the compatible feature set and 1326 * the incompatible feature set is that if there is a bit set 1327 * in the incompatible feature set that the kernel doesn't 1328 * know about, it should refuse to mount the filesystem. 1329 * 1330 * e2fsck's requirements are more strict; if it doesn't know 1331 * about a feature in either the compatible or incompatible 1332 * feature set, it must abort and not try to meddle with 1333 * things it doesn't understand... 1334 */ 1335 __le32 s_first_ino; /* First non-reserved inode */ 1336 __le16 s_inode_size; /* size of inode structure */ 1337 __le16 s_block_group_nr; /* block group # of this superblock */ 1338 __le32 s_feature_compat; /* compatible feature set */ 1339 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1340 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1341 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1342 /*78*/ char s_volume_name[16]; /* volume name */ 1343 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1344 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1345 /* 1346 * Performance hints. Directory preallocation should only 1347 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1348 */ 1349 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1350 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1351 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1352 /* 1353 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1354 */ 1355 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1356 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1357 __le32 s_journal_dev; /* device number of journal file */ 1358 __le32 s_last_orphan; /* start of list of inodes to delete */ 1359 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1360 __u8 s_def_hash_version; /* Default hash version to use */ 1361 __u8 s_jnl_backup_type; 1362 __le16 s_desc_size; /* size of group descriptor */ 1363 /*100*/ __le32 s_default_mount_opts; 1364 __le32 s_first_meta_bg; /* First metablock block group */ 1365 __le32 s_mkfs_time; /* When the filesystem was created */ 1366 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1367 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1368 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1369 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1370 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1371 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1372 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1373 __le32 s_flags; /* Miscellaneous flags */ 1374 __le16 s_raid_stride; /* RAID stride */ 1375 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1376 __le64 s_mmp_block; /* Block for multi-mount protection */ 1377 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1378 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1379 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1380 __u8 s_encryption_level; /* versioning level for encryption */ 1381 __u8 s_reserved_pad; /* Padding to next 32bits */ 1382 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1383 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1384 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1385 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1386 snapshot's future use */ 1387 __le32 s_snapshot_list; /* inode number of the head of the 1388 on-disk snapshot list */ 1389 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1390 __le32 s_error_count; /* number of fs errors */ 1391 __le32 s_first_error_time; /* first time an error happened */ 1392 __le32 s_first_error_ino; /* inode involved in first error */ 1393 __le64 s_first_error_block; /* block involved of first error */ 1394 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1395 __le32 s_first_error_line; /* line number where error happened */ 1396 __le32 s_last_error_time; /* most recent time of an error */ 1397 __le32 s_last_error_ino; /* inode involved in last error */ 1398 __le32 s_last_error_line; /* line number where error happened */ 1399 __le64 s_last_error_block; /* block involved of last error */ 1400 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1401 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1402 __u8 s_mount_opts[64]; 1403 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1404 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1405 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1406 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1407 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1408 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1409 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1410 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1411 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1412 __u8 s_wtime_hi; 1413 __u8 s_mtime_hi; 1414 __u8 s_mkfs_time_hi; 1415 __u8 s_lastcheck_hi; 1416 __u8 s_first_error_time_hi; 1417 __u8 s_last_error_time_hi; 1418 __u8 s_first_error_errcode; 1419 __u8 s_last_error_errcode; 1420 __le16 s_encoding; /* Filename charset encoding */ 1421 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1422 __le32 s_reserved[95]; /* Padding to the end of the block */ 1423 __le32 s_checksum; /* crc32c(superblock) */ 1424 }; 1425 1426 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1427 1428 #ifdef __KERNEL__ 1429 1430 #ifdef CONFIG_FS_ENCRYPTION 1431 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL) 1432 #else 1433 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1434 #endif 1435 1436 /* Number of quota types we support */ 1437 #define EXT4_MAXQUOTAS 3 1438 1439 #define EXT4_ENC_UTF8_12_1 1 1440 1441 /* 1442 * fourth extended-fs super-block data in memory 1443 */ 1444 struct ext4_sb_info { 1445 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1446 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1447 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1448 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1449 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1450 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1451 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1452 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1453 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1454 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1455 unsigned long s_overhead; /* # of fs overhead clusters */ 1456 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1457 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1458 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1459 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1460 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1461 struct buffer_head * __rcu *s_group_desc; 1462 unsigned int s_mount_opt; 1463 unsigned int s_mount_opt2; 1464 unsigned long s_mount_flags; 1465 unsigned int s_def_mount_opt; 1466 ext4_fsblk_t s_sb_block; 1467 atomic64_t s_resv_clusters; 1468 kuid_t s_resuid; 1469 kgid_t s_resgid; 1470 unsigned short s_mount_state; 1471 unsigned short s_pad; 1472 int s_addr_per_block_bits; 1473 int s_desc_per_block_bits; 1474 int s_inode_size; 1475 int s_first_ino; 1476 unsigned int s_inode_readahead_blks; 1477 unsigned int s_inode_goal; 1478 u32 s_hash_seed[4]; 1479 int s_def_hash_version; 1480 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1481 struct percpu_counter s_freeclusters_counter; 1482 struct percpu_counter s_freeinodes_counter; 1483 struct percpu_counter s_dirs_counter; 1484 struct percpu_counter s_dirtyclusters_counter; 1485 struct percpu_counter s_sra_exceeded_retry_limit; 1486 struct blockgroup_lock *s_blockgroup_lock; 1487 struct proc_dir_entry *s_proc; 1488 struct kobject s_kobj; 1489 struct completion s_kobj_unregister; 1490 struct super_block *s_sb; 1491 1492 /* Journaling */ 1493 struct journal_s *s_journal; 1494 struct list_head s_orphan; 1495 struct mutex s_orphan_lock; 1496 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1497 unsigned long s_commit_interval; 1498 u32 s_max_batch_time; 1499 u32 s_min_batch_time; 1500 struct block_device *s_journal_bdev; 1501 #ifdef CONFIG_QUOTA 1502 /* Names of quota files with journalled quota */ 1503 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1504 int s_jquota_fmt; /* Format of quota to use */ 1505 #endif 1506 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1507 struct ext4_system_blocks __rcu *s_system_blks; 1508 1509 #ifdef EXTENTS_STATS 1510 /* ext4 extents stats */ 1511 unsigned long s_ext_min; 1512 unsigned long s_ext_max; 1513 unsigned long s_depth_max; 1514 spinlock_t s_ext_stats_lock; 1515 unsigned long s_ext_blocks; 1516 unsigned long s_ext_extents; 1517 #endif 1518 1519 /* for buddy allocator */ 1520 struct ext4_group_info ** __rcu *s_group_info; 1521 struct inode *s_buddy_cache; 1522 spinlock_t s_md_lock; 1523 unsigned short *s_mb_offsets; 1524 unsigned int *s_mb_maxs; 1525 unsigned int s_group_info_size; 1526 unsigned int s_mb_free_pending; 1527 struct list_head s_freed_data_list; /* List of blocks to be freed 1528 after commit completed */ 1529 struct rb_root s_mb_avg_fragment_size_root; 1530 rwlock_t s_mb_rb_lock; 1531 struct list_head *s_mb_largest_free_orders; 1532 rwlock_t *s_mb_largest_free_orders_locks; 1533 1534 /* tunables */ 1535 unsigned long s_stripe; 1536 unsigned int s_mb_max_linear_groups; 1537 unsigned int s_mb_stream_request; 1538 unsigned int s_mb_max_to_scan; 1539 unsigned int s_mb_min_to_scan; 1540 unsigned int s_mb_stats; 1541 unsigned int s_mb_order2_reqs; 1542 unsigned int s_mb_group_prealloc; 1543 unsigned int s_mb_max_inode_prealloc; 1544 unsigned int s_max_dir_size_kb; 1545 /* where last allocation was done - for stream allocation */ 1546 unsigned long s_mb_last_group; 1547 unsigned long s_mb_last_start; 1548 unsigned int s_mb_prefetch; 1549 unsigned int s_mb_prefetch_limit; 1550 1551 /* stats for buddy allocator */ 1552 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1553 atomic_t s_bal_success; /* we found long enough chunks */ 1554 atomic_t s_bal_allocated; /* in blocks */ 1555 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1556 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1557 atomic_t s_bal_goals; /* goal hits */ 1558 atomic_t s_bal_breaks; /* too long searches */ 1559 atomic_t s_bal_2orders; /* 2^order hits */ 1560 atomic_t s_bal_cr0_bad_suggestions; 1561 atomic_t s_bal_cr1_bad_suggestions; 1562 atomic64_t s_bal_cX_groups_considered[4]; 1563 atomic64_t s_bal_cX_hits[4]; 1564 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */ 1565 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1566 atomic64_t s_mb_generation_time; 1567 atomic_t s_mb_lost_chunks; 1568 atomic_t s_mb_preallocated; 1569 atomic_t s_mb_discarded; 1570 atomic_t s_lock_busy; 1571 1572 /* locality groups */ 1573 struct ext4_locality_group __percpu *s_locality_groups; 1574 1575 /* for write statistics */ 1576 unsigned long s_sectors_written_start; 1577 u64 s_kbytes_written; 1578 1579 /* the size of zero-out chunk */ 1580 unsigned int s_extent_max_zeroout_kb; 1581 1582 unsigned int s_log_groups_per_flex; 1583 struct flex_groups * __rcu *s_flex_groups; 1584 ext4_group_t s_flex_groups_allocated; 1585 1586 /* workqueue for reserved extent conversions (buffered io) */ 1587 struct workqueue_struct *rsv_conversion_wq; 1588 1589 /* timer for periodic error stats printing */ 1590 struct timer_list s_err_report; 1591 1592 /* Lazy inode table initialization info */ 1593 struct ext4_li_request *s_li_request; 1594 /* Wait multiplier for lazy initialization thread */ 1595 unsigned int s_li_wait_mult; 1596 1597 /* Kernel thread for multiple mount protection */ 1598 struct task_struct *s_mmp_tsk; 1599 1600 /* record the last minlen when FITRIM is called. */ 1601 atomic_t s_last_trim_minblks; 1602 1603 /* Reference to checksum algorithm driver via cryptoapi */ 1604 struct crypto_shash *s_chksum_driver; 1605 1606 /* Precomputed FS UUID checksum for seeding other checksums */ 1607 __u32 s_csum_seed; 1608 1609 /* Reclaim extents from extent status tree */ 1610 struct shrinker s_es_shrinker; 1611 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1612 long s_es_nr_inode; 1613 struct ext4_es_stats s_es_stats; 1614 struct mb_cache *s_ea_block_cache; 1615 struct mb_cache *s_ea_inode_cache; 1616 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1617 1618 /* Ratelimit ext4 messages. */ 1619 struct ratelimit_state s_err_ratelimit_state; 1620 struct ratelimit_state s_warning_ratelimit_state; 1621 struct ratelimit_state s_msg_ratelimit_state; 1622 atomic_t s_warning_count; 1623 atomic_t s_msg_count; 1624 1625 /* Encryption policy for '-o test_dummy_encryption' */ 1626 struct fscrypt_dummy_policy s_dummy_enc_policy; 1627 1628 /* 1629 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1630 * or EXTENTS flag. 1631 */ 1632 struct percpu_rw_semaphore s_writepages_rwsem; 1633 struct dax_device *s_daxdev; 1634 #ifdef CONFIG_EXT4_DEBUG 1635 unsigned long s_simulate_fail; 1636 #endif 1637 /* Record the errseq of the backing block device */ 1638 errseq_t s_bdev_wb_err; 1639 spinlock_t s_bdev_wb_lock; 1640 1641 /* Information about errors that happened during this mount */ 1642 spinlock_t s_error_lock; 1643 int s_add_error_count; 1644 int s_first_error_code; 1645 __u32 s_first_error_line; 1646 __u32 s_first_error_ino; 1647 __u64 s_first_error_block; 1648 const char *s_first_error_func; 1649 time64_t s_first_error_time; 1650 int s_last_error_code; 1651 __u32 s_last_error_line; 1652 __u32 s_last_error_ino; 1653 __u64 s_last_error_block; 1654 const char *s_last_error_func; 1655 time64_t s_last_error_time; 1656 /* 1657 * If we are in a context where we cannot update error information in 1658 * the on-disk superblock, we queue this work to do it. 1659 */ 1660 struct work_struct s_error_work; 1661 1662 /* Ext4 fast commit stuff */ 1663 atomic_t s_fc_subtid; 1664 atomic_t s_fc_ineligible_updates; 1665 /* 1666 * After commit starts, the main queue gets locked, and the further 1667 * updates get added in the staging queue. 1668 */ 1669 #define FC_Q_MAIN 0 1670 #define FC_Q_STAGING 1 1671 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1672 * that have data changes in them. 1673 */ 1674 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1675 unsigned int s_fc_bytes; 1676 /* 1677 * Main fast commit lock. This lock protects accesses to the 1678 * following fields: 1679 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1680 */ 1681 spinlock_t s_fc_lock; 1682 struct buffer_head *s_fc_bh; 1683 struct ext4_fc_stats s_fc_stats; 1684 u64 s_fc_avg_commit_time; 1685 #ifdef CONFIG_EXT4_DEBUG 1686 int s_fc_debug_max_replay; 1687 #endif 1688 struct ext4_fc_replay_state s_fc_replay_state; 1689 }; 1690 1691 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1692 { 1693 return sb->s_fs_info; 1694 } 1695 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1696 { 1697 return container_of(inode, struct ext4_inode_info, vfs_inode); 1698 } 1699 1700 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1701 { 1702 return ino == EXT4_ROOT_INO || 1703 (ino >= EXT4_FIRST_INO(sb) && 1704 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1705 } 1706 1707 /* 1708 * Returns: sbi->field[index] 1709 * Used to access an array element from the following sbi fields which require 1710 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1711 * - s_group_desc 1712 * - s_group_info 1713 * - s_flex_group 1714 */ 1715 #define sbi_array_rcu_deref(sbi, field, index) \ 1716 ({ \ 1717 typeof(*((sbi)->field)) _v; \ 1718 rcu_read_lock(); \ 1719 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1720 rcu_read_unlock(); \ 1721 _v; \ 1722 }) 1723 1724 /* 1725 * run-time mount flags 1726 */ 1727 enum { 1728 EXT4_MF_MNTDIR_SAMPLED, 1729 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1730 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1731 EXT4_MF_FC_COMMITTING /* File system underoing a fast 1732 * commit. 1733 */ 1734 }; 1735 1736 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1737 { 1738 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1739 } 1740 1741 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1742 { 1743 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1744 } 1745 1746 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1747 { 1748 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1749 } 1750 1751 1752 /* 1753 * Simulate_fail codes 1754 */ 1755 #define EXT4_SIM_BBITMAP_EIO 1 1756 #define EXT4_SIM_BBITMAP_CRC 2 1757 #define EXT4_SIM_IBITMAP_EIO 3 1758 #define EXT4_SIM_IBITMAP_CRC 4 1759 #define EXT4_SIM_INODE_EIO 5 1760 #define EXT4_SIM_INODE_CRC 6 1761 #define EXT4_SIM_DIRBLOCK_EIO 7 1762 #define EXT4_SIM_DIRBLOCK_CRC 8 1763 1764 static inline bool ext4_simulate_fail(struct super_block *sb, 1765 unsigned long code) 1766 { 1767 #ifdef CONFIG_EXT4_DEBUG 1768 struct ext4_sb_info *sbi = EXT4_SB(sb); 1769 1770 if (unlikely(sbi->s_simulate_fail == code)) { 1771 sbi->s_simulate_fail = 0; 1772 return true; 1773 } 1774 #endif 1775 return false; 1776 } 1777 1778 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1779 struct buffer_head *bh, 1780 unsigned long code) 1781 { 1782 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1783 clear_buffer_uptodate(bh); 1784 } 1785 1786 /* 1787 * Error number codes for s_{first,last}_error_errno 1788 * 1789 * Linux errno numbers are architecture specific, so we need to translate 1790 * them into something which is architecture independent. We don't define 1791 * codes for all errno's; just the ones which are most likely to be the cause 1792 * of an ext4_error() call. 1793 */ 1794 #define EXT4_ERR_UNKNOWN 1 1795 #define EXT4_ERR_EIO 2 1796 #define EXT4_ERR_ENOMEM 3 1797 #define EXT4_ERR_EFSBADCRC 4 1798 #define EXT4_ERR_EFSCORRUPTED 5 1799 #define EXT4_ERR_ENOSPC 6 1800 #define EXT4_ERR_ENOKEY 7 1801 #define EXT4_ERR_EROFS 8 1802 #define EXT4_ERR_EFBIG 9 1803 #define EXT4_ERR_EEXIST 10 1804 #define EXT4_ERR_ERANGE 11 1805 #define EXT4_ERR_EOVERFLOW 12 1806 #define EXT4_ERR_EBUSY 13 1807 #define EXT4_ERR_ENOTDIR 14 1808 #define EXT4_ERR_ENOTEMPTY 15 1809 #define EXT4_ERR_ESHUTDOWN 16 1810 #define EXT4_ERR_EFAULT 17 1811 1812 /* 1813 * Inode dynamic state flags 1814 */ 1815 enum { 1816 EXT4_STATE_JDATA, /* journaled data exists */ 1817 EXT4_STATE_NEW, /* inode is newly created */ 1818 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1819 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1820 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1821 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1822 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1823 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1824 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1825 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1826 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1827 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1828 }; 1829 1830 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1831 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1832 { \ 1833 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1834 } \ 1835 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1836 { \ 1837 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1838 } \ 1839 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1840 { \ 1841 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1842 } 1843 1844 /* Add these declarations here only so that these functions can be 1845 * found by name. Otherwise, they are very hard to locate. */ 1846 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1847 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1848 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1849 EXT4_INODE_BIT_FNS(flag, flags, 0) 1850 1851 /* Add these declarations here only so that these functions can be 1852 * found by name. Otherwise, they are very hard to locate. */ 1853 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1854 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1855 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1856 #if (BITS_PER_LONG < 64) 1857 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1858 1859 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1860 { 1861 (ei)->i_state_flags = 0; 1862 } 1863 #else 1864 EXT4_INODE_BIT_FNS(state, flags, 32) 1865 1866 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1867 { 1868 /* We depend on the fact that callers will set i_flags */ 1869 } 1870 #endif 1871 #else 1872 /* Assume that user mode programs are passing in an ext4fs superblock, not 1873 * a kernel struct super_block. This will allow us to call the feature-test 1874 * macros from user land. */ 1875 #define EXT4_SB(sb) (sb) 1876 #endif 1877 1878 static inline bool ext4_verity_in_progress(struct inode *inode) 1879 { 1880 return IS_ENABLED(CONFIG_FS_VERITY) && 1881 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1882 } 1883 1884 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1885 1886 /* 1887 * Codes for operating systems 1888 */ 1889 #define EXT4_OS_LINUX 0 1890 #define EXT4_OS_HURD 1 1891 #define EXT4_OS_MASIX 2 1892 #define EXT4_OS_FREEBSD 3 1893 #define EXT4_OS_LITES 4 1894 1895 /* 1896 * Revision levels 1897 */ 1898 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1899 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1900 1901 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1902 1903 #define EXT4_GOOD_OLD_INODE_SIZE 128 1904 1905 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1906 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1907 #define EXT4_TIMESTAMP_MIN S32_MIN 1908 1909 /* 1910 * Feature set definitions 1911 */ 1912 1913 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1914 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1915 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1916 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1917 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1918 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1919 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1920 /* 1921 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1922 * incompatible only if fast commit blocks are present in the FS. Since we 1923 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1924 * incompatible. We also have a JBD2 incompat feature, which gets set when 1925 * there are fast commit blocks present in the journal. 1926 */ 1927 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1928 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1929 1930 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1931 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1932 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1933 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1934 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1935 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1936 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1937 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1938 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1939 /* 1940 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1941 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1942 * all other data structures' checksums. However, the METADATA_CSUM and 1943 * GDT_CSUM bits are mutually exclusive. 1944 */ 1945 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1946 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1947 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1948 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1949 1950 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1951 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1952 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1953 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1954 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1955 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1956 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1957 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1958 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1959 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1960 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1961 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1962 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1963 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1964 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1965 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1966 1967 extern void ext4_update_dynamic_rev(struct super_block *sb); 1968 1969 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1970 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1971 { \ 1972 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1973 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1974 } \ 1975 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1976 { \ 1977 ext4_update_dynamic_rev(sb); \ 1978 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1979 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1980 } \ 1981 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1982 { \ 1983 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1984 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1985 } 1986 1987 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1988 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1989 { \ 1990 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1991 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1992 } \ 1993 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1994 { \ 1995 ext4_update_dynamic_rev(sb); \ 1996 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1997 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1998 } \ 1999 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2000 { \ 2001 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2002 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2003 } 2004 2005 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2006 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2007 { \ 2008 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2009 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2010 } \ 2011 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2012 { \ 2013 ext4_update_dynamic_rev(sb); \ 2014 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2015 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2016 } \ 2017 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2018 { \ 2019 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2020 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2021 } 2022 2023 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2024 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2025 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2026 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2027 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2028 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2029 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2030 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2031 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2032 2033 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2034 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2035 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2036 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2037 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2038 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2039 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2040 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2041 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2042 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2043 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2044 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2045 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2046 2047 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2048 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2049 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2050 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2051 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2052 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2053 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2054 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2055 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2056 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2057 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2058 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2059 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2060 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2061 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2062 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2063 2064 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2065 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2066 EXT4_FEATURE_INCOMPAT_META_BG) 2067 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2068 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2069 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2070 2071 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2072 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2073 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2074 EXT4_FEATURE_INCOMPAT_META_BG) 2075 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2076 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2077 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2078 2079 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2080 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2081 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2082 EXT4_FEATURE_INCOMPAT_META_BG| \ 2083 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2084 EXT4_FEATURE_INCOMPAT_64BIT| \ 2085 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2086 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2087 EXT4_FEATURE_INCOMPAT_MMP | \ 2088 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2089 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2090 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2091 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2092 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2093 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2094 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2095 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2096 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2097 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2098 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2099 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2100 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2101 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2102 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2103 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2104 EXT4_FEATURE_RO_COMPAT_VERITY) 2105 2106 #define EXTN_FEATURE_FUNCS(ver) \ 2107 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2108 { \ 2109 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2110 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2111 } \ 2112 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2113 { \ 2114 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2115 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2116 } \ 2117 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2118 { \ 2119 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2120 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2121 } 2122 2123 EXTN_FEATURE_FUNCS(2) 2124 EXTN_FEATURE_FUNCS(3) 2125 EXTN_FEATURE_FUNCS(4) 2126 2127 static inline bool ext4_has_compat_features(struct super_block *sb) 2128 { 2129 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2130 } 2131 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2132 { 2133 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2134 } 2135 static inline bool ext4_has_incompat_features(struct super_block *sb) 2136 { 2137 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2138 } 2139 2140 /* 2141 * Superblock flags 2142 */ 2143 #define EXT4_FLAGS_RESIZING 0 2144 #define EXT4_FLAGS_SHUTDOWN 1 2145 #define EXT4_FLAGS_BDEV_IS_DAX 2 2146 2147 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2148 { 2149 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2150 } 2151 2152 2153 /* 2154 * Default values for user and/or group using reserved blocks 2155 */ 2156 #define EXT4_DEF_RESUID 0 2157 #define EXT4_DEF_RESGID 0 2158 2159 /* 2160 * Default project ID 2161 */ 2162 #define EXT4_DEF_PROJID 0 2163 2164 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2165 2166 /* 2167 * Default mount options 2168 */ 2169 #define EXT4_DEFM_DEBUG 0x0001 2170 #define EXT4_DEFM_BSDGROUPS 0x0002 2171 #define EXT4_DEFM_XATTR_USER 0x0004 2172 #define EXT4_DEFM_ACL 0x0008 2173 #define EXT4_DEFM_UID16 0x0010 2174 #define EXT4_DEFM_JMODE 0x0060 2175 #define EXT4_DEFM_JMODE_DATA 0x0020 2176 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2177 #define EXT4_DEFM_JMODE_WBACK 0x0060 2178 #define EXT4_DEFM_NOBARRIER 0x0100 2179 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2180 #define EXT4_DEFM_DISCARD 0x0400 2181 #define EXT4_DEFM_NODELALLOC 0x0800 2182 2183 /* 2184 * Default journal batch times 2185 */ 2186 #define EXT4_DEF_MIN_BATCH_TIME 0 2187 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2188 2189 /* 2190 * Minimum number of groups in a flexgroup before we separate out 2191 * directories into the first block group of a flexgroup 2192 */ 2193 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2194 2195 /* 2196 * Structure of a directory entry 2197 */ 2198 #define EXT4_NAME_LEN 255 2199 2200 struct ext4_dir_entry { 2201 __le32 inode; /* Inode number */ 2202 __le16 rec_len; /* Directory entry length */ 2203 __le16 name_len; /* Name length */ 2204 char name[EXT4_NAME_LEN]; /* File name */ 2205 }; 2206 2207 2208 /* 2209 * Encrypted Casefolded entries require saving the hash on disk. This structure 2210 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2211 * boundary. 2212 */ 2213 struct ext4_dir_entry_hash { 2214 __le32 hash; 2215 __le32 minor_hash; 2216 }; 2217 2218 /* 2219 * The new version of the directory entry. Since EXT4 structures are 2220 * stored in intel byte order, and the name_len field could never be 2221 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2222 * file_type field. 2223 */ 2224 struct ext4_dir_entry_2 { 2225 __le32 inode; /* Inode number */ 2226 __le16 rec_len; /* Directory entry length */ 2227 __u8 name_len; /* Name length */ 2228 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2229 char name[EXT4_NAME_LEN]; /* File name */ 2230 }; 2231 2232 /* 2233 * Access the hashes at the end of ext4_dir_entry_2 2234 */ 2235 #define EXT4_DIRENT_HASHES(entry) \ 2236 ((struct ext4_dir_entry_hash *) \ 2237 (((void *)(entry)) + \ 2238 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2239 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2240 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2241 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2242 2243 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2244 { 2245 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2246 } 2247 2248 /* 2249 * This is a bogus directory entry at the end of each leaf block that 2250 * records checksums. 2251 */ 2252 struct ext4_dir_entry_tail { 2253 __le32 det_reserved_zero1; /* Pretend to be unused */ 2254 __le16 det_rec_len; /* 12 */ 2255 __u8 det_reserved_zero2; /* Zero name length */ 2256 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2257 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2258 }; 2259 2260 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2261 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2262 ((blocksize) - \ 2263 sizeof(struct ext4_dir_entry_tail)))) 2264 2265 /* 2266 * Ext4 directory file types. Only the low 3 bits are used. The 2267 * other bits are reserved for now. 2268 */ 2269 #define EXT4_FT_UNKNOWN 0 2270 #define EXT4_FT_REG_FILE 1 2271 #define EXT4_FT_DIR 2 2272 #define EXT4_FT_CHRDEV 3 2273 #define EXT4_FT_BLKDEV 4 2274 #define EXT4_FT_FIFO 5 2275 #define EXT4_FT_SOCK 6 2276 #define EXT4_FT_SYMLINK 7 2277 2278 #define EXT4_FT_MAX 8 2279 2280 #define EXT4_FT_DIR_CSUM 0xDE 2281 2282 /* 2283 * EXT4_DIR_PAD defines the directory entries boundaries 2284 * 2285 * NOTE: It must be a multiple of 4 2286 */ 2287 #define EXT4_DIR_PAD 4 2288 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2289 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2290 2291 /* 2292 * The rec_len is dependent on the type of directory. Directories that are 2293 * casefolded and encrypted need to store the hash as well, so we add room for 2294 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2295 * pass NULL for dir, as those entries do not use the extra fields. 2296 */ 2297 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2298 const struct inode *dir) 2299 { 2300 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2301 2302 if (dir && ext4_hash_in_dirent(dir)) 2303 rec_len += sizeof(struct ext4_dir_entry_hash); 2304 return (rec_len & ~EXT4_DIR_ROUND); 2305 } 2306 2307 /* 2308 * If we ever get support for fs block sizes > page_size, we'll need 2309 * to remove the #if statements in the next two functions... 2310 */ 2311 static inline unsigned int 2312 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2313 { 2314 unsigned len = le16_to_cpu(dlen); 2315 2316 #if (PAGE_SIZE >= 65536) 2317 if (len == EXT4_MAX_REC_LEN || len == 0) 2318 return blocksize; 2319 return (len & 65532) | ((len & 3) << 16); 2320 #else 2321 return len; 2322 #endif 2323 } 2324 2325 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2326 { 2327 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2328 BUG(); 2329 #if (PAGE_SIZE >= 65536) 2330 if (len < 65536) 2331 return cpu_to_le16(len); 2332 if (len == blocksize) { 2333 if (blocksize == 65536) 2334 return cpu_to_le16(EXT4_MAX_REC_LEN); 2335 else 2336 return cpu_to_le16(0); 2337 } 2338 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2339 #else 2340 return cpu_to_le16(len); 2341 #endif 2342 } 2343 2344 /* 2345 * Hash Tree Directory indexing 2346 * (c) Daniel Phillips, 2001 2347 */ 2348 2349 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2350 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2351 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2352 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2353 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2354 2355 /* Legal values for the dx_root hash_version field: */ 2356 2357 #define DX_HASH_LEGACY 0 2358 #define DX_HASH_HALF_MD4 1 2359 #define DX_HASH_TEA 2 2360 #define DX_HASH_LEGACY_UNSIGNED 3 2361 #define DX_HASH_HALF_MD4_UNSIGNED 4 2362 #define DX_HASH_TEA_UNSIGNED 5 2363 #define DX_HASH_SIPHASH 6 2364 2365 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2366 const void *address, unsigned int length) 2367 { 2368 struct { 2369 struct shash_desc shash; 2370 char ctx[4]; 2371 } desc; 2372 2373 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2374 2375 desc.shash.tfm = sbi->s_chksum_driver; 2376 *(u32 *)desc.ctx = crc; 2377 2378 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2379 2380 return *(u32 *)desc.ctx; 2381 } 2382 2383 #ifdef __KERNEL__ 2384 2385 /* hash info structure used by the directory hash */ 2386 struct dx_hash_info 2387 { 2388 u32 hash; 2389 u32 minor_hash; 2390 int hash_version; 2391 u32 *seed; 2392 }; 2393 2394 2395 /* 32 and 64 bit signed EOF for dx directories */ 2396 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2397 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2398 2399 2400 /* 2401 * Control parameters used by ext4_htree_next_block 2402 */ 2403 #define HASH_NB_ALWAYS 1 2404 2405 struct ext4_filename { 2406 const struct qstr *usr_fname; 2407 struct fscrypt_str disk_name; 2408 struct dx_hash_info hinfo; 2409 #ifdef CONFIG_FS_ENCRYPTION 2410 struct fscrypt_str crypto_buf; 2411 #endif 2412 #ifdef CONFIG_UNICODE 2413 struct fscrypt_str cf_name; 2414 #endif 2415 }; 2416 2417 #define fname_name(p) ((p)->disk_name.name) 2418 #define fname_usr_name(p) ((p)->usr_fname->name) 2419 #define fname_len(p) ((p)->disk_name.len) 2420 2421 /* 2422 * Describe an inode's exact location on disk and in memory 2423 */ 2424 struct ext4_iloc 2425 { 2426 struct buffer_head *bh; 2427 unsigned long offset; 2428 ext4_group_t block_group; 2429 }; 2430 2431 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2432 { 2433 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2434 } 2435 2436 static inline bool ext4_is_quota_file(struct inode *inode) 2437 { 2438 return IS_NOQUOTA(inode) && 2439 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2440 } 2441 2442 /* 2443 * This structure is stuffed into the struct file's private_data field 2444 * for directories. It is where we put information so that we can do 2445 * readdir operations in hash tree order. 2446 */ 2447 struct dir_private_info { 2448 struct rb_root root; 2449 struct rb_node *curr_node; 2450 struct fname *extra_fname; 2451 loff_t last_pos; 2452 __u32 curr_hash; 2453 __u32 curr_minor_hash; 2454 __u32 next_hash; 2455 }; 2456 2457 /* calculate the first block number of the group */ 2458 static inline ext4_fsblk_t 2459 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2460 { 2461 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2462 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2463 } 2464 2465 /* 2466 * Special error return code only used by dx_probe() and its callers. 2467 */ 2468 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2469 2470 /* htree levels for ext4 */ 2471 #define EXT4_HTREE_LEVEL_COMPAT 2 2472 #define EXT4_HTREE_LEVEL 3 2473 2474 static inline int ext4_dir_htree_level(struct super_block *sb) 2475 { 2476 return ext4_has_feature_largedir(sb) ? 2477 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2478 } 2479 2480 /* 2481 * Timeout and state flag for lazy initialization inode thread. 2482 */ 2483 #define EXT4_DEF_LI_WAIT_MULT 10 2484 #define EXT4_DEF_LI_MAX_START_DELAY 5 2485 #define EXT4_LAZYINIT_QUIT 0x0001 2486 #define EXT4_LAZYINIT_RUNNING 0x0002 2487 2488 /* 2489 * Lazy inode table initialization info 2490 */ 2491 struct ext4_lazy_init { 2492 unsigned long li_state; 2493 struct list_head li_request_list; 2494 struct mutex li_list_mtx; 2495 }; 2496 2497 enum ext4_li_mode { 2498 EXT4_LI_MODE_PREFETCH_BBITMAP, 2499 EXT4_LI_MODE_ITABLE, 2500 }; 2501 2502 struct ext4_li_request { 2503 struct super_block *lr_super; 2504 enum ext4_li_mode lr_mode; 2505 ext4_group_t lr_first_not_zeroed; 2506 ext4_group_t lr_next_group; 2507 struct list_head lr_request; 2508 unsigned long lr_next_sched; 2509 unsigned long lr_timeout; 2510 }; 2511 2512 struct ext4_features { 2513 struct kobject f_kobj; 2514 struct completion f_kobj_unregister; 2515 }; 2516 2517 /* 2518 * This structure will be used for multiple mount protection. It will be 2519 * written into the block number saved in the s_mmp_block field in the 2520 * superblock. Programs that check MMP should assume that if 2521 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2522 * to use the filesystem, regardless of how old the timestamp is. 2523 */ 2524 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2525 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2526 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2527 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2528 2529 struct mmp_struct { 2530 __le32 mmp_magic; /* Magic number for MMP */ 2531 __le32 mmp_seq; /* Sequence no. updated periodically */ 2532 2533 /* 2534 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2535 * purposes and do not affect the correctness of the algorithm 2536 */ 2537 __le64 mmp_time; /* Time last updated */ 2538 char mmp_nodename[64]; /* Node which last updated MMP block */ 2539 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2540 2541 /* 2542 * mmp_check_interval is used to verify if the MMP block has been 2543 * updated on the block device. The value is updated based on the 2544 * maximum time to write the MMP block during an update cycle. 2545 */ 2546 __le16 mmp_check_interval; 2547 2548 __le16 mmp_pad1; 2549 __le32 mmp_pad2[226]; 2550 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2551 }; 2552 2553 /* arguments passed to the mmp thread */ 2554 struct mmpd_data { 2555 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2556 struct super_block *sb; /* super block of the fs */ 2557 }; 2558 2559 /* 2560 * Check interval multiplier 2561 * The MMP block is written every update interval and initially checked every 2562 * update interval x the multiplier (the value is then adapted based on the 2563 * write latency). The reason is that writes can be delayed under load and we 2564 * don't want readers to incorrectly assume that the filesystem is no longer 2565 * in use. 2566 */ 2567 #define EXT4_MMP_CHECK_MULT 2UL 2568 2569 /* 2570 * Minimum interval for MMP checking in seconds. 2571 */ 2572 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2573 2574 /* 2575 * Maximum interval for MMP checking in seconds. 2576 */ 2577 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2578 2579 /* 2580 * Function prototypes 2581 */ 2582 2583 /* 2584 * Ok, these declarations are also in <linux/kernel.h> but none of the 2585 * ext4 source programs needs to include it so they are duplicated here. 2586 */ 2587 # define NORET_TYPE /**/ 2588 # define ATTRIB_NORET __attribute__((noreturn)) 2589 # define NORET_AND noreturn, 2590 2591 /* bitmap.c */ 2592 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2593 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2594 struct ext4_group_desc *gdp, 2595 struct buffer_head *bh, int sz); 2596 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2597 struct ext4_group_desc *gdp, 2598 struct buffer_head *bh, int sz); 2599 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2600 struct ext4_group_desc *gdp, 2601 struct buffer_head *bh); 2602 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2603 struct ext4_group_desc *gdp, 2604 struct buffer_head *bh); 2605 2606 /* balloc.c */ 2607 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2608 ext4_fsblk_t blocknr, 2609 ext4_group_t *blockgrpp, 2610 ext4_grpblk_t *offsetp); 2611 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2612 ext4_fsblk_t block); 2613 2614 extern unsigned int ext4_block_group(struct super_block *sb, 2615 ext4_fsblk_t blocknr); 2616 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2617 ext4_fsblk_t blocknr); 2618 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2619 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2620 ext4_group_t group); 2621 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2622 ext4_fsblk_t goal, 2623 unsigned int flags, 2624 unsigned long *count, 2625 int *errp); 2626 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2627 s64 nclusters, unsigned int flags); 2628 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2629 extern void ext4_check_blocks_bitmap(struct super_block *); 2630 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2631 ext4_group_t block_group, 2632 struct buffer_head ** bh); 2633 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2634 2635 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2636 ext4_group_t block_group, 2637 bool ignore_locked); 2638 extern int ext4_wait_block_bitmap(struct super_block *sb, 2639 ext4_group_t block_group, 2640 struct buffer_head *bh); 2641 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2642 ext4_group_t block_group); 2643 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2644 ext4_group_t block_group, 2645 struct ext4_group_desc *gdp); 2646 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2647 2648 #ifdef CONFIG_UNICODE 2649 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2650 const struct qstr *iname, 2651 struct ext4_filename *fname); 2652 #endif 2653 2654 #ifdef CONFIG_FS_ENCRYPTION 2655 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst, 2656 const struct fscrypt_name *src) 2657 { 2658 memset(dst, 0, sizeof(*dst)); 2659 2660 dst->usr_fname = src->usr_fname; 2661 dst->disk_name = src->disk_name; 2662 dst->hinfo.hash = src->hash; 2663 dst->hinfo.minor_hash = src->minor_hash; 2664 dst->crypto_buf = src->crypto_buf; 2665 } 2666 2667 static inline int ext4_fname_setup_filename(struct inode *dir, 2668 const struct qstr *iname, 2669 int lookup, 2670 struct ext4_filename *fname) 2671 { 2672 struct fscrypt_name name; 2673 int err; 2674 2675 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2676 if (err) 2677 return err; 2678 2679 ext4_fname_from_fscrypt_name(fname, &name); 2680 2681 #ifdef CONFIG_UNICODE 2682 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2683 #endif 2684 return err; 2685 } 2686 2687 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2688 struct dentry *dentry, 2689 struct ext4_filename *fname) 2690 { 2691 struct fscrypt_name name; 2692 int err; 2693 2694 err = fscrypt_prepare_lookup(dir, dentry, &name); 2695 if (err) 2696 return err; 2697 2698 ext4_fname_from_fscrypt_name(fname, &name); 2699 2700 #ifdef CONFIG_UNICODE 2701 err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname); 2702 #endif 2703 return err; 2704 } 2705 2706 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2707 { 2708 struct fscrypt_name name; 2709 2710 name.crypto_buf = fname->crypto_buf; 2711 fscrypt_free_filename(&name); 2712 2713 fname->crypto_buf.name = NULL; 2714 fname->usr_fname = NULL; 2715 fname->disk_name.name = NULL; 2716 2717 #ifdef CONFIG_UNICODE 2718 kfree(fname->cf_name.name); 2719 fname->cf_name.name = NULL; 2720 #endif 2721 } 2722 #else /* !CONFIG_FS_ENCRYPTION */ 2723 static inline int ext4_fname_setup_filename(struct inode *dir, 2724 const struct qstr *iname, 2725 int lookup, 2726 struct ext4_filename *fname) 2727 { 2728 int err = 0; 2729 fname->usr_fname = iname; 2730 fname->disk_name.name = (unsigned char *) iname->name; 2731 fname->disk_name.len = iname->len; 2732 2733 #ifdef CONFIG_UNICODE 2734 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2735 #endif 2736 2737 return err; 2738 } 2739 2740 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2741 struct dentry *dentry, 2742 struct ext4_filename *fname) 2743 { 2744 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2745 } 2746 2747 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2748 { 2749 #ifdef CONFIG_UNICODE 2750 kfree(fname->cf_name.name); 2751 fname->cf_name.name = NULL; 2752 #endif 2753 } 2754 #endif /* !CONFIG_FS_ENCRYPTION */ 2755 2756 /* dir.c */ 2757 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2758 struct file *, 2759 struct ext4_dir_entry_2 *, 2760 struct buffer_head *, char *, int, 2761 unsigned int); 2762 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2763 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2764 (de), (bh), (buf), (size), (offset))) 2765 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2766 __u32 minor_hash, 2767 struct ext4_dir_entry_2 *dirent, 2768 struct fscrypt_str *ent_name); 2769 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2770 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2771 struct buffer_head *bh, 2772 void *buf, int buf_size, 2773 struct ext4_filename *fname, 2774 struct ext4_dir_entry_2 **dest_de); 2775 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2776 struct ext4_dir_entry_2 *de, 2777 int buf_size, 2778 struct ext4_filename *fname); 2779 static inline void ext4_update_dx_flag(struct inode *inode) 2780 { 2781 if (!ext4_has_feature_dir_index(inode->i_sb) && 2782 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2783 /* ext4_iget() should have caught this... */ 2784 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2785 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2786 } 2787 } 2788 static const unsigned char ext4_filetype_table[] = { 2789 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2790 }; 2791 2792 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2793 { 2794 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2795 return DT_UNKNOWN; 2796 2797 return ext4_filetype_table[filetype]; 2798 } 2799 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2800 void *buf, int buf_size); 2801 2802 /* fsync.c */ 2803 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2804 2805 /* hash.c */ 2806 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2807 struct dx_hash_info *hinfo); 2808 2809 /* ialloc.c */ 2810 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2811 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *, 2812 struct inode *, umode_t, 2813 const struct qstr *qstr, __u32 goal, 2814 uid_t *owner, __u32 i_flags, 2815 int handle_type, unsigned int line_no, 2816 int nblocks); 2817 2818 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2819 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \ 2820 (goal), (owner), i_flags, 0, 0, 0) 2821 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \ 2822 type, nblocks) \ 2823 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2824 0, (type), __LINE__, (nblocks)) 2825 2826 2827 extern void ext4_free_inode(handle_t *, struct inode *); 2828 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2829 extern unsigned long ext4_count_free_inodes(struct super_block *); 2830 extern unsigned long ext4_count_dirs(struct super_block *); 2831 extern void ext4_check_inodes_bitmap(struct super_block *); 2832 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2833 extern int ext4_init_inode_table(struct super_block *sb, 2834 ext4_group_t group, int barrier); 2835 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2836 2837 /* fast_commit.c */ 2838 int ext4_fc_info_show(struct seq_file *seq, void *v); 2839 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2840 void ext4_fc_init_inode(struct inode *inode); 2841 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2842 ext4_lblk_t end); 2843 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2844 struct dentry *dentry); 2845 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2846 struct dentry *dentry); 2847 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2848 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2849 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2850 struct dentry *dentry); 2851 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2852 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2853 void ext4_fc_mark_ineligible(struct super_block *sb, int reason); 2854 void ext4_fc_start_ineligible(struct super_block *sb, int reason); 2855 void ext4_fc_stop_ineligible(struct super_block *sb); 2856 void ext4_fc_start_update(struct inode *inode); 2857 void ext4_fc_stop_update(struct inode *inode); 2858 void ext4_fc_del(struct inode *inode); 2859 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2860 void ext4_fc_replay_cleanup(struct super_block *sb); 2861 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2862 int __init ext4_fc_init_dentry_cache(void); 2863 2864 /* mballoc.c */ 2865 extern const struct seq_operations ext4_mb_seq_groups_ops; 2866 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2867 extern long ext4_mb_stats; 2868 extern long ext4_mb_max_to_scan; 2869 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2870 extern int ext4_mb_init(struct super_block *); 2871 extern int ext4_mb_release(struct super_block *); 2872 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2873 struct ext4_allocation_request *, int *); 2874 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2875 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2876 extern int __init ext4_init_mballoc(void); 2877 extern void ext4_exit_mballoc(void); 2878 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2879 ext4_group_t group, 2880 unsigned int nr, int *cnt); 2881 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2882 unsigned int nr); 2883 2884 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2885 struct buffer_head *bh, ext4_fsblk_t block, 2886 unsigned long count, int flags); 2887 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2888 ext4_group_t ngroups); 2889 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2890 ext4_group_t i, struct ext4_group_desc *desc); 2891 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2892 ext4_fsblk_t block, unsigned long count); 2893 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2894 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2895 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2896 int len, int state); 2897 2898 /* inode.c */ 2899 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2900 struct ext4_inode_info *ei); 2901 int ext4_inode_is_fast_symlink(struct inode *inode); 2902 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2903 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2904 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2905 bool wait, struct buffer_head **bhs); 2906 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2907 struct buffer_head *bh_result, int create); 2908 int ext4_get_block(struct inode *inode, sector_t iblock, 2909 struct buffer_head *bh_result, int create); 2910 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2911 struct buffer_head *bh, int create); 2912 int ext4_walk_page_buffers(handle_t *handle, 2913 struct buffer_head *head, 2914 unsigned from, 2915 unsigned to, 2916 int *partial, 2917 int (*fn)(handle_t *handle, 2918 struct buffer_head *bh)); 2919 int do_journal_get_write_access(handle_t *handle, 2920 struct buffer_head *bh); 2921 #define FALL_BACK_TO_NONDELALLOC 1 2922 #define CONVERT_INLINE_DATA 2 2923 2924 typedef enum { 2925 EXT4_IGET_NORMAL = 0, 2926 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2927 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2928 } ext4_iget_flags; 2929 2930 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2931 ext4_iget_flags flags, const char *function, 2932 unsigned int line); 2933 2934 #define ext4_iget(sb, ino, flags) \ 2935 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2936 2937 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2938 extern int ext4_setattr(struct user_namespace *, struct dentry *, 2939 struct iattr *); 2940 extern int ext4_getattr(struct user_namespace *, const struct path *, 2941 struct kstat *, u32, unsigned int); 2942 extern void ext4_evict_inode(struct inode *); 2943 extern void ext4_clear_inode(struct inode *); 2944 extern int ext4_file_getattr(struct user_namespace *, const struct path *, 2945 struct kstat *, u32, unsigned int); 2946 extern int ext4_sync_inode(handle_t *, struct inode *); 2947 extern void ext4_dirty_inode(struct inode *, int); 2948 extern int ext4_change_inode_journal_flag(struct inode *, int); 2949 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2950 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2951 struct ext4_iloc *iloc); 2952 extern int ext4_inode_attach_jinode(struct inode *inode); 2953 extern int ext4_can_truncate(struct inode *inode); 2954 extern int ext4_truncate(struct inode *); 2955 extern int ext4_break_layouts(struct inode *); 2956 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2957 extern void ext4_set_inode_flags(struct inode *, bool init); 2958 extern int ext4_alloc_da_blocks(struct inode *inode); 2959 extern void ext4_set_aops(struct inode *inode); 2960 extern int ext4_writepage_trans_blocks(struct inode *); 2961 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2962 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2963 loff_t lstart, loff_t lend); 2964 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2965 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2966 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2967 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2968 extern void ext4_da_release_space(struct inode *inode, int to_free); 2969 extern void ext4_da_update_reserve_space(struct inode *inode, 2970 int used, int quota_claim); 2971 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2972 ext4_fsblk_t pblk, ext4_lblk_t len); 2973 2974 /* indirect.c */ 2975 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2976 struct ext4_map_blocks *map, int flags); 2977 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2978 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2979 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2980 ext4_lblk_t start, ext4_lblk_t end); 2981 2982 /* ioctl.c */ 2983 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2984 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2985 int ext4_fileattr_set(struct user_namespace *mnt_userns, 2986 struct dentry *dentry, struct fileattr *fa); 2987 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 2988 extern void ext4_reset_inode_seed(struct inode *inode); 2989 2990 /* migrate.c */ 2991 extern int ext4_ext_migrate(struct inode *); 2992 extern int ext4_ind_migrate(struct inode *inode); 2993 2994 /* namei.c */ 2995 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 2996 struct inode *inode); 2997 extern int ext4_dirblock_csum_verify(struct inode *inode, 2998 struct buffer_head *bh); 2999 extern int ext4_orphan_add(handle_t *, struct inode *); 3000 extern int ext4_orphan_del(handle_t *, struct inode *); 3001 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3002 __u32 start_minor_hash, __u32 *next_hash); 3003 extern int ext4_search_dir(struct buffer_head *bh, 3004 char *search_buf, 3005 int buf_size, 3006 struct inode *dir, 3007 struct ext4_filename *fname, 3008 unsigned int offset, 3009 struct ext4_dir_entry_2 **res_dir); 3010 extern int ext4_generic_delete_entry(struct inode *dir, 3011 struct ext4_dir_entry_2 *de_del, 3012 struct buffer_head *bh, 3013 void *entry_buf, 3014 int buf_size, 3015 int csum_size); 3016 extern bool ext4_empty_dir(struct inode *inode); 3017 3018 /* resize.c */ 3019 extern void ext4_kvfree_array_rcu(void *to_free); 3020 extern int ext4_group_add(struct super_block *sb, 3021 struct ext4_new_group_data *input); 3022 extern int ext4_group_extend(struct super_block *sb, 3023 struct ext4_super_block *es, 3024 ext4_fsblk_t n_blocks_count); 3025 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3026 3027 /* super.c */ 3028 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3029 sector_t block, int op_flags); 3030 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3031 sector_t block); 3032 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags, 3033 bh_end_io_t *end_io); 3034 extern int ext4_read_bh(struct buffer_head *bh, int op_flags, 3035 bh_end_io_t *end_io); 3036 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait); 3037 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3038 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3039 extern int ext4_calculate_overhead(struct super_block *sb); 3040 extern void ext4_superblock_csum_set(struct super_block *sb); 3041 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3042 ext4_group_t ngroup); 3043 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3044 char nbuf[16]); 3045 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3046 ext4_group_t block_group, 3047 unsigned int flags); 3048 3049 extern __printf(7, 8) 3050 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3051 int, __u64, const char *, ...); 3052 extern __printf(6, 7) 3053 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3054 ext4_fsblk_t, int, const char *, ...); 3055 extern __printf(5, 6) 3056 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3057 const char *, ...); 3058 extern void __ext4_std_error(struct super_block *, const char *, 3059 unsigned int, int); 3060 extern __printf(4, 5) 3061 void __ext4_warning(struct super_block *, const char *, unsigned int, 3062 const char *, ...); 3063 extern __printf(4, 5) 3064 void __ext4_warning_inode(const struct inode *inode, const char *function, 3065 unsigned int line, const char *fmt, ...); 3066 extern __printf(3, 4) 3067 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3068 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3069 const char *, unsigned int, const char *); 3070 extern __printf(7, 8) 3071 void __ext4_grp_locked_error(const char *, unsigned int, 3072 struct super_block *, ext4_group_t, 3073 unsigned long, ext4_fsblk_t, 3074 const char *, ...); 3075 3076 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3077 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3078 3079 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3080 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3081 3082 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3083 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3084 (fmt), ## a) 3085 3086 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3087 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3088 3089 #define ext4_abort(sb, err, fmt, a...) \ 3090 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3091 3092 #ifdef CONFIG_PRINTK 3093 3094 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3095 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3096 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3097 __ext4_error_inode((inode), (func), (line), (block), \ 3098 (err), (fmt), ##__VA_ARGS__) 3099 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3100 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3101 #define ext4_error(sb, fmt, ...) \ 3102 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3103 ##__VA_ARGS__) 3104 #define ext4_error_err(sb, err, fmt, ...) \ 3105 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3106 ##__VA_ARGS__) 3107 #define ext4_warning(sb, fmt, ...) \ 3108 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3109 #define ext4_warning_inode(inode, fmt, ...) \ 3110 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3111 #define ext4_msg(sb, level, fmt, ...) \ 3112 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3113 #define dump_mmp_msg(sb, mmp, msg) \ 3114 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3115 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3116 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3117 fmt, ##__VA_ARGS__) 3118 3119 #else 3120 3121 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3122 do { \ 3123 no_printk(fmt, ##__VA_ARGS__); \ 3124 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3125 } while (0) 3126 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3127 do { \ 3128 no_printk(fmt, ##__VA_ARGS__); \ 3129 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3130 } while (0) 3131 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3132 do { \ 3133 no_printk(fmt, ##__VA_ARGS__); \ 3134 __ext4_error_file(file, "", 0, block, " "); \ 3135 } while (0) 3136 #define ext4_error(sb, fmt, ...) \ 3137 do { \ 3138 no_printk(fmt, ##__VA_ARGS__); \ 3139 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3140 } while (0) 3141 #define ext4_error_err(sb, err, fmt, ...) \ 3142 do { \ 3143 no_printk(fmt, ##__VA_ARGS__); \ 3144 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3145 } while (0) 3146 #define ext4_warning(sb, fmt, ...) \ 3147 do { \ 3148 no_printk(fmt, ##__VA_ARGS__); \ 3149 __ext4_warning(sb, "", 0, " "); \ 3150 } while (0) 3151 #define ext4_warning_inode(inode, fmt, ...) \ 3152 do { \ 3153 no_printk(fmt, ##__VA_ARGS__); \ 3154 __ext4_warning_inode(inode, "", 0, " "); \ 3155 } while (0) 3156 #define ext4_msg(sb, level, fmt, ...) \ 3157 do { \ 3158 no_printk(fmt, ##__VA_ARGS__); \ 3159 __ext4_msg(sb, "", " "); \ 3160 } while (0) 3161 #define dump_mmp_msg(sb, mmp, msg) \ 3162 __dump_mmp_msg(sb, mmp, "", 0, "") 3163 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3164 do { \ 3165 no_printk(fmt, ##__VA_ARGS__); \ 3166 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3167 } while (0) 3168 3169 #endif 3170 3171 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3172 struct ext4_group_desc *bg); 3173 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3174 struct ext4_group_desc *bg); 3175 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3176 struct ext4_group_desc *bg); 3177 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3178 struct ext4_group_desc *bg); 3179 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3180 struct ext4_group_desc *bg); 3181 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3182 struct ext4_group_desc *bg); 3183 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3184 struct ext4_group_desc *bg); 3185 extern void ext4_block_bitmap_set(struct super_block *sb, 3186 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3187 extern void ext4_inode_bitmap_set(struct super_block *sb, 3188 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3189 extern void ext4_inode_table_set(struct super_block *sb, 3190 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3191 extern void ext4_free_group_clusters_set(struct super_block *sb, 3192 struct ext4_group_desc *bg, 3193 __u32 count); 3194 extern void ext4_free_inodes_set(struct super_block *sb, 3195 struct ext4_group_desc *bg, __u32 count); 3196 extern void ext4_used_dirs_set(struct super_block *sb, 3197 struct ext4_group_desc *bg, __u32 count); 3198 extern void ext4_itable_unused_set(struct super_block *sb, 3199 struct ext4_group_desc *bg, __u32 count); 3200 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3201 struct ext4_group_desc *gdp); 3202 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3203 struct ext4_group_desc *gdp); 3204 extern int ext4_register_li_request(struct super_block *sb, 3205 ext4_group_t first_not_zeroed); 3206 3207 static inline int ext4_has_metadata_csum(struct super_block *sb) 3208 { 3209 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3210 !EXT4_SB(sb)->s_chksum_driver); 3211 3212 return ext4_has_feature_metadata_csum(sb) && 3213 (EXT4_SB(sb)->s_chksum_driver != NULL); 3214 } 3215 3216 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3217 { 3218 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3219 } 3220 3221 #define ext4_read_incompat_64bit_val(es, name) \ 3222 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3223 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3224 le32_to_cpu(es->name##_lo)) 3225 3226 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3227 { 3228 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3229 } 3230 3231 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3232 { 3233 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3234 } 3235 3236 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3237 { 3238 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3239 } 3240 3241 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3242 ext4_fsblk_t blk) 3243 { 3244 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3245 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3246 } 3247 3248 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3249 ext4_fsblk_t blk) 3250 { 3251 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3252 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3253 } 3254 3255 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3256 ext4_fsblk_t blk) 3257 { 3258 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3259 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3260 } 3261 3262 static inline loff_t ext4_isize(struct super_block *sb, 3263 struct ext4_inode *raw_inode) 3264 { 3265 if (ext4_has_feature_largedir(sb) || 3266 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3267 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3268 le32_to_cpu(raw_inode->i_size_lo); 3269 3270 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3271 } 3272 3273 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3274 { 3275 raw_inode->i_size_lo = cpu_to_le32(i_size); 3276 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3277 } 3278 3279 static inline 3280 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3281 ext4_group_t group) 3282 { 3283 struct ext4_group_info **grp_info; 3284 long indexv, indexh; 3285 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3286 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3287 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3288 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3289 return grp_info[indexh]; 3290 } 3291 3292 /* 3293 * Reading s_groups_count requires using smp_rmb() afterwards. See 3294 * the locking protocol documented in the comments of ext4_group_add() 3295 * in resize.c 3296 */ 3297 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3298 { 3299 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3300 3301 smp_rmb(); 3302 return ngroups; 3303 } 3304 3305 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3306 ext4_group_t block_group) 3307 { 3308 return block_group >> sbi->s_log_groups_per_flex; 3309 } 3310 3311 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3312 { 3313 return 1 << sbi->s_log_groups_per_flex; 3314 } 3315 3316 #define ext4_std_error(sb, errno) \ 3317 do { \ 3318 if ((errno)) \ 3319 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3320 } while (0) 3321 3322 #ifdef CONFIG_SMP 3323 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3324 * counters. So we need to make sure we have free clusters more 3325 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3326 */ 3327 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3328 #else 3329 #define EXT4_FREECLUSTERS_WATERMARK 0 3330 #endif 3331 3332 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 3333 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3334 { 3335 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3336 !inode_is_locked(inode)); 3337 down_write(&EXT4_I(inode)->i_data_sem); 3338 if (newsize > EXT4_I(inode)->i_disksize) 3339 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3340 up_write(&EXT4_I(inode)->i_data_sem); 3341 } 3342 3343 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 3344 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3345 { 3346 int changed = 0; 3347 3348 if (newsize > inode->i_size) { 3349 i_size_write(inode, newsize); 3350 changed = 1; 3351 } 3352 if (newsize > EXT4_I(inode)->i_disksize) { 3353 ext4_update_i_disksize(inode, newsize); 3354 changed |= 2; 3355 } 3356 return changed; 3357 } 3358 3359 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3360 loff_t len); 3361 3362 struct ext4_group_info { 3363 unsigned long bb_state; 3364 #ifdef AGGRESSIVE_CHECK 3365 unsigned long bb_check_counter; 3366 #endif 3367 struct rb_root bb_free_root; 3368 ext4_grpblk_t bb_first_free; /* first free block */ 3369 ext4_grpblk_t bb_free; /* total free blocks */ 3370 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3371 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3372 ext4_group_t bb_group; /* Group number */ 3373 struct list_head bb_prealloc_list; 3374 #ifdef DOUBLE_CHECK 3375 void *bb_bitmap; 3376 #endif 3377 struct rw_semaphore alloc_sem; 3378 struct rb_node bb_avg_fragment_size_rb; 3379 struct list_head bb_largest_free_order_node; 3380 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3381 * regions, index is order. 3382 * bb_counters[3] = 5 means 3383 * 5 free 8-block regions. */ 3384 }; 3385 3386 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3387 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3388 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3389 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3390 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3391 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3392 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3393 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3394 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3395 3396 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3397 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3398 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3399 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3400 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3401 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3402 3403 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3404 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3405 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3406 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3407 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3408 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3409 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3410 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3411 3412 #define EXT4_MAX_CONTENTION 8 3413 #define EXT4_CONTENTION_THRESHOLD 2 3414 3415 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3416 ext4_group_t group) 3417 { 3418 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3419 } 3420 3421 /* 3422 * Returns true if the filesystem is busy enough that attempts to 3423 * access the block group locks has run into contention. 3424 */ 3425 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3426 { 3427 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3428 } 3429 3430 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3431 { 3432 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3433 if (spin_trylock(lock)) 3434 /* 3435 * We're able to grab the lock right away, so drop the 3436 * lock contention counter. 3437 */ 3438 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3439 else { 3440 /* 3441 * The lock is busy, so bump the contention counter, 3442 * and then wait on the spin lock. 3443 */ 3444 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3445 EXT4_MAX_CONTENTION); 3446 spin_lock(lock); 3447 } 3448 } 3449 3450 static inline void ext4_unlock_group(struct super_block *sb, 3451 ext4_group_t group) 3452 { 3453 spin_unlock(ext4_group_lock_ptr(sb, group)); 3454 } 3455 3456 #ifdef CONFIG_QUOTA 3457 static inline bool ext4_quota_capable(struct super_block *sb) 3458 { 3459 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3460 } 3461 3462 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3463 { 3464 struct ext4_sb_info *sbi = EXT4_SB(sb); 3465 3466 return (ext4_has_feature_quota(sb) || 3467 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3468 } 3469 #endif 3470 3471 /* 3472 * Block validity checking 3473 */ 3474 #define ext4_check_indirect_blockref(inode, bh) \ 3475 ext4_check_blockref(__func__, __LINE__, inode, \ 3476 (__le32 *)(bh)->b_data, \ 3477 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3478 3479 #define ext4_ind_check_inode(inode) \ 3480 ext4_check_blockref(__func__, __LINE__, inode, \ 3481 EXT4_I(inode)->i_data, \ 3482 EXT4_NDIR_BLOCKS) 3483 3484 /* 3485 * Inodes and files operations 3486 */ 3487 3488 /* dir.c */ 3489 extern const struct file_operations ext4_dir_operations; 3490 3491 /* file.c */ 3492 extern const struct inode_operations ext4_file_inode_operations; 3493 extern const struct file_operations ext4_file_operations; 3494 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3495 3496 /* inline.c */ 3497 extern int ext4_get_max_inline_size(struct inode *inode); 3498 extern int ext4_find_inline_data_nolock(struct inode *inode); 3499 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3500 unsigned int len); 3501 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3502 3503 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3504 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3505 struct inode *inode, 3506 loff_t pos, unsigned len, 3507 unsigned flags, 3508 struct page **pagep); 3509 extern int ext4_write_inline_data_end(struct inode *inode, 3510 loff_t pos, unsigned len, 3511 unsigned copied, 3512 struct page *page); 3513 extern struct buffer_head * 3514 ext4_journalled_write_inline_data(struct inode *inode, 3515 unsigned len, 3516 struct page *page); 3517 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3518 struct inode *inode, 3519 loff_t pos, unsigned len, 3520 unsigned flags, 3521 struct page **pagep, 3522 void **fsdata); 3523 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3524 unsigned len, unsigned copied, 3525 struct page *page); 3526 extern int ext4_try_add_inline_entry(handle_t *handle, 3527 struct ext4_filename *fname, 3528 struct inode *dir, struct inode *inode); 3529 extern int ext4_try_create_inline_dir(handle_t *handle, 3530 struct inode *parent, 3531 struct inode *inode); 3532 extern int ext4_read_inline_dir(struct file *filp, 3533 struct dir_context *ctx, 3534 int *has_inline_data); 3535 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3536 struct inode *dir, ext4_lblk_t block, 3537 struct dx_hash_info *hinfo, 3538 __u32 start_hash, __u32 start_minor_hash, 3539 int *has_inline_data); 3540 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3541 struct ext4_filename *fname, 3542 struct ext4_dir_entry_2 **res_dir, 3543 int *has_inline_data); 3544 extern int ext4_delete_inline_entry(handle_t *handle, 3545 struct inode *dir, 3546 struct ext4_dir_entry_2 *de_del, 3547 struct buffer_head *bh, 3548 int *has_inline_data); 3549 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3550 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3551 struct ext4_dir_entry_2 **parent_de, 3552 int *retval); 3553 extern int ext4_inline_data_fiemap(struct inode *inode, 3554 struct fiemap_extent_info *fieinfo, 3555 int *has_inline, __u64 start, __u64 len); 3556 3557 struct iomap; 3558 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3559 3560 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3561 3562 extern int ext4_convert_inline_data(struct inode *inode); 3563 3564 static inline int ext4_has_inline_data(struct inode *inode) 3565 { 3566 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3567 EXT4_I(inode)->i_inline_off; 3568 } 3569 3570 /* namei.c */ 3571 extern const struct inode_operations ext4_dir_inode_operations; 3572 extern const struct inode_operations ext4_special_inode_operations; 3573 extern struct dentry *ext4_get_parent(struct dentry *child); 3574 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3575 struct ext4_dir_entry_2 *de, 3576 int blocksize, int csum_size, 3577 unsigned int parent_ino, int dotdot_real_len); 3578 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3579 unsigned int blocksize); 3580 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3581 struct buffer_head *bh); 3582 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name, 3583 struct inode *inode); 3584 extern int __ext4_link(struct inode *dir, struct inode *inode, 3585 struct dentry *dentry); 3586 3587 #define S_SHIFT 12 3588 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3589 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3590 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3591 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3592 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3593 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3594 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3595 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3596 }; 3597 3598 static inline void ext4_set_de_type(struct super_block *sb, 3599 struct ext4_dir_entry_2 *de, 3600 umode_t mode) { 3601 if (ext4_has_feature_filetype(sb)) 3602 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3603 } 3604 3605 /* readpages.c */ 3606 extern int ext4_mpage_readpages(struct inode *inode, 3607 struct readahead_control *rac, struct page *page); 3608 extern int __init ext4_init_post_read_processing(void); 3609 extern void ext4_exit_post_read_processing(void); 3610 3611 /* symlink.c */ 3612 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3613 extern const struct inode_operations ext4_symlink_inode_operations; 3614 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3615 3616 /* sysfs.c */ 3617 extern int ext4_register_sysfs(struct super_block *sb); 3618 extern void ext4_unregister_sysfs(struct super_block *sb); 3619 extern int __init ext4_init_sysfs(void); 3620 extern void ext4_exit_sysfs(void); 3621 3622 /* block_validity */ 3623 extern void ext4_release_system_zone(struct super_block *sb); 3624 extern int ext4_setup_system_zone(struct super_block *sb); 3625 extern int __init ext4_init_system_zone(void); 3626 extern void ext4_exit_system_zone(void); 3627 extern int ext4_inode_block_valid(struct inode *inode, 3628 ext4_fsblk_t start_blk, 3629 unsigned int count); 3630 extern int ext4_check_blockref(const char *, unsigned int, 3631 struct inode *, __le32 *, unsigned int); 3632 3633 /* extents.c */ 3634 struct ext4_ext_path; 3635 struct ext4_extent; 3636 3637 /* 3638 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3639 * __le32. 3640 */ 3641 #define EXT_MAX_BLOCKS 0xffffffff 3642 3643 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3644 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3645 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3646 struct ext4_map_blocks *map, int flags); 3647 extern int ext4_ext_truncate(handle_t *, struct inode *); 3648 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3649 ext4_lblk_t end); 3650 extern void ext4_ext_init(struct super_block *); 3651 extern void ext4_ext_release(struct super_block *); 3652 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3653 loff_t len); 3654 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3655 loff_t offset, ssize_t len); 3656 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3657 ext4_io_end_t *io_end); 3658 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3659 struct ext4_map_blocks *map, int flags); 3660 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3661 int num, 3662 struct ext4_ext_path *path); 3663 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3664 struct ext4_ext_path **, 3665 struct ext4_extent *, int); 3666 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3667 struct ext4_ext_path **, 3668 int flags); 3669 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3670 extern int ext4_ext_check_inode(struct inode *inode); 3671 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3672 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3673 __u64 start, __u64 len); 3674 extern int ext4_get_es_cache(struct inode *inode, 3675 struct fiemap_extent_info *fieinfo, 3676 __u64 start, __u64 len); 3677 extern int ext4_ext_precache(struct inode *inode); 3678 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3679 struct inode *inode2, ext4_lblk_t lblk1, 3680 ext4_lblk_t lblk2, ext4_lblk_t count, 3681 int mark_unwritten,int *err); 3682 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3683 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3684 int check_cred, int restart_cred, 3685 int revoke_cred); 3686 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3687 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3688 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3689 int len, int unwritten, ext4_fsblk_t pblk); 3690 extern int ext4_ext_clear_bb(struct inode *inode); 3691 3692 3693 /* move_extent.c */ 3694 extern void ext4_double_down_write_data_sem(struct inode *first, 3695 struct inode *second); 3696 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3697 struct inode *donor_inode); 3698 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3699 __u64 start_orig, __u64 start_donor, 3700 __u64 len, __u64 *moved_len); 3701 3702 /* page-io.c */ 3703 extern int __init ext4_init_pageio(void); 3704 extern void ext4_exit_pageio(void); 3705 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3706 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3707 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3708 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3709 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3710 struct writeback_control *wbc); 3711 extern void ext4_end_io_rsv_work(struct work_struct *work); 3712 extern void ext4_io_submit(struct ext4_io_submit *io); 3713 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3714 struct page *page, 3715 int len, 3716 bool keep_towrite); 3717 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3718 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3719 3720 /* mmp.c */ 3721 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3722 3723 /* verity.c */ 3724 extern const struct fsverity_operations ext4_verityops; 3725 3726 /* 3727 * Add new method to test whether block and inode bitmaps are properly 3728 * initialized. With uninit_bg reading the block from disk is not enough 3729 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3730 */ 3731 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3732 3733 static inline int bitmap_uptodate(struct buffer_head *bh) 3734 { 3735 return (buffer_uptodate(bh) && 3736 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3737 } 3738 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3739 { 3740 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3741 } 3742 3743 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3744 3745 /* For ioend & aio unwritten conversion wait queues */ 3746 #define EXT4_WQ_HASH_SZ 37 3747 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3748 EXT4_WQ_HASH_SZ]) 3749 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3750 3751 extern int ext4_resize_begin(struct super_block *sb); 3752 extern void ext4_resize_end(struct super_block *sb); 3753 3754 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3755 struct ext4_io_end *io_end) 3756 { 3757 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3758 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3759 atomic_inc(&EXT4_I(inode)->i_unwritten); 3760 } 3761 } 3762 3763 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3764 { 3765 struct inode *inode = io_end->inode; 3766 3767 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3768 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3769 /* Wake up anyone waiting on unwritten extent conversion */ 3770 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3771 wake_up_all(ext4_ioend_wq(inode)); 3772 } 3773 } 3774 3775 extern const struct iomap_ops ext4_iomap_ops; 3776 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3777 extern const struct iomap_ops ext4_iomap_report_ops; 3778 3779 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3780 { 3781 /* 3782 * If the buffer has the write error flag, we have failed 3783 * to write out data in the block. In this case, we don't 3784 * have to read the block because we may read the old data 3785 * successfully. 3786 */ 3787 if (!buffer_uptodate(bh) && buffer_write_io_error(bh)) 3788 set_buffer_uptodate(bh); 3789 return buffer_uptodate(bh); 3790 } 3791 3792 #endif /* __KERNEL__ */ 3793 3794 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3795 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3796 3797 #endif /* _EXT4_H */ 3798