1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/refcount.h> 21 #include <linux/types.h> 22 #include <linux/blkdev.h> 23 #include <linux/magic.h> 24 #include <linux/jbd2.h> 25 #include <linux/quota.h> 26 #include <linux/rwsem.h> 27 #include <linux/rbtree.h> 28 #include <linux/seqlock.h> 29 #include <linux/mutex.h> 30 #include <linux/timer.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #include <linux/fiemap.h> 40 #ifdef __KERNEL__ 41 #include <linux/compat.h> 42 #endif 43 #include <uapi/linux/ext4.h> 44 45 #include <linux/fscrypt.h> 46 #include <linux/fsverity.h> 47 48 #include <linux/compiler.h> 49 50 /* 51 * The fourth extended filesystem constants/structures 52 */ 53 54 /* 55 * with AGGRESSIVE_CHECK allocator runs consistency checks over 56 * structures. these checks slow things down a lot 57 */ 58 #define AGGRESSIVE_CHECK__ 59 60 /* 61 * with DOUBLE_CHECK defined mballoc creates persistent in-core 62 * bitmaps, maintains and uses them to check for double allocations 63 */ 64 #define DOUBLE_CHECK__ 65 66 /* 67 * Define EXT4FS_DEBUG to produce debug messages 68 */ 69 #undef EXT4FS_DEBUG 70 71 /* 72 * Debug code 73 */ 74 #ifdef EXT4FS_DEBUG 75 #define ext4_debug(f, a...) \ 76 do { \ 77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 78 __FILE__, __LINE__, __func__); \ 79 printk(KERN_DEBUG f, ## a); \ 80 } while (0) 81 #else 82 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 83 #endif 84 85 /* 86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 87 */ 88 #define EXT_DEBUG__ 89 90 /* 91 * Dynamic printk for controlled extents debugging. 92 */ 93 #ifdef CONFIG_EXT4_DEBUG 94 #define ext_debug(ino, fmt, ...) \ 95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 96 current->comm, task_pid_nr(current), \ 97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 98 __func__, ##__VA_ARGS__) 99 #else 100 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 101 #endif 102 103 #define ASSERT(assert) \ 104 do { \ 105 if (unlikely(!(assert))) { \ 106 printk(KERN_EMERG \ 107 "Assertion failure in %s() at %s:%d: '%s'\n", \ 108 __func__, __FILE__, __LINE__, #assert); \ 109 BUG(); \ 110 } \ 111 } while (0) 112 113 /* data type for block offset of block group */ 114 typedef int ext4_grpblk_t; 115 116 /* data type for filesystem-wide blocks number */ 117 typedef unsigned long long ext4_fsblk_t; 118 119 /* data type for file logical block number */ 120 typedef __u32 ext4_lblk_t; 121 122 /* data type for block group number */ 123 typedef unsigned int ext4_group_t; 124 125 enum SHIFT_DIRECTION { 126 SHIFT_LEFT = 0, 127 SHIFT_RIGHT, 128 }; 129 130 /* 131 * Flags used in mballoc's allocation_context flags field. 132 * 133 * Also used to show what's going on for debugging purposes when the 134 * flag field is exported via the traceport interface 135 */ 136 137 /* prefer goal again. length */ 138 #define EXT4_MB_HINT_MERGE 0x0001 139 /* blocks already reserved */ 140 #define EXT4_MB_HINT_RESERVED 0x0002 141 /* metadata is being allocated */ 142 #define EXT4_MB_HINT_METADATA 0x0004 143 /* first blocks in the file */ 144 #define EXT4_MB_HINT_FIRST 0x0008 145 /* search for the best chunk */ 146 #define EXT4_MB_HINT_BEST 0x0010 147 /* data is being allocated */ 148 #define EXT4_MB_HINT_DATA 0x0020 149 /* don't preallocate (for tails) */ 150 #define EXT4_MB_HINT_NOPREALLOC 0x0040 151 /* allocate for locality group */ 152 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 153 /* allocate goal blocks or none */ 154 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 155 /* goal is meaningful */ 156 #define EXT4_MB_HINT_TRY_GOAL 0x0200 157 /* blocks already pre-reserved by delayed allocation */ 158 #define EXT4_MB_DELALLOC_RESERVED 0x0400 159 /* We are doing stream allocation */ 160 #define EXT4_MB_STREAM_ALLOC 0x0800 161 /* Use reserved root blocks if needed */ 162 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 163 /* Use blocks from reserved pool */ 164 #define EXT4_MB_USE_RESERVED 0x2000 165 /* Do strict check for free blocks while retrying block allocation */ 166 #define EXT4_MB_STRICT_CHECK 0x4000 167 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 168 #define EXT4_MB_CR0_OPTIMIZED 0x8000 169 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 170 #define EXT4_MB_CR1_OPTIMIZED 0x00010000 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 refcount_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 == (1U << 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 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 596 struct compat_ext4_new_group_input { 597 u32 group; 598 compat_u64 block_bitmap; 599 compat_u64 inode_bitmap; 600 compat_u64 inode_table; 601 u32 blocks_count; 602 u16 reserved_blocks; 603 u16 unused; 604 }; 605 #endif 606 607 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 608 struct ext4_new_group_data { 609 __u32 group; 610 __u64 block_bitmap; 611 __u64 inode_bitmap; 612 __u64 inode_table; 613 __u32 blocks_count; 614 __u16 reserved_blocks; 615 __u16 mdata_blocks; 616 __u32 free_clusters_count; 617 }; 618 619 /* Indexes used to index group tables in ext4_new_group_data */ 620 enum { 621 BLOCK_BITMAP = 0, /* block bitmap */ 622 INODE_BITMAP, /* inode bitmap */ 623 INODE_TABLE, /* inode tables */ 624 GROUP_TABLE_COUNT, 625 }; 626 627 /* 628 * Flags used by ext4_map_blocks() 629 */ 630 /* Allocate any needed blocks and/or convert an unwritten 631 extent to be an initialized ext4 */ 632 #define EXT4_GET_BLOCKS_CREATE 0x0001 633 /* Request the creation of an unwritten extent */ 634 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 635 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 636 EXT4_GET_BLOCKS_CREATE) 637 /* Caller is from the delayed allocation writeout path 638 * finally doing the actual allocation of delayed blocks */ 639 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 640 /* caller is from the direct IO path, request to creation of an 641 unwritten extents if not allocated, split the unwritten 642 extent if blocks has been preallocated already*/ 643 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 644 #define EXT4_GET_BLOCKS_CONVERT 0x0010 645 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 646 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 647 /* Convert extent to initialized after IO complete */ 648 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 649 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 650 /* Eventual metadata allocation (due to growing extent tree) 651 * should not fail, so try to use reserved blocks for that.*/ 652 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 653 /* Don't normalize allocation size (used for fallocate) */ 654 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 655 /* Convert written extents to unwritten */ 656 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 657 /* Write zeros to newly created written extents */ 658 #define EXT4_GET_BLOCKS_ZERO 0x0200 659 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 660 EXT4_GET_BLOCKS_ZERO) 661 /* Caller will submit data before dropping transaction handle. This 662 * allows jbd2 to avoid submitting data before commit. */ 663 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 664 /* Caller is in the atomic contex, find extent if it has been cached */ 665 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 666 667 /* 668 * The bit position of these flags must not overlap with any of the 669 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 670 * read_extent_tree_block(), ext4_split_extent_at(), 671 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 672 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 673 * caching the extents when reading from the extent tree while a 674 * truncate or punch hole operation is in progress. 675 */ 676 #define EXT4_EX_NOCACHE 0x40000000 677 #define EXT4_EX_FORCE_CACHE 0x20000000 678 #define EXT4_EX_NOFAIL 0x10000000 679 680 /* 681 * Flags used by ext4_free_blocks 682 */ 683 #define EXT4_FREE_BLOCKS_METADATA 0x0001 684 #define EXT4_FREE_BLOCKS_FORGET 0x0002 685 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 686 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 687 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 688 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 689 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 690 691 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 692 /* 693 * ioctl commands in 32 bit emulation 694 */ 695 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 696 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 697 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 698 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 699 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 700 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 701 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 702 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 703 #endif 704 705 /* Max physical block we can address w/o extents */ 706 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 707 708 /* Max logical block we can support */ 709 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 710 711 /* 712 * Structure of an inode on the disk 713 */ 714 struct ext4_inode { 715 __le16 i_mode; /* File mode */ 716 __le16 i_uid; /* Low 16 bits of Owner Uid */ 717 __le32 i_size_lo; /* Size in bytes */ 718 __le32 i_atime; /* Access time */ 719 __le32 i_ctime; /* Inode Change time */ 720 __le32 i_mtime; /* Modification time */ 721 __le32 i_dtime; /* Deletion Time */ 722 __le16 i_gid; /* Low 16 bits of Group Id */ 723 __le16 i_links_count; /* Links count */ 724 __le32 i_blocks_lo; /* Blocks count */ 725 __le32 i_flags; /* File flags */ 726 union { 727 struct { 728 __le32 l_i_version; 729 } linux1; 730 struct { 731 __u32 h_i_translator; 732 } hurd1; 733 struct { 734 __u32 m_i_reserved1; 735 } masix1; 736 } osd1; /* OS dependent 1 */ 737 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 738 __le32 i_generation; /* File version (for NFS) */ 739 __le32 i_file_acl_lo; /* File ACL */ 740 __le32 i_size_high; 741 __le32 i_obso_faddr; /* Obsoleted fragment address */ 742 union { 743 struct { 744 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 745 __le16 l_i_file_acl_high; 746 __le16 l_i_uid_high; /* these 2 fields */ 747 __le16 l_i_gid_high; /* were reserved2[0] */ 748 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 749 __le16 l_i_reserved; 750 } linux2; 751 struct { 752 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 753 __u16 h_i_mode_high; 754 __u16 h_i_uid_high; 755 __u16 h_i_gid_high; 756 __u32 h_i_author; 757 } hurd2; 758 struct { 759 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 760 __le16 m_i_file_acl_high; 761 __u32 m_i_reserved2[2]; 762 } masix2; 763 } osd2; /* OS dependent 2 */ 764 __le16 i_extra_isize; 765 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 766 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 767 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 768 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 769 __le32 i_crtime; /* File Creation time */ 770 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 771 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 772 __le32 i_projid; /* Project ID */ 773 }; 774 775 #define EXT4_EPOCH_BITS 2 776 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 777 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 778 779 /* 780 * Extended fields will fit into an inode if the filesystem was formatted 781 * with large inodes (-I 256 or larger) and there are not currently any EAs 782 * consuming all of the available space. For new inodes we always reserve 783 * enough space for the kernel's known extended fields, but for inodes 784 * created with an old kernel this might not have been the case. None of 785 * the extended inode fields is critical for correct filesystem operation. 786 * This macro checks if a certain field fits in the inode. Note that 787 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 788 */ 789 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 790 ((offsetof(typeof(*ext4_inode), field) + \ 791 sizeof((ext4_inode)->field)) \ 792 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 793 (einode)->i_extra_isize)) \ 794 795 /* 796 * We use an encoding that preserves the times for extra epoch "00": 797 * 798 * extra msb of adjust for signed 799 * epoch 32-bit 32-bit tv_sec to 800 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 801 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 802 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 803 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 804 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 805 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 806 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 807 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 808 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 809 * 810 * Note that previous versions of the kernel on 64-bit systems would 811 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 812 * 1970. e2fsck will correct this, assuming that it is run on the 813 * affected filesystem before 2242. 814 */ 815 816 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 817 { 818 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 819 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 820 } 821 822 static inline void ext4_decode_extra_time(struct timespec64 *time, 823 __le32 extra) 824 { 825 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 826 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 827 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 828 } 829 830 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 831 do { \ 832 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 833 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 834 (raw_inode)->xtime ## _extra = \ 835 ext4_encode_extra_time(&(inode)->xtime); \ 836 } \ 837 else \ 838 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 839 } while (0) 840 841 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 842 do { \ 843 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 844 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 845 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 846 (raw_inode)->xtime ## _extra = \ 847 ext4_encode_extra_time(&(einode)->xtime); \ 848 } while (0) 849 850 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 851 do { \ 852 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 853 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 854 ext4_decode_extra_time(&(inode)->xtime, \ 855 raw_inode->xtime ## _extra); \ 856 } \ 857 else \ 858 (inode)->xtime.tv_nsec = 0; \ 859 } while (0) 860 861 862 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 863 do { \ 864 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 865 (einode)->xtime.tv_sec = \ 866 (signed)le32_to_cpu((raw_inode)->xtime); \ 867 else \ 868 (einode)->xtime.tv_sec = 0; \ 869 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 870 ext4_decode_extra_time(&(einode)->xtime, \ 871 raw_inode->xtime ## _extra); \ 872 else \ 873 (einode)->xtime.tv_nsec = 0; \ 874 } while (0) 875 876 #define i_disk_version osd1.linux1.l_i_version 877 878 #if defined(__KERNEL__) || defined(__linux__) 879 #define i_reserved1 osd1.linux1.l_i_reserved1 880 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 881 #define i_blocks_high osd2.linux2.l_i_blocks_high 882 #define i_uid_low i_uid 883 #define i_gid_low i_gid 884 #define i_uid_high osd2.linux2.l_i_uid_high 885 #define i_gid_high osd2.linux2.l_i_gid_high 886 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 887 888 #elif defined(__GNU__) 889 890 #define i_translator osd1.hurd1.h_i_translator 891 #define i_uid_high osd2.hurd2.h_i_uid_high 892 #define i_gid_high osd2.hurd2.h_i_gid_high 893 #define i_author osd2.hurd2.h_i_author 894 895 #elif defined(__masix__) 896 897 #define i_reserved1 osd1.masix1.m_i_reserved1 898 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 899 #define i_reserved2 osd2.masix2.m_i_reserved2 900 901 #endif /* defined(__KERNEL__) || defined(__linux__) */ 902 903 #include "extents_status.h" 904 #include "fast_commit.h" 905 906 /* 907 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 908 * 909 * These are needed to avoid lockdep false positives when we need to 910 * allocate blocks to the quota inode during ext4_map_blocks(), while 911 * holding i_data_sem for a normal (non-quota) inode. Since we don't 912 * do quota tracking for the quota inode, this avoids deadlock (as 913 * well as infinite recursion, since it isn't turtles all the way 914 * down...) 915 * 916 * I_DATA_SEM_NORMAL - Used for most inodes 917 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 918 * where the second inode has larger inode number 919 * than the first 920 * I_DATA_SEM_QUOTA - Used for quota inodes only 921 */ 922 enum { 923 I_DATA_SEM_NORMAL = 0, 924 I_DATA_SEM_OTHER, 925 I_DATA_SEM_QUOTA, 926 }; 927 928 929 /* 930 * fourth extended file system inode data in memory 931 */ 932 struct ext4_inode_info { 933 __le32 i_data[15]; /* unconverted */ 934 __u32 i_dtime; 935 ext4_fsblk_t i_file_acl; 936 937 /* 938 * i_block_group is the number of the block group which contains 939 * this file's inode. Constant across the lifetime of the inode, 940 * it is used for making block allocation decisions - we try to 941 * place a file's data blocks near its inode block, and new inodes 942 * near to their parent directory's inode. 943 */ 944 ext4_group_t i_block_group; 945 ext4_lblk_t i_dir_start_lookup; 946 #if (BITS_PER_LONG < 64) 947 unsigned long i_state_flags; /* Dynamic state flags */ 948 #endif 949 unsigned long i_flags; 950 951 /* 952 * Extended attributes can be read independently of the main file 953 * data. Taking i_rwsem even when reading would cause contention 954 * between readers of EAs and writers of regular file data, so 955 * instead we synchronize on xattr_sem when reading or changing 956 * EAs. 957 */ 958 struct rw_semaphore xattr_sem; 959 960 /* 961 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 962 * i_orphan is used. 963 */ 964 union { 965 struct list_head i_orphan; /* unlinked but open inodes */ 966 unsigned int i_orphan_idx; /* Index in orphan file */ 967 }; 968 969 /* Fast commit related info */ 970 971 /* For tracking dentry create updates */ 972 struct list_head i_fc_dilist; 973 struct list_head i_fc_list; /* 974 * inodes that need fast commit 975 * protected by sbi->s_fc_lock. 976 */ 977 978 /* Start of lblk range that needs to be committed in this fast commit */ 979 ext4_lblk_t i_fc_lblk_start; 980 981 /* End of lblk range that needs to be committed in this fast commit */ 982 ext4_lblk_t i_fc_lblk_len; 983 984 /* Number of ongoing updates on this inode */ 985 atomic_t i_fc_updates; 986 987 /* Fast commit wait queue for this inode */ 988 wait_queue_head_t i_fc_wait; 989 990 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 991 struct mutex i_fc_lock; 992 993 /* 994 * i_disksize keeps track of what the inode size is ON DISK, not 995 * in memory. During truncate, i_size is set to the new size by 996 * the VFS prior to calling ext4_truncate(), but the filesystem won't 997 * set i_disksize to 0 until the truncate is actually under way. 998 * 999 * The intent is that i_disksize always represents the blocks which 1000 * are used by this file. This allows recovery to restart truncate 1001 * on orphans if we crash during truncate. We actually write i_disksize 1002 * into the on-disk inode when writing inodes out, instead of i_size. 1003 * 1004 * The only time when i_disksize and i_size may be different is when 1005 * a truncate is in progress. The only things which change i_disksize 1006 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1007 */ 1008 loff_t i_disksize; 1009 1010 /* 1011 * i_data_sem is for serialising ext4_truncate() against 1012 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1013 * data tree are chopped off during truncate. We can't do that in 1014 * ext4 because whenever we perform intermediate commits during 1015 * truncate, the inode and all the metadata blocks *must* be in a 1016 * consistent state which allows truncation of the orphans to restart 1017 * during recovery. Hence we must fix the get_block-vs-truncate race 1018 * by other means, so we have i_data_sem. 1019 */ 1020 struct rw_semaphore i_data_sem; 1021 struct inode vfs_inode; 1022 struct jbd2_inode *jinode; 1023 1024 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1025 1026 /* 1027 * File creation time. Its function is same as that of 1028 * struct timespec64 i_{a,c,m}time in the generic inode. 1029 */ 1030 struct timespec64 i_crtime; 1031 1032 /* mballoc */ 1033 atomic_t i_prealloc_active; 1034 struct rb_root i_prealloc_node; 1035 rwlock_t i_prealloc_lock; 1036 1037 /* extents status tree */ 1038 struct ext4_es_tree i_es_tree; 1039 rwlock_t i_es_lock; 1040 struct list_head i_es_list; 1041 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1042 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1043 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1044 extents to shrink. Protected by 1045 i_es_lock */ 1046 1047 /* ialloc */ 1048 ext4_group_t i_last_alloc_group; 1049 1050 /* allocation reservation info for delalloc */ 1051 /* In case of bigalloc, this refer to clusters rather than blocks */ 1052 unsigned int i_reserved_data_blocks; 1053 1054 /* pending cluster reservations for bigalloc file systems */ 1055 struct ext4_pending_tree i_pending_tree; 1056 1057 /* on-disk additional length */ 1058 __u16 i_extra_isize; 1059 1060 /* Indicate the inline data space. */ 1061 u16 i_inline_off; 1062 u16 i_inline_size; 1063 1064 #ifdef CONFIG_QUOTA 1065 /* quota space reservation, managed internally by quota code */ 1066 qsize_t i_reserved_quota; 1067 #endif 1068 1069 /* Lock protecting lists below */ 1070 spinlock_t i_completed_io_lock; 1071 /* 1072 * Completed IOs that need unwritten extents handling and have 1073 * transaction reserved 1074 */ 1075 struct list_head i_rsv_conversion_list; 1076 struct work_struct i_rsv_conversion_work; 1077 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1078 1079 spinlock_t i_block_reservation_lock; 1080 1081 /* 1082 * Transactions that contain inode's metadata needed to complete 1083 * fsync and fdatasync, respectively. 1084 */ 1085 tid_t i_sync_tid; 1086 tid_t i_datasync_tid; 1087 1088 #ifdef CONFIG_QUOTA 1089 struct dquot *i_dquot[MAXQUOTAS]; 1090 #endif 1091 1092 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1093 __u32 i_csum_seed; 1094 1095 kprojid_t i_projid; 1096 }; 1097 1098 /* 1099 * File system states 1100 */ 1101 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1102 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1103 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1104 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1105 1106 /* 1107 * Misc. filesystem flags 1108 */ 1109 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1110 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1111 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1112 1113 /* 1114 * Mount flags set via mount options or defaults 1115 */ 1116 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1117 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1118 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1119 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1120 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1121 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1122 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1123 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1124 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1125 #ifdef CONFIG_FS_DAX 1126 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1127 #else 1128 #define EXT4_MOUNT_DAX_ALWAYS 0 1129 #endif 1130 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1131 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1132 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1133 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1134 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1135 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1136 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1137 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1138 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1139 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1140 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1141 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1142 * enable enforcement for hidden 1143 * quota files */ 1144 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1145 * enforcement for hidden quota 1146 * files */ 1147 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1148 * enforcement */ 1149 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1150 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1151 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1152 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1153 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1154 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1155 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1156 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1157 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1158 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1159 1160 /* 1161 * Mount flags set either automatically (could not be set by mount option) 1162 * based on per file system feature or property or in special cases such as 1163 * distinguishing between explicit mount option definition and default. 1164 */ 1165 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1166 specified delalloc */ 1167 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1168 size of blocksize * 8 1169 blocks */ 1170 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1171 file systems */ 1172 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1173 specified journal checksum */ 1174 1175 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1176 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1177 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1178 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1179 * scanning in mballoc 1180 */ 1181 1182 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1183 ~EXT4_MOUNT_##opt 1184 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1185 EXT4_MOUNT_##opt 1186 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1187 EXT4_MOUNT_##opt) 1188 1189 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1190 ~EXT4_MOUNT2_##opt 1191 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1192 EXT4_MOUNT2_##opt 1193 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1194 EXT4_MOUNT2_##opt) 1195 1196 #define ext4_test_and_set_bit __test_and_set_bit_le 1197 #define ext4_set_bit __set_bit_le 1198 #define ext4_set_bit_atomic ext2_set_bit_atomic 1199 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1200 #define ext4_clear_bit __clear_bit_le 1201 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1202 #define ext4_test_bit test_bit_le 1203 #define ext4_find_next_zero_bit find_next_zero_bit_le 1204 #define ext4_find_next_bit find_next_bit_le 1205 1206 extern void mb_set_bits(void *bm, int cur, int len); 1207 1208 /* 1209 * Maximal mount counts between two filesystem checks 1210 */ 1211 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1212 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1213 1214 /* 1215 * Behaviour when detecting errors 1216 */ 1217 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1218 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1219 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1220 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1221 1222 /* Metadata checksum algorithm codes */ 1223 #define EXT4_CRC32C_CHKSUM 1 1224 1225 #define EXT4_LABEL_MAX 16 1226 1227 /* 1228 * Structure of the super block 1229 */ 1230 struct ext4_super_block { 1231 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1232 __le32 s_blocks_count_lo; /* Blocks count */ 1233 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1234 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1235 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1236 __le32 s_first_data_block; /* First Data Block */ 1237 __le32 s_log_block_size; /* Block size */ 1238 __le32 s_log_cluster_size; /* Allocation cluster size */ 1239 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1240 __le32 s_clusters_per_group; /* # Clusters per group */ 1241 __le32 s_inodes_per_group; /* # Inodes per group */ 1242 __le32 s_mtime; /* Mount time */ 1243 /*30*/ __le32 s_wtime; /* Write time */ 1244 __le16 s_mnt_count; /* Mount count */ 1245 __le16 s_max_mnt_count; /* Maximal mount count */ 1246 __le16 s_magic; /* Magic signature */ 1247 __le16 s_state; /* File system state */ 1248 __le16 s_errors; /* Behaviour when detecting errors */ 1249 __le16 s_minor_rev_level; /* minor revision level */ 1250 /*40*/ __le32 s_lastcheck; /* time of last check */ 1251 __le32 s_checkinterval; /* max. time between checks */ 1252 __le32 s_creator_os; /* OS */ 1253 __le32 s_rev_level; /* Revision level */ 1254 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1255 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1256 /* 1257 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1258 * 1259 * Note: the difference between the compatible feature set and 1260 * the incompatible feature set is that if there is a bit set 1261 * in the incompatible feature set that the kernel doesn't 1262 * know about, it should refuse to mount the filesystem. 1263 * 1264 * e2fsck's requirements are more strict; if it doesn't know 1265 * about a feature in either the compatible or incompatible 1266 * feature set, it must abort and not try to meddle with 1267 * things it doesn't understand... 1268 */ 1269 __le32 s_first_ino; /* First non-reserved inode */ 1270 __le16 s_inode_size; /* size of inode structure */ 1271 __le16 s_block_group_nr; /* block group # of this superblock */ 1272 __le32 s_feature_compat; /* compatible feature set */ 1273 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1274 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1275 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1276 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1277 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1278 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1279 /* 1280 * Performance hints. Directory preallocation should only 1281 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1282 */ 1283 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1284 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1285 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1286 /* 1287 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1288 */ 1289 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1290 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1291 __le32 s_journal_dev; /* device number of journal file */ 1292 __le32 s_last_orphan; /* start of list of inodes to delete */ 1293 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1294 __u8 s_def_hash_version; /* Default hash version to use */ 1295 __u8 s_jnl_backup_type; 1296 __le16 s_desc_size; /* size of group descriptor */ 1297 /*100*/ __le32 s_default_mount_opts; 1298 __le32 s_first_meta_bg; /* First metablock block group */ 1299 __le32 s_mkfs_time; /* When the filesystem was created */ 1300 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1301 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1302 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1303 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1304 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1305 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1306 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1307 __le32 s_flags; /* Miscellaneous flags */ 1308 __le16 s_raid_stride; /* RAID stride */ 1309 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1310 __le64 s_mmp_block; /* Block for multi-mount protection */ 1311 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1312 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1313 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1314 __u8 s_encryption_level; /* versioning level for encryption */ 1315 __u8 s_reserved_pad; /* Padding to next 32bits */ 1316 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1317 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1318 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1319 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1320 snapshot's future use */ 1321 __le32 s_snapshot_list; /* inode number of the head of the 1322 on-disk snapshot list */ 1323 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1324 __le32 s_error_count; /* number of fs errors */ 1325 __le32 s_first_error_time; /* first time an error happened */ 1326 __le32 s_first_error_ino; /* inode involved in first error */ 1327 __le64 s_first_error_block; /* block involved of first error */ 1328 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1329 __le32 s_first_error_line; /* line number where error happened */ 1330 __le32 s_last_error_time; /* most recent time of an error */ 1331 __le32 s_last_error_ino; /* inode involved in last error */ 1332 __le32 s_last_error_line; /* line number where error happened */ 1333 __le64 s_last_error_block; /* block involved of last error */ 1334 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1335 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1336 __u8 s_mount_opts[64]; 1337 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1338 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1339 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1340 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1341 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1342 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1343 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1344 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1345 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1346 __u8 s_wtime_hi; 1347 __u8 s_mtime_hi; 1348 __u8 s_mkfs_time_hi; 1349 __u8 s_lastcheck_hi; 1350 __u8 s_first_error_time_hi; 1351 __u8 s_last_error_time_hi; 1352 __u8 s_first_error_errcode; 1353 __u8 s_last_error_errcode; 1354 __le16 s_encoding; /* Filename charset encoding */ 1355 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1356 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1357 __le32 s_reserved[94]; /* Padding to the end of the block */ 1358 __le32 s_checksum; /* crc32c(superblock) */ 1359 }; 1360 1361 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1362 1363 #ifdef __KERNEL__ 1364 1365 /* Number of quota types we support */ 1366 #define EXT4_MAXQUOTAS 3 1367 1368 #define EXT4_ENC_UTF8_12_1 1 1369 1370 /* Types of ext4 journal triggers */ 1371 enum ext4_journal_trigger_type { 1372 EXT4_JTR_ORPHAN_FILE, 1373 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1374 }; 1375 1376 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1377 1378 struct ext4_journal_trigger { 1379 struct jbd2_buffer_trigger_type tr_triggers; 1380 struct super_block *sb; 1381 }; 1382 1383 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1384 struct jbd2_buffer_trigger_type *trigger) 1385 { 1386 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1387 } 1388 1389 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1390 1391 /* Structure at the tail of orphan block */ 1392 struct ext4_orphan_block_tail { 1393 __le32 ob_magic; 1394 __le32 ob_checksum; 1395 }; 1396 1397 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1398 { 1399 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1400 sizeof(u32); 1401 } 1402 1403 struct ext4_orphan_block { 1404 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1405 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1406 }; 1407 1408 /* 1409 * Info about orphan file. 1410 */ 1411 struct ext4_orphan_info { 1412 int of_blocks; /* Number of orphan blocks in a file */ 1413 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1414 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1415 * file blocks */ 1416 }; 1417 1418 /* 1419 * fourth extended-fs super-block data in memory 1420 */ 1421 struct ext4_sb_info { 1422 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1423 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1424 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1425 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1426 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1427 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1428 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1429 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1430 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1431 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1432 unsigned long s_overhead; /* # of fs overhead clusters */ 1433 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1434 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1435 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1436 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1437 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1438 struct buffer_head * __rcu *s_group_desc; 1439 unsigned int s_mount_opt; 1440 unsigned int s_mount_opt2; 1441 unsigned long s_mount_flags; 1442 unsigned int s_def_mount_opt; 1443 unsigned int s_def_mount_opt2; 1444 ext4_fsblk_t s_sb_block; 1445 atomic64_t s_resv_clusters; 1446 kuid_t s_resuid; 1447 kgid_t s_resgid; 1448 unsigned short s_mount_state; 1449 unsigned short s_pad; 1450 int s_addr_per_block_bits; 1451 int s_desc_per_block_bits; 1452 int s_inode_size; 1453 int s_first_ino; 1454 unsigned int s_inode_readahead_blks; 1455 unsigned int s_inode_goal; 1456 u32 s_hash_seed[4]; 1457 int s_def_hash_version; 1458 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1459 struct percpu_counter s_freeclusters_counter; 1460 struct percpu_counter s_freeinodes_counter; 1461 struct percpu_counter s_dirs_counter; 1462 struct percpu_counter s_dirtyclusters_counter; 1463 struct percpu_counter s_sra_exceeded_retry_limit; 1464 struct blockgroup_lock *s_blockgroup_lock; 1465 struct proc_dir_entry *s_proc; 1466 struct kobject s_kobj; 1467 struct completion s_kobj_unregister; 1468 struct super_block *s_sb; 1469 struct buffer_head *s_mmp_bh; 1470 1471 /* Journaling */ 1472 struct journal_s *s_journal; 1473 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1474 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1475 struct list_head s_orphan; /* List of orphaned inodes in on disk 1476 list */ 1477 struct ext4_orphan_info s_orphan_info; 1478 unsigned long s_commit_interval; 1479 u32 s_max_batch_time; 1480 u32 s_min_batch_time; 1481 struct block_device *s_journal_bdev; 1482 #ifdef CONFIG_QUOTA 1483 /* Names of quota files with journalled quota */ 1484 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1485 int s_jquota_fmt; /* Format of quota to use */ 1486 #endif 1487 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1488 struct ext4_system_blocks __rcu *s_system_blks; 1489 1490 #ifdef EXTENTS_STATS 1491 /* ext4 extents stats */ 1492 unsigned long s_ext_min; 1493 unsigned long s_ext_max; 1494 unsigned long s_depth_max; 1495 spinlock_t s_ext_stats_lock; 1496 unsigned long s_ext_blocks; 1497 unsigned long s_ext_extents; 1498 #endif 1499 1500 /* for buddy allocator */ 1501 struct ext4_group_info ** __rcu *s_group_info; 1502 struct inode *s_buddy_cache; 1503 spinlock_t s_md_lock; 1504 unsigned short *s_mb_offsets; 1505 unsigned int *s_mb_maxs; 1506 unsigned int s_group_info_size; 1507 unsigned int s_mb_free_pending; 1508 struct list_head s_freed_data_list; /* List of blocks to be freed 1509 after commit completed */ 1510 struct list_head s_discard_list; 1511 struct work_struct s_discard_work; 1512 atomic_t s_retry_alloc_pending; 1513 struct list_head *s_mb_avg_fragment_size; 1514 rwlock_t *s_mb_avg_fragment_size_locks; 1515 struct list_head *s_mb_largest_free_orders; 1516 rwlock_t *s_mb_largest_free_orders_locks; 1517 1518 /* tunables */ 1519 unsigned long s_stripe; 1520 unsigned int s_mb_max_linear_groups; 1521 unsigned int s_mb_stream_request; 1522 unsigned int s_mb_max_to_scan; 1523 unsigned int s_mb_min_to_scan; 1524 unsigned int s_mb_stats; 1525 unsigned int s_mb_order2_reqs; 1526 unsigned int s_mb_group_prealloc; 1527 unsigned int s_max_dir_size_kb; 1528 /* where last allocation was done - for stream allocation */ 1529 unsigned long s_mb_last_group; 1530 unsigned long s_mb_last_start; 1531 unsigned int s_mb_prefetch; 1532 unsigned int s_mb_prefetch_limit; 1533 1534 /* stats for buddy allocator */ 1535 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1536 atomic_t s_bal_success; /* we found long enough chunks */ 1537 atomic_t s_bal_allocated; /* in blocks */ 1538 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1539 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1540 atomic_t s_bal_goals; /* goal hits */ 1541 atomic_t s_bal_breaks; /* too long searches */ 1542 atomic_t s_bal_2orders; /* 2^order hits */ 1543 atomic_t s_bal_cr0_bad_suggestions; 1544 atomic_t s_bal_cr1_bad_suggestions; 1545 atomic64_t s_bal_cX_groups_considered[4]; 1546 atomic64_t s_bal_cX_hits[4]; 1547 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */ 1548 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1549 atomic64_t s_mb_generation_time; 1550 atomic_t s_mb_lost_chunks; 1551 atomic_t s_mb_preallocated; 1552 atomic_t s_mb_discarded; 1553 atomic_t s_lock_busy; 1554 1555 /* locality groups */ 1556 struct ext4_locality_group __percpu *s_locality_groups; 1557 1558 /* for write statistics */ 1559 unsigned long s_sectors_written_start; 1560 u64 s_kbytes_written; 1561 1562 /* the size of zero-out chunk */ 1563 unsigned int s_extent_max_zeroout_kb; 1564 1565 unsigned int s_log_groups_per_flex; 1566 struct flex_groups * __rcu *s_flex_groups; 1567 ext4_group_t s_flex_groups_allocated; 1568 1569 /* workqueue for reserved extent conversions (buffered io) */ 1570 struct workqueue_struct *rsv_conversion_wq; 1571 1572 /* timer for periodic error stats printing */ 1573 struct timer_list s_err_report; 1574 1575 /* Lazy inode table initialization info */ 1576 struct ext4_li_request *s_li_request; 1577 /* Wait multiplier for lazy initialization thread */ 1578 unsigned int s_li_wait_mult; 1579 1580 /* Kernel thread for multiple mount protection */ 1581 struct task_struct *s_mmp_tsk; 1582 1583 /* record the last minlen when FITRIM is called. */ 1584 unsigned long s_last_trim_minblks; 1585 1586 /* Reference to checksum algorithm driver via cryptoapi */ 1587 struct crypto_shash *s_chksum_driver; 1588 1589 /* Precomputed FS UUID checksum for seeding other checksums */ 1590 __u32 s_csum_seed; 1591 1592 /* Reclaim extents from extent status tree */ 1593 struct shrinker s_es_shrinker; 1594 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1595 long s_es_nr_inode; 1596 struct ext4_es_stats s_es_stats; 1597 struct mb_cache *s_ea_block_cache; 1598 struct mb_cache *s_ea_inode_cache; 1599 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1600 1601 /* Journal triggers for checksum computation */ 1602 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1603 1604 /* Ratelimit ext4 messages. */ 1605 struct ratelimit_state s_err_ratelimit_state; 1606 struct ratelimit_state s_warning_ratelimit_state; 1607 struct ratelimit_state s_msg_ratelimit_state; 1608 atomic_t s_warning_count; 1609 atomic_t s_msg_count; 1610 1611 /* Encryption policy for '-o test_dummy_encryption' */ 1612 struct fscrypt_dummy_policy s_dummy_enc_policy; 1613 1614 /* 1615 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1616 * or EXTENTS flag. 1617 */ 1618 struct percpu_rw_semaphore s_writepages_rwsem; 1619 struct dax_device *s_daxdev; 1620 u64 s_dax_part_off; 1621 #ifdef CONFIG_EXT4_DEBUG 1622 unsigned long s_simulate_fail; 1623 #endif 1624 /* Record the errseq of the backing block device */ 1625 errseq_t s_bdev_wb_err; 1626 spinlock_t s_bdev_wb_lock; 1627 1628 /* Information about errors that happened during this mount */ 1629 spinlock_t s_error_lock; 1630 int s_add_error_count; 1631 int s_first_error_code; 1632 __u32 s_first_error_line; 1633 __u32 s_first_error_ino; 1634 __u64 s_first_error_block; 1635 const char *s_first_error_func; 1636 time64_t s_first_error_time; 1637 int s_last_error_code; 1638 __u32 s_last_error_line; 1639 __u32 s_last_error_ino; 1640 __u64 s_last_error_block; 1641 const char *s_last_error_func; 1642 time64_t s_last_error_time; 1643 /* 1644 * If we are in a context where we cannot update error information in 1645 * the on-disk superblock, we queue this work to do it. 1646 */ 1647 struct work_struct s_error_work; 1648 1649 /* Ext4 fast commit sub transaction ID */ 1650 atomic_t s_fc_subtid; 1651 1652 /* 1653 * After commit starts, the main queue gets locked, and the further 1654 * updates get added in the staging queue. 1655 */ 1656 #define FC_Q_MAIN 0 1657 #define FC_Q_STAGING 1 1658 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1659 * that have data changes in them. 1660 */ 1661 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1662 unsigned int s_fc_bytes; 1663 /* 1664 * Main fast commit lock. This lock protects accesses to the 1665 * following fields: 1666 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1667 */ 1668 spinlock_t s_fc_lock; 1669 struct buffer_head *s_fc_bh; 1670 struct ext4_fc_stats s_fc_stats; 1671 tid_t s_fc_ineligible_tid; 1672 #ifdef CONFIG_EXT4_DEBUG 1673 int s_fc_debug_max_replay; 1674 #endif 1675 struct ext4_fc_replay_state s_fc_replay_state; 1676 }; 1677 1678 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1679 { 1680 return sb->s_fs_info; 1681 } 1682 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1683 { 1684 return container_of(inode, struct ext4_inode_info, vfs_inode); 1685 } 1686 1687 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1688 { 1689 return ino == EXT4_ROOT_INO || 1690 (ino >= EXT4_FIRST_INO(sb) && 1691 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1692 } 1693 1694 /* 1695 * Returns: sbi->field[index] 1696 * Used to access an array element from the following sbi fields which require 1697 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1698 * - s_group_desc 1699 * - s_group_info 1700 * - s_flex_group 1701 */ 1702 #define sbi_array_rcu_deref(sbi, field, index) \ 1703 ({ \ 1704 typeof(*((sbi)->field)) _v; \ 1705 rcu_read_lock(); \ 1706 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1707 rcu_read_unlock(); \ 1708 _v; \ 1709 }) 1710 1711 /* 1712 * run-time mount flags 1713 */ 1714 enum { 1715 EXT4_MF_MNTDIR_SAMPLED, 1716 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1717 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1718 }; 1719 1720 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1721 { 1722 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1723 } 1724 1725 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1726 { 1727 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1728 } 1729 1730 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1731 { 1732 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1733 } 1734 1735 1736 /* 1737 * Simulate_fail codes 1738 */ 1739 #define EXT4_SIM_BBITMAP_EIO 1 1740 #define EXT4_SIM_BBITMAP_CRC 2 1741 #define EXT4_SIM_IBITMAP_EIO 3 1742 #define EXT4_SIM_IBITMAP_CRC 4 1743 #define EXT4_SIM_INODE_EIO 5 1744 #define EXT4_SIM_INODE_CRC 6 1745 #define EXT4_SIM_DIRBLOCK_EIO 7 1746 #define EXT4_SIM_DIRBLOCK_CRC 8 1747 1748 static inline bool ext4_simulate_fail(struct super_block *sb, 1749 unsigned long code) 1750 { 1751 #ifdef CONFIG_EXT4_DEBUG 1752 struct ext4_sb_info *sbi = EXT4_SB(sb); 1753 1754 if (unlikely(sbi->s_simulate_fail == code)) { 1755 sbi->s_simulate_fail = 0; 1756 return true; 1757 } 1758 #endif 1759 return false; 1760 } 1761 1762 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1763 struct buffer_head *bh, 1764 unsigned long code) 1765 { 1766 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1767 clear_buffer_uptodate(bh); 1768 } 1769 1770 /* 1771 * Error number codes for s_{first,last}_error_errno 1772 * 1773 * Linux errno numbers are architecture specific, so we need to translate 1774 * them into something which is architecture independent. We don't define 1775 * codes for all errno's; just the ones which are most likely to be the cause 1776 * of an ext4_error() call. 1777 */ 1778 #define EXT4_ERR_UNKNOWN 1 1779 #define EXT4_ERR_EIO 2 1780 #define EXT4_ERR_ENOMEM 3 1781 #define EXT4_ERR_EFSBADCRC 4 1782 #define EXT4_ERR_EFSCORRUPTED 5 1783 #define EXT4_ERR_ENOSPC 6 1784 #define EXT4_ERR_ENOKEY 7 1785 #define EXT4_ERR_EROFS 8 1786 #define EXT4_ERR_EFBIG 9 1787 #define EXT4_ERR_EEXIST 10 1788 #define EXT4_ERR_ERANGE 11 1789 #define EXT4_ERR_EOVERFLOW 12 1790 #define EXT4_ERR_EBUSY 13 1791 #define EXT4_ERR_ENOTDIR 14 1792 #define EXT4_ERR_ENOTEMPTY 15 1793 #define EXT4_ERR_ESHUTDOWN 16 1794 #define EXT4_ERR_EFAULT 17 1795 1796 /* 1797 * Inode dynamic state flags 1798 */ 1799 enum { 1800 EXT4_STATE_NEW, /* inode is newly created */ 1801 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1802 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1803 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1804 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1805 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1806 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1807 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1808 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1809 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1810 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1811 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1812 }; 1813 1814 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1815 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1816 { \ 1817 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1818 } \ 1819 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1820 { \ 1821 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1822 } \ 1823 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1824 { \ 1825 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1826 } 1827 1828 /* Add these declarations here only so that these functions can be 1829 * found by name. Otherwise, they are very hard to locate. */ 1830 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1831 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1832 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1833 EXT4_INODE_BIT_FNS(flag, flags, 0) 1834 1835 /* Add these declarations here only so that these functions can be 1836 * found by name. Otherwise, they are very hard to locate. */ 1837 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1838 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1839 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1840 #if (BITS_PER_LONG < 64) 1841 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1842 1843 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1844 { 1845 (ei)->i_state_flags = 0; 1846 } 1847 #else 1848 EXT4_INODE_BIT_FNS(state, flags, 32) 1849 1850 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1851 { 1852 /* We depend on the fact that callers will set i_flags */ 1853 } 1854 #endif 1855 #else 1856 /* Assume that user mode programs are passing in an ext4fs superblock, not 1857 * a kernel struct super_block. This will allow us to call the feature-test 1858 * macros from user land. */ 1859 #define EXT4_SB(sb) (sb) 1860 #endif 1861 1862 static inline bool ext4_verity_in_progress(struct inode *inode) 1863 { 1864 return IS_ENABLED(CONFIG_FS_VERITY) && 1865 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1866 } 1867 1868 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1869 1870 /* 1871 * Codes for operating systems 1872 */ 1873 #define EXT4_OS_LINUX 0 1874 #define EXT4_OS_HURD 1 1875 #define EXT4_OS_MASIX 2 1876 #define EXT4_OS_FREEBSD 3 1877 #define EXT4_OS_LITES 4 1878 1879 /* 1880 * Revision levels 1881 */ 1882 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1883 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1884 1885 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1886 1887 #define EXT4_GOOD_OLD_INODE_SIZE 128 1888 1889 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1890 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1891 #define EXT4_TIMESTAMP_MIN S32_MIN 1892 1893 /* 1894 * Feature set definitions 1895 */ 1896 1897 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1898 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1899 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1900 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1901 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1902 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1903 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1904 /* 1905 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1906 * incompatible only if fast commit blocks are present in the FS. Since we 1907 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1908 * incompatible. We also have a JBD2 incompat feature, which gets set when 1909 * there are fast commit blocks present in the journal. 1910 */ 1911 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1912 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1913 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 1914 1915 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1916 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1917 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1918 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1919 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1920 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1921 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1922 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1923 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1924 /* 1925 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1926 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1927 * all other data structures' checksums. However, the METADATA_CSUM and 1928 * GDT_CSUM bits are mutually exclusive. 1929 */ 1930 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1931 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1932 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1933 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1934 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 1935 non-empty */ 1936 1937 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1938 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1939 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1940 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1941 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1942 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1943 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1944 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1945 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1946 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1947 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1948 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1949 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1950 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1951 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1952 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1953 1954 extern void ext4_update_dynamic_rev(struct super_block *sb); 1955 1956 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1957 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1958 { \ 1959 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1960 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1961 } \ 1962 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1963 { \ 1964 ext4_update_dynamic_rev(sb); \ 1965 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1966 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1967 } \ 1968 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1969 { \ 1970 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1971 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1972 } 1973 1974 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1975 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1976 { \ 1977 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1978 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1979 } \ 1980 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1981 { \ 1982 ext4_update_dynamic_rev(sb); \ 1983 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1984 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1985 } \ 1986 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1987 { \ 1988 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1989 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1990 } 1991 1992 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1993 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1994 { \ 1995 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1996 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1997 } \ 1998 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1999 { \ 2000 ext4_update_dynamic_rev(sb); \ 2001 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2002 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2003 } \ 2004 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2005 { \ 2006 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2007 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2008 } 2009 2010 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2011 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2012 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2013 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2014 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2015 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2016 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2017 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2018 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2019 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2020 2021 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2022 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2023 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2024 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2025 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2026 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2027 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2028 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2029 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2030 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2031 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2032 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2033 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2034 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2035 2036 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2037 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2038 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2039 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2040 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2041 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2042 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2043 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2044 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2045 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2046 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2047 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2048 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2049 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2050 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2051 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2052 2053 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2054 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2055 EXT4_FEATURE_INCOMPAT_META_BG) 2056 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2057 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2058 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2059 2060 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2061 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2062 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2063 EXT4_FEATURE_INCOMPAT_META_BG) 2064 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2065 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2066 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2067 2068 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2069 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2070 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2071 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2072 EXT4_FEATURE_INCOMPAT_META_BG| \ 2073 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2074 EXT4_FEATURE_INCOMPAT_64BIT| \ 2075 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2076 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2077 EXT4_FEATURE_INCOMPAT_MMP | \ 2078 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2079 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2080 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2081 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2082 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2083 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2084 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2085 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2086 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2087 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2088 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2089 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2090 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2091 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2092 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2093 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2094 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2095 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2096 2097 #define EXTN_FEATURE_FUNCS(ver) \ 2098 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2099 { \ 2100 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2101 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2102 } \ 2103 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2104 { \ 2105 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2106 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2107 } \ 2108 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2109 { \ 2110 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2111 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2112 } 2113 2114 EXTN_FEATURE_FUNCS(2) 2115 EXTN_FEATURE_FUNCS(3) 2116 EXTN_FEATURE_FUNCS(4) 2117 2118 static inline bool ext4_has_compat_features(struct super_block *sb) 2119 { 2120 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2121 } 2122 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2123 { 2124 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2125 } 2126 static inline bool ext4_has_incompat_features(struct super_block *sb) 2127 { 2128 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2129 } 2130 2131 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2132 2133 /* 2134 * Superblock flags 2135 */ 2136 #define EXT4_FLAGS_RESIZING 0 2137 #define EXT4_FLAGS_SHUTDOWN 1 2138 #define EXT4_FLAGS_BDEV_IS_DAX 2 2139 2140 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2141 { 2142 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2143 } 2144 2145 /* 2146 * Default values for user and/or group using reserved blocks 2147 */ 2148 #define EXT4_DEF_RESUID 0 2149 #define EXT4_DEF_RESGID 0 2150 2151 /* 2152 * Default project ID 2153 */ 2154 #define EXT4_DEF_PROJID 0 2155 2156 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2157 2158 /* 2159 * Default mount options 2160 */ 2161 #define EXT4_DEFM_DEBUG 0x0001 2162 #define EXT4_DEFM_BSDGROUPS 0x0002 2163 #define EXT4_DEFM_XATTR_USER 0x0004 2164 #define EXT4_DEFM_ACL 0x0008 2165 #define EXT4_DEFM_UID16 0x0010 2166 #define EXT4_DEFM_JMODE 0x0060 2167 #define EXT4_DEFM_JMODE_DATA 0x0020 2168 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2169 #define EXT4_DEFM_JMODE_WBACK 0x0060 2170 #define EXT4_DEFM_NOBARRIER 0x0100 2171 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2172 #define EXT4_DEFM_DISCARD 0x0400 2173 #define EXT4_DEFM_NODELALLOC 0x0800 2174 2175 /* 2176 * Default journal batch times 2177 */ 2178 #define EXT4_DEF_MIN_BATCH_TIME 0 2179 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2180 2181 /* 2182 * Minimum number of groups in a flexgroup before we separate out 2183 * directories into the first block group of a flexgroup 2184 */ 2185 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2186 2187 /* 2188 * Structure of a directory entry 2189 */ 2190 #define EXT4_NAME_LEN 255 2191 /* 2192 * Base length of the ext4 directory entry excluding the name length 2193 */ 2194 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2195 2196 struct ext4_dir_entry { 2197 __le32 inode; /* Inode number */ 2198 __le16 rec_len; /* Directory entry length */ 2199 __le16 name_len; /* Name length */ 2200 char name[EXT4_NAME_LEN]; /* File name */ 2201 }; 2202 2203 2204 /* 2205 * Encrypted Casefolded entries require saving the hash on disk. This structure 2206 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2207 * boundary. 2208 */ 2209 struct ext4_dir_entry_hash { 2210 __le32 hash; 2211 __le32 minor_hash; 2212 }; 2213 2214 /* 2215 * The new version of the directory entry. Since EXT4 structures are 2216 * stored in intel byte order, and the name_len field could never be 2217 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2218 * file_type field. 2219 */ 2220 struct ext4_dir_entry_2 { 2221 __le32 inode; /* Inode number */ 2222 __le16 rec_len; /* Directory entry length */ 2223 __u8 name_len; /* Name length */ 2224 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2225 char name[EXT4_NAME_LEN]; /* File name */ 2226 }; 2227 2228 /* 2229 * Access the hashes at the end of ext4_dir_entry_2 2230 */ 2231 #define EXT4_DIRENT_HASHES(entry) \ 2232 ((struct ext4_dir_entry_hash *) \ 2233 (((void *)(entry)) + \ 2234 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2235 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2236 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2237 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2238 2239 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2240 { 2241 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2242 } 2243 2244 /* 2245 * This is a bogus directory entry at the end of each leaf block that 2246 * records checksums. 2247 */ 2248 struct ext4_dir_entry_tail { 2249 __le32 det_reserved_zero1; /* Pretend to be unused */ 2250 __le16 det_rec_len; /* 12 */ 2251 __u8 det_reserved_zero2; /* Zero name length */ 2252 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2253 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2254 }; 2255 2256 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2257 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2258 ((blocksize) - \ 2259 sizeof(struct ext4_dir_entry_tail)))) 2260 2261 /* 2262 * Ext4 directory file types. Only the low 3 bits are used. The 2263 * other bits are reserved for now. 2264 */ 2265 #define EXT4_FT_UNKNOWN 0 2266 #define EXT4_FT_REG_FILE 1 2267 #define EXT4_FT_DIR 2 2268 #define EXT4_FT_CHRDEV 3 2269 #define EXT4_FT_BLKDEV 4 2270 #define EXT4_FT_FIFO 5 2271 #define EXT4_FT_SOCK 6 2272 #define EXT4_FT_SYMLINK 7 2273 2274 #define EXT4_FT_MAX 8 2275 2276 #define EXT4_FT_DIR_CSUM 0xDE 2277 2278 /* 2279 * EXT4_DIR_PAD defines the directory entries boundaries 2280 * 2281 * NOTE: It must be a multiple of 4 2282 */ 2283 #define EXT4_DIR_PAD 4 2284 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2285 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2286 2287 /* 2288 * The rec_len is dependent on the type of directory. Directories that are 2289 * casefolded and encrypted need to store the hash as well, so we add room for 2290 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2291 * pass NULL for dir, as those entries do not use the extra fields. 2292 */ 2293 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2294 const struct inode *dir) 2295 { 2296 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2297 2298 if (dir && ext4_hash_in_dirent(dir)) 2299 rec_len += sizeof(struct ext4_dir_entry_hash); 2300 return (rec_len & ~EXT4_DIR_ROUND); 2301 } 2302 2303 /* 2304 * If we ever get support for fs block sizes > page_size, we'll need 2305 * to remove the #if statements in the next two functions... 2306 */ 2307 static inline unsigned int 2308 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2309 { 2310 unsigned len = le16_to_cpu(dlen); 2311 2312 #if (PAGE_SIZE >= 65536) 2313 if (len == EXT4_MAX_REC_LEN || len == 0) 2314 return blocksize; 2315 return (len & 65532) | ((len & 3) << 16); 2316 #else 2317 return len; 2318 #endif 2319 } 2320 2321 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2322 { 2323 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2324 #if (PAGE_SIZE >= 65536) 2325 if (len < 65536) 2326 return cpu_to_le16(len); 2327 if (len == blocksize) { 2328 if (blocksize == 65536) 2329 return cpu_to_le16(EXT4_MAX_REC_LEN); 2330 else 2331 return cpu_to_le16(0); 2332 } 2333 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2334 #else 2335 return cpu_to_le16(len); 2336 #endif 2337 } 2338 2339 /* 2340 * Hash Tree Directory indexing 2341 * (c) Daniel Phillips, 2001 2342 */ 2343 2344 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2345 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2346 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2347 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2348 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2349 2350 /* Legal values for the dx_root hash_version field: */ 2351 2352 #define DX_HASH_LEGACY 0 2353 #define DX_HASH_HALF_MD4 1 2354 #define DX_HASH_TEA 2 2355 #define DX_HASH_LEGACY_UNSIGNED 3 2356 #define DX_HASH_HALF_MD4_UNSIGNED 4 2357 #define DX_HASH_TEA_UNSIGNED 5 2358 #define DX_HASH_SIPHASH 6 2359 2360 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2361 const void *address, unsigned int length) 2362 { 2363 struct { 2364 struct shash_desc shash; 2365 char ctx[4]; 2366 } desc; 2367 2368 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2369 2370 desc.shash.tfm = sbi->s_chksum_driver; 2371 *(u32 *)desc.ctx = crc; 2372 2373 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2374 2375 return *(u32 *)desc.ctx; 2376 } 2377 2378 #ifdef __KERNEL__ 2379 2380 /* hash info structure used by the directory hash */ 2381 struct dx_hash_info 2382 { 2383 u32 hash; 2384 u32 minor_hash; 2385 int hash_version; 2386 u32 *seed; 2387 }; 2388 2389 2390 /* 32 and 64 bit signed EOF for dx directories */ 2391 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2392 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2393 2394 2395 /* 2396 * Control parameters used by ext4_htree_next_block 2397 */ 2398 #define HASH_NB_ALWAYS 1 2399 2400 struct ext4_filename { 2401 const struct qstr *usr_fname; 2402 struct fscrypt_str disk_name; 2403 struct dx_hash_info hinfo; 2404 #ifdef CONFIG_FS_ENCRYPTION 2405 struct fscrypt_str crypto_buf; 2406 #endif 2407 #if IS_ENABLED(CONFIG_UNICODE) 2408 struct fscrypt_str cf_name; 2409 #endif 2410 }; 2411 2412 #define fname_name(p) ((p)->disk_name.name) 2413 #define fname_usr_name(p) ((p)->usr_fname->name) 2414 #define fname_len(p) ((p)->disk_name.len) 2415 2416 /* 2417 * Describe an inode's exact location on disk and in memory 2418 */ 2419 struct ext4_iloc 2420 { 2421 struct buffer_head *bh; 2422 unsigned long offset; 2423 ext4_group_t block_group; 2424 }; 2425 2426 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2427 { 2428 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2429 } 2430 2431 static inline bool ext4_is_quota_file(struct inode *inode) 2432 { 2433 return IS_NOQUOTA(inode) && 2434 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2435 } 2436 2437 /* 2438 * This structure is stuffed into the struct file's private_data field 2439 * for directories. It is where we put information so that we can do 2440 * readdir operations in hash tree order. 2441 */ 2442 struct dir_private_info { 2443 struct rb_root root; 2444 struct rb_node *curr_node; 2445 struct fname *extra_fname; 2446 loff_t last_pos; 2447 __u32 curr_hash; 2448 __u32 curr_minor_hash; 2449 __u32 next_hash; 2450 }; 2451 2452 /* calculate the first block number of the group */ 2453 static inline ext4_fsblk_t 2454 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2455 { 2456 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2457 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2458 } 2459 2460 /* 2461 * Special error return code only used by dx_probe() and its callers. 2462 */ 2463 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2464 2465 /* htree levels for ext4 */ 2466 #define EXT4_HTREE_LEVEL_COMPAT 2 2467 #define EXT4_HTREE_LEVEL 3 2468 2469 static inline int ext4_dir_htree_level(struct super_block *sb) 2470 { 2471 return ext4_has_feature_largedir(sb) ? 2472 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2473 } 2474 2475 /* 2476 * Timeout and state flag for lazy initialization inode thread. 2477 */ 2478 #define EXT4_DEF_LI_WAIT_MULT 10 2479 #define EXT4_DEF_LI_MAX_START_DELAY 5 2480 #define EXT4_LAZYINIT_QUIT 0x0001 2481 #define EXT4_LAZYINIT_RUNNING 0x0002 2482 2483 /* 2484 * Lazy inode table initialization info 2485 */ 2486 struct ext4_lazy_init { 2487 unsigned long li_state; 2488 struct list_head li_request_list; 2489 struct mutex li_list_mtx; 2490 }; 2491 2492 enum ext4_li_mode { 2493 EXT4_LI_MODE_PREFETCH_BBITMAP, 2494 EXT4_LI_MODE_ITABLE, 2495 }; 2496 2497 struct ext4_li_request { 2498 struct super_block *lr_super; 2499 enum ext4_li_mode lr_mode; 2500 ext4_group_t lr_first_not_zeroed; 2501 ext4_group_t lr_next_group; 2502 struct list_head lr_request; 2503 unsigned long lr_next_sched; 2504 unsigned long lr_timeout; 2505 }; 2506 2507 struct ext4_features { 2508 struct kobject f_kobj; 2509 struct completion f_kobj_unregister; 2510 }; 2511 2512 /* 2513 * This structure will be used for multiple mount protection. It will be 2514 * written into the block number saved in the s_mmp_block field in the 2515 * superblock. Programs that check MMP should assume that if 2516 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2517 * to use the filesystem, regardless of how old the timestamp is. 2518 */ 2519 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2520 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2521 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2522 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2523 2524 struct mmp_struct { 2525 __le32 mmp_magic; /* Magic number for MMP */ 2526 __le32 mmp_seq; /* Sequence no. updated periodically */ 2527 2528 /* 2529 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2530 * purposes and do not affect the correctness of the algorithm 2531 */ 2532 __le64 mmp_time; /* Time last updated */ 2533 char mmp_nodename[64]; /* Node which last updated MMP block */ 2534 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2535 2536 /* 2537 * mmp_check_interval is used to verify if the MMP block has been 2538 * updated on the block device. The value is updated based on the 2539 * maximum time to write the MMP block during an update cycle. 2540 */ 2541 __le16 mmp_check_interval; 2542 2543 __le16 mmp_pad1; 2544 __le32 mmp_pad2[226]; 2545 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2546 }; 2547 2548 /* arguments passed to the mmp thread */ 2549 struct mmpd_data { 2550 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2551 struct super_block *sb; /* super block of the fs */ 2552 }; 2553 2554 /* 2555 * Check interval multiplier 2556 * The MMP block is written every update interval and initially checked every 2557 * update interval x the multiplier (the value is then adapted based on the 2558 * write latency). The reason is that writes can be delayed under load and we 2559 * don't want readers to incorrectly assume that the filesystem is no longer 2560 * in use. 2561 */ 2562 #define EXT4_MMP_CHECK_MULT 2UL 2563 2564 /* 2565 * Minimum interval for MMP checking in seconds. 2566 */ 2567 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2568 2569 /* 2570 * Maximum interval for MMP checking in seconds. 2571 */ 2572 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2573 2574 /* 2575 * Function prototypes 2576 */ 2577 2578 /* 2579 * Ok, these declarations are also in <linux/kernel.h> but none of the 2580 * ext4 source programs needs to include it so they are duplicated here. 2581 */ 2582 # define NORET_TYPE /**/ 2583 # define ATTRIB_NORET __attribute__((noreturn)) 2584 # define NORET_AND noreturn, 2585 2586 /* bitmap.c */ 2587 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2588 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2589 struct ext4_group_desc *gdp, 2590 struct buffer_head *bh, int sz); 2591 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2592 struct ext4_group_desc *gdp, 2593 struct buffer_head *bh, int sz); 2594 void ext4_block_bitmap_csum_set(struct super_block *sb, 2595 struct ext4_group_desc *gdp, 2596 struct buffer_head *bh); 2597 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2598 struct ext4_group_desc *gdp, 2599 struct buffer_head *bh); 2600 2601 /* balloc.c */ 2602 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2603 ext4_fsblk_t blocknr, 2604 ext4_group_t *blockgrpp, 2605 ext4_grpblk_t *offsetp); 2606 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2607 ext4_fsblk_t block); 2608 2609 extern unsigned int ext4_block_group(struct super_block *sb, 2610 ext4_fsblk_t blocknr); 2611 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2612 ext4_fsblk_t blocknr); 2613 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2614 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2615 ext4_group_t group); 2616 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2617 ext4_fsblk_t goal, 2618 unsigned int flags, 2619 unsigned long *count, 2620 int *errp); 2621 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2622 s64 nclusters, unsigned int flags); 2623 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2624 extern void ext4_check_blocks_bitmap(struct super_block *); 2625 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2626 ext4_group_t block_group, 2627 struct buffer_head ** bh); 2628 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2629 2630 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2631 ext4_group_t block_group, 2632 bool ignore_locked); 2633 extern int ext4_wait_block_bitmap(struct super_block *sb, 2634 ext4_group_t block_group, 2635 struct buffer_head *bh); 2636 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2637 ext4_group_t block_group); 2638 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2639 ext4_group_t block_group, 2640 struct ext4_group_desc *gdp); 2641 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2642 2643 #if IS_ENABLED(CONFIG_UNICODE) 2644 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2645 const struct qstr *iname, 2646 struct ext4_filename *fname); 2647 #endif 2648 2649 /* ext4 encryption related stuff goes here crypto.c */ 2650 #ifdef CONFIG_FS_ENCRYPTION 2651 extern const struct fscrypt_operations ext4_cryptops; 2652 2653 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2654 int lookup, struct ext4_filename *fname); 2655 2656 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2657 struct ext4_filename *fname); 2658 2659 void ext4_fname_free_filename(struct ext4_filename *fname); 2660 2661 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2662 2663 #else /* !CONFIG_FS_ENCRYPTION */ 2664 static inline int ext4_fname_setup_filename(struct inode *dir, 2665 const struct qstr *iname, 2666 int lookup, 2667 struct ext4_filename *fname) 2668 { 2669 int err = 0; 2670 fname->usr_fname = iname; 2671 fname->disk_name.name = (unsigned char *) iname->name; 2672 fname->disk_name.len = iname->len; 2673 2674 #if IS_ENABLED(CONFIG_UNICODE) 2675 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2676 #endif 2677 2678 return err; 2679 } 2680 2681 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2682 struct dentry *dentry, 2683 struct ext4_filename *fname) 2684 { 2685 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2686 } 2687 2688 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2689 { 2690 #if IS_ENABLED(CONFIG_UNICODE) 2691 kfree(fname->cf_name.name); 2692 fname->cf_name.name = NULL; 2693 #endif 2694 } 2695 2696 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2697 void __user *arg) 2698 { 2699 return -EOPNOTSUPP; 2700 } 2701 #endif /* !CONFIG_FS_ENCRYPTION */ 2702 2703 /* dir.c */ 2704 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2705 struct file *, 2706 struct ext4_dir_entry_2 *, 2707 struct buffer_head *, char *, int, 2708 unsigned int); 2709 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2710 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2711 (de), (bh), (buf), (size), (offset))) 2712 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2713 __u32 minor_hash, 2714 struct ext4_dir_entry_2 *dirent, 2715 struct fscrypt_str *ent_name); 2716 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2717 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2718 struct buffer_head *bh, 2719 void *buf, int buf_size, 2720 struct ext4_filename *fname, 2721 struct ext4_dir_entry_2 **dest_de); 2722 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2723 struct ext4_dir_entry_2 *de, 2724 int buf_size, 2725 struct ext4_filename *fname); 2726 static inline void ext4_update_dx_flag(struct inode *inode) 2727 { 2728 if (!ext4_has_feature_dir_index(inode->i_sb) && 2729 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2730 /* ext4_iget() should have caught this... */ 2731 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2732 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2733 } 2734 } 2735 static const unsigned char ext4_filetype_table[] = { 2736 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2737 }; 2738 2739 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2740 { 2741 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2742 return DT_UNKNOWN; 2743 2744 return ext4_filetype_table[filetype]; 2745 } 2746 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2747 void *buf, int buf_size); 2748 2749 /* fsync.c */ 2750 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2751 2752 /* hash.c */ 2753 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2754 struct dx_hash_info *hinfo); 2755 2756 /* ialloc.c */ 2757 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2758 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2759 struct inode *, umode_t, 2760 const struct qstr *qstr, __u32 goal, 2761 uid_t *owner, __u32 i_flags, 2762 int handle_type, unsigned int line_no, 2763 int nblocks); 2764 2765 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2766 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2767 (goal), (owner), i_flags, 0, 0, 0) 2768 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2769 type, nblocks) \ 2770 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2771 0, (type), __LINE__, (nblocks)) 2772 2773 2774 extern void ext4_free_inode(handle_t *, struct inode *); 2775 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2776 extern unsigned long ext4_count_free_inodes(struct super_block *); 2777 extern unsigned long ext4_count_dirs(struct super_block *); 2778 extern void ext4_check_inodes_bitmap(struct super_block *); 2779 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2780 extern int ext4_init_inode_table(struct super_block *sb, 2781 ext4_group_t group, int barrier); 2782 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2783 2784 /* fast_commit.c */ 2785 int ext4_fc_info_show(struct seq_file *seq, void *v); 2786 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2787 void ext4_fc_init_inode(struct inode *inode); 2788 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2789 ext4_lblk_t end); 2790 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2791 struct dentry *dentry); 2792 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2793 struct dentry *dentry); 2794 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2795 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2796 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2797 struct dentry *dentry); 2798 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2799 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2800 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2801 void ext4_fc_start_update(struct inode *inode); 2802 void ext4_fc_stop_update(struct inode *inode); 2803 void ext4_fc_del(struct inode *inode); 2804 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2805 void ext4_fc_replay_cleanup(struct super_block *sb); 2806 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2807 int __init ext4_fc_init_dentry_cache(void); 2808 void ext4_fc_destroy_dentry_cache(void); 2809 int ext4_fc_record_regions(struct super_block *sb, int ino, 2810 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2811 int len, int replay); 2812 2813 /* mballoc.c */ 2814 extern const struct seq_operations ext4_mb_seq_groups_ops; 2815 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2816 extern long ext4_mb_stats; 2817 extern long ext4_mb_max_to_scan; 2818 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2819 extern int ext4_mb_init(struct super_block *); 2820 extern int ext4_mb_release(struct super_block *); 2821 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2822 struct ext4_allocation_request *, int *); 2823 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2824 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2825 extern int __init ext4_init_mballoc(void); 2826 extern void ext4_exit_mballoc(void); 2827 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2828 ext4_group_t group, 2829 unsigned int nr, int *cnt); 2830 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2831 unsigned int nr); 2832 2833 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2834 struct buffer_head *bh, ext4_fsblk_t block, 2835 unsigned long count, int flags); 2836 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2837 ext4_group_t ngroups); 2838 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2839 ext4_group_t i, struct ext4_group_desc *desc); 2840 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2841 ext4_fsblk_t block, unsigned long count); 2842 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2843 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2844 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2845 int len, int state); 2846 2847 /* inode.c */ 2848 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2849 struct ext4_inode_info *ei); 2850 int ext4_inode_is_fast_symlink(struct inode *inode); 2851 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2852 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2853 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2854 bool wait, struct buffer_head **bhs); 2855 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2856 struct buffer_head *bh_result, int create); 2857 int ext4_get_block(struct inode *inode, sector_t iblock, 2858 struct buffer_head *bh_result, int create); 2859 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2860 struct buffer_head *bh, int create); 2861 int ext4_walk_page_buffers(handle_t *handle, 2862 struct inode *inode, 2863 struct buffer_head *head, 2864 unsigned from, 2865 unsigned to, 2866 int *partial, 2867 int (*fn)(handle_t *handle, struct inode *inode, 2868 struct buffer_head *bh)); 2869 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2870 struct buffer_head *bh); 2871 #define FALL_BACK_TO_NONDELALLOC 1 2872 #define CONVERT_INLINE_DATA 2 2873 2874 typedef enum { 2875 EXT4_IGET_NORMAL = 0, 2876 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2877 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2878 EXT4_IGET_BAD = 0x0004 /* Allow to iget a bad inode */ 2879 } ext4_iget_flags; 2880 2881 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2882 ext4_iget_flags flags, const char *function, 2883 unsigned int line); 2884 2885 #define ext4_iget(sb, ino, flags) \ 2886 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2887 2888 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2889 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2890 struct iattr *); 2891 extern u32 ext4_dio_alignment(struct inode *inode); 2892 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2893 struct kstat *, u32, unsigned int); 2894 extern void ext4_evict_inode(struct inode *); 2895 extern void ext4_clear_inode(struct inode *); 2896 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2897 struct kstat *, u32, unsigned int); 2898 extern int ext4_sync_inode(handle_t *, struct inode *); 2899 extern void ext4_dirty_inode(struct inode *, int); 2900 extern int ext4_change_inode_journal_flag(struct inode *, int); 2901 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2902 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2903 struct ext4_iloc *iloc); 2904 extern int ext4_inode_attach_jinode(struct inode *inode); 2905 extern int ext4_can_truncate(struct inode *inode); 2906 extern int ext4_truncate(struct inode *); 2907 extern int ext4_break_layouts(struct inode *); 2908 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2909 extern void ext4_set_inode_flags(struct inode *, bool init); 2910 extern int ext4_alloc_da_blocks(struct inode *inode); 2911 extern void ext4_set_aops(struct inode *inode); 2912 extern int ext4_writepage_trans_blocks(struct inode *); 2913 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 2914 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2915 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2916 loff_t lstart, loff_t lend); 2917 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2918 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2919 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2920 extern void ext4_da_release_space(struct inode *inode, int to_free); 2921 extern void ext4_da_update_reserve_space(struct inode *inode, 2922 int used, int quota_claim); 2923 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2924 ext4_fsblk_t pblk, ext4_lblk_t len); 2925 2926 /* indirect.c */ 2927 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2928 struct ext4_map_blocks *map, int flags); 2929 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2930 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2931 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2932 ext4_lblk_t start, ext4_lblk_t end); 2933 2934 /* ioctl.c */ 2935 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2936 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2937 int ext4_fileattr_set(struct mnt_idmap *idmap, 2938 struct dentry *dentry, struct fileattr *fa); 2939 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 2940 extern void ext4_reset_inode_seed(struct inode *inode); 2941 int ext4_update_overhead(struct super_block *sb, bool force); 2942 2943 /* migrate.c */ 2944 extern int ext4_ext_migrate(struct inode *); 2945 extern int ext4_ind_migrate(struct inode *inode); 2946 2947 /* namei.c */ 2948 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 2949 struct inode *inode); 2950 extern int ext4_dirblock_csum_verify(struct inode *inode, 2951 struct buffer_head *bh); 2952 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2953 __u32 start_minor_hash, __u32 *next_hash); 2954 extern int ext4_search_dir(struct buffer_head *bh, 2955 char *search_buf, 2956 int buf_size, 2957 struct inode *dir, 2958 struct ext4_filename *fname, 2959 unsigned int offset, 2960 struct ext4_dir_entry_2 **res_dir); 2961 extern int ext4_generic_delete_entry(struct inode *dir, 2962 struct ext4_dir_entry_2 *de_del, 2963 struct buffer_head *bh, 2964 void *entry_buf, 2965 int buf_size, 2966 int csum_size); 2967 extern bool ext4_empty_dir(struct inode *inode); 2968 2969 /* resize.c */ 2970 extern void ext4_kvfree_array_rcu(void *to_free); 2971 extern int ext4_group_add(struct super_block *sb, 2972 struct ext4_new_group_data *input); 2973 extern int ext4_group_extend(struct super_block *sb, 2974 struct ext4_super_block *es, 2975 ext4_fsblk_t n_blocks_count); 2976 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2977 extern unsigned int ext4_list_backups(struct super_block *sb, 2978 unsigned int *three, unsigned int *five, 2979 unsigned int *seven); 2980 2981 /* super.c */ 2982 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2983 sector_t block, blk_opf_t op_flags); 2984 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 2985 sector_t block); 2986 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 2987 bh_end_io_t *end_io); 2988 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 2989 bh_end_io_t *end_io); 2990 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 2991 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 2992 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2993 extern int ext4_calculate_overhead(struct super_block *sb); 2994 extern __le32 ext4_superblock_csum(struct super_block *sb, 2995 struct ext4_super_block *es); 2996 extern void ext4_superblock_csum_set(struct super_block *sb); 2997 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2998 ext4_group_t ngroup); 2999 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3000 char nbuf[16]); 3001 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3002 ext4_group_t block_group, 3003 unsigned int flags); 3004 3005 extern __printf(7, 8) 3006 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3007 int, __u64, const char *, ...); 3008 extern __printf(6, 7) 3009 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3010 ext4_fsblk_t, int, const char *, ...); 3011 extern __printf(5, 6) 3012 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3013 const char *, ...); 3014 extern void __ext4_std_error(struct super_block *, const char *, 3015 unsigned int, int); 3016 extern __printf(4, 5) 3017 void __ext4_warning(struct super_block *, const char *, unsigned int, 3018 const char *, ...); 3019 extern __printf(4, 5) 3020 void __ext4_warning_inode(const struct inode *inode, const char *function, 3021 unsigned int line, const char *fmt, ...); 3022 extern __printf(3, 4) 3023 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3024 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3025 const char *, unsigned int, const char *); 3026 extern __printf(7, 8) 3027 void __ext4_grp_locked_error(const char *, unsigned int, 3028 struct super_block *, ext4_group_t, 3029 unsigned long, ext4_fsblk_t, 3030 const char *, ...); 3031 3032 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3033 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3034 3035 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3036 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3037 3038 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3039 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3040 (fmt), ## a) 3041 3042 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3043 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3044 3045 #define ext4_abort(sb, err, fmt, a...) \ 3046 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3047 3048 #ifdef CONFIG_PRINTK 3049 3050 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3051 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3052 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3053 __ext4_error_inode((inode), (func), (line), (block), \ 3054 (err), (fmt), ##__VA_ARGS__) 3055 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3056 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3057 #define ext4_error(sb, fmt, ...) \ 3058 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3059 ##__VA_ARGS__) 3060 #define ext4_error_err(sb, err, fmt, ...) \ 3061 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3062 ##__VA_ARGS__) 3063 #define ext4_warning(sb, fmt, ...) \ 3064 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3065 #define ext4_warning_inode(inode, fmt, ...) \ 3066 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3067 #define ext4_msg(sb, level, fmt, ...) \ 3068 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3069 #define dump_mmp_msg(sb, mmp, msg) \ 3070 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3071 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3072 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3073 fmt, ##__VA_ARGS__) 3074 3075 #else 3076 3077 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3078 do { \ 3079 no_printk(fmt, ##__VA_ARGS__); \ 3080 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3081 } while (0) 3082 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3083 do { \ 3084 no_printk(fmt, ##__VA_ARGS__); \ 3085 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3086 } while (0) 3087 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3088 do { \ 3089 no_printk(fmt, ##__VA_ARGS__); \ 3090 __ext4_error_file(file, "", 0, block, " "); \ 3091 } while (0) 3092 #define ext4_error(sb, fmt, ...) \ 3093 do { \ 3094 no_printk(fmt, ##__VA_ARGS__); \ 3095 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3096 } while (0) 3097 #define ext4_error_err(sb, err, fmt, ...) \ 3098 do { \ 3099 no_printk(fmt, ##__VA_ARGS__); \ 3100 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3101 } while (0) 3102 #define ext4_warning(sb, fmt, ...) \ 3103 do { \ 3104 no_printk(fmt, ##__VA_ARGS__); \ 3105 __ext4_warning(sb, "", 0, " "); \ 3106 } while (0) 3107 #define ext4_warning_inode(inode, fmt, ...) \ 3108 do { \ 3109 no_printk(fmt, ##__VA_ARGS__); \ 3110 __ext4_warning_inode(inode, "", 0, " "); \ 3111 } while (0) 3112 #define ext4_msg(sb, level, fmt, ...) \ 3113 do { \ 3114 no_printk(fmt, ##__VA_ARGS__); \ 3115 __ext4_msg(sb, "", " "); \ 3116 } while (0) 3117 #define dump_mmp_msg(sb, mmp, msg) \ 3118 __dump_mmp_msg(sb, mmp, "", 0, "") 3119 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3120 do { \ 3121 no_printk(fmt, ##__VA_ARGS__); \ 3122 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3123 } while (0) 3124 3125 #endif 3126 3127 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3128 struct ext4_group_desc *bg); 3129 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3130 struct ext4_group_desc *bg); 3131 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3132 struct ext4_group_desc *bg); 3133 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3134 struct ext4_group_desc *bg); 3135 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3136 struct ext4_group_desc *bg); 3137 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3138 struct ext4_group_desc *bg); 3139 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3140 struct ext4_group_desc *bg); 3141 extern void ext4_block_bitmap_set(struct super_block *sb, 3142 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3143 extern void ext4_inode_bitmap_set(struct super_block *sb, 3144 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3145 extern void ext4_inode_table_set(struct super_block *sb, 3146 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3147 extern void ext4_free_group_clusters_set(struct super_block *sb, 3148 struct ext4_group_desc *bg, 3149 __u32 count); 3150 extern void ext4_free_inodes_set(struct super_block *sb, 3151 struct ext4_group_desc *bg, __u32 count); 3152 extern void ext4_used_dirs_set(struct super_block *sb, 3153 struct ext4_group_desc *bg, __u32 count); 3154 extern void ext4_itable_unused_set(struct super_block *sb, 3155 struct ext4_group_desc *bg, __u32 count); 3156 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3157 struct ext4_group_desc *gdp); 3158 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3159 struct ext4_group_desc *gdp); 3160 extern int ext4_register_li_request(struct super_block *sb, 3161 ext4_group_t first_not_zeroed); 3162 3163 static inline int ext4_has_metadata_csum(struct super_block *sb) 3164 { 3165 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3166 !EXT4_SB(sb)->s_chksum_driver); 3167 3168 return ext4_has_feature_metadata_csum(sb) && 3169 (EXT4_SB(sb)->s_chksum_driver != NULL); 3170 } 3171 3172 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3173 { 3174 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3175 } 3176 3177 #define ext4_read_incompat_64bit_val(es, name) \ 3178 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3179 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3180 le32_to_cpu(es->name##_lo)) 3181 3182 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3183 { 3184 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3185 } 3186 3187 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3188 { 3189 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3190 } 3191 3192 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3193 { 3194 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3195 } 3196 3197 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3198 ext4_fsblk_t blk) 3199 { 3200 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3201 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3202 } 3203 3204 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3205 ext4_fsblk_t blk) 3206 { 3207 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3208 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3209 } 3210 3211 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3212 ext4_fsblk_t blk) 3213 { 3214 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3215 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3216 } 3217 3218 static inline loff_t ext4_isize(struct super_block *sb, 3219 struct ext4_inode *raw_inode) 3220 { 3221 if (ext4_has_feature_largedir(sb) || 3222 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3223 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3224 le32_to_cpu(raw_inode->i_size_lo); 3225 3226 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3227 } 3228 3229 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3230 { 3231 raw_inode->i_size_lo = cpu_to_le32(i_size); 3232 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3233 } 3234 3235 static inline 3236 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3237 ext4_group_t group) 3238 { 3239 struct ext4_group_info **grp_info; 3240 long indexv, indexh; 3241 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3242 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3243 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3244 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3245 return grp_info[indexh]; 3246 } 3247 3248 /* 3249 * Reading s_groups_count requires using smp_rmb() afterwards. See 3250 * the locking protocol documented in the comments of ext4_group_add() 3251 * in resize.c 3252 */ 3253 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3254 { 3255 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3256 3257 smp_rmb(); 3258 return ngroups; 3259 } 3260 3261 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3262 ext4_group_t block_group) 3263 { 3264 return block_group >> sbi->s_log_groups_per_flex; 3265 } 3266 3267 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3268 { 3269 return 1 << sbi->s_log_groups_per_flex; 3270 } 3271 3272 #define ext4_std_error(sb, errno) \ 3273 do { \ 3274 if ((errno)) \ 3275 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3276 } while (0) 3277 3278 #ifdef CONFIG_SMP 3279 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3280 * counters. So we need to make sure we have free clusters more 3281 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3282 */ 3283 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3284 #else 3285 #define EXT4_FREECLUSTERS_WATERMARK 0 3286 #endif 3287 3288 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3289 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3290 { 3291 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3292 !inode_is_locked(inode)); 3293 down_write(&EXT4_I(inode)->i_data_sem); 3294 if (newsize > EXT4_I(inode)->i_disksize) 3295 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3296 up_write(&EXT4_I(inode)->i_data_sem); 3297 } 3298 3299 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3300 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3301 { 3302 int changed = 0; 3303 3304 if (newsize > inode->i_size) { 3305 i_size_write(inode, newsize); 3306 changed = 1; 3307 } 3308 if (newsize > EXT4_I(inode)->i_disksize) { 3309 ext4_update_i_disksize(inode, newsize); 3310 changed |= 2; 3311 } 3312 return changed; 3313 } 3314 3315 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3316 loff_t len); 3317 3318 struct ext4_group_info { 3319 unsigned long bb_state; 3320 #ifdef AGGRESSIVE_CHECK 3321 unsigned long bb_check_counter; 3322 #endif 3323 struct rb_root bb_free_root; 3324 ext4_grpblk_t bb_first_free; /* first free block */ 3325 ext4_grpblk_t bb_free; /* total free blocks */ 3326 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3327 int bb_avg_fragment_size_order; /* order of average 3328 fragment in BG */ 3329 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3330 ext4_group_t bb_group; /* Group number */ 3331 struct list_head bb_prealloc_list; 3332 #ifdef DOUBLE_CHECK 3333 void *bb_bitmap; 3334 #endif 3335 struct rw_semaphore alloc_sem; 3336 struct list_head bb_avg_fragment_size_node; 3337 struct list_head bb_largest_free_order_node; 3338 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3339 * regions, index is order. 3340 * bb_counters[3] = 5 means 3341 * 5 free 8-block regions. */ 3342 }; 3343 3344 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3345 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3346 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3347 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3348 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3349 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3350 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3351 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3352 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3353 3354 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3355 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3356 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3357 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3358 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3359 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3360 3361 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3362 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3363 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3364 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3365 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3366 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3367 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3368 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3369 3370 #define EXT4_MAX_CONTENTION 8 3371 #define EXT4_CONTENTION_THRESHOLD 2 3372 3373 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3374 ext4_group_t group) 3375 { 3376 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3377 } 3378 3379 /* 3380 * Returns true if the filesystem is busy enough that attempts to 3381 * access the block group locks has run into contention. 3382 */ 3383 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3384 { 3385 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3386 } 3387 3388 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3389 { 3390 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3391 if (spin_trylock(lock)) 3392 /* 3393 * We're able to grab the lock right away, so drop the 3394 * lock contention counter. 3395 */ 3396 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3397 else { 3398 /* 3399 * The lock is busy, so bump the contention counter, 3400 * and then wait on the spin lock. 3401 */ 3402 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3403 EXT4_MAX_CONTENTION); 3404 spin_lock(lock); 3405 } 3406 } 3407 3408 static inline void ext4_unlock_group(struct super_block *sb, 3409 ext4_group_t group) 3410 { 3411 spin_unlock(ext4_group_lock_ptr(sb, group)); 3412 } 3413 3414 #ifdef CONFIG_QUOTA 3415 static inline bool ext4_quota_capable(struct super_block *sb) 3416 { 3417 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3418 } 3419 3420 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3421 { 3422 struct ext4_sb_info *sbi = EXT4_SB(sb); 3423 3424 return (ext4_has_feature_quota(sb) || 3425 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3426 } 3427 int ext4_enable_quotas(struct super_block *sb); 3428 #endif 3429 3430 /* 3431 * Block validity checking 3432 */ 3433 #define ext4_check_indirect_blockref(inode, bh) \ 3434 ext4_check_blockref(__func__, __LINE__, inode, \ 3435 (__le32 *)(bh)->b_data, \ 3436 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3437 3438 #define ext4_ind_check_inode(inode) \ 3439 ext4_check_blockref(__func__, __LINE__, inode, \ 3440 EXT4_I(inode)->i_data, \ 3441 EXT4_NDIR_BLOCKS) 3442 3443 /* 3444 * Inodes and files operations 3445 */ 3446 3447 /* dir.c */ 3448 extern const struct file_operations ext4_dir_operations; 3449 3450 /* file.c */ 3451 extern const struct inode_operations ext4_file_inode_operations; 3452 extern const struct file_operations ext4_file_operations; 3453 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3454 3455 /* inline.c */ 3456 extern int ext4_get_max_inline_size(struct inode *inode); 3457 extern int ext4_find_inline_data_nolock(struct inode *inode); 3458 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3459 unsigned int len); 3460 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3461 3462 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3463 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3464 struct inode *inode, 3465 loff_t pos, unsigned len, 3466 struct page **pagep); 3467 extern int ext4_write_inline_data_end(struct inode *inode, 3468 loff_t pos, unsigned len, 3469 unsigned copied, 3470 struct page *page); 3471 extern struct buffer_head * 3472 ext4_journalled_write_inline_data(struct inode *inode, 3473 unsigned len, 3474 struct page *page); 3475 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3476 struct inode *inode, 3477 loff_t pos, unsigned len, 3478 struct page **pagep, 3479 void **fsdata); 3480 extern int ext4_try_add_inline_entry(handle_t *handle, 3481 struct ext4_filename *fname, 3482 struct inode *dir, struct inode *inode); 3483 extern int ext4_try_create_inline_dir(handle_t *handle, 3484 struct inode *parent, 3485 struct inode *inode); 3486 extern int ext4_read_inline_dir(struct file *filp, 3487 struct dir_context *ctx, 3488 int *has_inline_data); 3489 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3490 struct inode *dir, ext4_lblk_t block, 3491 struct dx_hash_info *hinfo, 3492 __u32 start_hash, __u32 start_minor_hash, 3493 int *has_inline_data); 3494 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3495 struct ext4_filename *fname, 3496 struct ext4_dir_entry_2 **res_dir, 3497 int *has_inline_data); 3498 extern int ext4_delete_inline_entry(handle_t *handle, 3499 struct inode *dir, 3500 struct ext4_dir_entry_2 *de_del, 3501 struct buffer_head *bh, 3502 int *has_inline_data); 3503 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3504 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3505 struct ext4_dir_entry_2 **parent_de, 3506 int *retval); 3507 extern void *ext4_read_inline_link(struct inode *inode); 3508 3509 struct iomap; 3510 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3511 3512 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3513 3514 extern int ext4_convert_inline_data(struct inode *inode); 3515 3516 static inline int ext4_has_inline_data(struct inode *inode) 3517 { 3518 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3519 EXT4_I(inode)->i_inline_off; 3520 } 3521 3522 /* namei.c */ 3523 extern const struct inode_operations ext4_dir_inode_operations; 3524 extern const struct inode_operations ext4_special_inode_operations; 3525 extern struct dentry *ext4_get_parent(struct dentry *child); 3526 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3527 struct ext4_dir_entry_2 *de, 3528 int blocksize, int csum_size, 3529 unsigned int parent_ino, int dotdot_real_len); 3530 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3531 unsigned int blocksize); 3532 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3533 struct buffer_head *bh); 3534 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3535 struct inode *inode, struct dentry *dentry); 3536 extern int __ext4_link(struct inode *dir, struct inode *inode, 3537 struct dentry *dentry); 3538 3539 #define S_SHIFT 12 3540 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3541 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3542 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3543 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3544 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3545 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3546 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3547 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3548 }; 3549 3550 static inline void ext4_set_de_type(struct super_block *sb, 3551 struct ext4_dir_entry_2 *de, 3552 umode_t mode) { 3553 if (ext4_has_feature_filetype(sb)) 3554 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3555 } 3556 3557 /* readpages.c */ 3558 extern int ext4_mpage_readpages(struct inode *inode, 3559 struct readahead_control *rac, struct folio *folio); 3560 extern int __init ext4_init_post_read_processing(void); 3561 extern void ext4_exit_post_read_processing(void); 3562 3563 /* symlink.c */ 3564 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3565 extern const struct inode_operations ext4_symlink_inode_operations; 3566 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3567 3568 /* sysfs.c */ 3569 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3570 extern int ext4_register_sysfs(struct super_block *sb); 3571 extern void ext4_unregister_sysfs(struct super_block *sb); 3572 extern int __init ext4_init_sysfs(void); 3573 extern void ext4_exit_sysfs(void); 3574 3575 /* block_validity */ 3576 extern void ext4_release_system_zone(struct super_block *sb); 3577 extern int ext4_setup_system_zone(struct super_block *sb); 3578 extern int __init ext4_init_system_zone(void); 3579 extern void ext4_exit_system_zone(void); 3580 extern int ext4_inode_block_valid(struct inode *inode, 3581 ext4_fsblk_t start_blk, 3582 unsigned int count); 3583 extern int ext4_check_blockref(const char *, unsigned int, 3584 struct inode *, __le32 *, unsigned int); 3585 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3586 ext4_fsblk_t start_blk, unsigned int count); 3587 3588 3589 /* extents.c */ 3590 struct ext4_ext_path; 3591 struct ext4_extent; 3592 3593 /* 3594 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3595 * __le32. 3596 */ 3597 #define EXT_MAX_BLOCKS 0xffffffff 3598 3599 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3600 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3601 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3602 struct ext4_map_blocks *map, int flags); 3603 extern int ext4_ext_truncate(handle_t *, struct inode *); 3604 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3605 ext4_lblk_t end); 3606 extern void ext4_ext_init(struct super_block *); 3607 extern void ext4_ext_release(struct super_block *); 3608 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3609 loff_t len); 3610 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3611 loff_t offset, ssize_t len); 3612 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3613 ext4_io_end_t *io_end); 3614 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3615 struct ext4_map_blocks *map, int flags); 3616 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3617 int num, 3618 struct ext4_ext_path *path); 3619 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3620 struct ext4_ext_path **, 3621 struct ext4_extent *, int); 3622 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3623 struct ext4_ext_path **, 3624 int flags); 3625 extern void ext4_free_ext_path(struct ext4_ext_path *); 3626 extern int ext4_ext_check_inode(struct inode *inode); 3627 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3628 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3629 __u64 start, __u64 len); 3630 extern int ext4_get_es_cache(struct inode *inode, 3631 struct fiemap_extent_info *fieinfo, 3632 __u64 start, __u64 len); 3633 extern int ext4_ext_precache(struct inode *inode); 3634 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3635 struct inode *inode2, ext4_lblk_t lblk1, 3636 ext4_lblk_t lblk2, ext4_lblk_t count, 3637 int mark_unwritten,int *err); 3638 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3639 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3640 int check_cred, int restart_cred, 3641 int revoke_cred); 3642 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3643 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3644 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3645 int len, int unwritten, ext4_fsblk_t pblk); 3646 extern int ext4_ext_clear_bb(struct inode *inode); 3647 3648 3649 /* move_extent.c */ 3650 extern void ext4_double_down_write_data_sem(struct inode *first, 3651 struct inode *second); 3652 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3653 struct inode *donor_inode); 3654 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3655 __u64 start_orig, __u64 start_donor, 3656 __u64 len, __u64 *moved_len); 3657 3658 /* page-io.c */ 3659 extern int __init ext4_init_pageio(void); 3660 extern void ext4_exit_pageio(void); 3661 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3662 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3663 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3664 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3665 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3666 struct writeback_control *wbc); 3667 extern void ext4_end_io_rsv_work(struct work_struct *work); 3668 extern void ext4_io_submit(struct ext4_io_submit *io); 3669 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3670 size_t len); 3671 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3672 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3673 3674 /* mmp.c */ 3675 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3676 3677 /* mmp.c */ 3678 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3679 3680 /* verity.c */ 3681 extern const struct fsverity_operations ext4_verityops; 3682 3683 /* orphan.c */ 3684 extern int ext4_orphan_add(handle_t *, struct inode *); 3685 extern int ext4_orphan_del(handle_t *, struct inode *); 3686 extern void ext4_orphan_cleanup(struct super_block *sb, 3687 struct ext4_super_block *es); 3688 extern void ext4_release_orphan_info(struct super_block *sb); 3689 extern int ext4_init_orphan_info(struct super_block *sb); 3690 extern int ext4_orphan_file_empty(struct super_block *sb); 3691 extern void ext4_orphan_file_block_trigger( 3692 struct jbd2_buffer_trigger_type *triggers, 3693 struct buffer_head *bh, 3694 void *data, size_t size); 3695 3696 /* 3697 * Add new method to test whether block and inode bitmaps are properly 3698 * initialized. With uninit_bg reading the block from disk is not enough 3699 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3700 */ 3701 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3702 3703 static inline int bitmap_uptodate(struct buffer_head *bh) 3704 { 3705 return (buffer_uptodate(bh) && 3706 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3707 } 3708 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3709 { 3710 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3711 } 3712 3713 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3714 3715 /* For ioend & aio unwritten conversion wait queues */ 3716 #define EXT4_WQ_HASH_SZ 37 3717 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3718 EXT4_WQ_HASH_SZ]) 3719 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3720 3721 extern int ext4_resize_begin(struct super_block *sb); 3722 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3723 3724 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3725 struct ext4_io_end *io_end) 3726 { 3727 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3728 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3729 atomic_inc(&EXT4_I(inode)->i_unwritten); 3730 } 3731 } 3732 3733 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3734 { 3735 struct inode *inode = io_end->inode; 3736 3737 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3738 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3739 /* Wake up anyone waiting on unwritten extent conversion */ 3740 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3741 wake_up_all(ext4_ioend_wq(inode)); 3742 } 3743 } 3744 3745 extern const struct iomap_ops ext4_iomap_ops; 3746 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3747 extern const struct iomap_ops ext4_iomap_report_ops; 3748 3749 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3750 { 3751 /* 3752 * If the buffer has the write error flag, we have failed 3753 * to write out data in the block. In this case, we don't 3754 * have to read the block because we may read the old data 3755 * successfully. 3756 */ 3757 if (buffer_write_io_error(bh)) 3758 set_buffer_uptodate(bh); 3759 return buffer_uptodate(bh); 3760 } 3761 3762 #endif /* __KERNEL__ */ 3763 3764 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3765 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3766 3767 #endif /* _EXT4_H */ 3768