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 * I_DATA_SEM_EA - Used for ea_inodes only 922 */ 923 enum { 924 I_DATA_SEM_NORMAL = 0, 925 I_DATA_SEM_OTHER, 926 I_DATA_SEM_QUOTA, 927 I_DATA_SEM_EA 928 }; 929 930 931 /* 932 * fourth extended file system inode data in memory 933 */ 934 struct ext4_inode_info { 935 __le32 i_data[15]; /* unconverted */ 936 __u32 i_dtime; 937 ext4_fsblk_t i_file_acl; 938 939 /* 940 * i_block_group is the number of the block group which contains 941 * this file's inode. Constant across the lifetime of the inode, 942 * it is used for making block allocation decisions - we try to 943 * place a file's data blocks near its inode block, and new inodes 944 * near to their parent directory's inode. 945 */ 946 ext4_group_t i_block_group; 947 ext4_lblk_t i_dir_start_lookup; 948 #if (BITS_PER_LONG < 64) 949 unsigned long i_state_flags; /* Dynamic state flags */ 950 #endif 951 unsigned long i_flags; 952 953 /* 954 * Extended attributes can be read independently of the main file 955 * data. Taking i_rwsem even when reading would cause contention 956 * between readers of EAs and writers of regular file data, so 957 * instead we synchronize on xattr_sem when reading or changing 958 * EAs. 959 */ 960 struct rw_semaphore xattr_sem; 961 962 /* 963 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 964 * i_orphan is used. 965 */ 966 union { 967 struct list_head i_orphan; /* unlinked but open inodes */ 968 unsigned int i_orphan_idx; /* Index in orphan file */ 969 }; 970 971 /* Fast commit related info */ 972 973 /* For tracking dentry create updates */ 974 struct list_head i_fc_dilist; 975 struct list_head i_fc_list; /* 976 * inodes that need fast commit 977 * protected by sbi->s_fc_lock. 978 */ 979 980 /* Start of lblk range that needs to be committed in this fast commit */ 981 ext4_lblk_t i_fc_lblk_start; 982 983 /* End of lblk range that needs to be committed in this fast commit */ 984 ext4_lblk_t i_fc_lblk_len; 985 986 /* Number of ongoing updates on this inode */ 987 atomic_t i_fc_updates; 988 989 /* Fast commit wait queue for this inode */ 990 wait_queue_head_t i_fc_wait; 991 992 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 993 struct mutex i_fc_lock; 994 995 /* 996 * i_disksize keeps track of what the inode size is ON DISK, not 997 * in memory. During truncate, i_size is set to the new size by 998 * the VFS prior to calling ext4_truncate(), but the filesystem won't 999 * set i_disksize to 0 until the truncate is actually under way. 1000 * 1001 * The intent is that i_disksize always represents the blocks which 1002 * are used by this file. This allows recovery to restart truncate 1003 * on orphans if we crash during truncate. We actually write i_disksize 1004 * into the on-disk inode when writing inodes out, instead of i_size. 1005 * 1006 * The only time when i_disksize and i_size may be different is when 1007 * a truncate is in progress. The only things which change i_disksize 1008 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1009 */ 1010 loff_t i_disksize; 1011 1012 /* 1013 * i_data_sem is for serialising ext4_truncate() against 1014 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1015 * data tree are chopped off during truncate. We can't do that in 1016 * ext4 because whenever we perform intermediate commits during 1017 * truncate, the inode and all the metadata blocks *must* be in a 1018 * consistent state which allows truncation of the orphans to restart 1019 * during recovery. Hence we must fix the get_block-vs-truncate race 1020 * by other means, so we have i_data_sem. 1021 */ 1022 struct rw_semaphore i_data_sem; 1023 struct inode vfs_inode; 1024 struct jbd2_inode *jinode; 1025 1026 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1027 1028 /* 1029 * File creation time. Its function is same as that of 1030 * struct timespec64 i_{a,c,m}time in the generic inode. 1031 */ 1032 struct timespec64 i_crtime; 1033 1034 /* mballoc */ 1035 atomic_t i_prealloc_active; 1036 struct rb_root i_prealloc_node; 1037 rwlock_t i_prealloc_lock; 1038 1039 /* extents status tree */ 1040 struct ext4_es_tree i_es_tree; 1041 rwlock_t i_es_lock; 1042 struct list_head i_es_list; 1043 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1044 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1045 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1046 extents to shrink. Protected by 1047 i_es_lock */ 1048 1049 /* ialloc */ 1050 ext4_group_t i_last_alloc_group; 1051 1052 /* allocation reservation info for delalloc */ 1053 /* In case of bigalloc, this refer to clusters rather than blocks */ 1054 unsigned int i_reserved_data_blocks; 1055 1056 /* pending cluster reservations for bigalloc file systems */ 1057 struct ext4_pending_tree i_pending_tree; 1058 1059 /* on-disk additional length */ 1060 __u16 i_extra_isize; 1061 1062 /* Indicate the inline data space. */ 1063 u16 i_inline_off; 1064 u16 i_inline_size; 1065 1066 #ifdef CONFIG_QUOTA 1067 /* quota space reservation, managed internally by quota code */ 1068 qsize_t i_reserved_quota; 1069 #endif 1070 1071 /* Lock protecting lists below */ 1072 spinlock_t i_completed_io_lock; 1073 /* 1074 * Completed IOs that need unwritten extents handling and have 1075 * transaction reserved 1076 */ 1077 struct list_head i_rsv_conversion_list; 1078 struct work_struct i_rsv_conversion_work; 1079 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1080 1081 spinlock_t i_block_reservation_lock; 1082 1083 /* 1084 * Transactions that contain inode's metadata needed to complete 1085 * fsync and fdatasync, respectively. 1086 */ 1087 tid_t i_sync_tid; 1088 tid_t i_datasync_tid; 1089 1090 #ifdef CONFIG_QUOTA 1091 struct dquot *i_dquot[MAXQUOTAS]; 1092 #endif 1093 1094 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1095 __u32 i_csum_seed; 1096 1097 kprojid_t i_projid; 1098 }; 1099 1100 /* 1101 * File system states 1102 */ 1103 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1104 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1105 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1106 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1107 1108 /* 1109 * Misc. filesystem flags 1110 */ 1111 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1112 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1113 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1114 1115 /* 1116 * Mount flags set via mount options or defaults 1117 */ 1118 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1119 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1120 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1121 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1122 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1123 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1124 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1125 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1126 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1127 #ifdef CONFIG_FS_DAX 1128 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1129 #else 1130 #define EXT4_MOUNT_DAX_ALWAYS 0 1131 #endif 1132 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1133 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1134 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1135 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1136 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1137 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1138 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1139 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1140 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1141 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1142 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1143 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1144 * enable enforcement for hidden 1145 * quota files */ 1146 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1147 * enforcement for hidden quota 1148 * files */ 1149 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1150 * enforcement */ 1151 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1152 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1153 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1154 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1155 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1156 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1157 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1158 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1159 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1160 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1161 1162 /* 1163 * Mount flags set either automatically (could not be set by mount option) 1164 * based on per file system feature or property or in special cases such as 1165 * distinguishing between explicit mount option definition and default. 1166 */ 1167 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1168 specified delalloc */ 1169 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1170 size of blocksize * 8 1171 blocks */ 1172 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1173 file systems */ 1174 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1175 specified journal checksum */ 1176 1177 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1178 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1179 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1180 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1181 * scanning in mballoc 1182 */ 1183 1184 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1185 ~EXT4_MOUNT_##opt 1186 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1187 EXT4_MOUNT_##opt 1188 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1189 EXT4_MOUNT_##opt) 1190 1191 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1192 ~EXT4_MOUNT2_##opt 1193 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1194 EXT4_MOUNT2_##opt 1195 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1196 EXT4_MOUNT2_##opt) 1197 1198 #define ext4_test_and_set_bit __test_and_set_bit_le 1199 #define ext4_set_bit __set_bit_le 1200 #define ext4_set_bit_atomic ext2_set_bit_atomic 1201 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1202 #define ext4_clear_bit __clear_bit_le 1203 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1204 #define ext4_test_bit test_bit_le 1205 #define ext4_find_next_zero_bit find_next_zero_bit_le 1206 #define ext4_find_next_bit find_next_bit_le 1207 1208 extern void mb_set_bits(void *bm, int cur, int len); 1209 1210 /* 1211 * Maximal mount counts between two filesystem checks 1212 */ 1213 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1214 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1215 1216 /* 1217 * Behaviour when detecting errors 1218 */ 1219 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1220 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1221 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1222 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1223 1224 /* Metadata checksum algorithm codes */ 1225 #define EXT4_CRC32C_CHKSUM 1 1226 1227 #define EXT4_LABEL_MAX 16 1228 1229 /* 1230 * Structure of the super block 1231 */ 1232 struct ext4_super_block { 1233 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1234 __le32 s_blocks_count_lo; /* Blocks count */ 1235 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1236 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1237 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1238 __le32 s_first_data_block; /* First Data Block */ 1239 __le32 s_log_block_size; /* Block size */ 1240 __le32 s_log_cluster_size; /* Allocation cluster size */ 1241 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1242 __le32 s_clusters_per_group; /* # Clusters per group */ 1243 __le32 s_inodes_per_group; /* # Inodes per group */ 1244 __le32 s_mtime; /* Mount time */ 1245 /*30*/ __le32 s_wtime; /* Write time */ 1246 __le16 s_mnt_count; /* Mount count */ 1247 __le16 s_max_mnt_count; /* Maximal mount count */ 1248 __le16 s_magic; /* Magic signature */ 1249 __le16 s_state; /* File system state */ 1250 __le16 s_errors; /* Behaviour when detecting errors */ 1251 __le16 s_minor_rev_level; /* minor revision level */ 1252 /*40*/ __le32 s_lastcheck; /* time of last check */ 1253 __le32 s_checkinterval; /* max. time between checks */ 1254 __le32 s_creator_os; /* OS */ 1255 __le32 s_rev_level; /* Revision level */ 1256 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1257 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1258 /* 1259 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1260 * 1261 * Note: the difference between the compatible feature set and 1262 * the incompatible feature set is that if there is a bit set 1263 * in the incompatible feature set that the kernel doesn't 1264 * know about, it should refuse to mount the filesystem. 1265 * 1266 * e2fsck's requirements are more strict; if it doesn't know 1267 * about a feature in either the compatible or incompatible 1268 * feature set, it must abort and not try to meddle with 1269 * things it doesn't understand... 1270 */ 1271 __le32 s_first_ino; /* First non-reserved inode */ 1272 __le16 s_inode_size; /* size of inode structure */ 1273 __le16 s_block_group_nr; /* block group # of this superblock */ 1274 __le32 s_feature_compat; /* compatible feature set */ 1275 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1276 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1277 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1278 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1279 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1280 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1281 /* 1282 * Performance hints. Directory preallocation should only 1283 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1284 */ 1285 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1286 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1287 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1288 /* 1289 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1290 */ 1291 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1292 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1293 __le32 s_journal_dev; /* device number of journal file */ 1294 __le32 s_last_orphan; /* start of list of inodes to delete */ 1295 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1296 __u8 s_def_hash_version; /* Default hash version to use */ 1297 __u8 s_jnl_backup_type; 1298 __le16 s_desc_size; /* size of group descriptor */ 1299 /*100*/ __le32 s_default_mount_opts; 1300 __le32 s_first_meta_bg; /* First metablock block group */ 1301 __le32 s_mkfs_time; /* When the filesystem was created */ 1302 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1303 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1304 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1305 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1306 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1307 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1308 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1309 __le32 s_flags; /* Miscellaneous flags */ 1310 __le16 s_raid_stride; /* RAID stride */ 1311 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1312 __le64 s_mmp_block; /* Block for multi-mount protection */ 1313 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1314 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1315 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1316 __u8 s_encryption_level; /* versioning level for encryption */ 1317 __u8 s_reserved_pad; /* Padding to next 32bits */ 1318 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1319 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1320 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1321 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1322 snapshot's future use */ 1323 __le32 s_snapshot_list; /* inode number of the head of the 1324 on-disk snapshot list */ 1325 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1326 __le32 s_error_count; /* number of fs errors */ 1327 __le32 s_first_error_time; /* first time an error happened */ 1328 __le32 s_first_error_ino; /* inode involved in first error */ 1329 __le64 s_first_error_block; /* block involved of first error */ 1330 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1331 __le32 s_first_error_line; /* line number where error happened */ 1332 __le32 s_last_error_time; /* most recent time of an error */ 1333 __le32 s_last_error_ino; /* inode involved in last error */ 1334 __le32 s_last_error_line; /* line number where error happened */ 1335 __le64 s_last_error_block; /* block involved of last error */ 1336 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1337 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1338 __u8 s_mount_opts[64]; 1339 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1340 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1341 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1342 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1343 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1344 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1345 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1346 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1347 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1348 __u8 s_wtime_hi; 1349 __u8 s_mtime_hi; 1350 __u8 s_mkfs_time_hi; 1351 __u8 s_lastcheck_hi; 1352 __u8 s_first_error_time_hi; 1353 __u8 s_last_error_time_hi; 1354 __u8 s_first_error_errcode; 1355 __u8 s_last_error_errcode; 1356 __le16 s_encoding; /* Filename charset encoding */ 1357 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1358 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1359 __le32 s_reserved[94]; /* Padding to the end of the block */ 1360 __le32 s_checksum; /* crc32c(superblock) */ 1361 }; 1362 1363 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1364 1365 #ifdef __KERNEL__ 1366 1367 /* Number of quota types we support */ 1368 #define EXT4_MAXQUOTAS 3 1369 1370 #define EXT4_ENC_UTF8_12_1 1 1371 1372 /* Types of ext4 journal triggers */ 1373 enum ext4_journal_trigger_type { 1374 EXT4_JTR_ORPHAN_FILE, 1375 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1376 }; 1377 1378 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1379 1380 struct ext4_journal_trigger { 1381 struct jbd2_buffer_trigger_type tr_triggers; 1382 struct super_block *sb; 1383 }; 1384 1385 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1386 struct jbd2_buffer_trigger_type *trigger) 1387 { 1388 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1389 } 1390 1391 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1392 1393 /* Structure at the tail of orphan block */ 1394 struct ext4_orphan_block_tail { 1395 __le32 ob_magic; 1396 __le32 ob_checksum; 1397 }; 1398 1399 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1400 { 1401 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1402 sizeof(u32); 1403 } 1404 1405 struct ext4_orphan_block { 1406 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1407 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1408 }; 1409 1410 /* 1411 * Info about orphan file. 1412 */ 1413 struct ext4_orphan_info { 1414 int of_blocks; /* Number of orphan blocks in a file */ 1415 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1416 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1417 * file blocks */ 1418 }; 1419 1420 /* 1421 * fourth extended-fs super-block data in memory 1422 */ 1423 struct ext4_sb_info { 1424 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1425 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1426 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1427 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1428 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1429 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1430 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1431 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1432 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1433 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1434 unsigned long s_overhead; /* # of fs overhead clusters */ 1435 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1436 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1437 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1438 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1439 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1440 struct buffer_head * __rcu *s_group_desc; 1441 unsigned int s_mount_opt; 1442 unsigned int s_mount_opt2; 1443 unsigned long s_mount_flags; 1444 unsigned int s_def_mount_opt; 1445 unsigned int s_def_mount_opt2; 1446 ext4_fsblk_t s_sb_block; 1447 atomic64_t s_resv_clusters; 1448 kuid_t s_resuid; 1449 kgid_t s_resgid; 1450 unsigned short s_mount_state; 1451 unsigned short s_pad; 1452 int s_addr_per_block_bits; 1453 int s_desc_per_block_bits; 1454 int s_inode_size; 1455 int s_first_ino; 1456 unsigned int s_inode_readahead_blks; 1457 unsigned int s_inode_goal; 1458 u32 s_hash_seed[4]; 1459 int s_def_hash_version; 1460 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1461 struct percpu_counter s_freeclusters_counter; 1462 struct percpu_counter s_freeinodes_counter; 1463 struct percpu_counter s_dirs_counter; 1464 struct percpu_counter s_dirtyclusters_counter; 1465 struct percpu_counter s_sra_exceeded_retry_limit; 1466 struct blockgroup_lock *s_blockgroup_lock; 1467 struct proc_dir_entry *s_proc; 1468 struct kobject s_kobj; 1469 struct completion s_kobj_unregister; 1470 struct super_block *s_sb; 1471 struct buffer_head *s_mmp_bh; 1472 1473 /* Journaling */ 1474 struct journal_s *s_journal; 1475 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1476 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1477 struct list_head s_orphan; /* List of orphaned inodes in on disk 1478 list */ 1479 struct ext4_orphan_info s_orphan_info; 1480 unsigned long s_commit_interval; 1481 u32 s_max_batch_time; 1482 u32 s_min_batch_time; 1483 struct block_device *s_journal_bdev; 1484 #ifdef CONFIG_QUOTA 1485 /* Names of quota files with journalled quota */ 1486 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1487 int s_jquota_fmt; /* Format of quota to use */ 1488 #endif 1489 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1490 struct ext4_system_blocks __rcu *s_system_blks; 1491 1492 #ifdef EXTENTS_STATS 1493 /* ext4 extents stats */ 1494 unsigned long s_ext_min; 1495 unsigned long s_ext_max; 1496 unsigned long s_depth_max; 1497 spinlock_t s_ext_stats_lock; 1498 unsigned long s_ext_blocks; 1499 unsigned long s_ext_extents; 1500 #endif 1501 1502 /* for buddy allocator */ 1503 struct ext4_group_info ** __rcu *s_group_info; 1504 struct inode *s_buddy_cache; 1505 spinlock_t s_md_lock; 1506 unsigned short *s_mb_offsets; 1507 unsigned int *s_mb_maxs; 1508 unsigned int s_group_info_size; 1509 unsigned int s_mb_free_pending; 1510 struct list_head s_freed_data_list; /* List of blocks to be freed 1511 after commit completed */ 1512 struct list_head s_discard_list; 1513 struct work_struct s_discard_work; 1514 atomic_t s_retry_alloc_pending; 1515 struct list_head *s_mb_avg_fragment_size; 1516 rwlock_t *s_mb_avg_fragment_size_locks; 1517 struct list_head *s_mb_largest_free_orders; 1518 rwlock_t *s_mb_largest_free_orders_locks; 1519 1520 /* tunables */ 1521 unsigned long s_stripe; 1522 unsigned int s_mb_max_linear_groups; 1523 unsigned int s_mb_stream_request; 1524 unsigned int s_mb_max_to_scan; 1525 unsigned int s_mb_min_to_scan; 1526 unsigned int s_mb_stats; 1527 unsigned int s_mb_order2_reqs; 1528 unsigned int s_mb_group_prealloc; 1529 unsigned int s_max_dir_size_kb; 1530 /* where last allocation was done - for stream allocation */ 1531 unsigned long s_mb_last_group; 1532 unsigned long s_mb_last_start; 1533 unsigned int s_mb_prefetch; 1534 unsigned int s_mb_prefetch_limit; 1535 1536 /* stats for buddy allocator */ 1537 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1538 atomic_t s_bal_success; /* we found long enough chunks */ 1539 atomic_t s_bal_allocated; /* in blocks */ 1540 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1541 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1542 atomic_t s_bal_goals; /* goal hits */ 1543 atomic_t s_bal_breaks; /* too long searches */ 1544 atomic_t s_bal_2orders; /* 2^order hits */ 1545 atomic_t s_bal_cr0_bad_suggestions; 1546 atomic_t s_bal_cr1_bad_suggestions; 1547 atomic64_t s_bal_cX_groups_considered[4]; 1548 atomic64_t s_bal_cX_hits[4]; 1549 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */ 1550 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1551 atomic64_t s_mb_generation_time; 1552 atomic_t s_mb_lost_chunks; 1553 atomic_t s_mb_preallocated; 1554 atomic_t s_mb_discarded; 1555 atomic_t s_lock_busy; 1556 1557 /* locality groups */ 1558 struct ext4_locality_group __percpu *s_locality_groups; 1559 1560 /* for write statistics */ 1561 unsigned long s_sectors_written_start; 1562 u64 s_kbytes_written; 1563 1564 /* the size of zero-out chunk */ 1565 unsigned int s_extent_max_zeroout_kb; 1566 1567 unsigned int s_log_groups_per_flex; 1568 struct flex_groups * __rcu *s_flex_groups; 1569 ext4_group_t s_flex_groups_allocated; 1570 1571 /* workqueue for reserved extent conversions (buffered io) */ 1572 struct workqueue_struct *rsv_conversion_wq; 1573 1574 /* timer for periodic error stats printing */ 1575 struct timer_list s_err_report; 1576 1577 /* Lazy inode table initialization info */ 1578 struct ext4_li_request *s_li_request; 1579 /* Wait multiplier for lazy initialization thread */ 1580 unsigned int s_li_wait_mult; 1581 1582 /* Kernel thread for multiple mount protection */ 1583 struct task_struct *s_mmp_tsk; 1584 1585 /* record the last minlen when FITRIM is called. */ 1586 unsigned long s_last_trim_minblks; 1587 1588 /* Reference to checksum algorithm driver via cryptoapi */ 1589 struct crypto_shash *s_chksum_driver; 1590 1591 /* Precomputed FS UUID checksum for seeding other checksums */ 1592 __u32 s_csum_seed; 1593 1594 /* Reclaim extents from extent status tree */ 1595 struct shrinker s_es_shrinker; 1596 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1597 long s_es_nr_inode; 1598 struct ext4_es_stats s_es_stats; 1599 struct mb_cache *s_ea_block_cache; 1600 struct mb_cache *s_ea_inode_cache; 1601 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1602 1603 /* Journal triggers for checksum computation */ 1604 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1605 1606 /* Ratelimit ext4 messages. */ 1607 struct ratelimit_state s_err_ratelimit_state; 1608 struct ratelimit_state s_warning_ratelimit_state; 1609 struct ratelimit_state s_msg_ratelimit_state; 1610 atomic_t s_warning_count; 1611 atomic_t s_msg_count; 1612 1613 /* Encryption policy for '-o test_dummy_encryption' */ 1614 struct fscrypt_dummy_policy s_dummy_enc_policy; 1615 1616 /* 1617 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1618 * or EXTENTS flag. 1619 */ 1620 struct percpu_rw_semaphore s_writepages_rwsem; 1621 struct dax_device *s_daxdev; 1622 u64 s_dax_part_off; 1623 #ifdef CONFIG_EXT4_DEBUG 1624 unsigned long s_simulate_fail; 1625 #endif 1626 /* Record the errseq of the backing block device */ 1627 errseq_t s_bdev_wb_err; 1628 spinlock_t s_bdev_wb_lock; 1629 1630 /* Information about errors that happened during this mount */ 1631 spinlock_t s_error_lock; 1632 int s_add_error_count; 1633 int s_first_error_code; 1634 __u32 s_first_error_line; 1635 __u32 s_first_error_ino; 1636 __u64 s_first_error_block; 1637 const char *s_first_error_func; 1638 time64_t s_first_error_time; 1639 int s_last_error_code; 1640 __u32 s_last_error_line; 1641 __u32 s_last_error_ino; 1642 __u64 s_last_error_block; 1643 const char *s_last_error_func; 1644 time64_t s_last_error_time; 1645 /* 1646 * If we are in a context where we cannot update error information in 1647 * the on-disk superblock, we queue this work to do it. 1648 */ 1649 struct work_struct s_error_work; 1650 1651 /* Ext4 fast commit sub transaction ID */ 1652 atomic_t s_fc_subtid; 1653 1654 /* 1655 * After commit starts, the main queue gets locked, and the further 1656 * updates get added in the staging queue. 1657 */ 1658 #define FC_Q_MAIN 0 1659 #define FC_Q_STAGING 1 1660 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1661 * that have data changes in them. 1662 */ 1663 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1664 unsigned int s_fc_bytes; 1665 /* 1666 * Main fast commit lock. This lock protects accesses to the 1667 * following fields: 1668 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1669 */ 1670 spinlock_t s_fc_lock; 1671 struct buffer_head *s_fc_bh; 1672 struct ext4_fc_stats s_fc_stats; 1673 tid_t s_fc_ineligible_tid; 1674 #ifdef CONFIG_EXT4_DEBUG 1675 int s_fc_debug_max_replay; 1676 #endif 1677 struct ext4_fc_replay_state s_fc_replay_state; 1678 }; 1679 1680 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1681 { 1682 return sb->s_fs_info; 1683 } 1684 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1685 { 1686 return container_of(inode, struct ext4_inode_info, vfs_inode); 1687 } 1688 1689 static inline int ext4_writepages_down_read(struct super_block *sb) 1690 { 1691 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1692 return memalloc_nofs_save(); 1693 } 1694 1695 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1696 { 1697 memalloc_nofs_restore(ctx); 1698 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1699 } 1700 1701 static inline int ext4_writepages_down_write(struct super_block *sb) 1702 { 1703 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1704 return memalloc_nofs_save(); 1705 } 1706 1707 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1708 { 1709 memalloc_nofs_restore(ctx); 1710 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1711 } 1712 1713 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1714 { 1715 return ino == EXT4_ROOT_INO || 1716 (ino >= EXT4_FIRST_INO(sb) && 1717 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1718 } 1719 1720 /* 1721 * Returns: sbi->field[index] 1722 * Used to access an array element from the following sbi fields which require 1723 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1724 * - s_group_desc 1725 * - s_group_info 1726 * - s_flex_group 1727 */ 1728 #define sbi_array_rcu_deref(sbi, field, index) \ 1729 ({ \ 1730 typeof(*((sbi)->field)) _v; \ 1731 rcu_read_lock(); \ 1732 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1733 rcu_read_unlock(); \ 1734 _v; \ 1735 }) 1736 1737 /* 1738 * run-time mount flags 1739 */ 1740 enum { 1741 EXT4_MF_MNTDIR_SAMPLED, 1742 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1743 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1744 }; 1745 1746 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1747 { 1748 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1749 } 1750 1751 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1752 { 1753 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1754 } 1755 1756 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1757 { 1758 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1759 } 1760 1761 1762 /* 1763 * Simulate_fail codes 1764 */ 1765 #define EXT4_SIM_BBITMAP_EIO 1 1766 #define EXT4_SIM_BBITMAP_CRC 2 1767 #define EXT4_SIM_IBITMAP_EIO 3 1768 #define EXT4_SIM_IBITMAP_CRC 4 1769 #define EXT4_SIM_INODE_EIO 5 1770 #define EXT4_SIM_INODE_CRC 6 1771 #define EXT4_SIM_DIRBLOCK_EIO 7 1772 #define EXT4_SIM_DIRBLOCK_CRC 8 1773 1774 static inline bool ext4_simulate_fail(struct super_block *sb, 1775 unsigned long code) 1776 { 1777 #ifdef CONFIG_EXT4_DEBUG 1778 struct ext4_sb_info *sbi = EXT4_SB(sb); 1779 1780 if (unlikely(sbi->s_simulate_fail == code)) { 1781 sbi->s_simulate_fail = 0; 1782 return true; 1783 } 1784 #endif 1785 return false; 1786 } 1787 1788 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1789 struct buffer_head *bh, 1790 unsigned long code) 1791 { 1792 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1793 clear_buffer_uptodate(bh); 1794 } 1795 1796 /* 1797 * Error number codes for s_{first,last}_error_errno 1798 * 1799 * Linux errno numbers are architecture specific, so we need to translate 1800 * them into something which is architecture independent. We don't define 1801 * codes for all errno's; just the ones which are most likely to be the cause 1802 * of an ext4_error() call. 1803 */ 1804 #define EXT4_ERR_UNKNOWN 1 1805 #define EXT4_ERR_EIO 2 1806 #define EXT4_ERR_ENOMEM 3 1807 #define EXT4_ERR_EFSBADCRC 4 1808 #define EXT4_ERR_EFSCORRUPTED 5 1809 #define EXT4_ERR_ENOSPC 6 1810 #define EXT4_ERR_ENOKEY 7 1811 #define EXT4_ERR_EROFS 8 1812 #define EXT4_ERR_EFBIG 9 1813 #define EXT4_ERR_EEXIST 10 1814 #define EXT4_ERR_ERANGE 11 1815 #define EXT4_ERR_EOVERFLOW 12 1816 #define EXT4_ERR_EBUSY 13 1817 #define EXT4_ERR_ENOTDIR 14 1818 #define EXT4_ERR_ENOTEMPTY 15 1819 #define EXT4_ERR_ESHUTDOWN 16 1820 #define EXT4_ERR_EFAULT 17 1821 1822 /* 1823 * Inode dynamic state flags 1824 */ 1825 enum { 1826 EXT4_STATE_NEW, /* inode is newly created */ 1827 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1828 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1829 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1830 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1831 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1832 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1833 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1834 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1835 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1836 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1837 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1838 }; 1839 1840 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1841 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1842 { \ 1843 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1844 } \ 1845 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1846 { \ 1847 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1848 } \ 1849 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1850 { \ 1851 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1852 } 1853 1854 /* Add these declarations here only so that these functions can be 1855 * found by name. Otherwise, they are very hard to locate. */ 1856 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1857 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1858 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1859 EXT4_INODE_BIT_FNS(flag, flags, 0) 1860 1861 /* Add these declarations here only so that these functions can be 1862 * found by name. Otherwise, they are very hard to locate. */ 1863 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1864 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1865 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1866 #if (BITS_PER_LONG < 64) 1867 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1868 1869 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1870 { 1871 (ei)->i_state_flags = 0; 1872 } 1873 #else 1874 EXT4_INODE_BIT_FNS(state, flags, 32) 1875 1876 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1877 { 1878 /* We depend on the fact that callers will set i_flags */ 1879 } 1880 #endif 1881 #else 1882 /* Assume that user mode programs are passing in an ext4fs superblock, not 1883 * a kernel struct super_block. This will allow us to call the feature-test 1884 * macros from user land. */ 1885 #define EXT4_SB(sb) (sb) 1886 #endif 1887 1888 static inline bool ext4_verity_in_progress(struct inode *inode) 1889 { 1890 return IS_ENABLED(CONFIG_FS_VERITY) && 1891 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1892 } 1893 1894 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1895 1896 /* 1897 * Codes for operating systems 1898 */ 1899 #define EXT4_OS_LINUX 0 1900 #define EXT4_OS_HURD 1 1901 #define EXT4_OS_MASIX 2 1902 #define EXT4_OS_FREEBSD 3 1903 #define EXT4_OS_LITES 4 1904 1905 /* 1906 * Revision levels 1907 */ 1908 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1909 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1910 1911 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1912 1913 #define EXT4_GOOD_OLD_INODE_SIZE 128 1914 1915 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1916 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1917 #define EXT4_TIMESTAMP_MIN S32_MIN 1918 1919 /* 1920 * Feature set definitions 1921 */ 1922 1923 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1924 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1925 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1926 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1927 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1928 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1929 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1930 /* 1931 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1932 * incompatible only if fast commit blocks are present in the FS. Since we 1933 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1934 * incompatible. We also have a JBD2 incompat feature, which gets set when 1935 * there are fast commit blocks present in the journal. 1936 */ 1937 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1938 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1939 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 1940 1941 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1942 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1943 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1944 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1945 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1946 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1947 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1948 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1949 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1950 /* 1951 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1952 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1953 * all other data structures' checksums. However, the METADATA_CSUM and 1954 * GDT_CSUM bits are mutually exclusive. 1955 */ 1956 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1957 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1958 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1959 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1960 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 1961 non-empty */ 1962 1963 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1964 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1965 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1966 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1967 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1968 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1969 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1970 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1971 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1972 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1973 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1974 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1975 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1976 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1977 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1978 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1979 1980 extern void ext4_update_dynamic_rev(struct super_block *sb); 1981 1982 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1983 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1984 { \ 1985 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1986 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1987 } \ 1988 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1989 { \ 1990 ext4_update_dynamic_rev(sb); \ 1991 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1992 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1993 } \ 1994 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1995 { \ 1996 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1997 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1998 } 1999 2000 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2001 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2002 { \ 2003 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2004 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2005 } \ 2006 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2007 { \ 2008 ext4_update_dynamic_rev(sb); \ 2009 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2010 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2011 } \ 2012 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2013 { \ 2014 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2015 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2016 } 2017 2018 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2019 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2020 { \ 2021 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2022 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2023 } \ 2024 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2025 { \ 2026 ext4_update_dynamic_rev(sb); \ 2027 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2028 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2029 } \ 2030 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2031 { \ 2032 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2033 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2034 } 2035 2036 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2037 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2038 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2039 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2040 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2041 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2042 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2043 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2044 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2045 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2046 2047 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2048 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2049 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2050 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2051 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2052 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2053 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2054 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2055 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2056 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2057 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2058 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2059 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2060 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2061 2062 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2063 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2064 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2065 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2066 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2067 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2068 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2069 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2070 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2071 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2072 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2073 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2074 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2075 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2076 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2077 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2078 2079 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2080 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2081 EXT4_FEATURE_INCOMPAT_META_BG) 2082 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2083 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2084 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2085 2086 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2087 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2088 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2089 EXT4_FEATURE_INCOMPAT_META_BG) 2090 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2091 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2092 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2093 2094 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2095 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2096 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2097 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2098 EXT4_FEATURE_INCOMPAT_META_BG| \ 2099 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2100 EXT4_FEATURE_INCOMPAT_64BIT| \ 2101 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2102 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2103 EXT4_FEATURE_INCOMPAT_MMP | \ 2104 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2105 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2106 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2107 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2108 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2109 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2110 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2111 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2112 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2113 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2114 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2115 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2116 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2117 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2118 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2119 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2120 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2121 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2122 2123 #define EXTN_FEATURE_FUNCS(ver) \ 2124 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2125 { \ 2126 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2127 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2128 } \ 2129 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2130 { \ 2131 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2132 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2133 } \ 2134 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2135 { \ 2136 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2137 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2138 } 2139 2140 EXTN_FEATURE_FUNCS(2) 2141 EXTN_FEATURE_FUNCS(3) 2142 EXTN_FEATURE_FUNCS(4) 2143 2144 static inline bool ext4_has_compat_features(struct super_block *sb) 2145 { 2146 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2147 } 2148 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2149 { 2150 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2151 } 2152 static inline bool ext4_has_incompat_features(struct super_block *sb) 2153 { 2154 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2155 } 2156 2157 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2158 2159 /* 2160 * Superblock flags 2161 */ 2162 #define EXT4_FLAGS_RESIZING 0 2163 #define EXT4_FLAGS_SHUTDOWN 1 2164 #define EXT4_FLAGS_BDEV_IS_DAX 2 2165 2166 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2167 { 2168 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2169 } 2170 2171 /* 2172 * Default values for user and/or group using reserved blocks 2173 */ 2174 #define EXT4_DEF_RESUID 0 2175 #define EXT4_DEF_RESGID 0 2176 2177 /* 2178 * Default project ID 2179 */ 2180 #define EXT4_DEF_PROJID 0 2181 2182 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2183 2184 /* 2185 * Default mount options 2186 */ 2187 #define EXT4_DEFM_DEBUG 0x0001 2188 #define EXT4_DEFM_BSDGROUPS 0x0002 2189 #define EXT4_DEFM_XATTR_USER 0x0004 2190 #define EXT4_DEFM_ACL 0x0008 2191 #define EXT4_DEFM_UID16 0x0010 2192 #define EXT4_DEFM_JMODE 0x0060 2193 #define EXT4_DEFM_JMODE_DATA 0x0020 2194 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2195 #define EXT4_DEFM_JMODE_WBACK 0x0060 2196 #define EXT4_DEFM_NOBARRIER 0x0100 2197 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2198 #define EXT4_DEFM_DISCARD 0x0400 2199 #define EXT4_DEFM_NODELALLOC 0x0800 2200 2201 /* 2202 * Default journal batch times 2203 */ 2204 #define EXT4_DEF_MIN_BATCH_TIME 0 2205 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2206 2207 /* 2208 * Minimum number of groups in a flexgroup before we separate out 2209 * directories into the first block group of a flexgroup 2210 */ 2211 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2212 2213 /* 2214 * Structure of a directory entry 2215 */ 2216 #define EXT4_NAME_LEN 255 2217 /* 2218 * Base length of the ext4 directory entry excluding the name length 2219 */ 2220 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2221 2222 struct ext4_dir_entry { 2223 __le32 inode; /* Inode number */ 2224 __le16 rec_len; /* Directory entry length */ 2225 __le16 name_len; /* Name length */ 2226 char name[EXT4_NAME_LEN]; /* File name */ 2227 }; 2228 2229 2230 /* 2231 * Encrypted Casefolded entries require saving the hash on disk. This structure 2232 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2233 * boundary. 2234 */ 2235 struct ext4_dir_entry_hash { 2236 __le32 hash; 2237 __le32 minor_hash; 2238 }; 2239 2240 /* 2241 * The new version of the directory entry. Since EXT4 structures are 2242 * stored in intel byte order, and the name_len field could never be 2243 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2244 * file_type field. 2245 */ 2246 struct ext4_dir_entry_2 { 2247 __le32 inode; /* Inode number */ 2248 __le16 rec_len; /* Directory entry length */ 2249 __u8 name_len; /* Name length */ 2250 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2251 char name[EXT4_NAME_LEN]; /* File name */ 2252 }; 2253 2254 /* 2255 * Access the hashes at the end of ext4_dir_entry_2 2256 */ 2257 #define EXT4_DIRENT_HASHES(entry) \ 2258 ((struct ext4_dir_entry_hash *) \ 2259 (((void *)(entry)) + \ 2260 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2261 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2262 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2263 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2264 2265 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2266 { 2267 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2268 } 2269 2270 /* 2271 * This is a bogus directory entry at the end of each leaf block that 2272 * records checksums. 2273 */ 2274 struct ext4_dir_entry_tail { 2275 __le32 det_reserved_zero1; /* Pretend to be unused */ 2276 __le16 det_rec_len; /* 12 */ 2277 __u8 det_reserved_zero2; /* Zero name length */ 2278 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2279 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2280 }; 2281 2282 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2283 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2284 ((blocksize) - \ 2285 sizeof(struct ext4_dir_entry_tail)))) 2286 2287 /* 2288 * Ext4 directory file types. Only the low 3 bits are used. The 2289 * other bits are reserved for now. 2290 */ 2291 #define EXT4_FT_UNKNOWN 0 2292 #define EXT4_FT_REG_FILE 1 2293 #define EXT4_FT_DIR 2 2294 #define EXT4_FT_CHRDEV 3 2295 #define EXT4_FT_BLKDEV 4 2296 #define EXT4_FT_FIFO 5 2297 #define EXT4_FT_SOCK 6 2298 #define EXT4_FT_SYMLINK 7 2299 2300 #define EXT4_FT_MAX 8 2301 2302 #define EXT4_FT_DIR_CSUM 0xDE 2303 2304 /* 2305 * EXT4_DIR_PAD defines the directory entries boundaries 2306 * 2307 * NOTE: It must be a multiple of 4 2308 */ 2309 #define EXT4_DIR_PAD 4 2310 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2311 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2312 2313 /* 2314 * The rec_len is dependent on the type of directory. Directories that are 2315 * casefolded and encrypted need to store the hash as well, so we add room for 2316 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2317 * pass NULL for dir, as those entries do not use the extra fields. 2318 */ 2319 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2320 const struct inode *dir) 2321 { 2322 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2323 2324 if (dir && ext4_hash_in_dirent(dir)) 2325 rec_len += sizeof(struct ext4_dir_entry_hash); 2326 return (rec_len & ~EXT4_DIR_ROUND); 2327 } 2328 2329 /* 2330 * If we ever get support for fs block sizes > page_size, we'll need 2331 * to remove the #if statements in the next two functions... 2332 */ 2333 static inline unsigned int 2334 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2335 { 2336 unsigned len = le16_to_cpu(dlen); 2337 2338 #if (PAGE_SIZE >= 65536) 2339 if (len == EXT4_MAX_REC_LEN || len == 0) 2340 return blocksize; 2341 return (len & 65532) | ((len & 3) << 16); 2342 #else 2343 return len; 2344 #endif 2345 } 2346 2347 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2348 { 2349 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2350 #if (PAGE_SIZE >= 65536) 2351 if (len < 65536) 2352 return cpu_to_le16(len); 2353 if (len == blocksize) { 2354 if (blocksize == 65536) 2355 return cpu_to_le16(EXT4_MAX_REC_LEN); 2356 else 2357 return cpu_to_le16(0); 2358 } 2359 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2360 #else 2361 return cpu_to_le16(len); 2362 #endif 2363 } 2364 2365 /* 2366 * Hash Tree Directory indexing 2367 * (c) Daniel Phillips, 2001 2368 */ 2369 2370 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2371 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2372 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2373 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2374 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2375 2376 /* Legal values for the dx_root hash_version field: */ 2377 2378 #define DX_HASH_LEGACY 0 2379 #define DX_HASH_HALF_MD4 1 2380 #define DX_HASH_TEA 2 2381 #define DX_HASH_LEGACY_UNSIGNED 3 2382 #define DX_HASH_HALF_MD4_UNSIGNED 4 2383 #define DX_HASH_TEA_UNSIGNED 5 2384 #define DX_HASH_SIPHASH 6 2385 2386 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2387 const void *address, unsigned int length) 2388 { 2389 struct { 2390 struct shash_desc shash; 2391 char ctx[4]; 2392 } desc; 2393 2394 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2395 2396 desc.shash.tfm = sbi->s_chksum_driver; 2397 *(u32 *)desc.ctx = crc; 2398 2399 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2400 2401 return *(u32 *)desc.ctx; 2402 } 2403 2404 #ifdef __KERNEL__ 2405 2406 /* hash info structure used by the directory hash */ 2407 struct dx_hash_info 2408 { 2409 u32 hash; 2410 u32 minor_hash; 2411 int hash_version; 2412 u32 *seed; 2413 }; 2414 2415 2416 /* 32 and 64 bit signed EOF for dx directories */ 2417 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2418 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2419 2420 2421 /* 2422 * Control parameters used by ext4_htree_next_block 2423 */ 2424 #define HASH_NB_ALWAYS 1 2425 2426 struct ext4_filename { 2427 const struct qstr *usr_fname; 2428 struct fscrypt_str disk_name; 2429 struct dx_hash_info hinfo; 2430 #ifdef CONFIG_FS_ENCRYPTION 2431 struct fscrypt_str crypto_buf; 2432 #endif 2433 #if IS_ENABLED(CONFIG_UNICODE) 2434 struct fscrypt_str cf_name; 2435 #endif 2436 }; 2437 2438 #define fname_name(p) ((p)->disk_name.name) 2439 #define fname_usr_name(p) ((p)->usr_fname->name) 2440 #define fname_len(p) ((p)->disk_name.len) 2441 2442 /* 2443 * Describe an inode's exact location on disk and in memory 2444 */ 2445 struct ext4_iloc 2446 { 2447 struct buffer_head *bh; 2448 unsigned long offset; 2449 ext4_group_t block_group; 2450 }; 2451 2452 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2453 { 2454 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2455 } 2456 2457 static inline bool ext4_is_quota_file(struct inode *inode) 2458 { 2459 return IS_NOQUOTA(inode) && 2460 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2461 } 2462 2463 /* 2464 * This structure is stuffed into the struct file's private_data field 2465 * for directories. It is where we put information so that we can do 2466 * readdir operations in hash tree order. 2467 */ 2468 struct dir_private_info { 2469 struct rb_root root; 2470 struct rb_node *curr_node; 2471 struct fname *extra_fname; 2472 loff_t last_pos; 2473 __u32 curr_hash; 2474 __u32 curr_minor_hash; 2475 __u32 next_hash; 2476 }; 2477 2478 /* calculate the first block number of the group */ 2479 static inline ext4_fsblk_t 2480 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2481 { 2482 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2483 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2484 } 2485 2486 /* 2487 * Special error return code only used by dx_probe() and its callers. 2488 */ 2489 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2490 2491 /* htree levels for ext4 */ 2492 #define EXT4_HTREE_LEVEL_COMPAT 2 2493 #define EXT4_HTREE_LEVEL 3 2494 2495 static inline int ext4_dir_htree_level(struct super_block *sb) 2496 { 2497 return ext4_has_feature_largedir(sb) ? 2498 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2499 } 2500 2501 /* 2502 * Timeout and state flag for lazy initialization inode thread. 2503 */ 2504 #define EXT4_DEF_LI_WAIT_MULT 10 2505 #define EXT4_DEF_LI_MAX_START_DELAY 5 2506 #define EXT4_LAZYINIT_QUIT 0x0001 2507 #define EXT4_LAZYINIT_RUNNING 0x0002 2508 2509 /* 2510 * Lazy inode table initialization info 2511 */ 2512 struct ext4_lazy_init { 2513 unsigned long li_state; 2514 struct list_head li_request_list; 2515 struct mutex li_list_mtx; 2516 }; 2517 2518 enum ext4_li_mode { 2519 EXT4_LI_MODE_PREFETCH_BBITMAP, 2520 EXT4_LI_MODE_ITABLE, 2521 }; 2522 2523 struct ext4_li_request { 2524 struct super_block *lr_super; 2525 enum ext4_li_mode lr_mode; 2526 ext4_group_t lr_first_not_zeroed; 2527 ext4_group_t lr_next_group; 2528 struct list_head lr_request; 2529 unsigned long lr_next_sched; 2530 unsigned long lr_timeout; 2531 }; 2532 2533 struct ext4_features { 2534 struct kobject f_kobj; 2535 struct completion f_kobj_unregister; 2536 }; 2537 2538 /* 2539 * This structure will be used for multiple mount protection. It will be 2540 * written into the block number saved in the s_mmp_block field in the 2541 * superblock. Programs that check MMP should assume that if 2542 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2543 * to use the filesystem, regardless of how old the timestamp is. 2544 */ 2545 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2546 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2547 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2548 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2549 2550 struct mmp_struct { 2551 __le32 mmp_magic; /* Magic number for MMP */ 2552 __le32 mmp_seq; /* Sequence no. updated periodically */ 2553 2554 /* 2555 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2556 * purposes and do not affect the correctness of the algorithm 2557 */ 2558 __le64 mmp_time; /* Time last updated */ 2559 char mmp_nodename[64]; /* Node which last updated MMP block */ 2560 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2561 2562 /* 2563 * mmp_check_interval is used to verify if the MMP block has been 2564 * updated on the block device. The value is updated based on the 2565 * maximum time to write the MMP block during an update cycle. 2566 */ 2567 __le16 mmp_check_interval; 2568 2569 __le16 mmp_pad1; 2570 __le32 mmp_pad2[226]; 2571 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2572 }; 2573 2574 /* arguments passed to the mmp thread */ 2575 struct mmpd_data { 2576 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2577 struct super_block *sb; /* super block of the fs */ 2578 }; 2579 2580 /* 2581 * Check interval multiplier 2582 * The MMP block is written every update interval and initially checked every 2583 * update interval x the multiplier (the value is then adapted based on the 2584 * write latency). The reason is that writes can be delayed under load and we 2585 * don't want readers to incorrectly assume that the filesystem is no longer 2586 * in use. 2587 */ 2588 #define EXT4_MMP_CHECK_MULT 2UL 2589 2590 /* 2591 * Minimum interval for MMP checking in seconds. 2592 */ 2593 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2594 2595 /* 2596 * Maximum interval for MMP checking in seconds. 2597 */ 2598 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2599 2600 /* 2601 * Function prototypes 2602 */ 2603 2604 /* 2605 * Ok, these declarations are also in <linux/kernel.h> but none of the 2606 * ext4 source programs needs to include it so they are duplicated here. 2607 */ 2608 # define NORET_TYPE /**/ 2609 # define ATTRIB_NORET __attribute__((noreturn)) 2610 # define NORET_AND noreturn, 2611 2612 /* bitmap.c */ 2613 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2614 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2615 struct ext4_group_desc *gdp, 2616 struct buffer_head *bh, int sz); 2617 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2618 struct ext4_group_desc *gdp, 2619 struct buffer_head *bh, int sz); 2620 void ext4_block_bitmap_csum_set(struct super_block *sb, 2621 struct ext4_group_desc *gdp, 2622 struct buffer_head *bh); 2623 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2624 struct ext4_group_desc *gdp, 2625 struct buffer_head *bh); 2626 2627 /* balloc.c */ 2628 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2629 ext4_fsblk_t blocknr, 2630 ext4_group_t *blockgrpp, 2631 ext4_grpblk_t *offsetp); 2632 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2633 ext4_fsblk_t block); 2634 2635 extern unsigned int ext4_block_group(struct super_block *sb, 2636 ext4_fsblk_t blocknr); 2637 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2638 ext4_fsblk_t blocknr); 2639 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2640 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2641 ext4_group_t group); 2642 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2643 ext4_fsblk_t goal, 2644 unsigned int flags, 2645 unsigned long *count, 2646 int *errp); 2647 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2648 s64 nclusters, unsigned int flags); 2649 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2650 extern void ext4_check_blocks_bitmap(struct super_block *); 2651 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2652 ext4_group_t block_group, 2653 struct buffer_head ** bh); 2654 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2655 ext4_group_t group); 2656 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2657 2658 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2659 ext4_group_t block_group, 2660 bool ignore_locked); 2661 extern int ext4_wait_block_bitmap(struct super_block *sb, 2662 ext4_group_t block_group, 2663 struct buffer_head *bh); 2664 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2665 ext4_group_t block_group); 2666 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2667 ext4_group_t block_group, 2668 struct ext4_group_desc *gdp); 2669 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2670 2671 #if IS_ENABLED(CONFIG_UNICODE) 2672 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2673 const struct qstr *iname, 2674 struct ext4_filename *fname); 2675 #endif 2676 2677 /* ext4 encryption related stuff goes here crypto.c */ 2678 #ifdef CONFIG_FS_ENCRYPTION 2679 extern const struct fscrypt_operations ext4_cryptops; 2680 2681 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2682 int lookup, struct ext4_filename *fname); 2683 2684 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2685 struct ext4_filename *fname); 2686 2687 void ext4_fname_free_filename(struct ext4_filename *fname); 2688 2689 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2690 2691 #else /* !CONFIG_FS_ENCRYPTION */ 2692 static inline int ext4_fname_setup_filename(struct inode *dir, 2693 const struct qstr *iname, 2694 int lookup, 2695 struct ext4_filename *fname) 2696 { 2697 int err = 0; 2698 fname->usr_fname = iname; 2699 fname->disk_name.name = (unsigned char *) iname->name; 2700 fname->disk_name.len = iname->len; 2701 2702 #if IS_ENABLED(CONFIG_UNICODE) 2703 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2704 #endif 2705 2706 return err; 2707 } 2708 2709 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2710 struct dentry *dentry, 2711 struct ext4_filename *fname) 2712 { 2713 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2714 } 2715 2716 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2717 { 2718 #if IS_ENABLED(CONFIG_UNICODE) 2719 kfree(fname->cf_name.name); 2720 fname->cf_name.name = NULL; 2721 #endif 2722 } 2723 2724 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2725 void __user *arg) 2726 { 2727 return -EOPNOTSUPP; 2728 } 2729 #endif /* !CONFIG_FS_ENCRYPTION */ 2730 2731 /* dir.c */ 2732 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2733 struct file *, 2734 struct ext4_dir_entry_2 *, 2735 struct buffer_head *, char *, int, 2736 unsigned int); 2737 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2738 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2739 (de), (bh), (buf), (size), (offset))) 2740 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2741 __u32 minor_hash, 2742 struct ext4_dir_entry_2 *dirent, 2743 struct fscrypt_str *ent_name); 2744 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2745 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2746 struct buffer_head *bh, 2747 void *buf, int buf_size, 2748 struct ext4_filename *fname, 2749 struct ext4_dir_entry_2 **dest_de); 2750 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2751 struct ext4_dir_entry_2 *de, 2752 int buf_size, 2753 struct ext4_filename *fname); 2754 static inline void ext4_update_dx_flag(struct inode *inode) 2755 { 2756 if (!ext4_has_feature_dir_index(inode->i_sb) && 2757 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2758 /* ext4_iget() should have caught this... */ 2759 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2760 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2761 } 2762 } 2763 static const unsigned char ext4_filetype_table[] = { 2764 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2765 }; 2766 2767 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2768 { 2769 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2770 return DT_UNKNOWN; 2771 2772 return ext4_filetype_table[filetype]; 2773 } 2774 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2775 void *buf, int buf_size); 2776 2777 /* fsync.c */ 2778 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2779 2780 /* hash.c */ 2781 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2782 struct dx_hash_info *hinfo); 2783 2784 /* ialloc.c */ 2785 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2786 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2787 struct inode *, umode_t, 2788 const struct qstr *qstr, __u32 goal, 2789 uid_t *owner, __u32 i_flags, 2790 int handle_type, unsigned int line_no, 2791 int nblocks); 2792 2793 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2794 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2795 (goal), (owner), i_flags, 0, 0, 0) 2796 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2797 type, nblocks) \ 2798 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2799 0, (type), __LINE__, (nblocks)) 2800 2801 2802 extern void ext4_free_inode(handle_t *, struct inode *); 2803 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2804 extern unsigned long ext4_count_free_inodes(struct super_block *); 2805 extern unsigned long ext4_count_dirs(struct super_block *); 2806 extern void ext4_check_inodes_bitmap(struct super_block *); 2807 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2808 extern int ext4_init_inode_table(struct super_block *sb, 2809 ext4_group_t group, int barrier); 2810 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2811 2812 /* fast_commit.c */ 2813 int ext4_fc_info_show(struct seq_file *seq, void *v); 2814 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2815 void ext4_fc_init_inode(struct inode *inode); 2816 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2817 ext4_lblk_t end); 2818 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2819 struct dentry *dentry); 2820 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2821 struct dentry *dentry); 2822 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2823 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2824 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2825 struct dentry *dentry); 2826 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2827 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2828 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2829 void ext4_fc_start_update(struct inode *inode); 2830 void ext4_fc_stop_update(struct inode *inode); 2831 void ext4_fc_del(struct inode *inode); 2832 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2833 void ext4_fc_replay_cleanup(struct super_block *sb); 2834 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2835 int __init ext4_fc_init_dentry_cache(void); 2836 void ext4_fc_destroy_dentry_cache(void); 2837 int ext4_fc_record_regions(struct super_block *sb, int ino, 2838 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2839 int len, int replay); 2840 2841 /* mballoc.c */ 2842 extern const struct seq_operations ext4_mb_seq_groups_ops; 2843 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2844 extern long ext4_mb_stats; 2845 extern long ext4_mb_max_to_scan; 2846 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2847 extern int ext4_mb_init(struct super_block *); 2848 extern int ext4_mb_release(struct super_block *); 2849 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2850 struct ext4_allocation_request *, int *); 2851 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2852 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2853 extern int __init ext4_init_mballoc(void); 2854 extern void ext4_exit_mballoc(void); 2855 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2856 ext4_group_t group, 2857 unsigned int nr, int *cnt); 2858 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2859 unsigned int nr); 2860 2861 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2862 struct buffer_head *bh, ext4_fsblk_t block, 2863 unsigned long count, int flags); 2864 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2865 ext4_group_t ngroups); 2866 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2867 ext4_group_t i, struct ext4_group_desc *desc); 2868 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2869 ext4_fsblk_t block, unsigned long count); 2870 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2871 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2872 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2873 int len, int state); 2874 2875 /* inode.c */ 2876 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2877 struct ext4_inode_info *ei); 2878 int ext4_inode_is_fast_symlink(struct inode *inode); 2879 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2880 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2881 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2882 bool wait, struct buffer_head **bhs); 2883 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2884 struct buffer_head *bh_result, int create); 2885 int ext4_get_block(struct inode *inode, sector_t iblock, 2886 struct buffer_head *bh_result, int create); 2887 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2888 struct buffer_head *bh, int create); 2889 int ext4_walk_page_buffers(handle_t *handle, 2890 struct inode *inode, 2891 struct buffer_head *head, 2892 unsigned from, 2893 unsigned to, 2894 int *partial, 2895 int (*fn)(handle_t *handle, struct inode *inode, 2896 struct buffer_head *bh)); 2897 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2898 struct buffer_head *bh); 2899 #define FALL_BACK_TO_NONDELALLOC 1 2900 #define CONVERT_INLINE_DATA 2 2901 2902 typedef enum { 2903 EXT4_IGET_NORMAL = 0, 2904 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2905 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2906 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 2907 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 2908 } ext4_iget_flags; 2909 2910 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2911 ext4_iget_flags flags, const char *function, 2912 unsigned int line); 2913 2914 #define ext4_iget(sb, ino, flags) \ 2915 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2916 2917 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2918 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2919 struct iattr *); 2920 extern u32 ext4_dio_alignment(struct inode *inode); 2921 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2922 struct kstat *, u32, unsigned int); 2923 extern void ext4_evict_inode(struct inode *); 2924 extern void ext4_clear_inode(struct inode *); 2925 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2926 struct kstat *, u32, unsigned int); 2927 extern int ext4_sync_inode(handle_t *, struct inode *); 2928 extern void ext4_dirty_inode(struct inode *, int); 2929 extern int ext4_change_inode_journal_flag(struct inode *, int); 2930 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2931 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2932 struct ext4_iloc *iloc); 2933 extern int ext4_inode_attach_jinode(struct inode *inode); 2934 extern int ext4_can_truncate(struct inode *inode); 2935 extern int ext4_truncate(struct inode *); 2936 extern int ext4_break_layouts(struct inode *); 2937 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2938 extern void ext4_set_inode_flags(struct inode *, bool init); 2939 extern int ext4_alloc_da_blocks(struct inode *inode); 2940 extern void ext4_set_aops(struct inode *inode); 2941 extern int ext4_writepage_trans_blocks(struct inode *); 2942 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 2943 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2944 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2945 loff_t lstart, loff_t lend); 2946 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2947 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2948 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2949 extern void ext4_da_release_space(struct inode *inode, int to_free); 2950 extern void ext4_da_update_reserve_space(struct inode *inode, 2951 int used, int quota_claim); 2952 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2953 ext4_fsblk_t pblk, ext4_lblk_t len); 2954 2955 /* indirect.c */ 2956 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2957 struct ext4_map_blocks *map, int flags); 2958 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2959 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2960 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2961 ext4_lblk_t start, ext4_lblk_t end); 2962 2963 /* ioctl.c */ 2964 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2965 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2966 int ext4_fileattr_set(struct mnt_idmap *idmap, 2967 struct dentry *dentry, struct fileattr *fa); 2968 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 2969 extern void ext4_reset_inode_seed(struct inode *inode); 2970 int ext4_update_overhead(struct super_block *sb, bool force); 2971 2972 /* migrate.c */ 2973 extern int ext4_ext_migrate(struct inode *); 2974 extern int ext4_ind_migrate(struct inode *inode); 2975 2976 /* namei.c */ 2977 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 2978 struct inode *inode); 2979 extern int ext4_dirblock_csum_verify(struct inode *inode, 2980 struct buffer_head *bh); 2981 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2982 __u32 start_minor_hash, __u32 *next_hash); 2983 extern int ext4_search_dir(struct buffer_head *bh, 2984 char *search_buf, 2985 int buf_size, 2986 struct inode *dir, 2987 struct ext4_filename *fname, 2988 unsigned int offset, 2989 struct ext4_dir_entry_2 **res_dir); 2990 extern int ext4_generic_delete_entry(struct inode *dir, 2991 struct ext4_dir_entry_2 *de_del, 2992 struct buffer_head *bh, 2993 void *entry_buf, 2994 int buf_size, 2995 int csum_size); 2996 extern bool ext4_empty_dir(struct inode *inode); 2997 2998 /* resize.c */ 2999 extern void ext4_kvfree_array_rcu(void *to_free); 3000 extern int ext4_group_add(struct super_block *sb, 3001 struct ext4_new_group_data *input); 3002 extern int ext4_group_extend(struct super_block *sb, 3003 struct ext4_super_block *es, 3004 ext4_fsblk_t n_blocks_count); 3005 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3006 extern unsigned int ext4_list_backups(struct super_block *sb, 3007 unsigned int *three, unsigned int *five, 3008 unsigned int *seven); 3009 3010 /* super.c */ 3011 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3012 sector_t block, blk_opf_t op_flags); 3013 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3014 sector_t block); 3015 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3016 bh_end_io_t *end_io); 3017 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3018 bh_end_io_t *end_io); 3019 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3020 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3021 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3022 extern int ext4_calculate_overhead(struct super_block *sb); 3023 extern __le32 ext4_superblock_csum(struct super_block *sb, 3024 struct ext4_super_block *es); 3025 extern void ext4_superblock_csum_set(struct super_block *sb); 3026 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3027 ext4_group_t ngroup); 3028 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3029 char nbuf[16]); 3030 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3031 ext4_group_t block_group, 3032 unsigned int flags); 3033 3034 extern __printf(7, 8) 3035 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3036 int, __u64, const char *, ...); 3037 extern __printf(6, 7) 3038 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3039 ext4_fsblk_t, int, const char *, ...); 3040 extern __printf(5, 6) 3041 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3042 const char *, ...); 3043 extern void __ext4_std_error(struct super_block *, const char *, 3044 unsigned int, int); 3045 extern __printf(4, 5) 3046 void __ext4_warning(struct super_block *, const char *, unsigned int, 3047 const char *, ...); 3048 extern __printf(4, 5) 3049 void __ext4_warning_inode(const struct inode *inode, const char *function, 3050 unsigned int line, const char *fmt, ...); 3051 extern __printf(3, 4) 3052 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3053 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3054 const char *, unsigned int, const char *); 3055 extern __printf(7, 8) 3056 void __ext4_grp_locked_error(const char *, unsigned int, 3057 struct super_block *, ext4_group_t, 3058 unsigned long, ext4_fsblk_t, 3059 const char *, ...); 3060 3061 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3062 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3063 3064 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3065 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3066 3067 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3068 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3069 (fmt), ## a) 3070 3071 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3072 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3073 3074 #define ext4_abort(sb, err, fmt, a...) \ 3075 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3076 3077 #ifdef CONFIG_PRINTK 3078 3079 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3080 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3081 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3082 __ext4_error_inode((inode), (func), (line), (block), \ 3083 (err), (fmt), ##__VA_ARGS__) 3084 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3085 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3086 #define ext4_error(sb, fmt, ...) \ 3087 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3088 ##__VA_ARGS__) 3089 #define ext4_error_err(sb, err, fmt, ...) \ 3090 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3091 ##__VA_ARGS__) 3092 #define ext4_warning(sb, fmt, ...) \ 3093 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3094 #define ext4_warning_inode(inode, fmt, ...) \ 3095 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3096 #define ext4_msg(sb, level, fmt, ...) \ 3097 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3098 #define dump_mmp_msg(sb, mmp, msg) \ 3099 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3100 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3101 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3102 fmt, ##__VA_ARGS__) 3103 3104 #else 3105 3106 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3107 do { \ 3108 no_printk(fmt, ##__VA_ARGS__); \ 3109 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3110 } while (0) 3111 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3112 do { \ 3113 no_printk(fmt, ##__VA_ARGS__); \ 3114 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3115 } while (0) 3116 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3117 do { \ 3118 no_printk(fmt, ##__VA_ARGS__); \ 3119 __ext4_error_file(file, "", 0, block, " "); \ 3120 } while (0) 3121 #define ext4_error(sb, fmt, ...) \ 3122 do { \ 3123 no_printk(fmt, ##__VA_ARGS__); \ 3124 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3125 } while (0) 3126 #define ext4_error_err(sb, err, fmt, ...) \ 3127 do { \ 3128 no_printk(fmt, ##__VA_ARGS__); \ 3129 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3130 } while (0) 3131 #define ext4_warning(sb, fmt, ...) \ 3132 do { \ 3133 no_printk(fmt, ##__VA_ARGS__); \ 3134 __ext4_warning(sb, "", 0, " "); \ 3135 } while (0) 3136 #define ext4_warning_inode(inode, fmt, ...) \ 3137 do { \ 3138 no_printk(fmt, ##__VA_ARGS__); \ 3139 __ext4_warning_inode(inode, "", 0, " "); \ 3140 } while (0) 3141 #define ext4_msg(sb, level, fmt, ...) \ 3142 do { \ 3143 no_printk(fmt, ##__VA_ARGS__); \ 3144 __ext4_msg(sb, "", " "); \ 3145 } while (0) 3146 #define dump_mmp_msg(sb, mmp, msg) \ 3147 __dump_mmp_msg(sb, mmp, "", 0, "") 3148 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3149 do { \ 3150 no_printk(fmt, ##__VA_ARGS__); \ 3151 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3152 } while (0) 3153 3154 #endif 3155 3156 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3157 struct ext4_group_desc *bg); 3158 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3159 struct ext4_group_desc *bg); 3160 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3161 struct ext4_group_desc *bg); 3162 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3163 struct ext4_group_desc *bg); 3164 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3165 struct ext4_group_desc *bg); 3166 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3167 struct ext4_group_desc *bg); 3168 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3169 struct ext4_group_desc *bg); 3170 extern void ext4_block_bitmap_set(struct super_block *sb, 3171 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3172 extern void ext4_inode_bitmap_set(struct super_block *sb, 3173 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3174 extern void ext4_inode_table_set(struct super_block *sb, 3175 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3176 extern void ext4_free_group_clusters_set(struct super_block *sb, 3177 struct ext4_group_desc *bg, 3178 __u32 count); 3179 extern void ext4_free_inodes_set(struct super_block *sb, 3180 struct ext4_group_desc *bg, __u32 count); 3181 extern void ext4_used_dirs_set(struct super_block *sb, 3182 struct ext4_group_desc *bg, __u32 count); 3183 extern void ext4_itable_unused_set(struct super_block *sb, 3184 struct ext4_group_desc *bg, __u32 count); 3185 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3186 struct ext4_group_desc *gdp); 3187 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3188 struct ext4_group_desc *gdp); 3189 extern int ext4_register_li_request(struct super_block *sb, 3190 ext4_group_t first_not_zeroed); 3191 3192 static inline int ext4_has_metadata_csum(struct super_block *sb) 3193 { 3194 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3195 !EXT4_SB(sb)->s_chksum_driver); 3196 3197 return ext4_has_feature_metadata_csum(sb) && 3198 (EXT4_SB(sb)->s_chksum_driver != NULL); 3199 } 3200 3201 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3202 { 3203 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3204 } 3205 3206 #define ext4_read_incompat_64bit_val(es, name) \ 3207 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3208 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3209 le32_to_cpu(es->name##_lo)) 3210 3211 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3212 { 3213 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3214 } 3215 3216 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3217 { 3218 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3219 } 3220 3221 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3222 { 3223 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3224 } 3225 3226 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3227 ext4_fsblk_t blk) 3228 { 3229 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3230 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3231 } 3232 3233 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3234 ext4_fsblk_t blk) 3235 { 3236 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3237 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3238 } 3239 3240 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3241 ext4_fsblk_t blk) 3242 { 3243 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3244 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3245 } 3246 3247 static inline loff_t ext4_isize(struct super_block *sb, 3248 struct ext4_inode *raw_inode) 3249 { 3250 if (ext4_has_feature_largedir(sb) || 3251 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3252 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3253 le32_to_cpu(raw_inode->i_size_lo); 3254 3255 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3256 } 3257 3258 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3259 { 3260 raw_inode->i_size_lo = cpu_to_le32(i_size); 3261 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3262 } 3263 3264 /* 3265 * Reading s_groups_count requires using smp_rmb() afterwards. See 3266 * the locking protocol documented in the comments of ext4_group_add() 3267 * in resize.c 3268 */ 3269 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3270 { 3271 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3272 3273 smp_rmb(); 3274 return ngroups; 3275 } 3276 3277 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3278 ext4_group_t block_group) 3279 { 3280 return block_group >> sbi->s_log_groups_per_flex; 3281 } 3282 3283 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3284 { 3285 return 1 << sbi->s_log_groups_per_flex; 3286 } 3287 3288 #define ext4_std_error(sb, errno) \ 3289 do { \ 3290 if ((errno)) \ 3291 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3292 } while (0) 3293 3294 #ifdef CONFIG_SMP 3295 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3296 * counters. So we need to make sure we have free clusters more 3297 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3298 */ 3299 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3300 #else 3301 #define EXT4_FREECLUSTERS_WATERMARK 0 3302 #endif 3303 3304 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3305 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3306 { 3307 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3308 !inode_is_locked(inode)); 3309 down_write(&EXT4_I(inode)->i_data_sem); 3310 if (newsize > EXT4_I(inode)->i_disksize) 3311 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3312 up_write(&EXT4_I(inode)->i_data_sem); 3313 } 3314 3315 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3316 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3317 { 3318 int changed = 0; 3319 3320 if (newsize > inode->i_size) { 3321 i_size_write(inode, newsize); 3322 changed = 1; 3323 } 3324 if (newsize > EXT4_I(inode)->i_disksize) { 3325 ext4_update_i_disksize(inode, newsize); 3326 changed |= 2; 3327 } 3328 return changed; 3329 } 3330 3331 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3332 loff_t len); 3333 3334 struct ext4_group_info { 3335 unsigned long bb_state; 3336 #ifdef AGGRESSIVE_CHECK 3337 unsigned long bb_check_counter; 3338 #endif 3339 struct rb_root bb_free_root; 3340 ext4_grpblk_t bb_first_free; /* first free block */ 3341 ext4_grpblk_t bb_free; /* total free blocks */ 3342 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3343 int bb_avg_fragment_size_order; /* order of average 3344 fragment in BG */ 3345 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3346 ext4_group_t bb_group; /* Group number */ 3347 struct list_head bb_prealloc_list; 3348 #ifdef DOUBLE_CHECK 3349 void *bb_bitmap; 3350 #endif 3351 struct rw_semaphore alloc_sem; 3352 struct list_head bb_avg_fragment_size_node; 3353 struct list_head bb_largest_free_order_node; 3354 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3355 * regions, index is order. 3356 * bb_counters[3] = 5 means 3357 * 5 free 8-block regions. */ 3358 }; 3359 3360 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3361 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3362 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3363 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3364 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3365 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3366 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3367 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3368 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3369 3370 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3371 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3372 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3373 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3374 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3375 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3376 3377 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3378 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3379 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3380 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3381 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3382 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3383 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3384 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3385 3386 #define EXT4_MAX_CONTENTION 8 3387 #define EXT4_CONTENTION_THRESHOLD 2 3388 3389 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3390 ext4_group_t group) 3391 { 3392 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3393 } 3394 3395 /* 3396 * Returns true if the filesystem is busy enough that attempts to 3397 * access the block group locks has run into contention. 3398 */ 3399 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3400 { 3401 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3402 } 3403 3404 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3405 { 3406 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3407 if (spin_trylock(lock)) 3408 /* 3409 * We're able to grab the lock right away, so drop the 3410 * lock contention counter. 3411 */ 3412 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3413 else { 3414 /* 3415 * The lock is busy, so bump the contention counter, 3416 * and then wait on the spin lock. 3417 */ 3418 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3419 EXT4_MAX_CONTENTION); 3420 spin_lock(lock); 3421 } 3422 } 3423 3424 static inline void ext4_unlock_group(struct super_block *sb, 3425 ext4_group_t group) 3426 { 3427 spin_unlock(ext4_group_lock_ptr(sb, group)); 3428 } 3429 3430 #ifdef CONFIG_QUOTA 3431 static inline bool ext4_quota_capable(struct super_block *sb) 3432 { 3433 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3434 } 3435 3436 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3437 { 3438 struct ext4_sb_info *sbi = EXT4_SB(sb); 3439 3440 return (ext4_has_feature_quota(sb) || 3441 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3442 } 3443 int ext4_enable_quotas(struct super_block *sb); 3444 #endif 3445 3446 /* 3447 * Block validity checking 3448 */ 3449 #define ext4_check_indirect_blockref(inode, bh) \ 3450 ext4_check_blockref(__func__, __LINE__, inode, \ 3451 (__le32 *)(bh)->b_data, \ 3452 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3453 3454 #define ext4_ind_check_inode(inode) \ 3455 ext4_check_blockref(__func__, __LINE__, inode, \ 3456 EXT4_I(inode)->i_data, \ 3457 EXT4_NDIR_BLOCKS) 3458 3459 /* 3460 * Inodes and files operations 3461 */ 3462 3463 /* dir.c */ 3464 extern const struct file_operations ext4_dir_operations; 3465 3466 /* file.c */ 3467 extern const struct inode_operations ext4_file_inode_operations; 3468 extern const struct file_operations ext4_file_operations; 3469 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3470 3471 /* inline.c */ 3472 extern int ext4_get_max_inline_size(struct inode *inode); 3473 extern int ext4_find_inline_data_nolock(struct inode *inode); 3474 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3475 unsigned int len); 3476 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3477 3478 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3479 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3480 struct inode *inode, 3481 loff_t pos, unsigned len, 3482 struct page **pagep); 3483 extern int ext4_write_inline_data_end(struct inode *inode, 3484 loff_t pos, unsigned len, 3485 unsigned copied, 3486 struct page *page); 3487 extern struct buffer_head * 3488 ext4_journalled_write_inline_data(struct inode *inode, 3489 unsigned len, 3490 struct page *page); 3491 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3492 struct inode *inode, 3493 loff_t pos, unsigned len, 3494 struct page **pagep, 3495 void **fsdata); 3496 extern int ext4_try_add_inline_entry(handle_t *handle, 3497 struct ext4_filename *fname, 3498 struct inode *dir, struct inode *inode); 3499 extern int ext4_try_create_inline_dir(handle_t *handle, 3500 struct inode *parent, 3501 struct inode *inode); 3502 extern int ext4_read_inline_dir(struct file *filp, 3503 struct dir_context *ctx, 3504 int *has_inline_data); 3505 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3506 struct inode *dir, ext4_lblk_t block, 3507 struct dx_hash_info *hinfo, 3508 __u32 start_hash, __u32 start_minor_hash, 3509 int *has_inline_data); 3510 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3511 struct ext4_filename *fname, 3512 struct ext4_dir_entry_2 **res_dir, 3513 int *has_inline_data); 3514 extern int ext4_delete_inline_entry(handle_t *handle, 3515 struct inode *dir, 3516 struct ext4_dir_entry_2 *de_del, 3517 struct buffer_head *bh, 3518 int *has_inline_data); 3519 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3520 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3521 struct ext4_dir_entry_2 **parent_de, 3522 int *retval); 3523 extern void *ext4_read_inline_link(struct inode *inode); 3524 3525 struct iomap; 3526 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3527 3528 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3529 3530 extern int ext4_convert_inline_data(struct inode *inode); 3531 3532 static inline int ext4_has_inline_data(struct inode *inode) 3533 { 3534 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3535 EXT4_I(inode)->i_inline_off; 3536 } 3537 3538 /* namei.c */ 3539 extern const struct inode_operations ext4_dir_inode_operations; 3540 extern const struct inode_operations ext4_special_inode_operations; 3541 extern struct dentry *ext4_get_parent(struct dentry *child); 3542 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3543 struct ext4_dir_entry_2 *de, 3544 int blocksize, int csum_size, 3545 unsigned int parent_ino, int dotdot_real_len); 3546 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3547 unsigned int blocksize); 3548 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3549 struct buffer_head *bh); 3550 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3551 struct inode *inode, struct dentry *dentry); 3552 extern int __ext4_link(struct inode *dir, struct inode *inode, 3553 struct dentry *dentry); 3554 3555 #define S_SHIFT 12 3556 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3557 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3558 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3559 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3560 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3561 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3562 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3563 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3564 }; 3565 3566 static inline void ext4_set_de_type(struct super_block *sb, 3567 struct ext4_dir_entry_2 *de, 3568 umode_t mode) { 3569 if (ext4_has_feature_filetype(sb)) 3570 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3571 } 3572 3573 /* readpages.c */ 3574 extern int ext4_mpage_readpages(struct inode *inode, 3575 struct readahead_control *rac, struct folio *folio); 3576 extern int __init ext4_init_post_read_processing(void); 3577 extern void ext4_exit_post_read_processing(void); 3578 3579 /* symlink.c */ 3580 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3581 extern const struct inode_operations ext4_symlink_inode_operations; 3582 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3583 3584 /* sysfs.c */ 3585 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3586 extern int ext4_register_sysfs(struct super_block *sb); 3587 extern void ext4_unregister_sysfs(struct super_block *sb); 3588 extern int __init ext4_init_sysfs(void); 3589 extern void ext4_exit_sysfs(void); 3590 3591 /* block_validity */ 3592 extern void ext4_release_system_zone(struct super_block *sb); 3593 extern int ext4_setup_system_zone(struct super_block *sb); 3594 extern int __init ext4_init_system_zone(void); 3595 extern void ext4_exit_system_zone(void); 3596 extern int ext4_inode_block_valid(struct inode *inode, 3597 ext4_fsblk_t start_blk, 3598 unsigned int count); 3599 extern int ext4_check_blockref(const char *, unsigned int, 3600 struct inode *, __le32 *, unsigned int); 3601 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3602 ext4_fsblk_t start_blk, unsigned int count); 3603 3604 3605 /* extents.c */ 3606 struct ext4_ext_path; 3607 struct ext4_extent; 3608 3609 /* 3610 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3611 * __le32. 3612 */ 3613 #define EXT_MAX_BLOCKS 0xffffffff 3614 3615 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3616 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3617 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3618 struct ext4_map_blocks *map, int flags); 3619 extern int ext4_ext_truncate(handle_t *, struct inode *); 3620 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3621 ext4_lblk_t end); 3622 extern void ext4_ext_init(struct super_block *); 3623 extern void ext4_ext_release(struct super_block *); 3624 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3625 loff_t len); 3626 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3627 loff_t offset, ssize_t len); 3628 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3629 ext4_io_end_t *io_end); 3630 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3631 struct ext4_map_blocks *map, int flags); 3632 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3633 int num, 3634 struct ext4_ext_path *path); 3635 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3636 struct ext4_ext_path **, 3637 struct ext4_extent *, int); 3638 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3639 struct ext4_ext_path **, 3640 int flags); 3641 extern void ext4_free_ext_path(struct ext4_ext_path *); 3642 extern int ext4_ext_check_inode(struct inode *inode); 3643 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3644 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3645 __u64 start, __u64 len); 3646 extern int ext4_get_es_cache(struct inode *inode, 3647 struct fiemap_extent_info *fieinfo, 3648 __u64 start, __u64 len); 3649 extern int ext4_ext_precache(struct inode *inode); 3650 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3651 struct inode *inode2, ext4_lblk_t lblk1, 3652 ext4_lblk_t lblk2, ext4_lblk_t count, 3653 int mark_unwritten,int *err); 3654 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3655 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3656 int check_cred, int restart_cred, 3657 int revoke_cred); 3658 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3659 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3660 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3661 int len, int unwritten, ext4_fsblk_t pblk); 3662 extern int ext4_ext_clear_bb(struct inode *inode); 3663 3664 3665 /* move_extent.c */ 3666 extern void ext4_double_down_write_data_sem(struct inode *first, 3667 struct inode *second); 3668 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3669 struct inode *donor_inode); 3670 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3671 __u64 start_orig, __u64 start_donor, 3672 __u64 len, __u64 *moved_len); 3673 3674 /* page-io.c */ 3675 extern int __init ext4_init_pageio(void); 3676 extern void ext4_exit_pageio(void); 3677 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3678 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3679 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3680 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3681 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3682 struct writeback_control *wbc); 3683 extern void ext4_end_io_rsv_work(struct work_struct *work); 3684 extern void ext4_io_submit(struct ext4_io_submit *io); 3685 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3686 size_t len); 3687 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3688 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3689 3690 /* mmp.c */ 3691 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3692 3693 /* mmp.c */ 3694 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3695 3696 /* verity.c */ 3697 extern const struct fsverity_operations ext4_verityops; 3698 3699 /* orphan.c */ 3700 extern int ext4_orphan_add(handle_t *, struct inode *); 3701 extern int ext4_orphan_del(handle_t *, struct inode *); 3702 extern void ext4_orphan_cleanup(struct super_block *sb, 3703 struct ext4_super_block *es); 3704 extern void ext4_release_orphan_info(struct super_block *sb); 3705 extern int ext4_init_orphan_info(struct super_block *sb); 3706 extern int ext4_orphan_file_empty(struct super_block *sb); 3707 extern void ext4_orphan_file_block_trigger( 3708 struct jbd2_buffer_trigger_type *triggers, 3709 struct buffer_head *bh, 3710 void *data, size_t size); 3711 3712 /* 3713 * Add new method to test whether block and inode bitmaps are properly 3714 * initialized. With uninit_bg reading the block from disk is not enough 3715 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3716 */ 3717 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3718 3719 static inline int bitmap_uptodate(struct buffer_head *bh) 3720 { 3721 return (buffer_uptodate(bh) && 3722 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3723 } 3724 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3725 { 3726 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3727 } 3728 3729 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3730 3731 /* For ioend & aio unwritten conversion wait queues */ 3732 #define EXT4_WQ_HASH_SZ 37 3733 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3734 EXT4_WQ_HASH_SZ]) 3735 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3736 3737 extern int ext4_resize_begin(struct super_block *sb); 3738 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3739 3740 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3741 struct ext4_io_end *io_end) 3742 { 3743 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3744 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3745 atomic_inc(&EXT4_I(inode)->i_unwritten); 3746 } 3747 } 3748 3749 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3750 { 3751 struct inode *inode = io_end->inode; 3752 3753 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3754 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3755 /* Wake up anyone waiting on unwritten extent conversion */ 3756 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3757 wake_up_all(ext4_ioend_wq(inode)); 3758 } 3759 } 3760 3761 extern const struct iomap_ops ext4_iomap_ops; 3762 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3763 extern const struct iomap_ops ext4_iomap_report_ops; 3764 3765 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3766 { 3767 /* 3768 * If the buffer has the write error flag, we have failed 3769 * to write out data in the block. In this case, we don't 3770 * have to read the block because we may read the old data 3771 * successfully. 3772 */ 3773 if (buffer_write_io_error(bh)) 3774 set_buffer_uptodate(bh); 3775 return buffer_uptodate(bh); 3776 } 3777 3778 #endif /* __KERNEL__ */ 3779 3780 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3781 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3782 3783 #endif /* _EXT4_H */ 3784