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 <linux/crc32c.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 * For each criteria, mballoc has slightly different way of finding 132 * the required blocks nad usually, higher the criteria the slower the 133 * allocation. We start at lower criterias and keep falling back to 134 * higher ones if we are not able to find any blocks. Lower (earlier) 135 * criteria are faster. 136 */ 137 enum criteria { 138 /* 139 * Used when number of blocks needed is a power of 2. This 140 * doesn't trigger any disk IO except prefetch and is the 141 * fastest criteria. 142 */ 143 CR_POWER2_ALIGNED, 144 145 /* 146 * Tries to lookup in-memory data structures to find the most 147 * suitable group that satisfies goal request. No disk IO 148 * except block prefetch. 149 */ 150 CR_GOAL_LEN_FAST, 151 152 /* 153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal 154 * length to the best available length for faster allocation. 155 */ 156 CR_BEST_AVAIL_LEN, 157 158 /* 159 * Reads each block group sequentially, performing disk IO if 160 * necessary, to find find_suitable block group. Tries to 161 * allocate goal length but might trim the request if nothing 162 * is found after enough tries. 163 */ 164 CR_GOAL_LEN_SLOW, 165 166 /* 167 * Finds the first free set of blocks and allocates 168 * those. This is only used in rare cases when 169 * CR_GOAL_LEN_SLOW also fails to allocate anything. 170 */ 171 CR_ANY_FREE, 172 173 /* 174 * Number of criterias defined. 175 */ 176 EXT4_MB_NUM_CRS 177 }; 178 179 /* 180 * Flags used in mballoc's allocation_context flags field. 181 * 182 * Also used to show what's going on for debugging purposes when the 183 * flag field is exported via the traceport interface 184 */ 185 186 /* prefer goal again. length */ 187 #define EXT4_MB_HINT_MERGE 0x0001 188 /* blocks already reserved */ 189 #define EXT4_MB_HINT_RESERVED 0x0002 190 /* metadata is being allocated */ 191 #define EXT4_MB_HINT_METADATA 0x0004 192 /* first blocks in the file */ 193 #define EXT4_MB_HINT_FIRST 0x0008 194 /* search for the best chunk */ 195 #define EXT4_MB_HINT_BEST 0x0010 196 /* data is being allocated */ 197 #define EXT4_MB_HINT_DATA 0x0020 198 /* don't preallocate (for tails) */ 199 #define EXT4_MB_HINT_NOPREALLOC 0x0040 200 /* allocate for locality group */ 201 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 202 /* allocate goal blocks or none */ 203 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 204 /* goal is meaningful */ 205 #define EXT4_MB_HINT_TRY_GOAL 0x0200 206 /* blocks already pre-reserved by delayed allocation */ 207 #define EXT4_MB_DELALLOC_RESERVED 0x0400 208 /* We are doing stream allocation */ 209 #define EXT4_MB_STREAM_ALLOC 0x0800 210 /* Use reserved root blocks if needed */ 211 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 212 /* Use blocks from reserved pool */ 213 #define EXT4_MB_USE_RESERVED 0x2000 214 /* Do strict check for free blocks while retrying block allocation */ 215 #define EXT4_MB_STRICT_CHECK 0x4000 216 /* Large fragment size list lookup succeeded at least once for 217 * CR_POWER2_ALIGNED */ 218 #define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000 219 /* Avg fragment size rb tree lookup succeeded at least once for 220 * CR_GOAL_LEN_FAST */ 221 #define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000 222 /* Avg fragment size rb tree lookup succeeded at least once for 223 * CR_BEST_AVAIL_LEN */ 224 #define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000 225 226 struct ext4_allocation_request { 227 /* target inode for block we're allocating */ 228 struct inode *inode; 229 /* how many blocks we want to allocate */ 230 unsigned int len; 231 /* logical block in target inode */ 232 ext4_lblk_t logical; 233 /* the closest logical allocated block to the left */ 234 ext4_lblk_t lleft; 235 /* the closest logical allocated block to the right */ 236 ext4_lblk_t lright; 237 /* phys. target (a hint) */ 238 ext4_fsblk_t goal; 239 /* phys. block for the closest logical allocated block to the left */ 240 ext4_fsblk_t pleft; 241 /* phys. block for the closest logical allocated block to the right */ 242 ext4_fsblk_t pright; 243 /* flags. see above EXT4_MB_HINT_* */ 244 unsigned int flags; 245 }; 246 247 /* 248 * Logical to physical block mapping, used by ext4_map_blocks() 249 * 250 * This structure is used to pass requests into ext4_map_blocks() as 251 * well as to store the information returned by ext4_map_blocks(). It 252 * takes less room on the stack than a struct buffer_head. 253 */ 254 #define EXT4_MAP_NEW BIT(BH_New) 255 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 256 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 257 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 258 #define EXT4_MAP_DELAYED BIT(BH_Delay) 259 /* 260 * This is for use in ext4_map_query_blocks() for a special case where we can 261 * have a physically and logically contiguous blocks split across two leaf 262 * nodes instead of a single extent. This is required in case of atomic writes 263 * to know whether the returned extent is last in leaf. If yes, then lookup for 264 * next in leaf block in ext4_map_query_blocks_next_in_leaf(). 265 * - This is never going to be added to any buffer head state. 266 * - We use the next available bit after BH_BITMAP_UPTODATE. 267 */ 268 #define EXT4_MAP_QUERY_LAST_IN_LEAF BIT(BH_BITMAP_UPTODATE + 1) 269 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 270 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 271 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF) 272 273 struct ext4_map_blocks { 274 ext4_fsblk_t m_pblk; 275 ext4_lblk_t m_lblk; 276 unsigned int m_len; 277 unsigned int m_flags; 278 }; 279 280 /* 281 * Block validity checking, system zone rbtree. 282 */ 283 struct ext4_system_blocks { 284 struct rb_root root; 285 struct rcu_head rcu; 286 }; 287 288 /* 289 * Flags for ext4_io_end->flags 290 */ 291 #define EXT4_IO_END_UNWRITTEN 0x0001 292 #define EXT4_IO_END_FAILED 0x0002 293 294 #define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED) 295 296 struct ext4_io_end_vec { 297 struct list_head list; /* list of io_end_vec */ 298 loff_t offset; /* offset in the file */ 299 ssize_t size; /* size of the extent */ 300 }; 301 302 /* 303 * For converting unwritten extents on a work queue. 'handle' is used for 304 * buffered writeback. 305 */ 306 typedef struct ext4_io_end { 307 struct list_head list; /* per-file finished IO list */ 308 handle_t *handle; /* handle reserved for extent 309 * conversion */ 310 struct inode *inode; /* file being written to */ 311 struct bio *bio; /* Linked list of completed 312 * bios covering the extent */ 313 unsigned int flag; /* unwritten or not */ 314 refcount_t count; /* reference counter */ 315 struct list_head list_vec; /* list of ext4_io_end_vec */ 316 } ext4_io_end_t; 317 318 struct ext4_io_submit { 319 struct writeback_control *io_wbc; 320 struct bio *io_bio; 321 ext4_io_end_t *io_end; 322 sector_t io_next_block; 323 }; 324 325 /* 326 * Special inodes numbers 327 */ 328 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 329 #define EXT4_ROOT_INO 2 /* Root inode */ 330 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 331 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 332 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 333 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 334 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 335 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 336 337 /* First non-reserved inode for old ext4 filesystems */ 338 #define EXT4_GOOD_OLD_FIRST_INO 11 339 340 /* 341 * Maximal count of links to a file 342 */ 343 #define EXT4_LINK_MAX 65000 344 345 /* 346 * Macro-instructions used to manage several block sizes 347 */ 348 #define EXT4_MIN_BLOCK_SIZE 1024 349 #define EXT4_MAX_BLOCK_SIZE 65536 350 #define EXT4_MIN_BLOCK_LOG_SIZE 10 351 #define EXT4_MAX_BLOCK_LOG_SIZE 16 352 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 353 #ifdef __KERNEL__ 354 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 355 #else 356 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 357 #endif 358 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 359 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 360 EXT4_SB(s)->s_cluster_bits) 361 #ifdef __KERNEL__ 362 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 363 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 364 #else 365 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 366 #endif 367 #ifdef __KERNEL__ 368 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 369 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 370 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 371 #else 372 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 373 EXT4_GOOD_OLD_INODE_SIZE : \ 374 (s)->s_inode_size) 375 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 376 EXT4_GOOD_OLD_FIRST_INO : \ 377 (s)->s_first_ino) 378 #endif 379 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 380 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 381 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 382 blkbits)) 383 #define EXT4_B_TO_LBLK(inode, offset) \ 384 (round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits) 385 386 /* Translate a block number to a cluster number */ 387 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 388 /* Translate a cluster number to a block number */ 389 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 390 /* Translate # of blks to # of clusters */ 391 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 392 (sbi)->s_cluster_bits) 393 /* Mask out the low bits to get the starting block of the cluster */ 394 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 395 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 396 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 397 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 398 /* Fill in the low bits to get the last block of the cluster */ 399 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 400 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 401 /* Get the cluster offset */ 402 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 403 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 404 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 405 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 406 407 /* 408 * Structure of a blocks group descriptor 409 */ 410 struct ext4_group_desc 411 { 412 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 413 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 414 __le32 bg_inode_table_lo; /* Inodes table block */ 415 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 416 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 417 __le16 bg_used_dirs_count_lo; /* Directories count */ 418 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 419 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 420 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 421 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 422 __le16 bg_itable_unused_lo; /* Unused inodes count */ 423 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 424 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 425 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 426 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 427 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 428 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 429 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 430 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 431 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 432 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 433 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 434 __u32 bg_reserved; 435 }; 436 437 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 438 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 439 sizeof(__le16)) 440 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 441 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 442 sizeof(__le16)) 443 444 /* 445 * Structure of a flex block group info 446 */ 447 448 struct flex_groups { 449 atomic64_t free_clusters; 450 atomic_t free_inodes; 451 atomic_t used_dirs; 452 }; 453 454 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 455 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 456 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 457 458 /* 459 * Macro-instructions used to manage group descriptors 460 */ 461 #define EXT4_MIN_DESC_SIZE 32 462 #define EXT4_MIN_DESC_SIZE_64BIT 64 463 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 464 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 465 #ifdef __KERNEL__ 466 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 467 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 468 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 469 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 470 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 471 #else 472 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 473 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 474 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 475 #endif 476 477 /* 478 * Constants relative to the data blocks 479 */ 480 #define EXT4_NDIR_BLOCKS 12 481 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 482 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 483 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 484 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 485 486 /* 487 * Inode flags 488 */ 489 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 490 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 491 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 492 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 493 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 494 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 495 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 496 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 497 /* Reserved for compression usage... */ 498 #define EXT4_DIRTY_FL 0x00000100 499 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 500 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 501 /* nb: was previously EXT2_ECOMPR_FL */ 502 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 503 /* End compression flags --- maybe not all used */ 504 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 505 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 506 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 507 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 508 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 509 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 510 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 511 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 512 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 513 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 514 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 515 516 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 517 518 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 519 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 520 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 521 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 522 523 /* User modifiable flags */ 524 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 525 EXT4_UNRM_FL | \ 526 EXT4_COMPR_FL | \ 527 EXT4_SYNC_FL | \ 528 EXT4_IMMUTABLE_FL | \ 529 EXT4_APPEND_FL | \ 530 EXT4_NODUMP_FL | \ 531 EXT4_NOATIME_FL | \ 532 EXT4_JOURNAL_DATA_FL | \ 533 EXT4_NOTAIL_FL | \ 534 EXT4_DIRSYNC_FL | \ 535 EXT4_TOPDIR_FL | \ 536 EXT4_EXTENTS_FL | \ 537 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 538 EXT4_DAX_FL | \ 539 EXT4_PROJINHERIT_FL | \ 540 EXT4_CASEFOLD_FL) 541 542 /* User visible flags */ 543 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 544 EXT4_DIRTY_FL | \ 545 EXT4_COMPRBLK_FL | \ 546 EXT4_NOCOMPR_FL | \ 547 EXT4_ENCRYPT_FL | \ 548 EXT4_INDEX_FL | \ 549 EXT4_VERITY_FL | \ 550 EXT4_INLINE_DATA_FL) 551 552 /* Flags that should be inherited by new inodes from their parent. */ 553 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 554 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 555 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 556 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 557 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 558 EXT4_DAX_FL) 559 560 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 561 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 562 EXT4_PROJINHERIT_FL)) 563 564 /* Flags that are appropriate for non-directories/regular files. */ 565 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 566 567 /* The only flags that should be swapped */ 568 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 569 570 /* Flags which are mutually exclusive to DAX */ 571 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 572 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 573 574 /* Mask out flags that are inappropriate for the given type of inode. */ 575 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 576 { 577 if (S_ISDIR(mode)) 578 return flags; 579 else if (S_ISREG(mode)) 580 return flags & EXT4_REG_FLMASK; 581 else 582 return flags & EXT4_OTHER_FLMASK; 583 } 584 585 /* 586 * Inode flags used for atomic set/get 587 */ 588 enum { 589 EXT4_INODE_SECRM = 0, /* Secure deletion */ 590 EXT4_INODE_UNRM = 1, /* Undelete */ 591 EXT4_INODE_COMPR = 2, /* Compress file */ 592 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 593 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 594 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 595 EXT4_INODE_NODUMP = 6, /* do not dump file */ 596 EXT4_INODE_NOATIME = 7, /* do not update atime */ 597 /* Reserved for compression usage... */ 598 EXT4_INODE_DIRTY = 8, 599 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 600 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 601 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 602 /* End compression flags --- maybe not all used */ 603 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 604 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 605 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 606 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 607 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 608 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 609 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 610 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 611 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 612 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 613 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 614 EXT4_INODE_DAX = 25, /* Inode is DAX */ 615 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 616 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 617 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 618 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 619 }; 620 621 /* 622 * Since it's pretty easy to mix up bit numbers and hex values, we use a 623 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 624 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 625 * any extra space in the compiled kernel image, otherwise, the build will fail. 626 * It's important that these values are the same, since we are using 627 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 628 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 629 * values found in ext2, ext3 and ext4 filesystems, and of course the values 630 * defined in e2fsprogs. 631 * 632 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 633 */ 634 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 635 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 636 637 static inline void ext4_check_flag_values(void) 638 { 639 CHECK_FLAG_VALUE(SECRM); 640 CHECK_FLAG_VALUE(UNRM); 641 CHECK_FLAG_VALUE(COMPR); 642 CHECK_FLAG_VALUE(SYNC); 643 CHECK_FLAG_VALUE(IMMUTABLE); 644 CHECK_FLAG_VALUE(APPEND); 645 CHECK_FLAG_VALUE(NODUMP); 646 CHECK_FLAG_VALUE(NOATIME); 647 CHECK_FLAG_VALUE(DIRTY); 648 CHECK_FLAG_VALUE(COMPRBLK); 649 CHECK_FLAG_VALUE(NOCOMPR); 650 CHECK_FLAG_VALUE(ENCRYPT); 651 CHECK_FLAG_VALUE(INDEX); 652 CHECK_FLAG_VALUE(IMAGIC); 653 CHECK_FLAG_VALUE(JOURNAL_DATA); 654 CHECK_FLAG_VALUE(NOTAIL); 655 CHECK_FLAG_VALUE(DIRSYNC); 656 CHECK_FLAG_VALUE(TOPDIR); 657 CHECK_FLAG_VALUE(HUGE_FILE); 658 CHECK_FLAG_VALUE(EXTENTS); 659 CHECK_FLAG_VALUE(VERITY); 660 CHECK_FLAG_VALUE(EA_INODE); 661 CHECK_FLAG_VALUE(INLINE_DATA); 662 CHECK_FLAG_VALUE(PROJINHERIT); 663 CHECK_FLAG_VALUE(CASEFOLD); 664 CHECK_FLAG_VALUE(RESERVED); 665 } 666 667 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 668 struct compat_ext4_new_group_input { 669 u32 group; 670 compat_u64 block_bitmap; 671 compat_u64 inode_bitmap; 672 compat_u64 inode_table; 673 u32 blocks_count; 674 u16 reserved_blocks; 675 u16 unused; 676 }; 677 #endif 678 679 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 680 struct ext4_new_group_data { 681 __u32 group; 682 __u64 block_bitmap; 683 __u64 inode_bitmap; 684 __u64 inode_table; 685 __u32 blocks_count; 686 __u16 reserved_blocks; 687 __u16 mdata_blocks; 688 __u32 free_clusters_count; 689 }; 690 691 /* Indexes used to index group tables in ext4_new_group_data */ 692 enum { 693 BLOCK_BITMAP = 0, /* block bitmap */ 694 INODE_BITMAP, /* inode bitmap */ 695 INODE_TABLE, /* inode tables */ 696 GROUP_TABLE_COUNT, 697 }; 698 699 /* 700 * Flags used by ext4_map_blocks() 701 */ 702 /* Allocate any needed blocks and/or convert an unwritten 703 extent to be an initialized ext4 */ 704 #define EXT4_GET_BLOCKS_CREATE 0x0001 705 /* Request the creation of an unwritten extent */ 706 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 707 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 708 EXT4_GET_BLOCKS_CREATE) 709 /* Caller is from the delayed allocation writeout path 710 * finally doing the actual allocation of delayed blocks */ 711 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 712 /* caller is from the direct IO path, request to creation of an 713 unwritten extents if not allocated, split the unwritten 714 extent if blocks has been preallocated already*/ 715 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 716 #define EXT4_GET_BLOCKS_CONVERT 0x0010 717 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 718 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 719 /* Eventual metadata allocation (due to growing extent tree) 720 * should not fail, so try to use reserved blocks for that.*/ 721 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 722 /* Don't normalize allocation size (used for fallocate) */ 723 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 724 /* Convert written extents to unwritten */ 725 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 726 /* Write zeros to newly created written extents */ 727 #define EXT4_GET_BLOCKS_ZERO 0x0200 728 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 729 EXT4_GET_BLOCKS_ZERO) 730 /* Caller is in the context of data submission, such as writeback, 731 * fsync, etc. Especially, in the generic writeback path, caller will 732 * submit data before dropping transaction handle. This allows jbd2 733 * to avoid submitting data before commit. */ 734 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 735 /* Convert extent to initialized after IO complete */ 736 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\ 737 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\ 738 EXT4_GET_BLOCKS_IO_SUBMIT) 739 /* Caller is in the atomic contex, find extent if it has been cached */ 740 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 741 /* 742 * Atomic write caller needs this to query in the slow path of mixed mapping 743 * case, when a contiguous extent can be split across two adjacent leaf nodes. 744 * Look EXT4_MAP_QUERY_LAST_IN_LEAF. 745 */ 746 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000 747 748 /* 749 * The bit position of these flags must not overlap with any of the 750 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 751 * read_extent_tree_block(), ext4_split_extent_at(), 752 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 753 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 754 * caching the extents when reading from the extent tree while a 755 * truncate or punch hole operation is in progress. 756 */ 757 #define EXT4_EX_NOCACHE 0x40000000 758 #define EXT4_EX_FORCE_CACHE 0x20000000 759 #define EXT4_EX_NOFAIL 0x10000000 760 /* 761 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to 762 * pass while lookup/querying of on disk extent tree. 763 */ 764 #define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\ 765 EXT4_EX_NOFAIL |\ 766 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF) 767 768 /* 769 * Flags used by ext4_free_blocks 770 */ 771 #define EXT4_FREE_BLOCKS_METADATA 0x0001 772 #define EXT4_FREE_BLOCKS_FORGET 0x0002 773 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 774 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 775 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 776 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 777 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 778 779 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 780 /* 781 * ioctl commands in 32 bit emulation 782 */ 783 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 784 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 785 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 786 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 787 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 788 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 789 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 790 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 791 #endif 792 793 /* Max physical block we can address w/o extents */ 794 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 795 796 /* Max logical block we can support */ 797 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 798 799 /* 800 * Structure of an inode on the disk 801 */ 802 struct ext4_inode { 803 __le16 i_mode; /* File mode */ 804 __le16 i_uid; /* Low 16 bits of Owner Uid */ 805 __le32 i_size_lo; /* Size in bytes */ 806 __le32 i_atime; /* Access time */ 807 __le32 i_ctime; /* Inode Change time */ 808 __le32 i_mtime; /* Modification time */ 809 __le32 i_dtime; /* Deletion Time */ 810 __le16 i_gid; /* Low 16 bits of Group Id */ 811 __le16 i_links_count; /* Links count */ 812 __le32 i_blocks_lo; /* Blocks count */ 813 __le32 i_flags; /* File flags */ 814 union { 815 struct { 816 __le32 l_i_version; 817 } linux1; 818 struct { 819 __u32 h_i_translator; 820 } hurd1; 821 struct { 822 __u32 m_i_reserved1; 823 } masix1; 824 } osd1; /* OS dependent 1 */ 825 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 826 __le32 i_generation; /* File version (for NFS) */ 827 __le32 i_file_acl_lo; /* File ACL */ 828 __le32 i_size_high; 829 __le32 i_obso_faddr; /* Obsoleted fragment address */ 830 union { 831 struct { 832 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 833 __le16 l_i_file_acl_high; 834 __le16 l_i_uid_high; /* these 2 fields */ 835 __le16 l_i_gid_high; /* were reserved2[0] */ 836 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 837 __le16 l_i_reserved; 838 } linux2; 839 struct { 840 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 841 __u16 h_i_mode_high; 842 __u16 h_i_uid_high; 843 __u16 h_i_gid_high; 844 __u32 h_i_author; 845 } hurd2; 846 struct { 847 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 848 __le16 m_i_file_acl_high; 849 __u32 m_i_reserved2[2]; 850 } masix2; 851 } osd2; /* OS dependent 2 */ 852 __le16 i_extra_isize; 853 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 854 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 855 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 856 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 857 __le32 i_crtime; /* File Creation time */ 858 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 859 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 860 __le32 i_projid; /* Project ID */ 861 }; 862 863 #define EXT4_EPOCH_BITS 2 864 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 865 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 866 867 /* 868 * Extended fields will fit into an inode if the filesystem was formatted 869 * with large inodes (-I 256 or larger) and there are not currently any EAs 870 * consuming all of the available space. For new inodes we always reserve 871 * enough space for the kernel's known extended fields, but for inodes 872 * created with an old kernel this might not have been the case. None of 873 * the extended inode fields is critical for correct filesystem operation. 874 * This macro checks if a certain field fits in the inode. Note that 875 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 876 */ 877 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 878 ((offsetof(typeof(*ext4_inode), field) + \ 879 sizeof((ext4_inode)->field)) \ 880 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 881 (einode)->i_extra_isize)) \ 882 883 /* 884 * We use an encoding that preserves the times for extra epoch "00": 885 * 886 * extra msb of adjust for signed 887 * epoch 32-bit 32-bit tv_sec to 888 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 889 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 890 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 891 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 892 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 893 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 894 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 895 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 896 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 897 * 898 * Note that previous versions of the kernel on 64-bit systems would 899 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 900 * 1970. e2fsck will correct this, assuming that it is run on the 901 * affected filesystem before 2242. 902 */ 903 904 static inline __le32 ext4_encode_extra_time(struct timespec64 ts) 905 { 906 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK; 907 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS)); 908 } 909 910 static inline struct timespec64 ext4_decode_extra_time(__le32 base, 911 __le32 extra) 912 { 913 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) }; 914 915 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 916 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 917 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 918 return ts; 919 } 920 921 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \ 922 do { \ 923 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 924 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \ 925 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \ 926 } else \ 927 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \ 928 } while (0) 929 930 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \ 931 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode)) 932 933 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \ 934 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode)) 935 936 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \ 937 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode)) 938 939 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 940 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 941 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \ 942 raw_inode, (einode)->xtime) 943 944 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \ 945 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \ 946 ext4_decode_extra_time((raw_inode)->xtime, \ 947 (raw_inode)->xtime ## _extra) : \ 948 (struct timespec64) { \ 949 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \ 950 }) 951 952 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \ 953 do { \ 954 inode_set_atime_to_ts(inode, \ 955 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \ 956 } while (0) 957 958 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \ 959 do { \ 960 inode_set_mtime_to_ts(inode, \ 961 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \ 962 } while (0) 963 964 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \ 965 do { \ 966 inode_set_ctime_to_ts(inode, \ 967 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \ 968 } while (0) 969 970 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 971 do { \ 972 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 973 (einode)->xtime = \ 974 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \ 975 raw_inode); \ 976 else \ 977 (einode)->xtime = (struct timespec64){0, 0}; \ 978 } while (0) 979 980 #define i_disk_version osd1.linux1.l_i_version 981 982 #if defined(__KERNEL__) || defined(__linux__) 983 #define i_reserved1 osd1.linux1.l_i_reserved1 984 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 985 #define i_blocks_high osd2.linux2.l_i_blocks_high 986 #define i_uid_low i_uid 987 #define i_gid_low i_gid 988 #define i_uid_high osd2.linux2.l_i_uid_high 989 #define i_gid_high osd2.linux2.l_i_gid_high 990 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 991 992 #elif defined(__GNU__) 993 994 #define i_translator osd1.hurd1.h_i_translator 995 #define i_uid_high osd2.hurd2.h_i_uid_high 996 #define i_gid_high osd2.hurd2.h_i_gid_high 997 #define i_author osd2.hurd2.h_i_author 998 999 #elif defined(__masix__) 1000 1001 #define i_reserved1 osd1.masix1.m_i_reserved1 1002 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 1003 #define i_reserved2 osd2.masix2.m_i_reserved2 1004 1005 #endif /* defined(__KERNEL__) || defined(__linux__) */ 1006 1007 #include "extents_status.h" 1008 #include "fast_commit.h" 1009 1010 /* 1011 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 1012 * 1013 * These are needed to avoid lockdep false positives when we need to 1014 * allocate blocks to the quota inode during ext4_map_blocks(), while 1015 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1016 * do quota tracking for the quota inode, this avoids deadlock (as 1017 * well as infinite recursion, since it isn't turtles all the way 1018 * down...) 1019 * 1020 * I_DATA_SEM_NORMAL - Used for most inodes 1021 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1022 * where the second inode has larger inode number 1023 * than the first 1024 * I_DATA_SEM_QUOTA - Used for quota inodes only 1025 * I_DATA_SEM_EA - Used for ea_inodes only 1026 */ 1027 enum { 1028 I_DATA_SEM_NORMAL = 0, 1029 I_DATA_SEM_OTHER, 1030 I_DATA_SEM_QUOTA, 1031 I_DATA_SEM_EA 1032 }; 1033 1034 1035 /* 1036 * fourth extended file system inode data in memory 1037 */ 1038 struct ext4_inode_info { 1039 __le32 i_data[15]; /* unconverted */ 1040 __u32 i_dtime; 1041 ext4_fsblk_t i_file_acl; 1042 1043 /* 1044 * i_block_group is the number of the block group which contains 1045 * this file's inode. Constant across the lifetime of the inode, 1046 * it is used for making block allocation decisions - we try to 1047 * place a file's data blocks near its inode block, and new inodes 1048 * near to their parent directory's inode. 1049 */ 1050 ext4_group_t i_block_group; 1051 ext4_lblk_t i_dir_start_lookup; 1052 #if (BITS_PER_LONG < 64) 1053 unsigned long i_state_flags; /* Dynamic state flags */ 1054 #endif 1055 unsigned long i_flags; 1056 1057 /* 1058 * Extended attributes can be read independently of the main file 1059 * data. Taking i_rwsem even when reading would cause contention 1060 * between readers of EAs and writers of regular file data, so 1061 * instead we synchronize on xattr_sem when reading or changing 1062 * EAs. 1063 */ 1064 struct rw_semaphore xattr_sem; 1065 1066 /* 1067 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1068 * i_orphan is used. 1069 */ 1070 union { 1071 struct list_head i_orphan; /* unlinked but open inodes */ 1072 unsigned int i_orphan_idx; /* Index in orphan file */ 1073 }; 1074 1075 /* Fast commit related info */ 1076 1077 /* For tracking dentry create updates */ 1078 struct list_head i_fc_dilist; 1079 struct list_head i_fc_list; /* 1080 * inodes that need fast commit 1081 * protected by sbi->s_fc_lock. 1082 */ 1083 1084 /* Start of lblk range that needs to be committed in this fast commit */ 1085 ext4_lblk_t i_fc_lblk_start; 1086 1087 /* End of lblk range that needs to be committed in this fast commit */ 1088 ext4_lblk_t i_fc_lblk_len; 1089 1090 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1091 1092 /* Fast commit wait queue for this inode */ 1093 wait_queue_head_t i_fc_wait; 1094 1095 /* 1096 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len 1097 * and inode's EXT4_FC_STATE_COMMITTING state bit. 1098 */ 1099 spinlock_t i_fc_lock; 1100 1101 /* 1102 * i_disksize keeps track of what the inode size is ON DISK, not 1103 * in memory. During truncate, i_size is set to the new size by 1104 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1105 * set i_disksize to 0 until the truncate is actually under way. 1106 * 1107 * The intent is that i_disksize always represents the blocks which 1108 * are used by this file. This allows recovery to restart truncate 1109 * on orphans if we crash during truncate. We actually write i_disksize 1110 * into the on-disk inode when writing inodes out, instead of i_size. 1111 * 1112 * The only time when i_disksize and i_size may be different is when 1113 * a truncate is in progress. The only things which change i_disksize 1114 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1115 */ 1116 loff_t i_disksize; 1117 1118 /* 1119 * i_data_sem is for serialising ext4_truncate() against 1120 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1121 * data tree are chopped off during truncate. We can't do that in 1122 * ext4 because whenever we perform intermediate commits during 1123 * truncate, the inode and all the metadata blocks *must* be in a 1124 * consistent state which allows truncation of the orphans to restart 1125 * during recovery. Hence we must fix the get_block-vs-truncate race 1126 * by other means, so we have i_data_sem. 1127 */ 1128 struct rw_semaphore i_data_sem; 1129 struct inode vfs_inode; 1130 struct jbd2_inode *jinode; 1131 1132 /* 1133 * File creation time. Its function is same as that of 1134 * struct timespec64 i_{a,c,m}time in the generic inode. 1135 */ 1136 struct timespec64 i_crtime; 1137 1138 /* mballoc */ 1139 atomic_t i_prealloc_active; 1140 1141 /* allocation reservation info for delalloc */ 1142 /* In case of bigalloc, this refer to clusters rather than blocks */ 1143 unsigned int i_reserved_data_blocks; 1144 struct rb_root i_prealloc_node; 1145 rwlock_t i_prealloc_lock; 1146 1147 /* extents status tree */ 1148 struct ext4_es_tree i_es_tree; 1149 rwlock_t i_es_lock; 1150 struct list_head i_es_list; 1151 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1152 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1153 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1154 extents to shrink. Protected by 1155 i_es_lock */ 1156 1157 /* ialloc */ 1158 ext4_group_t i_last_alloc_group; 1159 1160 /* pending cluster reservations for bigalloc file systems */ 1161 struct ext4_pending_tree i_pending_tree; 1162 1163 /* on-disk additional length */ 1164 __u16 i_extra_isize; 1165 1166 /* Indicate the inline data space. */ 1167 u16 i_inline_off; 1168 u16 i_inline_size; 1169 1170 #ifdef CONFIG_QUOTA 1171 /* quota space reservation, managed internally by quota code */ 1172 qsize_t i_reserved_quota; 1173 #endif 1174 spinlock_t i_block_reservation_lock; 1175 1176 /* Lock protecting lists below */ 1177 spinlock_t i_completed_io_lock; 1178 /* 1179 * Completed IOs that need unwritten extents handling and have 1180 * transaction reserved 1181 */ 1182 struct list_head i_rsv_conversion_list; 1183 struct work_struct i_rsv_conversion_work; 1184 1185 /* 1186 * Transactions that contain inode's metadata needed to complete 1187 * fsync and fdatasync, respectively. 1188 */ 1189 tid_t i_sync_tid; 1190 tid_t i_datasync_tid; 1191 1192 #ifdef CONFIG_QUOTA 1193 struct dquot __rcu *i_dquot[MAXQUOTAS]; 1194 #endif 1195 1196 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1197 __u32 i_csum_seed; 1198 1199 kprojid_t i_projid; 1200 }; 1201 1202 /* 1203 * File system states 1204 */ 1205 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1206 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1207 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1208 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1209 1210 /* 1211 * Misc. filesystem flags 1212 */ 1213 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1214 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1215 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1216 1217 /* 1218 * Mount flags set via mount options or defaults 1219 */ 1220 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1221 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1222 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1223 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1224 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1225 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1226 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1227 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1228 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1229 #ifdef CONFIG_FS_DAX 1230 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1231 #else 1232 #define EXT4_MOUNT_DAX_ALWAYS 0 1233 #endif 1234 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1235 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1236 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1237 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1238 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1239 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1240 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1241 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1242 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1243 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1244 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1245 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1246 * enable enforcement for hidden 1247 * quota files */ 1248 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1249 * enforcement for hidden quota 1250 * files */ 1251 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1252 * enforcement */ 1253 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1254 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1255 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1256 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1257 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1258 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1259 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1260 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1261 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1262 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1263 1264 /* 1265 * Mount flags set either automatically (could not be set by mount option) 1266 * based on per file system feature or property or in special cases such as 1267 * distinguishing between explicit mount option definition and default. 1268 */ 1269 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1270 specified delalloc */ 1271 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1272 size of blocksize * 8 1273 blocks */ 1274 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1275 file systems */ 1276 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1277 specified journal checksum */ 1278 1279 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1280 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1281 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1282 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1283 * scanning in mballoc 1284 */ 1285 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1286 1287 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1288 ~EXT4_MOUNT_##opt 1289 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1290 EXT4_MOUNT_##opt 1291 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1292 EXT4_MOUNT_##opt) 1293 1294 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1295 ~EXT4_MOUNT2_##opt 1296 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1297 EXT4_MOUNT2_##opt 1298 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1299 EXT4_MOUNT2_##opt) 1300 1301 #define ext4_test_and_set_bit __test_and_set_bit_le 1302 #define ext4_set_bit __set_bit_le 1303 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1304 #define ext4_clear_bit __clear_bit_le 1305 #define ext4_test_bit test_bit_le 1306 #define ext4_find_next_zero_bit find_next_zero_bit_le 1307 #define ext4_find_next_bit find_next_bit_le 1308 1309 extern void mb_set_bits(void *bm, int cur, int len); 1310 1311 /* 1312 * Maximal mount counts between two filesystem checks 1313 */ 1314 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1315 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1316 1317 /* 1318 * Behaviour when detecting errors 1319 */ 1320 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1321 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1322 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1323 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1324 1325 /* Metadata checksum algorithm codes */ 1326 #define EXT4_CRC32C_CHKSUM 1 1327 1328 #define EXT4_LABEL_MAX 16 1329 1330 /* 1331 * Structure of the super block 1332 */ 1333 struct ext4_super_block { 1334 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1335 __le32 s_blocks_count_lo; /* Blocks count */ 1336 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1337 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1338 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1339 __le32 s_first_data_block; /* First Data Block */ 1340 __le32 s_log_block_size; /* Block size */ 1341 __le32 s_log_cluster_size; /* Allocation cluster size */ 1342 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1343 __le32 s_clusters_per_group; /* # Clusters per group */ 1344 __le32 s_inodes_per_group; /* # Inodes per group */ 1345 __le32 s_mtime; /* Mount time */ 1346 /*30*/ __le32 s_wtime; /* Write time */ 1347 __le16 s_mnt_count; /* Mount count */ 1348 __le16 s_max_mnt_count; /* Maximal mount count */ 1349 __le16 s_magic; /* Magic signature */ 1350 __le16 s_state; /* File system state */ 1351 __le16 s_errors; /* Behaviour when detecting errors */ 1352 __le16 s_minor_rev_level; /* minor revision level */ 1353 /*40*/ __le32 s_lastcheck; /* time of last check */ 1354 __le32 s_checkinterval; /* max. time between checks */ 1355 __le32 s_creator_os; /* OS */ 1356 __le32 s_rev_level; /* Revision level */ 1357 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1358 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1359 /* 1360 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1361 * 1362 * Note: the difference between the compatible feature set and 1363 * the incompatible feature set is that if there is a bit set 1364 * in the incompatible feature set that the kernel doesn't 1365 * know about, it should refuse to mount the filesystem. 1366 * 1367 * e2fsck's requirements are more strict; if it doesn't know 1368 * about a feature in either the compatible or incompatible 1369 * feature set, it must abort and not try to meddle with 1370 * things it doesn't understand... 1371 */ 1372 __le32 s_first_ino; /* First non-reserved inode */ 1373 __le16 s_inode_size; /* size of inode structure */ 1374 __le16 s_block_group_nr; /* block group # of this superblock */ 1375 __le32 s_feature_compat; /* compatible feature set */ 1376 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1377 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1378 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1379 /*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */ 1380 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1381 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1382 /* 1383 * Performance hints. Directory preallocation should only 1384 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1385 */ 1386 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1387 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1388 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1389 /* 1390 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1391 */ 1392 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1393 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1394 __le32 s_journal_dev; /* device number of journal file */ 1395 __le32 s_last_orphan; /* start of list of inodes to delete */ 1396 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1397 __u8 s_def_hash_version; /* Default hash version to use */ 1398 __u8 s_jnl_backup_type; 1399 __le16 s_desc_size; /* size of group descriptor */ 1400 /*100*/ __le32 s_default_mount_opts; 1401 __le32 s_first_meta_bg; /* First metablock block group */ 1402 __le32 s_mkfs_time; /* When the filesystem was created */ 1403 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1404 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1405 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1406 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1407 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1408 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1409 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1410 __le32 s_flags; /* Miscellaneous flags */ 1411 __le16 s_raid_stride; /* RAID stride */ 1412 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1413 __le64 s_mmp_block; /* Block for multi-mount protection */ 1414 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1415 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1416 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1417 __u8 s_encryption_level; /* versioning level for encryption */ 1418 __u8 s_reserved_pad; /* Padding to next 32bits */ 1419 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1420 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1421 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1422 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1423 snapshot's future use */ 1424 __le32 s_snapshot_list; /* inode number of the head of the 1425 on-disk snapshot list */ 1426 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1427 __le32 s_error_count; /* number of fs errors */ 1428 __le32 s_first_error_time; /* first time an error happened */ 1429 __le32 s_first_error_ino; /* inode involved in first error */ 1430 __le64 s_first_error_block; /* block involved of first error */ 1431 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1432 __le32 s_first_error_line; /* line number where error happened */ 1433 __le32 s_last_error_time; /* most recent time of an error */ 1434 __le32 s_last_error_ino; /* inode involved in last error */ 1435 __le32 s_last_error_line; /* line number where error happened */ 1436 __le64 s_last_error_block; /* block involved of last error */ 1437 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1438 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1439 __u8 s_mount_opts[64]; 1440 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1441 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1442 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1443 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1444 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1445 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1446 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1447 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1448 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1449 __u8 s_wtime_hi; 1450 __u8 s_mtime_hi; 1451 __u8 s_mkfs_time_hi; 1452 __u8 s_lastcheck_hi; 1453 __u8 s_first_error_time_hi; 1454 __u8 s_last_error_time_hi; 1455 __u8 s_first_error_errcode; 1456 __u8 s_last_error_errcode; 1457 __le16 s_encoding; /* Filename charset encoding */ 1458 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1459 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1460 __le32 s_reserved[94]; /* Padding to the end of the block */ 1461 __le32 s_checksum; /* crc32c(superblock) */ 1462 }; 1463 1464 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1465 1466 #ifdef __KERNEL__ 1467 1468 /* Number of quota types we support */ 1469 #define EXT4_MAXQUOTAS 3 1470 1471 #define EXT4_ENC_UTF8_12_1 1 1472 1473 /* Types of ext4 journal triggers */ 1474 enum ext4_journal_trigger_type { 1475 EXT4_JTR_ORPHAN_FILE, 1476 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1477 }; 1478 1479 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1480 1481 struct ext4_journal_trigger { 1482 struct jbd2_buffer_trigger_type tr_triggers; 1483 struct super_block *sb; 1484 }; 1485 1486 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1487 struct jbd2_buffer_trigger_type *trigger) 1488 { 1489 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1490 } 1491 1492 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1493 1494 /* Structure at the tail of orphan block */ 1495 struct ext4_orphan_block_tail { 1496 __le32 ob_magic; 1497 __le32 ob_checksum; 1498 }; 1499 1500 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1501 { 1502 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1503 sizeof(u32); 1504 } 1505 1506 struct ext4_orphan_block { 1507 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1508 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1509 }; 1510 1511 /* 1512 * Info about orphan file. 1513 */ 1514 struct ext4_orphan_info { 1515 int of_blocks; /* Number of orphan blocks in a file */ 1516 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1517 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1518 * file blocks */ 1519 }; 1520 1521 /* 1522 * fourth extended-fs super-block data in memory 1523 */ 1524 struct ext4_sb_info { 1525 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1526 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1527 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1528 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1529 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1530 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1531 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1532 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1533 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1534 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1535 unsigned long s_overhead; /* # of fs overhead clusters */ 1536 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1537 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1538 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1539 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1540 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1541 /* Array of bh's for the block group descriptors */ 1542 struct buffer_head * __rcu *s_group_desc; 1543 unsigned int s_mount_opt; 1544 unsigned int s_mount_opt2; 1545 unsigned long s_mount_flags; 1546 unsigned int s_def_mount_opt; 1547 unsigned int s_def_mount_opt2; 1548 ext4_fsblk_t s_sb_block; 1549 atomic64_t s_resv_clusters; 1550 kuid_t s_resuid; 1551 kgid_t s_resgid; 1552 unsigned short s_mount_state; 1553 unsigned short s_pad; 1554 int s_addr_per_block_bits; 1555 int s_desc_per_block_bits; 1556 int s_inode_size; 1557 int s_first_ino; 1558 unsigned int s_inode_readahead_blks; 1559 unsigned int s_inode_goal; 1560 u32 s_hash_seed[4]; 1561 int s_def_hash_version; 1562 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1563 struct percpu_counter s_freeclusters_counter; 1564 struct percpu_counter s_freeinodes_counter; 1565 struct percpu_counter s_dirs_counter; 1566 struct percpu_counter s_dirtyclusters_counter; 1567 struct percpu_counter s_sra_exceeded_retry_limit; 1568 struct blockgroup_lock *s_blockgroup_lock; 1569 struct proc_dir_entry *s_proc; 1570 struct kobject s_kobj; 1571 struct completion s_kobj_unregister; 1572 struct super_block *s_sb; 1573 struct buffer_head *s_mmp_bh; 1574 1575 /* Journaling */ 1576 struct journal_s *s_journal; 1577 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1578 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1579 struct list_head s_orphan; /* List of orphaned inodes in on disk 1580 list */ 1581 struct ext4_orphan_info s_orphan_info; 1582 unsigned long s_commit_interval; 1583 u32 s_max_batch_time; 1584 u32 s_min_batch_time; 1585 struct file *s_journal_bdev_file; 1586 #ifdef CONFIG_QUOTA 1587 /* Names of quota files with journalled quota */ 1588 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1589 int s_jquota_fmt; /* Format of quota to use */ 1590 #endif 1591 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1592 struct ext4_system_blocks __rcu *s_system_blks; 1593 1594 #ifdef EXTENTS_STATS 1595 /* ext4 extents stats */ 1596 unsigned long s_ext_min; 1597 unsigned long s_ext_max; 1598 unsigned long s_depth_max; 1599 spinlock_t s_ext_stats_lock; 1600 unsigned long s_ext_blocks; 1601 unsigned long s_ext_extents; 1602 #endif 1603 1604 /* for buddy allocator */ 1605 struct ext4_group_info ** __rcu *s_group_info; 1606 struct inode *s_buddy_cache; 1607 spinlock_t s_md_lock; 1608 unsigned short *s_mb_offsets; 1609 unsigned int *s_mb_maxs; 1610 unsigned int s_group_info_size; 1611 unsigned int s_mb_free_pending; 1612 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1613 after commit completed */ 1614 struct list_head s_discard_list; 1615 struct work_struct s_discard_work; 1616 atomic_t s_retry_alloc_pending; 1617 struct list_head *s_mb_avg_fragment_size; 1618 rwlock_t *s_mb_avg_fragment_size_locks; 1619 struct list_head *s_mb_largest_free_orders; 1620 rwlock_t *s_mb_largest_free_orders_locks; 1621 1622 /* tunables */ 1623 unsigned long s_stripe; 1624 unsigned int s_mb_max_linear_groups; 1625 unsigned int s_mb_stream_request; 1626 unsigned int s_mb_max_to_scan; 1627 unsigned int s_mb_min_to_scan; 1628 unsigned int s_mb_stats; 1629 unsigned int s_mb_order2_reqs; 1630 unsigned int s_mb_group_prealloc; 1631 unsigned int s_max_dir_size_kb; 1632 /* where last allocation was done - for stream allocation */ 1633 unsigned long s_mb_last_group; 1634 unsigned long s_mb_last_start; 1635 unsigned int s_mb_prefetch; 1636 unsigned int s_mb_prefetch_limit; 1637 unsigned int s_mb_best_avail_max_trim_order; 1638 unsigned int s_sb_update_sec; 1639 unsigned int s_sb_update_kb; 1640 1641 /* stats for buddy allocator */ 1642 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1643 atomic_t s_bal_success; /* we found long enough chunks */ 1644 atomic_t s_bal_allocated; /* in blocks */ 1645 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1646 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1647 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1648 atomic_t s_bal_goals; /* goal hits */ 1649 atomic_t s_bal_len_goals; /* len goal hits */ 1650 atomic_t s_bal_breaks; /* too long searches */ 1651 atomic_t s_bal_2orders; /* 2^order hits */ 1652 atomic_t s_bal_p2_aligned_bad_suggestions; 1653 atomic_t s_bal_goal_fast_bad_suggestions; 1654 atomic_t s_bal_best_avail_bad_suggestions; 1655 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1656 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1657 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1658 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1659 atomic64_t s_mb_generation_time; 1660 atomic_t s_mb_lost_chunks; 1661 atomic_t s_mb_preallocated; 1662 atomic_t s_mb_discarded; 1663 atomic_t s_lock_busy; 1664 1665 /* locality groups */ 1666 struct ext4_locality_group __percpu *s_locality_groups; 1667 1668 /* for write statistics */ 1669 unsigned long s_sectors_written_start; 1670 u64 s_kbytes_written; 1671 1672 /* the size of zero-out chunk */ 1673 unsigned int s_extent_max_zeroout_kb; 1674 1675 unsigned int s_log_groups_per_flex; 1676 struct flex_groups * __rcu *s_flex_groups; 1677 ext4_group_t s_flex_groups_allocated; 1678 1679 /* workqueue for reserved extent conversions (buffered io) */ 1680 struct workqueue_struct *rsv_conversion_wq; 1681 1682 /* timer for periodic error stats printing */ 1683 struct timer_list s_err_report; 1684 1685 /* Lazy inode table initialization info */ 1686 struct ext4_li_request *s_li_request; 1687 /* Wait multiplier for lazy initialization thread */ 1688 unsigned int s_li_wait_mult; 1689 1690 /* Kernel thread for multiple mount protection */ 1691 struct task_struct *s_mmp_tsk; 1692 1693 /* record the last minlen when FITRIM is called. */ 1694 unsigned long s_last_trim_minblks; 1695 1696 /* Precomputed FS UUID checksum for seeding other checksums */ 1697 __u32 s_csum_seed; 1698 1699 /* Reclaim extents from extent status tree */ 1700 struct shrinker *s_es_shrinker; 1701 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1702 long s_es_nr_inode; 1703 struct ext4_es_stats s_es_stats; 1704 struct mb_cache *s_ea_block_cache; 1705 struct mb_cache *s_ea_inode_cache; 1706 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1707 1708 /* Journal triggers for checksum computation */ 1709 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1710 1711 /* Ratelimit ext4 messages. */ 1712 struct ratelimit_state s_err_ratelimit_state; 1713 struct ratelimit_state s_warning_ratelimit_state; 1714 struct ratelimit_state s_msg_ratelimit_state; 1715 atomic_t s_warning_count; 1716 atomic_t s_msg_count; 1717 1718 /* Encryption policy for '-o test_dummy_encryption' */ 1719 struct fscrypt_dummy_policy s_dummy_enc_policy; 1720 1721 /* 1722 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1723 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1724 * DIOREAD_NOLOCK mount options on remount. 1725 */ 1726 struct percpu_rw_semaphore s_writepages_rwsem; 1727 struct dax_device *s_daxdev; 1728 u64 s_dax_part_off; 1729 #ifdef CONFIG_EXT4_DEBUG 1730 unsigned long s_simulate_fail; 1731 #endif 1732 /* Record the errseq of the backing block device */ 1733 errseq_t s_bdev_wb_err; 1734 spinlock_t s_bdev_wb_lock; 1735 1736 /* Information about errors that happened during this mount */ 1737 spinlock_t s_error_lock; 1738 int s_add_error_count; 1739 int s_first_error_code; 1740 __u32 s_first_error_line; 1741 __u32 s_first_error_ino; 1742 __u64 s_first_error_block; 1743 const char *s_first_error_func; 1744 time64_t s_first_error_time; 1745 int s_last_error_code; 1746 __u32 s_last_error_line; 1747 __u32 s_last_error_ino; 1748 __u64 s_last_error_block; 1749 const char *s_last_error_func; 1750 time64_t s_last_error_time; 1751 /* 1752 * If we are in a context where we cannot update the on-disk 1753 * superblock, we queue the work here. This is used to update 1754 * the error information in the superblock, and for periodic 1755 * updates of the superblock called from the commit callback 1756 * function. 1757 */ 1758 struct work_struct s_sb_upd_work; 1759 1760 /* Atomic write unit values in bytes */ 1761 unsigned int s_awu_min; 1762 unsigned int s_awu_max; 1763 1764 /* Ext4 fast commit sub transaction ID */ 1765 atomic_t s_fc_subtid; 1766 1767 /* 1768 * After commit starts, the main queue gets locked, and the further 1769 * updates get added in the staging queue. 1770 */ 1771 #define FC_Q_MAIN 0 1772 #define FC_Q_STAGING 1 1773 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1774 * that have data changes in them. 1775 */ 1776 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1777 unsigned int s_fc_bytes; 1778 /* 1779 * Main fast commit lock. This lock protects accesses to the 1780 * following fields: 1781 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1782 */ 1783 struct mutex s_fc_lock; 1784 struct buffer_head *s_fc_bh; 1785 struct ext4_fc_stats s_fc_stats; 1786 tid_t s_fc_ineligible_tid; 1787 #ifdef CONFIG_EXT4_DEBUG 1788 int s_fc_debug_max_replay; 1789 #endif 1790 struct ext4_fc_replay_state s_fc_replay_state; 1791 }; 1792 1793 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1794 { 1795 return sb->s_fs_info; 1796 } 1797 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1798 { 1799 return container_of(inode, struct ext4_inode_info, vfs_inode); 1800 } 1801 1802 static inline int ext4_writepages_down_read(struct super_block *sb) 1803 { 1804 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1805 return memalloc_nofs_save(); 1806 } 1807 1808 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1809 { 1810 memalloc_nofs_restore(ctx); 1811 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1812 } 1813 1814 static inline int ext4_writepages_down_write(struct super_block *sb) 1815 { 1816 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1817 return memalloc_nofs_save(); 1818 } 1819 1820 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1821 { 1822 memalloc_nofs_restore(ctx); 1823 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1824 } 1825 1826 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1827 { 1828 return ino == EXT4_ROOT_INO || 1829 (ino >= EXT4_FIRST_INO(sb) && 1830 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1831 } 1832 1833 /* 1834 * Returns: sbi->field[index] 1835 * Used to access an array element from the following sbi fields which require 1836 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1837 * - s_group_desc 1838 * - s_group_info 1839 * - s_flex_group 1840 */ 1841 #define sbi_array_rcu_deref(sbi, field, index) \ 1842 ({ \ 1843 typeof(*((sbi)->field)) _v; \ 1844 rcu_read_lock(); \ 1845 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1846 rcu_read_unlock(); \ 1847 _v; \ 1848 }) 1849 1850 /* 1851 * run-time mount flags 1852 */ 1853 enum { 1854 EXT4_MF_MNTDIR_SAMPLED, 1855 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1856 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */ 1857 }; 1858 1859 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1860 { 1861 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1862 } 1863 1864 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1865 { 1866 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1867 } 1868 1869 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1870 { 1871 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1872 } 1873 1874 1875 /* 1876 * Simulate_fail codes 1877 */ 1878 #define EXT4_SIM_BBITMAP_EIO 1 1879 #define EXT4_SIM_BBITMAP_CRC 2 1880 #define EXT4_SIM_IBITMAP_EIO 3 1881 #define EXT4_SIM_IBITMAP_CRC 4 1882 #define EXT4_SIM_INODE_EIO 5 1883 #define EXT4_SIM_INODE_CRC 6 1884 #define EXT4_SIM_DIRBLOCK_EIO 7 1885 #define EXT4_SIM_DIRBLOCK_CRC 8 1886 1887 static inline bool ext4_simulate_fail(struct super_block *sb, 1888 unsigned long code) 1889 { 1890 #ifdef CONFIG_EXT4_DEBUG 1891 struct ext4_sb_info *sbi = EXT4_SB(sb); 1892 1893 if (unlikely(sbi->s_simulate_fail == code)) { 1894 sbi->s_simulate_fail = 0; 1895 return true; 1896 } 1897 #endif 1898 return false; 1899 } 1900 1901 /* 1902 * Error number codes for s_{first,last}_error_errno 1903 * 1904 * Linux errno numbers are architecture specific, so we need to translate 1905 * them into something which is architecture independent. We don't define 1906 * codes for all errno's; just the ones which are most likely to be the cause 1907 * of an ext4_error() call. 1908 */ 1909 #define EXT4_ERR_UNKNOWN 1 1910 #define EXT4_ERR_EIO 2 1911 #define EXT4_ERR_ENOMEM 3 1912 #define EXT4_ERR_EFSBADCRC 4 1913 #define EXT4_ERR_EFSCORRUPTED 5 1914 #define EXT4_ERR_ENOSPC 6 1915 #define EXT4_ERR_ENOKEY 7 1916 #define EXT4_ERR_EROFS 8 1917 #define EXT4_ERR_EFBIG 9 1918 #define EXT4_ERR_EEXIST 10 1919 #define EXT4_ERR_ERANGE 11 1920 #define EXT4_ERR_EOVERFLOW 12 1921 #define EXT4_ERR_EBUSY 13 1922 #define EXT4_ERR_ENOTDIR 14 1923 #define EXT4_ERR_ENOTEMPTY 15 1924 #define EXT4_ERR_ESHUTDOWN 16 1925 #define EXT4_ERR_EFAULT 17 1926 1927 /* 1928 * Inode dynamic state flags 1929 */ 1930 enum { 1931 EXT4_STATE_NEW, /* inode is newly created */ 1932 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1933 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1934 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1935 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1936 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1937 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1938 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1939 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1940 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1941 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1942 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */ 1943 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1944 }; 1945 1946 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1947 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1948 { \ 1949 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1950 } \ 1951 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1952 { \ 1953 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1954 } \ 1955 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1956 { \ 1957 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1958 } 1959 1960 /* Add these declarations here only so that these functions can be 1961 * found by name. Otherwise, they are very hard to locate. */ 1962 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1963 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1964 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1965 EXT4_INODE_BIT_FNS(flag, flags, 0) 1966 1967 /* Add these declarations here only so that these functions can be 1968 * found by name. Otherwise, they are very hard to locate. */ 1969 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1970 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1971 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1972 #if (BITS_PER_LONG < 64) 1973 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1974 1975 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1976 { 1977 (ei)->i_state_flags = 0; 1978 } 1979 #else 1980 EXT4_INODE_BIT_FNS(state, flags, 32) 1981 1982 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1983 { 1984 /* We depend on the fact that callers will set i_flags */ 1985 } 1986 #endif 1987 #else 1988 /* Assume that user mode programs are passing in an ext4fs superblock, not 1989 * a kernel struct super_block. This will allow us to call the feature-test 1990 * macros from user land. */ 1991 #define EXT4_SB(sb) (sb) 1992 #endif 1993 1994 static inline bool ext4_verity_in_progress(struct inode *inode) 1995 { 1996 return IS_ENABLED(CONFIG_FS_VERITY) && 1997 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1998 } 1999 2000 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 2001 2002 /* 2003 * Codes for operating systems 2004 */ 2005 #define EXT4_OS_LINUX 0 2006 #define EXT4_OS_HURD 1 2007 #define EXT4_OS_MASIX 2 2008 #define EXT4_OS_FREEBSD 3 2009 #define EXT4_OS_LITES 4 2010 2011 /* 2012 * Revision levels 2013 */ 2014 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 2015 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 2016 2017 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 2018 2019 #define EXT4_GOOD_OLD_INODE_SIZE 128 2020 2021 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 2022 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 2023 #define EXT4_TIMESTAMP_MIN S32_MIN 2024 2025 /* 2026 * Feature set definitions 2027 */ 2028 2029 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 2030 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2031 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2032 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2033 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2034 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2035 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2036 /* 2037 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2038 * incompatible only if fast commit blocks are present in the FS. Since we 2039 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2040 * incompatible. We also have a JBD2 incompat feature, which gets set when 2041 * there are fast commit blocks present in the journal. 2042 */ 2043 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2044 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2045 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2046 2047 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2048 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2049 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2050 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2051 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2052 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2053 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2054 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2055 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2056 /* 2057 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2058 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2059 * all other data structures' checksums. However, the METADATA_CSUM and 2060 * GDT_CSUM bits are mutually exclusive. 2061 */ 2062 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2063 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2064 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2065 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2066 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2067 non-empty */ 2068 2069 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2070 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2071 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2072 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2073 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2074 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2075 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2076 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2077 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2078 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2079 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2080 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2081 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2082 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2083 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2084 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2085 2086 extern void ext4_update_dynamic_rev(struct super_block *sb); 2087 2088 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2089 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2090 { \ 2091 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2092 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2093 } \ 2094 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2095 { \ 2096 ext4_update_dynamic_rev(sb); \ 2097 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2098 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2099 } \ 2100 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2101 { \ 2102 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2103 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2104 } 2105 2106 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2107 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2108 { \ 2109 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2110 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2111 } \ 2112 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2113 { \ 2114 ext4_update_dynamic_rev(sb); \ 2115 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2116 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2117 } \ 2118 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2119 { \ 2120 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2121 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2122 } 2123 2124 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2125 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2126 { \ 2127 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2128 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2129 } \ 2130 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2131 { \ 2132 ext4_update_dynamic_rev(sb); \ 2133 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2134 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2135 } \ 2136 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2137 { \ 2138 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2139 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2140 } 2141 2142 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2143 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2144 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2145 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2146 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2147 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2148 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2149 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2150 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2151 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2152 2153 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2154 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2155 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2156 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2157 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2158 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2159 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2160 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2161 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2162 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2163 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2164 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2165 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2166 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2167 2168 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2169 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2170 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2171 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2172 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2173 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2174 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2175 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2176 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2177 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2178 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2179 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2180 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2181 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2182 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2183 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2184 2185 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2186 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2187 EXT4_FEATURE_INCOMPAT_META_BG) 2188 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2189 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2190 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2191 2192 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2193 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2194 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2195 EXT4_FEATURE_INCOMPAT_META_BG) 2196 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2197 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2198 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2199 2200 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2201 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2202 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2203 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2204 EXT4_FEATURE_INCOMPAT_META_BG| \ 2205 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2206 EXT4_FEATURE_INCOMPAT_64BIT| \ 2207 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2208 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2209 EXT4_FEATURE_INCOMPAT_MMP | \ 2210 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2211 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2212 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2213 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2214 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2215 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2216 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2217 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2218 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2219 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2220 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2221 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2222 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2223 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2224 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2225 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2226 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2227 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2228 2229 #define EXTN_FEATURE_FUNCS(ver) \ 2230 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2231 { \ 2232 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2233 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2234 } \ 2235 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2236 { \ 2237 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2238 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2239 } \ 2240 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2241 { \ 2242 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2243 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2244 } 2245 2246 EXTN_FEATURE_FUNCS(2) 2247 EXTN_FEATURE_FUNCS(3) 2248 EXTN_FEATURE_FUNCS(4) 2249 2250 static inline bool ext4_has_compat_features(struct super_block *sb) 2251 { 2252 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2253 } 2254 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2255 { 2256 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2257 } 2258 static inline bool ext4_has_incompat_features(struct super_block *sb) 2259 { 2260 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2261 } 2262 2263 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2264 2265 /* 2266 * Superblock flags 2267 */ 2268 enum { 2269 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */ 2270 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */ 2271 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */ 2272 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */ 2273 }; 2274 2275 static inline int ext4_forced_shutdown(struct super_block *sb) 2276 { 2277 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2278 } 2279 2280 static inline int ext4_emergency_ro(struct super_block *sb) 2281 { 2282 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags); 2283 } 2284 2285 static inline int ext4_emergency_state(struct super_block *sb) 2286 { 2287 if (unlikely(ext4_forced_shutdown(sb))) 2288 return -EIO; 2289 if (unlikely(ext4_emergency_ro(sb))) 2290 return -EROFS; 2291 return 0; 2292 } 2293 2294 /* 2295 * Default values for user and/or group using reserved blocks 2296 */ 2297 #define EXT4_DEF_RESUID 0 2298 #define EXT4_DEF_RESGID 0 2299 2300 /* 2301 * Default project ID 2302 */ 2303 #define EXT4_DEF_PROJID 0 2304 2305 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2306 2307 /* 2308 * Default mount options 2309 */ 2310 #define EXT4_DEFM_DEBUG 0x0001 2311 #define EXT4_DEFM_BSDGROUPS 0x0002 2312 #define EXT4_DEFM_XATTR_USER 0x0004 2313 #define EXT4_DEFM_ACL 0x0008 2314 #define EXT4_DEFM_UID16 0x0010 2315 #define EXT4_DEFM_JMODE 0x0060 2316 #define EXT4_DEFM_JMODE_DATA 0x0020 2317 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2318 #define EXT4_DEFM_JMODE_WBACK 0x0060 2319 #define EXT4_DEFM_NOBARRIER 0x0100 2320 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2321 #define EXT4_DEFM_DISCARD 0x0400 2322 #define EXT4_DEFM_NODELALLOC 0x0800 2323 2324 /* 2325 * Default journal batch times and ioprio. 2326 */ 2327 #define EXT4_DEF_MIN_BATCH_TIME 0 2328 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2329 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) 2330 2331 2332 /* 2333 * Default values for superblock update 2334 */ 2335 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */ 2336 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */ 2337 2338 2339 /* 2340 * Minimum number of groups in a flexgroup before we separate out 2341 * directories into the first block group of a flexgroup 2342 */ 2343 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2344 2345 /* 2346 * Structure of a directory entry 2347 */ 2348 #define EXT4_NAME_LEN 255 2349 /* 2350 * Base length of the ext4 directory entry excluding the name length 2351 */ 2352 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2353 2354 struct ext4_dir_entry { 2355 __le32 inode; /* Inode number */ 2356 __le16 rec_len; /* Directory entry length */ 2357 __le16 name_len; /* Name length */ 2358 char name[EXT4_NAME_LEN]; /* File name */ 2359 }; 2360 2361 2362 /* 2363 * Encrypted Casefolded entries require saving the hash on disk. This structure 2364 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2365 * boundary. 2366 */ 2367 struct ext4_dir_entry_hash { 2368 __le32 hash; 2369 __le32 minor_hash; 2370 }; 2371 2372 /* 2373 * The new version of the directory entry. Since EXT4 structures are 2374 * stored in intel byte order, and the name_len field could never be 2375 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2376 * file_type field. 2377 */ 2378 struct ext4_dir_entry_2 { 2379 __le32 inode; /* Inode number */ 2380 __le16 rec_len; /* Directory entry length */ 2381 __u8 name_len; /* Name length */ 2382 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2383 char name[EXT4_NAME_LEN]; /* File name */ 2384 }; 2385 2386 /* 2387 * Access the hashes at the end of ext4_dir_entry_2 2388 */ 2389 #define EXT4_DIRENT_HASHES(entry) \ 2390 ((struct ext4_dir_entry_hash *) \ 2391 (((void *)(entry)) + \ 2392 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2393 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash) 2394 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2395 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash) 2396 2397 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2398 { 2399 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2400 } 2401 2402 /* 2403 * This is a bogus directory entry at the end of each leaf block that 2404 * records checksums. 2405 */ 2406 struct ext4_dir_entry_tail { 2407 __le32 det_reserved_zero1; /* Pretend to be unused */ 2408 __le16 det_rec_len; /* 12 */ 2409 __u8 det_reserved_zero2; /* Zero name length */ 2410 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2411 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2412 }; 2413 2414 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2415 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2416 ((blocksize) - \ 2417 sizeof(struct ext4_dir_entry_tail)))) 2418 2419 /* 2420 * Ext4 directory file types. Only the low 3 bits are used. The 2421 * other bits are reserved for now. 2422 */ 2423 #define EXT4_FT_UNKNOWN 0 2424 #define EXT4_FT_REG_FILE 1 2425 #define EXT4_FT_DIR 2 2426 #define EXT4_FT_CHRDEV 3 2427 #define EXT4_FT_BLKDEV 4 2428 #define EXT4_FT_FIFO 5 2429 #define EXT4_FT_SOCK 6 2430 #define EXT4_FT_SYMLINK 7 2431 2432 #define EXT4_FT_MAX 8 2433 2434 #define EXT4_FT_DIR_CSUM 0xDE 2435 2436 /* 2437 * EXT4_DIR_PAD defines the directory entries boundaries 2438 * 2439 * NOTE: It must be a multiple of 4 2440 */ 2441 #define EXT4_DIR_PAD 4 2442 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2443 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2444 2445 /* 2446 * The rec_len is dependent on the type of directory. Directories that are 2447 * casefolded and encrypted need to store the hash as well, so we add room for 2448 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2449 * pass NULL for dir, as those entries do not use the extra fields. 2450 */ 2451 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2452 const struct inode *dir) 2453 { 2454 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2455 2456 if (dir && ext4_hash_in_dirent(dir)) 2457 rec_len += sizeof(struct ext4_dir_entry_hash); 2458 return (rec_len & ~EXT4_DIR_ROUND); 2459 } 2460 2461 /* 2462 * If we ever get support for fs block sizes > page_size, we'll need 2463 * to remove the #if statements in the next two functions... 2464 */ 2465 static inline unsigned int 2466 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2467 { 2468 unsigned len = le16_to_cpu(dlen); 2469 2470 #if (PAGE_SIZE >= 65536) 2471 if (len == EXT4_MAX_REC_LEN || len == 0) 2472 return blocksize; 2473 return (len & 65532) | ((len & 3) << 16); 2474 #else 2475 return len; 2476 #endif 2477 } 2478 2479 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2480 { 2481 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2482 #if (PAGE_SIZE >= 65536) 2483 if (len < 65536) 2484 return cpu_to_le16(len); 2485 if (len == blocksize) { 2486 if (blocksize == 65536) 2487 return cpu_to_le16(EXT4_MAX_REC_LEN); 2488 else 2489 return cpu_to_le16(0); 2490 } 2491 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2492 #else 2493 return cpu_to_le16(len); 2494 #endif 2495 } 2496 2497 /* 2498 * Hash Tree Directory indexing 2499 * (c) Daniel Phillips, 2001 2500 */ 2501 2502 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2503 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2504 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2505 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2506 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2507 2508 /* Legal values for the dx_root hash_version field: */ 2509 2510 #define DX_HASH_LEGACY 0 2511 #define DX_HASH_HALF_MD4 1 2512 #define DX_HASH_TEA 2 2513 #define DX_HASH_LEGACY_UNSIGNED 3 2514 #define DX_HASH_HALF_MD4_UNSIGNED 4 2515 #define DX_HASH_TEA_UNSIGNED 5 2516 #define DX_HASH_SIPHASH 6 2517 #define DX_HASH_LAST DX_HASH_SIPHASH 2518 2519 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length) 2520 { 2521 return crc32c(crc, address, length); 2522 } 2523 2524 #ifdef __KERNEL__ 2525 2526 /* hash info structure used by the directory hash */ 2527 struct dx_hash_info 2528 { 2529 u32 hash; 2530 u32 minor_hash; 2531 int hash_version; 2532 u32 *seed; 2533 }; 2534 2535 2536 /* 32 and 64 bit signed EOF for dx directories */ 2537 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2538 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2539 2540 2541 /* 2542 * Control parameters used by ext4_htree_next_block 2543 */ 2544 #define HASH_NB_ALWAYS 1 2545 2546 struct ext4_filename { 2547 const struct qstr *usr_fname; 2548 struct fscrypt_str disk_name; 2549 struct dx_hash_info hinfo; 2550 #ifdef CONFIG_FS_ENCRYPTION 2551 struct fscrypt_str crypto_buf; 2552 #endif 2553 #if IS_ENABLED(CONFIG_UNICODE) 2554 struct qstr cf_name; 2555 #endif 2556 }; 2557 2558 #define fname_name(p) ((p)->disk_name.name) 2559 #define fname_usr_name(p) ((p)->usr_fname->name) 2560 #define fname_len(p) ((p)->disk_name.len) 2561 2562 /* 2563 * Describe an inode's exact location on disk and in memory 2564 */ 2565 struct ext4_iloc 2566 { 2567 struct buffer_head *bh; 2568 unsigned long offset; 2569 ext4_group_t block_group; 2570 }; 2571 2572 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2573 { 2574 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2575 } 2576 2577 static inline bool ext4_is_quota_file(struct inode *inode) 2578 { 2579 return IS_NOQUOTA(inode) && 2580 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2581 } 2582 2583 /* 2584 * This structure is stuffed into the struct file's private_data field 2585 * for directories. It is where we put information so that we can do 2586 * readdir operations in hash tree order. 2587 */ 2588 struct dir_private_info { 2589 struct rb_root root; 2590 struct rb_node *curr_node; 2591 struct fname *extra_fname; 2592 loff_t last_pos; 2593 __u32 curr_hash; 2594 __u32 curr_minor_hash; 2595 __u32 next_hash; 2596 u64 cookie; 2597 bool initialized; 2598 }; 2599 2600 /* calculate the first block number of the group */ 2601 static inline ext4_fsblk_t 2602 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2603 { 2604 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2605 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2606 } 2607 2608 /* 2609 * Special error return code only used by dx_probe() and its callers. 2610 */ 2611 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2612 2613 /* htree levels for ext4 */ 2614 #define EXT4_HTREE_LEVEL_COMPAT 2 2615 #define EXT4_HTREE_LEVEL 3 2616 2617 static inline int ext4_dir_htree_level(struct super_block *sb) 2618 { 2619 return ext4_has_feature_largedir(sb) ? 2620 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2621 } 2622 2623 /* 2624 * Timeout and state flag for lazy initialization inode thread. 2625 */ 2626 #define EXT4_DEF_LI_WAIT_MULT 10 2627 #define EXT4_DEF_LI_MAX_START_DELAY 5 2628 #define EXT4_LAZYINIT_QUIT 0x0001 2629 #define EXT4_LAZYINIT_RUNNING 0x0002 2630 2631 /* 2632 * Lazy inode table initialization info 2633 */ 2634 struct ext4_lazy_init { 2635 unsigned long li_state; 2636 struct list_head li_request_list; 2637 struct mutex li_list_mtx; 2638 }; 2639 2640 enum ext4_li_mode { 2641 EXT4_LI_MODE_PREFETCH_BBITMAP, 2642 EXT4_LI_MODE_ITABLE, 2643 }; 2644 2645 struct ext4_li_request { 2646 struct super_block *lr_super; 2647 enum ext4_li_mode lr_mode; 2648 ext4_group_t lr_first_not_zeroed; 2649 ext4_group_t lr_next_group; 2650 struct list_head lr_request; 2651 unsigned long lr_next_sched; 2652 unsigned long lr_timeout; 2653 }; 2654 2655 struct ext4_features { 2656 struct kobject f_kobj; 2657 struct completion f_kobj_unregister; 2658 }; 2659 2660 /* 2661 * This structure will be used for multiple mount protection. It will be 2662 * written into the block number saved in the s_mmp_block field in the 2663 * superblock. Programs that check MMP should assume that if 2664 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2665 * to use the filesystem, regardless of how old the timestamp is. 2666 */ 2667 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2668 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2669 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2670 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2671 2672 struct mmp_struct { 2673 __le32 mmp_magic; /* Magic number for MMP */ 2674 __le32 mmp_seq; /* Sequence no. updated periodically */ 2675 2676 /* 2677 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2678 * purposes and do not affect the correctness of the algorithm 2679 */ 2680 __le64 mmp_time; /* Time last updated */ 2681 char mmp_nodename[64]; /* Node which last updated MMP block */ 2682 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2683 2684 /* 2685 * mmp_check_interval is used to verify if the MMP block has been 2686 * updated on the block device. The value is updated based on the 2687 * maximum time to write the MMP block during an update cycle. 2688 */ 2689 __le16 mmp_check_interval; 2690 2691 __le16 mmp_pad1; 2692 __le32 mmp_pad2[226]; 2693 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2694 }; 2695 2696 /* arguments passed to the mmp thread */ 2697 struct mmpd_data { 2698 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2699 struct super_block *sb; /* super block of the fs */ 2700 }; 2701 2702 /* 2703 * Check interval multiplier 2704 * The MMP block is written every update interval and initially checked every 2705 * update interval x the multiplier (the value is then adapted based on the 2706 * write latency). The reason is that writes can be delayed under load and we 2707 * don't want readers to incorrectly assume that the filesystem is no longer 2708 * in use. 2709 */ 2710 #define EXT4_MMP_CHECK_MULT 2UL 2711 2712 /* 2713 * Minimum interval for MMP checking in seconds. 2714 */ 2715 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2716 2717 /* 2718 * Maximum interval for MMP checking in seconds. 2719 */ 2720 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2721 2722 /* 2723 * Function prototypes 2724 */ 2725 2726 /* 2727 * Ok, these declarations are also in <linux/kernel.h> but none of the 2728 * ext4 source programs needs to include it so they are duplicated here. 2729 */ 2730 # define NORET_TYPE /**/ 2731 # define ATTRIB_NORET __attribute__((noreturn)) 2732 # define NORET_AND noreturn, 2733 2734 /* bitmap.c */ 2735 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2736 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2737 struct ext4_group_desc *gdp, 2738 struct buffer_head *bh); 2739 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2740 struct ext4_group_desc *gdp, 2741 struct buffer_head *bh); 2742 void ext4_block_bitmap_csum_set(struct super_block *sb, 2743 struct ext4_group_desc *gdp, 2744 struct buffer_head *bh); 2745 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2746 struct ext4_group_desc *gdp, 2747 struct buffer_head *bh); 2748 2749 /* balloc.c */ 2750 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2751 ext4_fsblk_t blocknr, 2752 ext4_group_t *blockgrpp, 2753 ext4_grpblk_t *offsetp); 2754 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2755 ext4_fsblk_t block); 2756 2757 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2758 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2759 ext4_group_t group); 2760 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2761 ext4_fsblk_t goal, 2762 unsigned int flags, 2763 unsigned long *count, 2764 int *errp); 2765 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2766 s64 nclusters, unsigned int flags); 2767 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2768 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2769 ext4_group_t block_group, 2770 struct buffer_head ** bh); 2771 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2772 ext4_group_t group); 2773 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2774 2775 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2776 ext4_group_t block_group, 2777 bool ignore_locked); 2778 extern int ext4_wait_block_bitmap(struct super_block *sb, 2779 ext4_group_t block_group, 2780 struct buffer_head *bh); 2781 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2782 ext4_group_t block_group); 2783 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2784 ext4_group_t block_group, 2785 struct ext4_group_desc *gdp); 2786 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2787 2788 #if IS_ENABLED(CONFIG_UNICODE) 2789 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2790 const struct qstr *iname, 2791 struct ext4_filename *fname); 2792 2793 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2794 { 2795 kfree(fname->cf_name.name); 2796 fname->cf_name.name = NULL; 2797 } 2798 #else 2799 static inline int ext4_fname_setup_ci_filename(struct inode *dir, 2800 const struct qstr *iname, 2801 struct ext4_filename *fname) 2802 { 2803 return 0; 2804 } 2805 2806 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2807 { 2808 } 2809 #endif 2810 2811 /* ext4 encryption related stuff goes here crypto.c */ 2812 #ifdef CONFIG_FS_ENCRYPTION 2813 extern const struct fscrypt_operations ext4_cryptops; 2814 2815 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2816 int lookup, struct ext4_filename *fname); 2817 2818 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2819 struct ext4_filename *fname); 2820 2821 void ext4_fname_free_filename(struct ext4_filename *fname); 2822 2823 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2824 2825 #else /* !CONFIG_FS_ENCRYPTION */ 2826 static inline int ext4_fname_setup_filename(struct inode *dir, 2827 const struct qstr *iname, 2828 int lookup, 2829 struct ext4_filename *fname) 2830 { 2831 fname->usr_fname = iname; 2832 fname->disk_name.name = (unsigned char *) iname->name; 2833 fname->disk_name.len = iname->len; 2834 2835 return ext4_fname_setup_ci_filename(dir, iname, fname); 2836 } 2837 2838 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2839 struct dentry *dentry, 2840 struct ext4_filename *fname) 2841 { 2842 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2843 } 2844 2845 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2846 { 2847 ext4_fname_free_ci_filename(fname); 2848 } 2849 2850 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2851 void __user *arg) 2852 { 2853 return -EOPNOTSUPP; 2854 } 2855 #endif /* !CONFIG_FS_ENCRYPTION */ 2856 2857 /* dir.c */ 2858 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2859 struct file *, 2860 struct ext4_dir_entry_2 *, 2861 struct buffer_head *, char *, int, 2862 unsigned int); 2863 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2864 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2865 (de), (bh), (buf), (size), (offset))) 2866 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2867 __u32 minor_hash, 2868 struct ext4_dir_entry_2 *dirent, 2869 struct fscrypt_str *ent_name); 2870 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2871 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh, 2872 void *buf, int buf_size, 2873 struct ext4_filename *fname, 2874 struct ext4_dir_entry_2 **dest_de); 2875 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2876 struct ext4_dir_entry_2 *de, 2877 int buf_size, 2878 struct ext4_filename *fname); 2879 static inline void ext4_update_dx_flag(struct inode *inode) 2880 { 2881 if (!ext4_has_feature_dir_index(inode->i_sb) && 2882 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2883 /* ext4_iget() should have caught this... */ 2884 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2885 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2886 } 2887 } 2888 static const unsigned char ext4_filetype_table[] = { 2889 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2890 }; 2891 2892 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2893 { 2894 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2895 return DT_UNKNOWN; 2896 2897 return ext4_filetype_table[filetype]; 2898 } 2899 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2900 void *buf, int buf_size); 2901 2902 /* fsync.c */ 2903 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2904 2905 /* hash.c */ 2906 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2907 struct dx_hash_info *hinfo); 2908 2909 /* ialloc.c */ 2910 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2911 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2912 struct inode *, umode_t, 2913 const struct qstr *qstr, __u32 goal, 2914 uid_t *owner, __u32 i_flags, 2915 int handle_type, unsigned int line_no, 2916 int nblocks); 2917 2918 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2919 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2920 (goal), (owner), i_flags, 0, 0, 0) 2921 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2922 type, nblocks) \ 2923 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2924 0, (type), __LINE__, (nblocks)) 2925 2926 2927 extern void ext4_free_inode(handle_t *, struct inode *); 2928 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2929 extern unsigned long ext4_count_free_inodes(struct super_block *); 2930 extern unsigned long ext4_count_dirs(struct super_block *); 2931 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2932 extern int ext4_init_inode_table(struct super_block *sb, 2933 ext4_group_t group, int barrier); 2934 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2935 2936 /* fast_commit.c */ 2937 int ext4_fc_info_show(struct seq_file *seq, void *v); 2938 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2939 void ext4_fc_init_inode(struct inode *inode); 2940 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2941 ext4_lblk_t end); 2942 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2943 struct dentry *dentry); 2944 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2945 struct dentry *dentry); 2946 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2947 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2948 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2949 struct dentry *dentry); 2950 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2951 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2952 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2953 void ext4_fc_del(struct inode *inode); 2954 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2955 void ext4_fc_replay_cleanup(struct super_block *sb); 2956 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2957 int __init ext4_fc_init_dentry_cache(void); 2958 void ext4_fc_destroy_dentry_cache(void); 2959 int ext4_fc_record_regions(struct super_block *sb, int ino, 2960 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2961 int len, int replay); 2962 2963 /* mballoc.c */ 2964 extern const struct seq_operations ext4_mb_seq_groups_ops; 2965 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2966 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2967 extern int ext4_mb_init(struct super_block *); 2968 extern void ext4_mb_release(struct super_block *); 2969 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2970 struct ext4_allocation_request *, int *); 2971 extern void ext4_discard_preallocations(struct inode *); 2972 extern int __init ext4_init_mballoc(void); 2973 extern void ext4_exit_mballoc(void); 2974 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2975 ext4_group_t group, 2976 unsigned int nr, int *cnt); 2977 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2978 unsigned int nr); 2979 2980 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2981 struct buffer_head *bh, ext4_fsblk_t block, 2982 unsigned long count, int flags); 2983 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2984 ext4_group_t ngroups); 2985 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2986 ext4_group_t i, struct ext4_group_desc *desc); 2987 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2988 ext4_fsblk_t block, unsigned long count); 2989 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2990 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2991 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2992 int len, bool state); 2993 static inline bool ext4_mb_cr_expensive(enum criteria cr) 2994 { 2995 return cr >= CR_GOAL_LEN_SLOW; 2996 } 2997 2998 /* inode.c */ 2999 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 3000 struct ext4_inode_info *ei); 3001 int ext4_inode_is_fast_symlink(struct inode *inode); 3002 void ext4_check_map_extents_env(struct inode *inode); 3003 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 3004 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 3005 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 3006 bool wait, struct buffer_head **bhs); 3007 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 3008 struct buffer_head *bh_result, int create); 3009 int ext4_get_block(struct inode *inode, sector_t iblock, 3010 struct buffer_head *bh_result, int create); 3011 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 3012 struct buffer_head *bh, int create); 3013 int ext4_walk_page_buffers(handle_t *handle, 3014 struct inode *inode, 3015 struct buffer_head *head, 3016 unsigned from, 3017 unsigned to, 3018 int *partial, 3019 int (*fn)(handle_t *handle, struct inode *inode, 3020 struct buffer_head *bh)); 3021 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 3022 struct buffer_head *bh); 3023 bool ext4_should_enable_large_folio(struct inode *inode); 3024 #define FALL_BACK_TO_NONDELALLOC 1 3025 #define CONVERT_INLINE_DATA 2 3026 3027 typedef enum { 3028 EXT4_IGET_NORMAL = 0, 3029 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3030 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 3031 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 3032 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 3033 } ext4_iget_flags; 3034 3035 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3036 ext4_iget_flags flags, const char *function, 3037 unsigned int line); 3038 3039 #define ext4_iget(sb, ino, flags) \ 3040 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3041 3042 extern int ext4_write_inode(struct inode *, struct writeback_control *); 3043 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 3044 struct iattr *); 3045 extern u32 ext4_dio_alignment(struct inode *inode); 3046 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 3047 struct kstat *, u32, unsigned int); 3048 extern void ext4_evict_inode(struct inode *); 3049 extern void ext4_clear_inode(struct inode *); 3050 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 3051 struct kstat *, u32, unsigned int); 3052 extern void ext4_dirty_inode(struct inode *, int); 3053 extern int ext4_change_inode_journal_flag(struct inode *, int); 3054 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3055 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3056 struct ext4_iloc *iloc); 3057 extern int ext4_inode_attach_jinode(struct inode *inode); 3058 extern int ext4_can_truncate(struct inode *inode); 3059 extern int ext4_truncate(struct inode *); 3060 extern int ext4_break_layouts(struct inode *); 3061 extern int ext4_truncate_page_cache_block_range(struct inode *inode, 3062 loff_t start, loff_t end); 3063 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3064 extern void ext4_set_inode_flags(struct inode *, bool init); 3065 extern int ext4_alloc_da_blocks(struct inode *inode); 3066 extern void ext4_set_aops(struct inode *inode); 3067 extern int ext4_writepage_trans_blocks(struct inode *); 3068 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3069 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3070 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks, 3071 int pextents); 3072 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3073 loff_t lstart, loff_t lend); 3074 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3075 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3076 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3077 extern void ext4_da_release_space(struct inode *inode, int to_free); 3078 extern void ext4_da_update_reserve_space(struct inode *inode, 3079 int used, int quota_claim); 3080 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3081 ext4_fsblk_t pblk, ext4_lblk_t len); 3082 3083 static inline bool is_special_ino(struct super_block *sb, unsigned long ino) 3084 { 3085 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 3086 3087 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) || 3088 ino == le32_to_cpu(es->s_usr_quota_inum) || 3089 ino == le32_to_cpu(es->s_grp_quota_inum) || 3090 ino == le32_to_cpu(es->s_prj_quota_inum) || 3091 ino == le32_to_cpu(es->s_orphan_file_inum); 3092 } 3093 3094 /* indirect.c */ 3095 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3096 struct ext4_map_blocks *map, int flags); 3097 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3098 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3099 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3100 ext4_lblk_t start, ext4_lblk_t end); 3101 3102 /* ioctl.c */ 3103 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3104 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3105 int ext4_fileattr_set(struct mnt_idmap *idmap, 3106 struct dentry *dentry, struct fileattr *fa); 3107 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3108 extern void ext4_reset_inode_seed(struct inode *inode); 3109 int ext4_update_overhead(struct super_block *sb, bool force); 3110 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3111 3112 /* migrate.c */ 3113 extern int ext4_ext_migrate(struct inode *); 3114 extern int ext4_ind_migrate(struct inode *inode); 3115 3116 /* namei.c */ 3117 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3118 struct inode *inode); 3119 extern int ext4_dirblock_csum_verify(struct inode *inode, 3120 struct buffer_head *bh); 3121 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3122 __u32 start_minor_hash, __u32 *next_hash); 3123 extern int ext4_search_dir(struct buffer_head *bh, 3124 char *search_buf, 3125 int buf_size, 3126 struct inode *dir, 3127 struct ext4_filename *fname, 3128 unsigned int offset, 3129 struct ext4_dir_entry_2 **res_dir); 3130 extern int ext4_generic_delete_entry(struct inode *dir, 3131 struct ext4_dir_entry_2 *de_del, 3132 struct buffer_head *bh, 3133 void *entry_buf, 3134 int buf_size, 3135 int csum_size); 3136 extern bool ext4_empty_dir(struct inode *inode); 3137 3138 /* resize.c */ 3139 extern void ext4_kvfree_array_rcu(void *to_free); 3140 extern int ext4_group_add(struct super_block *sb, 3141 struct ext4_new_group_data *input); 3142 extern int ext4_group_extend(struct super_block *sb, 3143 struct ext4_super_block *es, 3144 ext4_fsblk_t n_blocks_count); 3145 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3146 extern unsigned int ext4_list_backups(struct super_block *sb, 3147 unsigned int *three, unsigned int *five, 3148 unsigned int *seven); 3149 3150 /* super.c */ 3151 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3152 sector_t block, blk_opf_t op_flags); 3153 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3154 sector_t block); 3155 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3156 bh_end_io_t *end_io, bool simu_fail); 3157 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3158 bh_end_io_t *end_io, bool simu_fail); 3159 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3160 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3161 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3162 extern int ext4_calculate_overhead(struct super_block *sb); 3163 extern __le32 ext4_superblock_csum(struct ext4_super_block *es); 3164 extern void ext4_superblock_csum_set(struct super_block *sb); 3165 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3166 ext4_group_t ngroup); 3167 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3168 char nbuf[16]); 3169 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3170 ext4_group_t block_group, 3171 unsigned int flags); 3172 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3173 ext4_group_t block_group); 3174 3175 extern __printf(7, 8) 3176 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3177 int, __u64, const char *, ...); 3178 extern __printf(6, 7) 3179 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3180 ext4_fsblk_t, int, const char *, ...); 3181 extern __printf(5, 6) 3182 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3183 const char *, ...); 3184 extern void __ext4_std_error(struct super_block *, const char *, 3185 unsigned int, int); 3186 extern __printf(4, 5) 3187 void __ext4_warning(struct super_block *, const char *, unsigned int, 3188 const char *, ...); 3189 extern __printf(4, 5) 3190 void __ext4_warning_inode(const struct inode *inode, const char *function, 3191 unsigned int line, const char *fmt, ...); 3192 extern __printf(3, 4) 3193 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3194 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3195 const char *, unsigned int, const char *); 3196 extern __printf(7, 8) 3197 void __ext4_grp_locked_error(const char *, unsigned int, 3198 struct super_block *, ext4_group_t, 3199 unsigned long, ext4_fsblk_t, 3200 const char *, ...); 3201 3202 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3203 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3204 3205 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3206 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3207 3208 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3209 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3210 (fmt), ## a) 3211 3212 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3213 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3214 3215 #define ext4_abort(sb, err, fmt, a...) \ 3216 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3217 3218 #ifdef CONFIG_PRINTK 3219 3220 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3221 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3222 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3223 __ext4_error_inode((inode), (func), (line), (block), \ 3224 (err), (fmt), ##__VA_ARGS__) 3225 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3226 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3227 #define ext4_error(sb, fmt, ...) \ 3228 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3229 ##__VA_ARGS__) 3230 #define ext4_error_err(sb, err, fmt, ...) \ 3231 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3232 ##__VA_ARGS__) 3233 #define ext4_warning(sb, fmt, ...) \ 3234 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3235 #define ext4_warning_inode(inode, fmt, ...) \ 3236 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3237 #define ext4_msg(sb, level, fmt, ...) \ 3238 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3239 #define dump_mmp_msg(sb, mmp, msg) \ 3240 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3241 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3242 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3243 fmt, ##__VA_ARGS__) 3244 3245 #else 3246 3247 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3248 do { \ 3249 no_printk(fmt, ##__VA_ARGS__); \ 3250 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3251 } while (0) 3252 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3253 do { \ 3254 no_printk(fmt, ##__VA_ARGS__); \ 3255 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3256 } while (0) 3257 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3258 do { \ 3259 no_printk(fmt, ##__VA_ARGS__); \ 3260 __ext4_error_file(file, "", 0, block, " "); \ 3261 } while (0) 3262 #define ext4_error(sb, fmt, ...) \ 3263 do { \ 3264 no_printk(fmt, ##__VA_ARGS__); \ 3265 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3266 } while (0) 3267 #define ext4_error_err(sb, err, fmt, ...) \ 3268 do { \ 3269 no_printk(fmt, ##__VA_ARGS__); \ 3270 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3271 } while (0) 3272 #define ext4_warning(sb, fmt, ...) \ 3273 do { \ 3274 no_printk(fmt, ##__VA_ARGS__); \ 3275 __ext4_warning(sb, "", 0, " "); \ 3276 } while (0) 3277 #define ext4_warning_inode(inode, fmt, ...) \ 3278 do { \ 3279 no_printk(fmt, ##__VA_ARGS__); \ 3280 __ext4_warning_inode(inode, "", 0, " "); \ 3281 } while (0) 3282 #define ext4_msg(sb, level, fmt, ...) \ 3283 do { \ 3284 no_printk(fmt, ##__VA_ARGS__); \ 3285 __ext4_msg(sb, "", " "); \ 3286 } while (0) 3287 #define dump_mmp_msg(sb, mmp, msg) \ 3288 __dump_mmp_msg(sb, mmp, "", 0, "") 3289 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3290 do { \ 3291 no_printk(fmt, ##__VA_ARGS__); \ 3292 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3293 } while (0) 3294 3295 #endif 3296 3297 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3298 struct ext4_group_desc *bg); 3299 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3300 struct ext4_group_desc *bg); 3301 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3302 struct ext4_group_desc *bg); 3303 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3304 struct ext4_group_desc *bg); 3305 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3306 struct ext4_group_desc *bg); 3307 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3308 struct ext4_group_desc *bg); 3309 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3310 struct ext4_group_desc *bg); 3311 extern void ext4_block_bitmap_set(struct super_block *sb, 3312 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3313 extern void ext4_inode_bitmap_set(struct super_block *sb, 3314 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3315 extern void ext4_inode_table_set(struct super_block *sb, 3316 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3317 extern void ext4_free_group_clusters_set(struct super_block *sb, 3318 struct ext4_group_desc *bg, 3319 __u32 count); 3320 extern void ext4_free_inodes_set(struct super_block *sb, 3321 struct ext4_group_desc *bg, __u32 count); 3322 extern void ext4_used_dirs_set(struct super_block *sb, 3323 struct ext4_group_desc *bg, __u32 count); 3324 extern void ext4_itable_unused_set(struct super_block *sb, 3325 struct ext4_group_desc *bg, __u32 count); 3326 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3327 struct ext4_group_desc *gdp); 3328 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3329 struct ext4_group_desc *gdp); 3330 extern int ext4_register_li_request(struct super_block *sb, 3331 ext4_group_t first_not_zeroed); 3332 3333 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3334 { 3335 return ext4_has_feature_gdt_csum(sb) || 3336 ext4_has_feature_metadata_csum(sb); 3337 } 3338 3339 #define ext4_read_incompat_64bit_val(es, name) \ 3340 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3341 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3342 le32_to_cpu(es->name##_lo)) 3343 3344 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3345 { 3346 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3347 } 3348 3349 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3350 { 3351 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3352 } 3353 3354 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3355 { 3356 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3357 } 3358 3359 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3360 ext4_fsblk_t blk) 3361 { 3362 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3363 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3364 } 3365 3366 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3367 ext4_fsblk_t blk) 3368 { 3369 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3370 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3371 } 3372 3373 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3374 ext4_fsblk_t blk) 3375 { 3376 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3377 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3378 } 3379 3380 static inline loff_t ext4_isize(struct super_block *sb, 3381 struct ext4_inode *raw_inode) 3382 { 3383 if (ext4_has_feature_largedir(sb) || 3384 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3385 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3386 le32_to_cpu(raw_inode->i_size_lo); 3387 3388 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3389 } 3390 3391 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3392 { 3393 raw_inode->i_size_lo = cpu_to_le32(i_size); 3394 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3395 } 3396 3397 /* 3398 * Reading s_groups_count requires using smp_rmb() afterwards. See 3399 * the locking protocol documented in the comments of ext4_group_add() 3400 * in resize.c 3401 */ 3402 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3403 { 3404 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3405 3406 smp_rmb(); 3407 return ngroups; 3408 } 3409 3410 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3411 ext4_group_t block_group) 3412 { 3413 return block_group >> sbi->s_log_groups_per_flex; 3414 } 3415 3416 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3417 { 3418 return 1 << sbi->s_log_groups_per_flex; 3419 } 3420 3421 static inline loff_t ext4_get_maxbytes(struct inode *inode) 3422 { 3423 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) 3424 return inode->i_sb->s_maxbytes; 3425 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; 3426 } 3427 3428 #define ext4_std_error(sb, errno) \ 3429 do { \ 3430 if ((errno)) \ 3431 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3432 } while (0) 3433 3434 #ifdef CONFIG_SMP 3435 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3436 * counters. So we need to make sure we have free clusters more 3437 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3438 */ 3439 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3440 #else 3441 #define EXT4_FREECLUSTERS_WATERMARK 0 3442 #endif 3443 3444 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3445 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3446 { 3447 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3448 !inode_is_locked(inode)); 3449 down_write(&EXT4_I(inode)->i_data_sem); 3450 if (newsize > EXT4_I(inode)->i_disksize) 3451 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3452 up_write(&EXT4_I(inode)->i_data_sem); 3453 } 3454 3455 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3456 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3457 { 3458 int changed = 0; 3459 3460 if (newsize > inode->i_size) { 3461 i_size_write(inode, newsize); 3462 changed = 1; 3463 } 3464 if (newsize > EXT4_I(inode)->i_disksize) { 3465 ext4_update_i_disksize(inode, newsize); 3466 changed |= 2; 3467 } 3468 return changed; 3469 } 3470 3471 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3472 loff_t len); 3473 3474 struct ext4_group_info { 3475 unsigned long bb_state; 3476 #ifdef AGGRESSIVE_CHECK 3477 unsigned long bb_check_counter; 3478 #endif 3479 struct rb_root bb_free_root; 3480 ext4_grpblk_t bb_first_free; /* first free block */ 3481 ext4_grpblk_t bb_free; /* total free blocks */ 3482 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3483 int bb_avg_fragment_size_order; /* order of average 3484 fragment in BG */ 3485 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3486 ext4_group_t bb_group; /* Group number */ 3487 struct list_head bb_prealloc_list; 3488 #ifdef DOUBLE_CHECK 3489 void *bb_bitmap; 3490 #endif 3491 struct rw_semaphore alloc_sem; 3492 struct list_head bb_avg_fragment_size_node; 3493 struct list_head bb_largest_free_order_node; 3494 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3495 * regions, index is order. 3496 * bb_counters[3] = 5 means 3497 * 5 free 8-block regions. */ 3498 }; 3499 3500 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3501 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3502 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3503 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3504 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3505 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3506 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3507 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3508 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3509 3510 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3511 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3512 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3513 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3514 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3515 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3516 3517 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3518 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3519 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3520 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3521 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3522 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3523 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3524 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3525 3526 #define EXT4_MAX_CONTENTION 8 3527 #define EXT4_CONTENTION_THRESHOLD 2 3528 3529 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3530 ext4_group_t group) 3531 { 3532 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3533 } 3534 3535 /* 3536 * Returns true if the filesystem is busy enough that attempts to 3537 * access the block group locks has run into contention. 3538 */ 3539 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3540 { 3541 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3542 } 3543 3544 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3545 { 3546 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3547 if (spin_trylock(lock)) 3548 /* 3549 * We're able to grab the lock right away, so drop the 3550 * lock contention counter. 3551 */ 3552 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3553 else { 3554 /* 3555 * The lock is busy, so bump the contention counter, 3556 * and then wait on the spin lock. 3557 */ 3558 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3559 EXT4_MAX_CONTENTION); 3560 spin_lock(lock); 3561 } 3562 } 3563 3564 static inline void ext4_unlock_group(struct super_block *sb, 3565 ext4_group_t group) 3566 { 3567 spin_unlock(ext4_group_lock_ptr(sb, group)); 3568 } 3569 3570 #ifdef CONFIG_QUOTA 3571 static inline bool ext4_quota_capable(struct super_block *sb) 3572 { 3573 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3574 } 3575 3576 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3577 { 3578 struct ext4_sb_info *sbi = EXT4_SB(sb); 3579 3580 return (ext4_has_feature_quota(sb) || 3581 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3582 } 3583 int ext4_enable_quotas(struct super_block *sb); 3584 #endif 3585 3586 /* 3587 * Block validity checking 3588 */ 3589 #define ext4_check_indirect_blockref(inode, bh) \ 3590 ext4_check_blockref(__func__, __LINE__, inode, \ 3591 (__le32 *)(bh)->b_data, \ 3592 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3593 3594 #define ext4_ind_check_inode(inode) \ 3595 ext4_check_blockref(__func__, __LINE__, inode, \ 3596 EXT4_I(inode)->i_data, \ 3597 EXT4_NDIR_BLOCKS) 3598 3599 /* 3600 * Inodes and files operations 3601 */ 3602 3603 /* dir.c */ 3604 extern const struct file_operations ext4_dir_operations; 3605 3606 /* file.c */ 3607 extern const struct inode_operations ext4_file_inode_operations; 3608 extern const struct file_operations ext4_file_operations; 3609 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3610 3611 /* inline.c */ 3612 extern int ext4_get_max_inline_size(struct inode *inode); 3613 extern int ext4_find_inline_data_nolock(struct inode *inode); 3614 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3615 3616 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3617 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3618 struct inode *inode, 3619 loff_t pos, unsigned len, 3620 struct folio **foliop); 3621 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3622 unsigned copied, struct folio *folio); 3623 extern int ext4_generic_write_inline_data(struct address_space *mapping, 3624 struct inode *inode, 3625 loff_t pos, unsigned len, 3626 struct folio **foliop, 3627 void **fsdata, bool da); 3628 extern int ext4_try_add_inline_entry(handle_t *handle, 3629 struct ext4_filename *fname, 3630 struct inode *dir, struct inode *inode); 3631 extern int ext4_try_create_inline_dir(handle_t *handle, 3632 struct inode *parent, 3633 struct inode *inode); 3634 extern int ext4_read_inline_dir(struct file *filp, 3635 struct dir_context *ctx, 3636 int *has_inline_data); 3637 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3638 struct inode *dir, ext4_lblk_t block, 3639 struct dx_hash_info *hinfo, 3640 __u32 start_hash, __u32 start_minor_hash, 3641 int *has_inline_data); 3642 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3643 struct ext4_filename *fname, 3644 struct ext4_dir_entry_2 **res_dir, 3645 int *has_inline_data); 3646 extern int ext4_delete_inline_entry(handle_t *handle, 3647 struct inode *dir, 3648 struct ext4_dir_entry_2 *de_del, 3649 struct buffer_head *bh, 3650 int *has_inline_data); 3651 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3652 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3653 struct ext4_dir_entry_2 **parent_de, 3654 int *retval); 3655 extern void *ext4_read_inline_link(struct inode *inode); 3656 3657 struct iomap; 3658 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3659 3660 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3661 3662 extern int ext4_convert_inline_data(struct inode *inode); 3663 3664 static inline int ext4_has_inline_data(struct inode *inode) 3665 { 3666 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3667 EXT4_I(inode)->i_inline_off; 3668 } 3669 3670 /* namei.c */ 3671 extern const struct inode_operations ext4_dir_inode_operations; 3672 extern const struct inode_operations ext4_special_inode_operations; 3673 extern struct dentry *ext4_get_parent(struct dentry *child); 3674 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3675 struct ext4_dir_entry_2 *de, 3676 int blocksize, int csum_size, 3677 unsigned int parent_ino, int dotdot_real_len); 3678 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3679 unsigned int blocksize); 3680 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3681 struct buffer_head *bh); 3682 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3683 struct inode *inode, struct dentry *dentry); 3684 extern int __ext4_link(struct inode *dir, struct inode *inode, 3685 struct dentry *dentry); 3686 3687 #define S_SHIFT 12 3688 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3689 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3690 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3691 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3692 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3693 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3694 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3695 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3696 }; 3697 3698 static inline void ext4_set_de_type(struct super_block *sb, 3699 struct ext4_dir_entry_2 *de, 3700 umode_t mode) { 3701 if (ext4_has_feature_filetype(sb)) 3702 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3703 } 3704 3705 /* readpages.c */ 3706 extern int ext4_mpage_readpages(struct inode *inode, 3707 struct readahead_control *rac, struct folio *folio); 3708 extern int __init ext4_init_post_read_processing(void); 3709 extern void ext4_exit_post_read_processing(void); 3710 3711 /* symlink.c */ 3712 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3713 extern const struct inode_operations ext4_symlink_inode_operations; 3714 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3715 3716 /* sysfs.c */ 3717 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3718 extern int ext4_register_sysfs(struct super_block *sb); 3719 extern void ext4_unregister_sysfs(struct super_block *sb); 3720 extern int __init ext4_init_sysfs(void); 3721 extern void ext4_exit_sysfs(void); 3722 3723 /* block_validity */ 3724 extern void ext4_release_system_zone(struct super_block *sb); 3725 extern int ext4_setup_system_zone(struct super_block *sb); 3726 extern int __init ext4_init_system_zone(void); 3727 extern void ext4_exit_system_zone(void); 3728 extern int ext4_inode_block_valid(struct inode *inode, 3729 ext4_fsblk_t start_blk, 3730 unsigned int count); 3731 extern int ext4_check_blockref(const char *, unsigned int, 3732 struct inode *, __le32 *, unsigned int); 3733 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3734 ext4_fsblk_t start_blk, unsigned int count); 3735 3736 3737 /* extents.c */ 3738 struct ext4_ext_path; 3739 struct ext4_extent; 3740 3741 /* 3742 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3743 * __le32. 3744 */ 3745 #define EXT_MAX_BLOCKS 0xffffffff 3746 3747 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3748 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3749 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3750 struct ext4_map_blocks *map, int flags); 3751 extern int ext4_ext_truncate(handle_t *, struct inode *); 3752 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3753 ext4_lblk_t end); 3754 extern void ext4_ext_init(struct super_block *); 3755 extern void ext4_ext_release(struct super_block *); 3756 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3757 loff_t len); 3758 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3759 loff_t offset, ssize_t len); 3760 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle, 3761 struct inode *inode, loff_t offset, ssize_t len); 3762 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3763 ext4_io_end_t *io_end); 3764 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3765 struct ext4_map_blocks *map, int flags); 3766 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3767 int num, 3768 struct ext4_ext_path *path); 3769 extern struct ext4_ext_path *ext4_ext_insert_extent( 3770 handle_t *handle, struct inode *inode, 3771 struct ext4_ext_path *path, 3772 struct ext4_extent *newext, int gb_flags); 3773 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3774 struct ext4_ext_path *, 3775 int flags); 3776 extern void ext4_free_ext_path(struct ext4_ext_path *); 3777 extern int ext4_ext_check_inode(struct inode *inode); 3778 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3779 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3780 __u64 start, __u64 len); 3781 extern int ext4_get_es_cache(struct inode *inode, 3782 struct fiemap_extent_info *fieinfo, 3783 __u64 start, __u64 len); 3784 extern int ext4_ext_precache(struct inode *inode); 3785 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3786 struct inode *inode2, ext4_lblk_t lblk1, 3787 ext4_lblk_t lblk2, ext4_lblk_t count, 3788 int mark_unwritten,int *err); 3789 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3790 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3791 int check_cred, int restart_cred, 3792 int revoke_cred); 3793 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3794 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3795 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3796 int len, int unwritten, ext4_fsblk_t pblk); 3797 extern int ext4_ext_clear_bb(struct inode *inode); 3798 3799 3800 /* move_extent.c */ 3801 extern void ext4_double_down_write_data_sem(struct inode *first, 3802 struct inode *second); 3803 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3804 struct inode *donor_inode); 3805 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3806 __u64 start_orig, __u64 start_donor, 3807 __u64 len, __u64 *moved_len); 3808 3809 /* page-io.c */ 3810 extern int __init ext4_init_pageio(void); 3811 extern void ext4_exit_pageio(void); 3812 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3813 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3814 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3815 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3816 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3817 struct writeback_control *wbc); 3818 extern void ext4_end_io_rsv_work(struct work_struct *work); 3819 extern void ext4_io_submit(struct ext4_io_submit *io); 3820 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3821 size_t len); 3822 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3823 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3824 3825 /* mmp.c */ 3826 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3827 3828 /* mmp.c */ 3829 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3830 3831 /* verity.c */ 3832 extern const struct fsverity_operations ext4_verityops; 3833 3834 /* orphan.c */ 3835 extern int ext4_orphan_add(handle_t *, struct inode *); 3836 extern int ext4_orphan_del(handle_t *, struct inode *); 3837 extern void ext4_orphan_cleanup(struct super_block *sb, 3838 struct ext4_super_block *es); 3839 extern void ext4_release_orphan_info(struct super_block *sb); 3840 extern int ext4_init_orphan_info(struct super_block *sb); 3841 extern int ext4_orphan_file_empty(struct super_block *sb); 3842 extern void ext4_orphan_file_block_trigger( 3843 struct jbd2_buffer_trigger_type *triggers, 3844 struct buffer_head *bh, 3845 void *data, size_t size); 3846 3847 /* 3848 * Add new method to test whether block and inode bitmaps are properly 3849 * initialized. With uninit_bg reading the block from disk is not enough 3850 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3851 */ 3852 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3853 3854 static inline int bitmap_uptodate(struct buffer_head *bh) 3855 { 3856 return (buffer_uptodate(bh) && 3857 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3858 } 3859 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3860 { 3861 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3862 } 3863 3864 extern int ext4_resize_begin(struct super_block *sb); 3865 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3866 3867 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end) 3868 { 3869 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) 3870 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3871 } 3872 3873 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3874 { 3875 if (io_end->flag & EXT4_IO_END_UNWRITTEN) 3876 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3877 } 3878 3879 extern const struct iomap_ops ext4_iomap_ops; 3880 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3881 extern const struct iomap_ops ext4_iomap_report_ops; 3882 3883 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3884 { 3885 /* 3886 * If the buffer has the write error flag, we have failed 3887 * to write out data in the block. In this case, we don't 3888 * have to read the block because we may read the old data 3889 * successfully. 3890 */ 3891 if (buffer_write_io_error(bh)) 3892 set_buffer_uptodate(bh); 3893 return buffer_uptodate(bh); 3894 } 3895 3896 static inline bool ext4_inode_can_atomic_write(struct inode *inode) 3897 { 3898 3899 return S_ISREG(inode->i_mode) && 3900 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) && 3901 EXT4_SB(inode->i_sb)->s_awu_min > 0; 3902 } 3903 3904 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio, 3905 loff_t pos, unsigned len, 3906 get_block_t *get_block); 3907 #endif /* __KERNEL__ */ 3908 3909 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3910 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3911 3912 #endif /* _EXT4_H */ 3913