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 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 /* first blocks in the file */ 189 #define EXT4_MB_HINT_FIRST 0x0008 190 /* data is being allocated */ 191 #define EXT4_MB_HINT_DATA 0x0020 192 /* don't preallocate (for tails) */ 193 #define EXT4_MB_HINT_NOPREALLOC 0x0040 194 /* allocate for locality group */ 195 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 196 /* allocate goal blocks or none */ 197 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 198 /* goal is meaningful */ 199 #define EXT4_MB_HINT_TRY_GOAL 0x0200 200 /* blocks already pre-reserved by delayed allocation */ 201 #define EXT4_MB_DELALLOC_RESERVED 0x0400 202 /* We are doing stream allocation */ 203 #define EXT4_MB_STREAM_ALLOC 0x0800 204 /* Use reserved root blocks if needed */ 205 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 206 /* Use blocks from reserved pool */ 207 #define EXT4_MB_USE_RESERVED 0x2000 208 /* Do strict check for free blocks while retrying block allocation */ 209 #define EXT4_MB_STRICT_CHECK 0x4000 210 211 struct ext4_allocation_request { 212 /* target inode for block we're allocating */ 213 struct inode *inode; 214 /* how many blocks we want to allocate */ 215 unsigned int len; 216 /* logical block in target inode */ 217 ext4_lblk_t logical; 218 /* the closest logical allocated block to the left */ 219 ext4_lblk_t lleft; 220 /* the closest logical allocated block to the right */ 221 ext4_lblk_t lright; 222 /* phys. target (a hint) */ 223 ext4_fsblk_t goal; 224 /* phys. block for the closest logical allocated block to the left */ 225 ext4_fsblk_t pleft; 226 /* phys. block for the closest logical allocated block to the right */ 227 ext4_fsblk_t pright; 228 /* flags. see above EXT4_MB_HINT_* */ 229 unsigned int flags; 230 }; 231 232 /* 233 * Logical to physical block mapping, used by ext4_map_blocks() 234 * 235 * This structure is used to pass requests into ext4_map_blocks() as 236 * well as to store the information returned by ext4_map_blocks(). It 237 * takes less room on the stack than a struct buffer_head. 238 */ 239 #define EXT4_MAP_NEW BIT(BH_New) 240 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 241 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 242 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 243 #define EXT4_MAP_DELAYED BIT(BH_Delay) 244 /* 245 * This is for use in ext4_map_query_blocks() for a special case where we can 246 * have a physically and logically contiguous blocks split across two leaf 247 * nodes instead of a single extent. This is required in case of atomic writes 248 * to know whether the returned extent is last in leaf. If yes, then lookup for 249 * next in leaf block in ext4_map_query_blocks_next_in_leaf(). 250 * - This is never going to be added to any buffer head state. 251 * - We use the next available bit after BH_BITMAP_UPTODATE. 252 */ 253 #define EXT4_MAP_QUERY_LAST_IN_LEAF BIT(BH_BITMAP_UPTODATE + 1) 254 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 255 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 256 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF) 257 258 struct ext4_map_blocks { 259 ext4_fsblk_t m_pblk; 260 ext4_lblk_t m_lblk; 261 unsigned int m_len; 262 unsigned int m_flags; 263 u64 m_seq; 264 }; 265 266 /* 267 * Block validity checking, system zone rbtree. 268 */ 269 struct ext4_system_blocks { 270 struct rb_root root; 271 struct rcu_head rcu; 272 }; 273 274 /* 275 * Flags for ext4_io_end->flags 276 */ 277 #define EXT4_IO_END_UNWRITTEN 0x0001 278 #define EXT4_IO_END_FAILED 0x0002 279 280 #define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED) 281 282 struct ext4_io_end_vec { 283 struct list_head list; /* list of io_end_vec */ 284 loff_t offset; /* offset in the file */ 285 ssize_t size; /* size of the extent */ 286 }; 287 288 /* 289 * For converting unwritten extents on a work queue. 'handle' is used for 290 * buffered writeback. 291 */ 292 typedef struct ext4_io_end { 293 struct list_head list; /* per-file finished IO list */ 294 handle_t *handle; /* handle reserved for extent 295 * conversion */ 296 struct inode *inode; /* file being written to */ 297 struct bio *bio; /* Linked list of completed 298 * bios covering the extent */ 299 unsigned int flag; /* unwritten or not */ 300 refcount_t count; /* reference counter */ 301 struct list_head list_vec; /* list of ext4_io_end_vec */ 302 } ext4_io_end_t; 303 304 struct ext4_io_submit { 305 struct writeback_control *io_wbc; 306 struct bio *io_bio; 307 ext4_io_end_t *io_end; 308 sector_t io_next_block; 309 }; 310 311 /* 312 * Special inodes numbers 313 */ 314 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 315 #define EXT4_ROOT_INO 2 /* Root inode */ 316 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 317 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 318 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 319 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 320 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 321 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 322 323 /* First non-reserved inode for old ext4 filesystems */ 324 #define EXT4_GOOD_OLD_FIRST_INO 11 325 326 /* 327 * Maximal count of links to a file 328 */ 329 #define EXT4_LINK_MAX 65000 330 331 /* 332 * Macro-instructions used to manage several block sizes 333 */ 334 #define EXT4_MIN_BLOCK_SIZE 1024 335 #define EXT4_MAX_BLOCK_SIZE 65536 336 #define EXT4_MIN_BLOCK_LOG_SIZE 10 337 #define EXT4_MAX_BLOCK_LOG_SIZE 16 338 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 339 #ifdef __KERNEL__ 340 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 341 #else 342 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 343 #endif 344 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 345 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 346 EXT4_SB(s)->s_cluster_bits) 347 #ifdef __KERNEL__ 348 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 349 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 350 #else 351 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 352 #endif 353 #ifdef __KERNEL__ 354 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 355 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 356 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 357 #else 358 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 359 EXT4_GOOD_OLD_INODE_SIZE : \ 360 (s)->s_inode_size) 361 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 362 EXT4_GOOD_OLD_FIRST_INO : \ 363 (s)->s_first_ino) 364 #endif 365 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 366 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 367 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 368 blkbits)) 369 #define EXT4_B_TO_LBLK(inode, offset) \ 370 (round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits) 371 372 /* Translate a block number to a cluster number */ 373 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 374 /* Translate a cluster number to a block number */ 375 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 376 /* Translate # of blks to # of clusters */ 377 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 378 (sbi)->s_cluster_bits) 379 /* Mask out the low bits to get the starting block of the cluster */ 380 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 381 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 382 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 383 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 384 /* Fill in the low bits to get the last block of the cluster */ 385 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 386 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 387 /* Get the cluster offset */ 388 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 389 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 390 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 391 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 392 393 /* 394 * Structure of a blocks group descriptor 395 */ 396 struct ext4_group_desc 397 { 398 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 399 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 400 __le32 bg_inode_table_lo; /* Inodes table block */ 401 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 402 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 403 __le16 bg_used_dirs_count_lo; /* Directories count */ 404 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 405 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 406 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 407 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 408 __le16 bg_itable_unused_lo; /* Unused inodes count */ 409 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 410 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 411 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 412 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 413 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 414 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 415 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 416 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 417 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 418 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 419 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 420 __u32 bg_reserved; 421 }; 422 423 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 424 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 425 sizeof(__le16)) 426 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 427 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 428 sizeof(__le16)) 429 430 /* 431 * Structure of a flex block group info 432 */ 433 434 struct flex_groups { 435 atomic64_t free_clusters; 436 atomic_t free_inodes; 437 atomic_t used_dirs; 438 }; 439 440 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 441 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 442 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 443 444 /* 445 * Macro-instructions used to manage group descriptors 446 */ 447 #define EXT4_MIN_DESC_SIZE 32 448 #define EXT4_MIN_DESC_SIZE_64BIT 64 449 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 450 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 451 #ifdef __KERNEL__ 452 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 453 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 454 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 455 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 456 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 457 #else 458 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 459 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 460 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 461 #endif 462 463 /* 464 * Constants relative to the data blocks 465 */ 466 #define EXT4_NDIR_BLOCKS 12 467 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 468 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 469 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 470 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 471 472 /* 473 * Inode flags 474 */ 475 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 476 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 477 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 478 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 479 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 480 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 481 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 482 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 483 /* Reserved for compression usage... */ 484 #define EXT4_DIRTY_FL 0x00000100 485 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 486 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 487 /* nb: was previously EXT2_ECOMPR_FL */ 488 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 489 /* End compression flags --- maybe not all used */ 490 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 491 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 492 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 493 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 494 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 495 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 496 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 497 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 498 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 499 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 500 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 501 502 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 503 504 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 505 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 506 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 507 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 508 509 /* User modifiable flags */ 510 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 511 EXT4_UNRM_FL | \ 512 EXT4_COMPR_FL | \ 513 EXT4_SYNC_FL | \ 514 EXT4_IMMUTABLE_FL | \ 515 EXT4_APPEND_FL | \ 516 EXT4_NODUMP_FL | \ 517 EXT4_NOATIME_FL | \ 518 EXT4_JOURNAL_DATA_FL | \ 519 EXT4_NOTAIL_FL | \ 520 EXT4_DIRSYNC_FL | \ 521 EXT4_TOPDIR_FL | \ 522 EXT4_EXTENTS_FL | \ 523 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 524 EXT4_DAX_FL | \ 525 EXT4_PROJINHERIT_FL | \ 526 EXT4_CASEFOLD_FL) 527 528 /* User visible flags */ 529 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 530 EXT4_DIRTY_FL | \ 531 EXT4_COMPRBLK_FL | \ 532 EXT4_NOCOMPR_FL | \ 533 EXT4_ENCRYPT_FL | \ 534 EXT4_INDEX_FL | \ 535 EXT4_VERITY_FL | \ 536 EXT4_INLINE_DATA_FL) 537 538 /* Flags that should be inherited by new inodes from their parent. */ 539 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 540 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 541 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 542 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 543 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 544 EXT4_DAX_FL) 545 546 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 547 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 548 EXT4_PROJINHERIT_FL)) 549 550 /* Flags that are appropriate for non-directories/regular files. */ 551 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 552 553 /* The only flags that should be swapped */ 554 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 555 556 /* Flags which are mutually exclusive to DAX */ 557 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 558 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 559 560 /* Mask out flags that are inappropriate for the given type of inode. */ 561 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 562 { 563 if (S_ISDIR(mode)) 564 return flags; 565 else if (S_ISREG(mode)) 566 return flags & EXT4_REG_FLMASK; 567 else 568 return flags & EXT4_OTHER_FLMASK; 569 } 570 571 /* 572 * Inode flags used for atomic set/get 573 */ 574 enum { 575 EXT4_INODE_SECRM = 0, /* Secure deletion */ 576 EXT4_INODE_UNRM = 1, /* Undelete */ 577 EXT4_INODE_COMPR = 2, /* Compress file */ 578 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 579 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 580 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 581 EXT4_INODE_NODUMP = 6, /* do not dump file */ 582 EXT4_INODE_NOATIME = 7, /* do not update atime */ 583 /* Reserved for compression usage... */ 584 EXT4_INODE_DIRTY = 8, 585 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 586 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 587 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 588 /* End compression flags --- maybe not all used */ 589 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 590 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 591 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 592 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 593 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 594 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 595 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 596 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 597 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 598 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 599 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 600 EXT4_INODE_DAX = 25, /* Inode is DAX */ 601 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 602 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 603 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 604 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 605 }; 606 607 /* 608 * Since it's pretty easy to mix up bit numbers and hex values, we use a 609 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 610 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 611 * any extra space in the compiled kernel image, otherwise, the build will fail. 612 * It's important that these values are the same, since we are using 613 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 614 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 615 * values found in ext2, ext3 and ext4 filesystems, and of course the values 616 * defined in e2fsprogs. 617 * 618 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 619 */ 620 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 621 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 622 623 static inline void ext4_check_flag_values(void) 624 { 625 CHECK_FLAG_VALUE(SECRM); 626 CHECK_FLAG_VALUE(UNRM); 627 CHECK_FLAG_VALUE(COMPR); 628 CHECK_FLAG_VALUE(SYNC); 629 CHECK_FLAG_VALUE(IMMUTABLE); 630 CHECK_FLAG_VALUE(APPEND); 631 CHECK_FLAG_VALUE(NODUMP); 632 CHECK_FLAG_VALUE(NOATIME); 633 CHECK_FLAG_VALUE(DIRTY); 634 CHECK_FLAG_VALUE(COMPRBLK); 635 CHECK_FLAG_VALUE(NOCOMPR); 636 CHECK_FLAG_VALUE(ENCRYPT); 637 CHECK_FLAG_VALUE(INDEX); 638 CHECK_FLAG_VALUE(IMAGIC); 639 CHECK_FLAG_VALUE(JOURNAL_DATA); 640 CHECK_FLAG_VALUE(NOTAIL); 641 CHECK_FLAG_VALUE(DIRSYNC); 642 CHECK_FLAG_VALUE(TOPDIR); 643 CHECK_FLAG_VALUE(HUGE_FILE); 644 CHECK_FLAG_VALUE(EXTENTS); 645 CHECK_FLAG_VALUE(VERITY); 646 CHECK_FLAG_VALUE(EA_INODE); 647 CHECK_FLAG_VALUE(INLINE_DATA); 648 CHECK_FLAG_VALUE(PROJINHERIT); 649 CHECK_FLAG_VALUE(CASEFOLD); 650 CHECK_FLAG_VALUE(RESERVED); 651 } 652 653 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 654 struct compat_ext4_new_group_input { 655 u32 group; 656 compat_u64 block_bitmap; 657 compat_u64 inode_bitmap; 658 compat_u64 inode_table; 659 u32 blocks_count; 660 u16 reserved_blocks; 661 u16 unused; 662 }; 663 #endif 664 665 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 666 struct ext4_new_group_data { 667 __u32 group; 668 __u64 block_bitmap; 669 __u64 inode_bitmap; 670 __u64 inode_table; 671 __u32 blocks_count; 672 __u16 reserved_blocks; 673 __u16 mdata_blocks; 674 __u32 free_clusters_count; 675 }; 676 677 /* Indexes used to index group tables in ext4_new_group_data */ 678 enum { 679 BLOCK_BITMAP = 0, /* block bitmap */ 680 INODE_BITMAP, /* inode bitmap */ 681 INODE_TABLE, /* inode tables */ 682 GROUP_TABLE_COUNT, 683 }; 684 685 /* 686 * Flags used by ext4_map_blocks() 687 */ 688 /* Allocate any needed blocks and/or convert an unwritten 689 extent to be an initialized ext4 */ 690 #define EXT4_GET_BLOCKS_CREATE 0x0001 691 /* Request the creation of an unwritten extent */ 692 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 693 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 694 EXT4_GET_BLOCKS_CREATE) 695 /* Caller is from the delayed allocation writeout path 696 * finally doing the actual allocation of delayed blocks */ 697 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 698 /* 699 * This means that we cannot merge newly allocated extents, and if we 700 * found an unwritten extent, we need to split it. 701 */ 702 #define EXT4_GET_BLOCKS_SPLIT_NOMERGE 0x0008 703 /* 704 * Caller is from the dio or dioread_nolock buffered IO, reqest to 705 * create an unwritten extent if it does not exist or split the 706 * found unwritten extent. Also do not merge the newly created 707 * unwritten extent, io end will convert unwritten to written, 708 * and try to merge the written extent. 709 */ 710 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_SPLIT_NOMERGE|\ 711 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 712 /* Convert unwritten extent to initialized. */ 713 #define EXT4_GET_BLOCKS_CONVERT 0x0010 714 /* Eventual metadata allocation (due to growing extent tree) 715 * should not fail, so try to use reserved blocks for that.*/ 716 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 717 /* Don't normalize allocation size (used for fallocate) */ 718 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 719 /* Convert written extents to unwritten */ 720 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 721 /* Write zeros to newly created written extents */ 722 #define EXT4_GET_BLOCKS_ZERO 0x0200 723 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 724 EXT4_GET_BLOCKS_ZERO) 725 /* Caller is in the context of data submission, such as writeback, 726 * fsync, etc. Especially, in the generic writeback path, caller will 727 * submit data before dropping transaction handle. This allows jbd2 728 * to avoid submitting data before commit. */ 729 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 730 /* Convert extent to initialized after IO complete */ 731 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\ 732 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\ 733 EXT4_GET_BLOCKS_IO_SUBMIT) 734 /* Caller is in the atomic contex, find extent if it has been cached */ 735 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 736 /* 737 * Atomic write caller needs this to query in the slow path of mixed mapping 738 * case, when a contiguous extent can be split across two adjacent leaf nodes. 739 * Look EXT4_MAP_QUERY_LAST_IN_LEAF. 740 */ 741 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000 742 743 /* 744 * The bit position of these flags must not overlap with any of the 745 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 746 * read_extent_tree_block(), ext4_split_extent_at(), 747 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 748 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 749 * caching the extents when reading from the extent tree while a 750 * truncate or punch hole operation is in progress. 751 */ 752 #define EXT4_EX_NOCACHE 0x40000000 753 #define EXT4_EX_FORCE_CACHE 0x20000000 754 #define EXT4_EX_NOFAIL 0x10000000 755 /* 756 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to 757 * pass while lookup/querying of on disk extent tree. 758 */ 759 #define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\ 760 EXT4_EX_NOFAIL |\ 761 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF) 762 763 /* 764 * Flags used by ext4_free_blocks 765 */ 766 #define EXT4_FREE_BLOCKS_METADATA 0x0001 767 #define EXT4_FREE_BLOCKS_FORGET 0x0002 768 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 769 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 770 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 771 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 772 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 773 774 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 775 /* 776 * ioctl commands in 32 bit emulation 777 */ 778 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 779 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 780 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 781 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 782 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 783 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 784 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 785 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 786 #endif 787 788 /* Max physical block we can address w/o extents */ 789 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 790 791 /* Max logical block we can support */ 792 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 793 794 /* 795 * Structure of an inode on the disk 796 */ 797 struct ext4_inode { 798 __le16 i_mode; /* File mode */ 799 __le16 i_uid; /* Low 16 bits of Owner Uid */ 800 __le32 i_size_lo; /* Size in bytes */ 801 __le32 i_atime; /* Access time */ 802 __le32 i_ctime; /* Inode Change time */ 803 __le32 i_mtime; /* Modification time */ 804 __le32 i_dtime; /* Deletion Time */ 805 __le16 i_gid; /* Low 16 bits of Group Id */ 806 __le16 i_links_count; /* Links count */ 807 __le32 i_blocks_lo; /* Blocks count */ 808 __le32 i_flags; /* File flags */ 809 union { 810 struct { 811 __le32 l_i_version; 812 } linux1; 813 struct { 814 __u32 h_i_translator; 815 } hurd1; 816 struct { 817 __u32 m_i_reserved1; 818 } masix1; 819 } osd1; /* OS dependent 1 */ 820 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 821 __le32 i_generation; /* File version (for NFS) */ 822 __le32 i_file_acl_lo; /* File ACL */ 823 __le32 i_size_high; 824 __le32 i_obso_faddr; /* Obsoleted fragment address */ 825 union { 826 struct { 827 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 828 __le16 l_i_file_acl_high; 829 __le16 l_i_uid_high; /* these 2 fields */ 830 __le16 l_i_gid_high; /* were reserved2[0] */ 831 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 832 __le16 l_i_reserved; 833 } linux2; 834 struct { 835 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 836 __u16 h_i_mode_high; 837 __u16 h_i_uid_high; 838 __u16 h_i_gid_high; 839 __u32 h_i_author; 840 } hurd2; 841 struct { 842 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 843 __le16 m_i_file_acl_high; 844 __u32 m_i_reserved2[2]; 845 } masix2; 846 } osd2; /* OS dependent 2 */ 847 __le16 i_extra_isize; 848 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 849 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 850 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 851 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 852 __le32 i_crtime; /* File Creation time */ 853 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 854 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 855 __le32 i_projid; /* Project ID */ 856 }; 857 858 #define EXT4_EPOCH_BITS 2 859 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 860 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 861 862 /* 863 * Extended fields will fit into an inode if the filesystem was formatted 864 * with large inodes (-I 256 or larger) and there are not currently any EAs 865 * consuming all of the available space. For new inodes we always reserve 866 * enough space for the kernel's known extended fields, but for inodes 867 * created with an old kernel this might not have been the case. None of 868 * the extended inode fields is critical for correct filesystem operation. 869 * This macro checks if a certain field fits in the inode. Note that 870 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 871 */ 872 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 873 ((offsetof(typeof(*ext4_inode), field) + \ 874 sizeof((ext4_inode)->field)) \ 875 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 876 (einode)->i_extra_isize)) \ 877 878 /* 879 * We use an encoding that preserves the times for extra epoch "00": 880 * 881 * extra msb of adjust for signed 882 * epoch 32-bit 32-bit tv_sec to 883 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 884 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 885 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 886 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 887 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 888 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 889 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 890 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 891 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 892 * 893 * Note that previous versions of the kernel on 64-bit systems would 894 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 895 * 1970. e2fsck will correct this, assuming that it is run on the 896 * affected filesystem before 2242. 897 */ 898 899 static inline __le32 ext4_encode_extra_time(struct timespec64 ts) 900 { 901 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK; 902 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS)); 903 } 904 905 static inline struct timespec64 ext4_decode_extra_time(__le32 base, 906 __le32 extra) 907 { 908 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) }; 909 910 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 911 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 912 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 913 return ts; 914 } 915 916 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \ 917 do { \ 918 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 919 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \ 920 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \ 921 } else \ 922 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \ 923 } while (0) 924 925 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \ 926 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode)) 927 928 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \ 929 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode)) 930 931 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \ 932 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode)) 933 934 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 935 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 936 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \ 937 raw_inode, (einode)->xtime) 938 939 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \ 940 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \ 941 ext4_decode_extra_time((raw_inode)->xtime, \ 942 (raw_inode)->xtime ## _extra) : \ 943 (struct timespec64) { \ 944 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \ 945 }) 946 947 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \ 948 do { \ 949 inode_set_atime_to_ts(inode, \ 950 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \ 951 } while (0) 952 953 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \ 954 do { \ 955 inode_set_mtime_to_ts(inode, \ 956 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \ 957 } while (0) 958 959 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \ 960 do { \ 961 inode_set_ctime_to_ts(inode, \ 962 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \ 963 } while (0) 964 965 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 966 do { \ 967 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 968 (einode)->xtime = \ 969 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \ 970 raw_inode); \ 971 else \ 972 (einode)->xtime = (struct timespec64){0, 0}; \ 973 } while (0) 974 975 #define i_disk_version osd1.linux1.l_i_version 976 977 #if defined(__KERNEL__) || defined(__linux__) 978 #define i_reserved1 osd1.linux1.l_i_reserved1 979 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 980 #define i_blocks_high osd2.linux2.l_i_blocks_high 981 #define i_uid_low i_uid 982 #define i_gid_low i_gid 983 #define i_uid_high osd2.linux2.l_i_uid_high 984 #define i_gid_high osd2.linux2.l_i_gid_high 985 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 986 987 #elif defined(__GNU__) 988 989 #define i_translator osd1.hurd1.h_i_translator 990 #define i_uid_high osd2.hurd2.h_i_uid_high 991 #define i_gid_high osd2.hurd2.h_i_gid_high 992 #define i_author osd2.hurd2.h_i_author 993 994 #elif defined(__masix__) 995 996 #define i_reserved1 osd1.masix1.m_i_reserved1 997 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 998 #define i_reserved2 osd2.masix2.m_i_reserved2 999 1000 #endif /* defined(__KERNEL__) || defined(__linux__) */ 1001 1002 #include "extents_status.h" 1003 #include "fast_commit.h" 1004 1005 /* 1006 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 1007 * 1008 * These are needed to avoid lockdep false positives when we need to 1009 * allocate blocks to the quota inode during ext4_map_blocks(), while 1010 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1011 * do quota tracking for the quota inode, this avoids deadlock (as 1012 * well as infinite recursion, since it isn't turtles all the way 1013 * down...) 1014 * 1015 * I_DATA_SEM_NORMAL - Used for most inodes 1016 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1017 * where the second inode has larger inode number 1018 * than the first 1019 * I_DATA_SEM_QUOTA - Used for quota inodes only 1020 * I_DATA_SEM_EA - Used for ea_inodes only 1021 */ 1022 enum { 1023 I_DATA_SEM_NORMAL = 0, 1024 I_DATA_SEM_OTHER, 1025 I_DATA_SEM_QUOTA, 1026 I_DATA_SEM_EA 1027 }; 1028 1029 1030 /* 1031 * fourth extended file system inode data in memory 1032 */ 1033 struct ext4_inode_info { 1034 __le32 i_data[15]; /* unconverted */ 1035 __u32 i_dtime; 1036 ext4_fsblk_t i_file_acl; 1037 1038 /* 1039 * i_block_group is the number of the block group which contains 1040 * this file's inode. Constant across the lifetime of the inode, 1041 * it is used for making block allocation decisions - we try to 1042 * place a file's data blocks near its inode block, and new inodes 1043 * near to their parent directory's inode. 1044 */ 1045 ext4_group_t i_block_group; 1046 ext4_lblk_t i_dir_start_lookup; 1047 #if (BITS_PER_LONG < 64) 1048 unsigned long i_state_flags; /* Dynamic state flags */ 1049 #endif 1050 unsigned long i_flags; 1051 1052 /* 1053 * Extended attributes can be read independently of the main file 1054 * data. Taking i_rwsem even when reading would cause contention 1055 * between readers of EAs and writers of regular file data, so 1056 * instead we synchronize on xattr_sem when reading or changing 1057 * EAs. 1058 */ 1059 struct rw_semaphore xattr_sem; 1060 1061 /* 1062 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1063 * i_orphan is used. 1064 */ 1065 union { 1066 struct list_head i_orphan; /* unlinked but open inodes */ 1067 unsigned int i_orphan_idx; /* Index in orphan file */ 1068 }; 1069 1070 /* Fast commit related info */ 1071 1072 /* For tracking dentry create updates */ 1073 struct list_head i_fc_dilist; 1074 struct list_head i_fc_list; /* 1075 * inodes that need fast commit 1076 * protected by sbi->s_fc_lock. 1077 */ 1078 1079 /* Start of lblk range that needs to be committed in this fast commit */ 1080 ext4_lblk_t i_fc_lblk_start; 1081 1082 /* End of lblk range that needs to be committed in this fast commit */ 1083 ext4_lblk_t i_fc_lblk_len; 1084 1085 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1086 1087 /* Fast commit wait queue for this inode */ 1088 wait_queue_head_t i_fc_wait; 1089 1090 /* 1091 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len 1092 * and inode's EXT4_FC_STATE_COMMITTING state bit. 1093 */ 1094 spinlock_t i_fc_lock; 1095 1096 /* 1097 * i_disksize keeps track of what the inode size is ON DISK, not 1098 * in memory. During truncate, i_size is set to the new size by 1099 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1100 * set i_disksize to 0 until the truncate is actually under way. 1101 * 1102 * The intent is that i_disksize always represents the blocks which 1103 * are used by this file. This allows recovery to restart truncate 1104 * on orphans if we crash during truncate. We actually write i_disksize 1105 * into the on-disk inode when writing inodes out, instead of i_size. 1106 * 1107 * The only time when i_disksize and i_size may be different is when 1108 * a truncate is in progress. The only things which change i_disksize 1109 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1110 */ 1111 loff_t i_disksize; 1112 1113 /* 1114 * i_data_sem is for serialising ext4_truncate() against 1115 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1116 * data tree are chopped off during truncate. We can't do that in 1117 * ext4 because whenever we perform intermediate commits during 1118 * truncate, the inode and all the metadata blocks *must* be in a 1119 * consistent state which allows truncation of the orphans to restart 1120 * during recovery. Hence we must fix the get_block-vs-truncate race 1121 * by other means, so we have i_data_sem. 1122 */ 1123 struct rw_semaphore i_data_sem; 1124 struct inode vfs_inode; 1125 struct jbd2_inode *jinode; 1126 1127 /* 1128 * File creation time. Its function is same as that of 1129 * struct timespec64 i_{a,c,m}time in the generic inode. 1130 */ 1131 struct timespec64 i_crtime; 1132 1133 /* mballoc */ 1134 atomic_t i_prealloc_active; 1135 1136 /* allocation reservation info for delalloc */ 1137 /* In case of bigalloc, this refer to clusters rather than blocks */ 1138 unsigned int i_reserved_data_blocks; 1139 struct rb_root i_prealloc_node; 1140 rwlock_t i_prealloc_lock; 1141 1142 /* extents status tree */ 1143 struct ext4_es_tree i_es_tree; 1144 rwlock_t i_es_lock; 1145 struct list_head i_es_list; 1146 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1147 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1148 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1149 extents to shrink. Protected by 1150 i_es_lock */ 1151 u64 i_es_seq; /* Change counter for extents. 1152 Protected by i_es_lock */ 1153 1154 /* ialloc */ 1155 ext4_group_t i_last_alloc_group; 1156 1157 /* pending cluster reservations for bigalloc file systems */ 1158 struct ext4_pending_tree i_pending_tree; 1159 1160 /* on-disk additional length */ 1161 __u16 i_extra_isize; 1162 1163 /* Indicate the inline data space. */ 1164 u16 i_inline_off; 1165 u16 i_inline_size; 1166 1167 #ifdef CONFIG_QUOTA 1168 /* quota space reservation, managed internally by quota code */ 1169 qsize_t i_reserved_quota; 1170 #endif 1171 spinlock_t i_block_reservation_lock; 1172 1173 /* Lock protecting lists below */ 1174 spinlock_t i_completed_io_lock; 1175 /* 1176 * Completed IOs that need unwritten extents handling and have 1177 * transaction reserved 1178 */ 1179 struct list_head i_rsv_conversion_list; 1180 struct work_struct i_rsv_conversion_work; 1181 1182 /* 1183 * Transactions that contain inode's metadata needed to complete 1184 * fsync and fdatasync, respectively. 1185 */ 1186 tid_t i_sync_tid; 1187 tid_t i_datasync_tid; 1188 1189 #ifdef CONFIG_QUOTA 1190 struct dquot __rcu *i_dquot[MAXQUOTAS]; 1191 #endif 1192 1193 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1194 __u32 i_csum_seed; 1195 1196 kprojid_t i_projid; 1197 1198 #ifdef CONFIG_FS_ENCRYPTION 1199 struct fscrypt_inode_info *i_crypt_info; 1200 #endif 1201 1202 #ifdef CONFIG_FS_VERITY 1203 struct fsverity_info *i_verity_info; 1204 #endif 1205 }; 1206 1207 /* 1208 * File system states 1209 */ 1210 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1211 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1212 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1213 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1214 1215 /* 1216 * Misc. filesystem flags 1217 */ 1218 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1219 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1220 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1221 1222 /* 1223 * Mount flags set via mount options or defaults 1224 */ 1225 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1226 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1227 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1228 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1229 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1230 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1231 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1232 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1233 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1234 #ifdef CONFIG_FS_DAX 1235 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1236 #else 1237 #define EXT4_MOUNT_DAX_ALWAYS 0 1238 #endif 1239 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1240 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1241 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1242 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1243 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1244 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1245 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1246 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1247 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1248 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1249 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1250 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1251 * enable enforcement for hidden 1252 * quota files */ 1253 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1254 * enforcement for hidden quota 1255 * files */ 1256 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1257 * enforcement */ 1258 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1259 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1260 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1261 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1262 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1263 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1264 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1265 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1266 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1267 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1268 1269 /* 1270 * Mount flags set either automatically (could not be set by mount option) 1271 * based on per file system feature or property or in special cases such as 1272 * distinguishing between explicit mount option definition and default. 1273 */ 1274 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1275 specified delalloc */ 1276 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1277 size of blocksize * 8 1278 blocks */ 1279 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1280 file systems */ 1281 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1282 specified journal checksum */ 1283 1284 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1285 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1286 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1287 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1288 * scanning in mballoc 1289 */ 1290 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1291 1292 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1293 ~EXT4_MOUNT_##opt 1294 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1295 EXT4_MOUNT_##opt 1296 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1297 EXT4_MOUNT_##opt) 1298 1299 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1300 ~EXT4_MOUNT2_##opt 1301 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1302 EXT4_MOUNT2_##opt 1303 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1304 EXT4_MOUNT2_##opt) 1305 1306 #define ext4_test_and_set_bit __test_and_set_bit_le 1307 #define ext4_set_bit __set_bit_le 1308 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1309 #define ext4_clear_bit __clear_bit_le 1310 #define ext4_test_bit test_bit_le 1311 #define ext4_find_next_zero_bit find_next_zero_bit_le 1312 #define ext4_find_next_bit find_next_bit_le 1313 1314 extern void mb_set_bits(void *bm, int cur, int len); 1315 1316 /* 1317 * Maximal mount counts between two filesystem checks 1318 */ 1319 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1320 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1321 1322 /* 1323 * Behaviour when detecting errors 1324 */ 1325 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1326 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1327 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1328 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1329 1330 /* Metadata checksum algorithm codes */ 1331 #define EXT4_CRC32C_CHKSUM 1 1332 1333 #define EXT4_LABEL_MAX 16 1334 1335 /* 1336 * Structure of the super block 1337 */ 1338 struct ext4_super_block { 1339 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1340 __le32 s_blocks_count_lo; /* Blocks count */ 1341 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1342 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1343 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1344 __le32 s_first_data_block; /* First Data Block */ 1345 __le32 s_log_block_size; /* Block size */ 1346 __le32 s_log_cluster_size; /* Allocation cluster size */ 1347 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1348 __le32 s_clusters_per_group; /* # Clusters per group */ 1349 __le32 s_inodes_per_group; /* # Inodes per group */ 1350 __le32 s_mtime; /* Mount time */ 1351 /*30*/ __le32 s_wtime; /* Write time */ 1352 __le16 s_mnt_count; /* Mount count */ 1353 __le16 s_max_mnt_count; /* Maximal mount count */ 1354 __le16 s_magic; /* Magic signature */ 1355 __le16 s_state; /* File system state */ 1356 __le16 s_errors; /* Behaviour when detecting errors */ 1357 __le16 s_minor_rev_level; /* minor revision level */ 1358 /*40*/ __le32 s_lastcheck; /* time of last check */ 1359 __le32 s_checkinterval; /* max. time between checks */ 1360 __le32 s_creator_os; /* OS */ 1361 __le32 s_rev_level; /* Revision level */ 1362 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1363 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1364 /* 1365 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1366 * 1367 * Note: the difference between the compatible feature set and 1368 * the incompatible feature set is that if there is a bit set 1369 * in the incompatible feature set that the kernel doesn't 1370 * know about, it should refuse to mount the filesystem. 1371 * 1372 * e2fsck's requirements are more strict; if it doesn't know 1373 * about a feature in either the compatible or incompatible 1374 * feature set, it must abort and not try to meddle with 1375 * things it doesn't understand... 1376 */ 1377 __le32 s_first_ino; /* First non-reserved inode */ 1378 __le16 s_inode_size; /* size of inode structure */ 1379 __le16 s_block_group_nr; /* block group # of this superblock */ 1380 __le32 s_feature_compat; /* compatible feature set */ 1381 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1382 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1383 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1384 /*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */ 1385 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1386 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1387 /* 1388 * Performance hints. Directory preallocation should only 1389 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1390 */ 1391 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1392 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1393 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1394 /* 1395 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1396 */ 1397 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1398 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1399 __le32 s_journal_dev; /* device number of journal file */ 1400 __le32 s_last_orphan; /* start of list of inodes to delete */ 1401 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1402 __u8 s_def_hash_version; /* Default hash version to use */ 1403 __u8 s_jnl_backup_type; 1404 __le16 s_desc_size; /* size of group descriptor */ 1405 /*100*/ __le32 s_default_mount_opts; 1406 __le32 s_first_meta_bg; /* First metablock block group */ 1407 __le32 s_mkfs_time; /* When the filesystem was created */ 1408 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1409 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1410 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1411 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1412 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1413 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1414 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1415 __le32 s_flags; /* Miscellaneous flags */ 1416 __le16 s_raid_stride; /* RAID stride */ 1417 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1418 __le64 s_mmp_block; /* Block for multi-mount protection */ 1419 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1420 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1421 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1422 __u8 s_encryption_level; /* versioning level for encryption */ 1423 __u8 s_reserved_pad; /* Padding to next 32bits */ 1424 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1425 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1426 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1427 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1428 snapshot's future use */ 1429 __le32 s_snapshot_list; /* inode number of the head of the 1430 on-disk snapshot list */ 1431 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1432 __le32 s_error_count; /* number of fs errors */ 1433 __le32 s_first_error_time; /* first time an error happened */ 1434 __le32 s_first_error_ino; /* inode involved in first error */ 1435 __le64 s_first_error_block; /* block involved of first error */ 1436 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1437 __le32 s_first_error_line; /* line number where error happened */ 1438 __le32 s_last_error_time; /* most recent time of an error */ 1439 __le32 s_last_error_ino; /* inode involved in last error */ 1440 __le32 s_last_error_line; /* line number where error happened */ 1441 __le64 s_last_error_block; /* block involved of last error */ 1442 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1443 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1444 __u8 s_mount_opts[64]; 1445 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1446 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1447 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1448 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1449 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1450 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1451 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1452 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1453 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1454 __u8 s_wtime_hi; 1455 __u8 s_mtime_hi; 1456 __u8 s_mkfs_time_hi; 1457 __u8 s_lastcheck_hi; 1458 __u8 s_first_error_time_hi; 1459 __u8 s_last_error_time_hi; 1460 __u8 s_first_error_errcode; 1461 __u8 s_last_error_errcode; 1462 __le16 s_encoding; /* Filename charset encoding */ 1463 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1464 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1465 __le16 s_def_resuid_hi; 1466 __le16 s_def_resgid_hi; 1467 __le32 s_reserved[93]; /* Padding to the end of the block */ 1468 __le32 s_checksum; /* crc32c(superblock) */ 1469 }; 1470 1471 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1472 1473 #ifdef __KERNEL__ 1474 1475 /* Number of quota types we support */ 1476 #define EXT4_MAXQUOTAS 3 1477 1478 #define EXT4_ENC_UTF8_12_1 1 1479 1480 /* Types of ext4 journal triggers */ 1481 enum ext4_journal_trigger_type { 1482 EXT4_JTR_ORPHAN_FILE, 1483 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1484 }; 1485 1486 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1487 1488 struct ext4_journal_trigger { 1489 struct jbd2_buffer_trigger_type tr_triggers; 1490 struct super_block *sb; 1491 }; 1492 1493 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1494 struct jbd2_buffer_trigger_type *trigger) 1495 { 1496 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1497 } 1498 1499 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1500 1501 /* Structure at the tail of orphan block */ 1502 struct ext4_orphan_block_tail { 1503 __le32 ob_magic; 1504 __le32 ob_checksum; 1505 }; 1506 1507 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1508 { 1509 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1510 sizeof(u32); 1511 } 1512 1513 struct ext4_orphan_block { 1514 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1515 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1516 }; 1517 1518 /* 1519 * Info about orphan file. 1520 */ 1521 struct ext4_orphan_info { 1522 int of_blocks; /* Number of orphan blocks in a file */ 1523 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1524 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1525 * file blocks */ 1526 }; 1527 1528 /* 1529 * fourth extended-fs super-block data in memory 1530 */ 1531 struct ext4_sb_info { 1532 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1533 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1534 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1535 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1536 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1537 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1538 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1539 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1540 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1541 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1542 unsigned long s_overhead; /* # of fs overhead clusters */ 1543 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1544 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1545 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1546 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1547 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1548 /* Array of bh's for the block group descriptors */ 1549 struct buffer_head * __rcu *s_group_desc; 1550 unsigned int s_mount_opt; 1551 unsigned int s_mount_opt2; 1552 unsigned long s_mount_flags; 1553 unsigned int s_def_mount_opt; 1554 unsigned int s_def_mount_opt2; 1555 ext4_fsblk_t s_sb_block; 1556 atomic64_t s_resv_clusters; 1557 kuid_t s_resuid; 1558 kgid_t s_resgid; 1559 unsigned short s_mount_state; 1560 unsigned short s_pad; 1561 int s_addr_per_block_bits; 1562 int s_desc_per_block_bits; 1563 int s_inode_size; 1564 int s_first_ino; 1565 unsigned int s_inode_readahead_blks; 1566 unsigned int s_inode_goal; 1567 u32 s_hash_seed[4]; 1568 int s_def_hash_version; 1569 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1570 struct percpu_counter s_freeclusters_counter; 1571 struct percpu_counter s_freeinodes_counter; 1572 struct percpu_counter s_dirs_counter; 1573 struct percpu_counter s_dirtyclusters_counter; 1574 struct percpu_counter s_sra_exceeded_retry_limit; 1575 struct blockgroup_lock *s_blockgroup_lock; 1576 struct proc_dir_entry *s_proc; 1577 struct kobject s_kobj; 1578 struct completion s_kobj_unregister; 1579 struct super_block *s_sb; 1580 struct buffer_head *s_mmp_bh; 1581 1582 /* Journaling */ 1583 struct journal_s *s_journal; 1584 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1585 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1586 struct list_head s_orphan; /* List of orphaned inodes in on disk 1587 list */ 1588 struct ext4_orphan_info s_orphan_info; 1589 unsigned long s_commit_interval; 1590 u32 s_max_batch_time; 1591 u32 s_min_batch_time; 1592 struct file *s_journal_bdev_file; 1593 #ifdef CONFIG_QUOTA 1594 /* Names of quota files with journalled quota */ 1595 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1596 int s_jquota_fmt; /* Format of quota to use */ 1597 #endif 1598 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1599 struct ext4_system_blocks __rcu *s_system_blks; 1600 1601 #ifdef EXTENTS_STATS 1602 /* ext4 extents stats */ 1603 unsigned long s_ext_min; 1604 unsigned long s_ext_max; 1605 unsigned long s_depth_max; 1606 spinlock_t s_ext_stats_lock; 1607 unsigned long s_ext_blocks; 1608 unsigned long s_ext_extents; 1609 #endif 1610 1611 /* for buddy allocator */ 1612 struct ext4_group_info ** __rcu *s_group_info; 1613 struct inode *s_buddy_cache; 1614 spinlock_t s_md_lock; 1615 unsigned short *s_mb_offsets; 1616 unsigned int *s_mb_maxs; 1617 unsigned int s_group_info_size; 1618 atomic_t s_mb_free_pending; 1619 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1620 after commit completed */ 1621 struct list_head s_discard_list; 1622 struct work_struct s_discard_work; 1623 atomic_t s_retry_alloc_pending; 1624 struct xarray *s_mb_avg_fragment_size; 1625 struct xarray *s_mb_largest_free_orders; 1626 1627 /* tunables */ 1628 unsigned long s_stripe; 1629 unsigned int s_mb_max_linear_groups; 1630 unsigned int s_mb_stream_request; 1631 unsigned int s_mb_max_to_scan; 1632 unsigned int s_mb_min_to_scan; 1633 unsigned int s_mb_stats; 1634 unsigned int s_mb_order2_reqs; 1635 unsigned int s_mb_group_prealloc; 1636 unsigned int s_max_dir_size_kb; 1637 unsigned int s_mb_prefetch; 1638 unsigned int s_mb_prefetch_limit; 1639 unsigned int s_mb_best_avail_max_trim_order; 1640 unsigned int s_sb_update_sec; 1641 unsigned int s_sb_update_kb; 1642 1643 /* where last allocation was done - for stream allocation */ 1644 ext4_group_t *s_mb_last_groups; 1645 unsigned int s_mb_nr_global_goals; 1646 1647 /* stats for buddy allocator */ 1648 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1649 atomic_t s_bal_success; /* we found long enough chunks */ 1650 atomic_t s_bal_allocated; /* in blocks */ 1651 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1652 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1653 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1654 atomic_t s_bal_goals; /* goal hits */ 1655 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */ 1656 atomic_t s_bal_len_goals; /* len goal hits */ 1657 atomic_t s_bal_breaks; /* too long searches */ 1658 atomic_t s_bal_2orders; /* 2^order hits */ 1659 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1660 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1661 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1662 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1663 atomic64_t s_mb_generation_time; 1664 atomic_t s_mb_lost_chunks; 1665 atomic_t s_mb_preallocated; 1666 atomic_t s_mb_discarded; 1667 atomic_t s_lock_busy; 1668 1669 /* locality groups */ 1670 struct ext4_locality_group __percpu *s_locality_groups; 1671 1672 /* for write statistics */ 1673 unsigned long s_sectors_written_start; 1674 u64 s_kbytes_written; 1675 1676 /* the size of zero-out chunk */ 1677 unsigned int s_extent_max_zeroout_kb; 1678 1679 unsigned int s_log_groups_per_flex; 1680 struct flex_groups * __rcu *s_flex_groups; 1681 ext4_group_t s_flex_groups_allocated; 1682 1683 /* workqueue for reserved extent conversions (buffered io) */ 1684 struct workqueue_struct *rsv_conversion_wq; 1685 1686 /* timer for periodic error stats printing */ 1687 struct timer_list s_err_report; 1688 1689 /* Lazy inode table initialization info */ 1690 struct ext4_li_request *s_li_request; 1691 /* Wait multiplier for lazy initialization thread */ 1692 unsigned int s_li_wait_mult; 1693 1694 /* Kernel thread for multiple mount protection */ 1695 struct task_struct *s_mmp_tsk; 1696 1697 /* record the last minlen when FITRIM is called. */ 1698 unsigned long s_last_trim_minblks; 1699 1700 /* Precomputed FS UUID checksum for seeding other checksums */ 1701 __u32 s_csum_seed; 1702 1703 /* Reclaim extents from extent status tree */ 1704 struct shrinker *s_es_shrinker; 1705 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1706 long s_es_nr_inode; 1707 struct ext4_es_stats s_es_stats; 1708 struct mb_cache *s_ea_block_cache; 1709 struct mb_cache *s_ea_inode_cache; 1710 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1711 1712 /* Journal triggers for checksum computation */ 1713 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1714 1715 /* Ratelimit ext4 messages. */ 1716 struct ratelimit_state s_err_ratelimit_state; 1717 struct ratelimit_state s_warning_ratelimit_state; 1718 struct ratelimit_state s_msg_ratelimit_state; 1719 atomic_t s_warning_count; 1720 atomic_t s_msg_count; 1721 1722 /* Encryption policy for '-o test_dummy_encryption' */ 1723 struct fscrypt_dummy_policy s_dummy_enc_policy; 1724 1725 /* 1726 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1727 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1728 * DIOREAD_NOLOCK mount options on remount. 1729 */ 1730 struct percpu_rw_semaphore s_writepages_rwsem; 1731 struct dax_device *s_daxdev; 1732 u64 s_dax_part_off; 1733 #ifdef CONFIG_EXT4_DEBUG 1734 unsigned long s_simulate_fail; 1735 #endif 1736 /* Record the errseq of the backing block device */ 1737 errseq_t s_bdev_wb_err; 1738 spinlock_t s_bdev_wb_lock; 1739 1740 /* Information about errors that happened during this mount */ 1741 spinlock_t s_error_lock; 1742 int s_add_error_count; 1743 int s_first_error_code; 1744 __u32 s_first_error_line; 1745 __u32 s_first_error_ino; 1746 __u64 s_first_error_block; 1747 const char *s_first_error_func; 1748 time64_t s_first_error_time; 1749 int s_last_error_code; 1750 __u32 s_last_error_line; 1751 __u32 s_last_error_ino; 1752 __u64 s_last_error_block; 1753 const char *s_last_error_func; 1754 time64_t s_last_error_time; 1755 /* 1756 * If we are in a context where we cannot update the on-disk 1757 * superblock, we queue the work here. This is used to update 1758 * the error information in the superblock, and for periodic 1759 * updates of the superblock called from the commit callback 1760 * function. 1761 */ 1762 struct work_struct s_sb_upd_work; 1763 1764 /* Atomic write unit values in bytes */ 1765 unsigned int s_awu_min; 1766 unsigned int s_awu_max; 1767 1768 /* Ext4 fast commit sub transaction ID */ 1769 atomic_t s_fc_subtid; 1770 1771 /* 1772 * After commit starts, the main queue gets locked, and the further 1773 * updates get added in the staging queue. 1774 */ 1775 #define FC_Q_MAIN 0 1776 #define FC_Q_STAGING 1 1777 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1778 * that have data changes in them. 1779 */ 1780 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1781 unsigned int s_fc_bytes; 1782 /* 1783 * Main fast commit lock. This lock protects accesses to the 1784 * following fields: 1785 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1786 */ 1787 struct mutex s_fc_lock; 1788 struct buffer_head *s_fc_bh; 1789 struct ext4_fc_stats s_fc_stats; 1790 tid_t s_fc_ineligible_tid; 1791 #ifdef CONFIG_EXT4_DEBUG 1792 int s_fc_debug_max_replay; 1793 #endif 1794 struct ext4_fc_replay_state s_fc_replay_state; 1795 }; 1796 1797 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1798 { 1799 return sb->s_fs_info; 1800 } 1801 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1802 { 1803 return container_of(inode, struct ext4_inode_info, vfs_inode); 1804 } 1805 1806 static inline int ext4_writepages_down_read(struct super_block *sb) 1807 { 1808 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1809 return memalloc_nofs_save(); 1810 } 1811 1812 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1813 { 1814 memalloc_nofs_restore(ctx); 1815 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1816 } 1817 1818 static inline int ext4_writepages_down_write(struct super_block *sb) 1819 { 1820 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1821 return memalloc_nofs_save(); 1822 } 1823 1824 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1825 { 1826 memalloc_nofs_restore(ctx); 1827 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1828 } 1829 1830 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1831 { 1832 return ino == EXT4_ROOT_INO || 1833 (ino >= EXT4_FIRST_INO(sb) && 1834 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1835 } 1836 1837 static inline int ext4_get_resuid(struct ext4_super_block *es) 1838 { 1839 return le16_to_cpu(es->s_def_resuid) | 1840 le16_to_cpu(es->s_def_resuid_hi) << 16; 1841 } 1842 1843 static inline int ext4_get_resgid(struct ext4_super_block *es) 1844 { 1845 return le16_to_cpu(es->s_def_resgid) | 1846 le16_to_cpu(es->s_def_resgid_hi) << 16; 1847 } 1848 1849 /* 1850 * Returns: sbi->field[index] 1851 * Used to access an array element from the following sbi fields which require 1852 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1853 * - s_group_desc 1854 * - s_group_info 1855 * - s_flex_group 1856 */ 1857 #define sbi_array_rcu_deref(sbi, field, index) \ 1858 ({ \ 1859 typeof(*((sbi)->field)) _v; \ 1860 rcu_read_lock(); \ 1861 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1862 rcu_read_unlock(); \ 1863 _v; \ 1864 }) 1865 1866 /* 1867 * run-time mount flags 1868 */ 1869 enum { 1870 EXT4_MF_MNTDIR_SAMPLED, 1871 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1872 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */ 1873 }; 1874 1875 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1876 { 1877 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1878 } 1879 1880 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1881 { 1882 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1883 } 1884 1885 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1886 { 1887 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1888 } 1889 1890 1891 /* 1892 * Simulate_fail codes 1893 */ 1894 #define EXT4_SIM_BBITMAP_EIO 1 1895 #define EXT4_SIM_BBITMAP_CRC 2 1896 #define EXT4_SIM_IBITMAP_EIO 3 1897 #define EXT4_SIM_IBITMAP_CRC 4 1898 #define EXT4_SIM_INODE_EIO 5 1899 #define EXT4_SIM_INODE_CRC 6 1900 #define EXT4_SIM_DIRBLOCK_EIO 7 1901 #define EXT4_SIM_DIRBLOCK_CRC 8 1902 1903 static inline bool ext4_simulate_fail(struct super_block *sb, 1904 unsigned long code) 1905 { 1906 #ifdef CONFIG_EXT4_DEBUG 1907 struct ext4_sb_info *sbi = EXT4_SB(sb); 1908 1909 if (unlikely(sbi->s_simulate_fail == code)) { 1910 sbi->s_simulate_fail = 0; 1911 return true; 1912 } 1913 #endif 1914 return false; 1915 } 1916 1917 /* 1918 * Error number codes for s_{first,last}_error_errno 1919 * 1920 * Linux errno numbers are architecture specific, so we need to translate 1921 * them into something which is architecture independent. We don't define 1922 * codes for all errno's; just the ones which are most likely to be the cause 1923 * of an ext4_error() call. 1924 */ 1925 #define EXT4_ERR_UNKNOWN 1 1926 #define EXT4_ERR_EIO 2 1927 #define EXT4_ERR_ENOMEM 3 1928 #define EXT4_ERR_EFSBADCRC 4 1929 #define EXT4_ERR_EFSCORRUPTED 5 1930 #define EXT4_ERR_ENOSPC 6 1931 #define EXT4_ERR_ENOKEY 7 1932 #define EXT4_ERR_EROFS 8 1933 #define EXT4_ERR_EFBIG 9 1934 #define EXT4_ERR_EEXIST 10 1935 #define EXT4_ERR_ERANGE 11 1936 #define EXT4_ERR_EOVERFLOW 12 1937 #define EXT4_ERR_EBUSY 13 1938 #define EXT4_ERR_ENOTDIR 14 1939 #define EXT4_ERR_ENOTEMPTY 15 1940 #define EXT4_ERR_ESHUTDOWN 16 1941 #define EXT4_ERR_EFAULT 17 1942 1943 /* 1944 * Inode dynamic state flags 1945 */ 1946 enum { 1947 EXT4_STATE_NEW, /* inode is newly created */ 1948 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1949 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1950 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1951 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1952 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1953 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1954 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1955 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1956 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1957 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1958 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */ 1959 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1960 }; 1961 1962 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1963 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1964 { \ 1965 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1966 } \ 1967 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1968 { \ 1969 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1970 } \ 1971 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1972 { \ 1973 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1974 } 1975 1976 /* Add these declarations here only so that these functions can be 1977 * found by name. Otherwise, they are very hard to locate. */ 1978 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1979 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1980 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1981 EXT4_INODE_BIT_FNS(flag, flags, 0) 1982 1983 /* Add these declarations here only so that these functions can be 1984 * found by name. Otherwise, they are very hard to locate. */ 1985 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1986 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1987 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1988 #if (BITS_PER_LONG < 64) 1989 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1990 1991 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1992 { 1993 (ei)->i_state_flags = 0; 1994 } 1995 #else 1996 EXT4_INODE_BIT_FNS(state, flags, 32) 1997 1998 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1999 { 2000 /* We depend on the fact that callers will set i_flags */ 2001 } 2002 #endif 2003 #else 2004 /* Assume that user mode programs are passing in an ext4fs superblock, not 2005 * a kernel struct super_block. This will allow us to call the feature-test 2006 * macros from user land. */ 2007 #define EXT4_SB(sb) (sb) 2008 #endif 2009 2010 static inline bool ext4_verity_in_progress(struct inode *inode) 2011 { 2012 return IS_ENABLED(CONFIG_FS_VERITY) && 2013 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 2014 } 2015 2016 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 2017 2018 /* 2019 * Check whether the inode is tracked as orphan (either in orphan file or 2020 * orphan list). 2021 */ 2022 static inline bool ext4_inode_orphan_tracked(struct inode *inode) 2023 { 2024 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) || 2025 !list_empty(&EXT4_I(inode)->i_orphan); 2026 } 2027 2028 /* 2029 * Codes for operating systems 2030 */ 2031 #define EXT4_OS_LINUX 0 2032 #define EXT4_OS_HURD 1 2033 #define EXT4_OS_MASIX 2 2034 #define EXT4_OS_FREEBSD 3 2035 #define EXT4_OS_LITES 4 2036 2037 /* 2038 * Revision levels 2039 */ 2040 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 2041 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 2042 2043 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 2044 2045 #define EXT4_GOOD_OLD_INODE_SIZE 128 2046 2047 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 2048 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 2049 #define EXT4_TIMESTAMP_MIN S32_MIN 2050 2051 /* 2052 * Feature set definitions 2053 */ 2054 2055 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 2056 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2057 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2058 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2059 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2060 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2061 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2062 /* 2063 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2064 * incompatible only if fast commit blocks are present in the FS. Since we 2065 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2066 * incompatible. We also have a JBD2 incompat feature, which gets set when 2067 * there are fast commit blocks present in the journal. 2068 */ 2069 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2070 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2071 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2072 2073 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2074 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2075 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2076 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2077 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2078 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2079 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2080 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2081 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2082 /* 2083 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2084 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2085 * all other data structures' checksums. However, the METADATA_CSUM and 2086 * GDT_CSUM bits are mutually exclusive. 2087 */ 2088 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2089 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2090 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2091 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2092 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2093 non-empty */ 2094 2095 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2096 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2097 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2098 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2099 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2100 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2101 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2102 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2103 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2104 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2105 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2106 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2107 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2108 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2109 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2110 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2111 2112 extern void ext4_update_dynamic_rev(struct super_block *sb); 2113 2114 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2115 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2116 { \ 2117 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2118 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2119 } \ 2120 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2121 { \ 2122 ext4_update_dynamic_rev(sb); \ 2123 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2124 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2125 } \ 2126 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2127 { \ 2128 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2129 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2130 } 2131 2132 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2133 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2134 { \ 2135 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2136 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2137 } \ 2138 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2139 { \ 2140 ext4_update_dynamic_rev(sb); \ 2141 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2142 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2143 } \ 2144 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2145 { \ 2146 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2147 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2148 } 2149 2150 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2151 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2152 { \ 2153 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2154 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2155 } \ 2156 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2157 { \ 2158 ext4_update_dynamic_rev(sb); \ 2159 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2160 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2161 } \ 2162 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2163 { \ 2164 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2165 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2166 } 2167 2168 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2169 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2170 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2171 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2172 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2173 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2174 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2175 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2176 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2177 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2178 2179 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2180 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2181 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2182 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2183 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2184 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2185 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2186 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2187 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2188 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2189 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2190 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2191 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2192 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2193 2194 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2195 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2196 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2197 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2198 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2199 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2200 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2201 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2202 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2203 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2204 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2205 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2206 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2207 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2208 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2209 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2210 2211 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2212 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2213 EXT4_FEATURE_INCOMPAT_META_BG) 2214 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2215 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2216 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2217 2218 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2219 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2220 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2221 EXT4_FEATURE_INCOMPAT_META_BG) 2222 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2223 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2224 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2225 2226 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2227 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2228 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2229 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2230 EXT4_FEATURE_INCOMPAT_META_BG| \ 2231 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2232 EXT4_FEATURE_INCOMPAT_64BIT| \ 2233 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2234 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2235 EXT4_FEATURE_INCOMPAT_MMP | \ 2236 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2237 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2238 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2239 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2240 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2241 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2242 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2243 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2244 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2245 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2246 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2247 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2248 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2249 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2250 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2251 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2252 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2253 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2254 2255 #define EXTN_FEATURE_FUNCS(ver) \ 2256 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2257 { \ 2258 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2259 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2260 } \ 2261 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2262 { \ 2263 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2264 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2265 } \ 2266 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2267 { \ 2268 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2269 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2270 } 2271 2272 EXTN_FEATURE_FUNCS(2) 2273 EXTN_FEATURE_FUNCS(3) 2274 EXTN_FEATURE_FUNCS(4) 2275 2276 static inline bool ext4_has_compat_features(struct super_block *sb) 2277 { 2278 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2279 } 2280 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2281 { 2282 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2283 } 2284 static inline bool ext4_has_incompat_features(struct super_block *sb) 2285 { 2286 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2287 } 2288 2289 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2290 2291 /* 2292 * Superblock flags 2293 */ 2294 enum { 2295 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */ 2296 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */ 2297 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */ 2298 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */ 2299 }; 2300 2301 static inline int ext4_forced_shutdown(struct super_block *sb) 2302 { 2303 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2304 } 2305 2306 static inline int ext4_emergency_ro(struct super_block *sb) 2307 { 2308 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags); 2309 } 2310 2311 static inline int ext4_emergency_state(struct super_block *sb) 2312 { 2313 if (unlikely(ext4_forced_shutdown(sb))) 2314 return -EIO; 2315 if (unlikely(ext4_emergency_ro(sb))) 2316 return -EROFS; 2317 return 0; 2318 } 2319 2320 /* 2321 * Default values for user and/or group using reserved blocks 2322 */ 2323 #define EXT4_DEF_RESUID 0 2324 #define EXT4_DEF_RESGID 0 2325 2326 /* 2327 * Default project ID 2328 */ 2329 #define EXT4_DEF_PROJID 0 2330 2331 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2332 2333 /* 2334 * Default mount options 2335 */ 2336 #define EXT4_DEFM_DEBUG 0x0001 2337 #define EXT4_DEFM_BSDGROUPS 0x0002 2338 #define EXT4_DEFM_XATTR_USER 0x0004 2339 #define EXT4_DEFM_ACL 0x0008 2340 #define EXT4_DEFM_UID16 0x0010 2341 #define EXT4_DEFM_JMODE 0x0060 2342 #define EXT4_DEFM_JMODE_DATA 0x0020 2343 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2344 #define EXT4_DEFM_JMODE_WBACK 0x0060 2345 #define EXT4_DEFM_NOBARRIER 0x0100 2346 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2347 #define EXT4_DEFM_DISCARD 0x0400 2348 #define EXT4_DEFM_NODELALLOC 0x0800 2349 2350 /* 2351 * Default journal batch times and ioprio. 2352 */ 2353 #define EXT4_DEF_MIN_BATCH_TIME 0 2354 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2355 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) 2356 2357 2358 /* 2359 * Default values for superblock update 2360 */ 2361 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */ 2362 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */ 2363 2364 2365 /* 2366 * Minimum number of groups in a flexgroup before we separate out 2367 * directories into the first block group of a flexgroup 2368 */ 2369 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2370 2371 /* 2372 * Structure of a directory entry 2373 */ 2374 #define EXT4_NAME_LEN 255 2375 /* 2376 * Base length of the ext4 directory entry excluding the name length 2377 */ 2378 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2379 2380 struct ext4_dir_entry { 2381 __le32 inode; /* Inode number */ 2382 __le16 rec_len; /* Directory entry length */ 2383 __le16 name_len; /* Name length */ 2384 char name[EXT4_NAME_LEN]; /* File name */ 2385 }; 2386 2387 2388 /* 2389 * Encrypted Casefolded entries require saving the hash on disk. This structure 2390 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2391 * boundary. 2392 */ 2393 struct ext4_dir_entry_hash { 2394 __le32 hash; 2395 __le32 minor_hash; 2396 }; 2397 2398 /* 2399 * The new version of the directory entry. Since EXT4 structures are 2400 * stored in intel byte order, and the name_len field could never be 2401 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2402 * file_type field. 2403 */ 2404 struct ext4_dir_entry_2 { 2405 __le32 inode; /* Inode number */ 2406 __le16 rec_len; /* Directory entry length */ 2407 __u8 name_len; /* Name length */ 2408 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2409 char name[EXT4_NAME_LEN]; /* File name */ 2410 }; 2411 2412 /* 2413 * Access the hashes at the end of ext4_dir_entry_2 2414 */ 2415 #define EXT4_DIRENT_HASHES(entry) \ 2416 ((struct ext4_dir_entry_hash *) \ 2417 (((void *)(entry)) + \ 2418 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2419 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash) 2420 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2421 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash) 2422 2423 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2424 { 2425 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2426 } 2427 2428 /* 2429 * This is a bogus directory entry at the end of each leaf block that 2430 * records checksums. 2431 */ 2432 struct ext4_dir_entry_tail { 2433 __le32 det_reserved_zero1; /* Pretend to be unused */ 2434 __le16 det_rec_len; /* 12 */ 2435 __u8 det_reserved_zero2; /* Zero name length */ 2436 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2437 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2438 }; 2439 2440 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2441 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2442 ((blocksize) - \ 2443 sizeof(struct ext4_dir_entry_tail)))) 2444 2445 /* 2446 * Ext4 directory file types. Only the low 3 bits are used. The 2447 * other bits are reserved for now. 2448 */ 2449 #define EXT4_FT_UNKNOWN 0 2450 #define EXT4_FT_REG_FILE 1 2451 #define EXT4_FT_DIR 2 2452 #define EXT4_FT_CHRDEV 3 2453 #define EXT4_FT_BLKDEV 4 2454 #define EXT4_FT_FIFO 5 2455 #define EXT4_FT_SOCK 6 2456 #define EXT4_FT_SYMLINK 7 2457 2458 #define EXT4_FT_MAX 8 2459 2460 #define EXT4_FT_DIR_CSUM 0xDE 2461 2462 /* 2463 * EXT4_DIR_PAD defines the directory entries boundaries 2464 * 2465 * NOTE: It must be a multiple of 4 2466 */ 2467 #define EXT4_DIR_PAD 4 2468 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2469 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2470 2471 /* 2472 * The rec_len is dependent on the type of directory. Directories that are 2473 * casefolded and encrypted need to store the hash as well, so we add room for 2474 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2475 * pass NULL for dir, as those entries do not use the extra fields. 2476 */ 2477 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2478 const struct inode *dir) 2479 { 2480 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2481 2482 if (dir && ext4_hash_in_dirent(dir)) 2483 rec_len += sizeof(struct ext4_dir_entry_hash); 2484 return (rec_len & ~EXT4_DIR_ROUND); 2485 } 2486 2487 /* 2488 * If we ever get support for fs block sizes > page_size, we'll need 2489 * to remove the #if statements in the next two functions... 2490 */ 2491 static inline unsigned int 2492 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2493 { 2494 unsigned len = le16_to_cpu(dlen); 2495 2496 #if (PAGE_SIZE >= 65536) 2497 if (len == EXT4_MAX_REC_LEN || len == 0) 2498 return blocksize; 2499 return (len & 65532) | ((len & 3) << 16); 2500 #else 2501 return len; 2502 #endif 2503 } 2504 2505 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2506 { 2507 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2508 #if (PAGE_SIZE >= 65536) 2509 if (len < 65536) 2510 return cpu_to_le16(len); 2511 if (len == blocksize) { 2512 if (blocksize == 65536) 2513 return cpu_to_le16(EXT4_MAX_REC_LEN); 2514 else 2515 return cpu_to_le16(0); 2516 } 2517 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2518 #else 2519 return cpu_to_le16(len); 2520 #endif 2521 } 2522 2523 /* 2524 * Hash Tree Directory indexing 2525 * (c) Daniel Phillips, 2001 2526 */ 2527 2528 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2529 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2530 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2531 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2532 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2533 2534 /* Legal values for the dx_root hash_version field: */ 2535 2536 #define DX_HASH_LEGACY 0 2537 #define DX_HASH_HALF_MD4 1 2538 #define DX_HASH_TEA 2 2539 #define DX_HASH_LEGACY_UNSIGNED 3 2540 #define DX_HASH_HALF_MD4_UNSIGNED 4 2541 #define DX_HASH_TEA_UNSIGNED 5 2542 #define DX_HASH_SIPHASH 6 2543 #define DX_HASH_LAST DX_HASH_SIPHASH 2544 2545 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length) 2546 { 2547 return crc32c(crc, address, length); 2548 } 2549 2550 #ifdef __KERNEL__ 2551 2552 /* hash info structure used by the directory hash */ 2553 struct dx_hash_info 2554 { 2555 u32 hash; 2556 u32 minor_hash; 2557 int hash_version; 2558 u32 *seed; 2559 }; 2560 2561 2562 /* 32 and 64 bit signed EOF for dx directories */ 2563 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2564 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2565 2566 2567 /* 2568 * Control parameters used by ext4_htree_next_block 2569 */ 2570 #define HASH_NB_ALWAYS 1 2571 2572 struct ext4_filename { 2573 const struct qstr *usr_fname; 2574 struct fscrypt_str disk_name; 2575 struct dx_hash_info hinfo; 2576 #ifdef CONFIG_FS_ENCRYPTION 2577 struct fscrypt_str crypto_buf; 2578 #endif 2579 #if IS_ENABLED(CONFIG_UNICODE) 2580 struct qstr cf_name; 2581 #endif 2582 }; 2583 2584 #define fname_name(p) ((p)->disk_name.name) 2585 #define fname_usr_name(p) ((p)->usr_fname->name) 2586 #define fname_len(p) ((p)->disk_name.len) 2587 2588 /* 2589 * Describe an inode's exact location on disk and in memory 2590 */ 2591 struct ext4_iloc 2592 { 2593 struct buffer_head *bh; 2594 unsigned long offset; 2595 ext4_group_t block_group; 2596 }; 2597 2598 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2599 { 2600 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2601 } 2602 2603 static inline bool ext4_is_quota_file(struct inode *inode) 2604 { 2605 return IS_NOQUOTA(inode) && 2606 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2607 } 2608 2609 /* 2610 * This structure is stuffed into the struct file's private_data field 2611 * for directories. It is where we put information so that we can do 2612 * readdir operations in hash tree order. 2613 */ 2614 struct dir_private_info { 2615 struct rb_root root; 2616 struct rb_node *curr_node; 2617 struct fname *extra_fname; 2618 loff_t last_pos; 2619 __u32 curr_hash; 2620 __u32 curr_minor_hash; 2621 __u32 next_hash; 2622 u64 cookie; 2623 bool initialized; 2624 }; 2625 2626 /* calculate the first block number of the group */ 2627 static inline ext4_fsblk_t 2628 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2629 { 2630 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2631 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2632 } 2633 2634 /* 2635 * Special error return code only used by dx_probe() and its callers. 2636 */ 2637 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2638 2639 /* htree levels for ext4 */ 2640 #define EXT4_HTREE_LEVEL_COMPAT 2 2641 #define EXT4_HTREE_LEVEL 3 2642 2643 static inline int ext4_dir_htree_level(struct super_block *sb) 2644 { 2645 return ext4_has_feature_largedir(sb) ? 2646 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2647 } 2648 2649 /* 2650 * Timeout and state flag for lazy initialization inode thread. 2651 */ 2652 #define EXT4_DEF_LI_WAIT_MULT 10 2653 #define EXT4_DEF_LI_MAX_START_DELAY 5 2654 #define EXT4_LAZYINIT_QUIT 0x0001 2655 #define EXT4_LAZYINIT_RUNNING 0x0002 2656 2657 /* 2658 * Lazy inode table initialization info 2659 */ 2660 struct ext4_lazy_init { 2661 unsigned long li_state; 2662 struct list_head li_request_list; 2663 struct mutex li_list_mtx; 2664 }; 2665 2666 enum ext4_li_mode { 2667 EXT4_LI_MODE_PREFETCH_BBITMAP, 2668 EXT4_LI_MODE_ITABLE, 2669 }; 2670 2671 struct ext4_li_request { 2672 struct super_block *lr_super; 2673 enum ext4_li_mode lr_mode; 2674 ext4_group_t lr_first_not_zeroed; 2675 ext4_group_t lr_next_group; 2676 struct list_head lr_request; 2677 unsigned long lr_next_sched; 2678 unsigned long lr_timeout; 2679 }; 2680 2681 struct ext4_features { 2682 struct kobject f_kobj; 2683 struct completion f_kobj_unregister; 2684 }; 2685 2686 /* 2687 * This structure will be used for multiple mount protection. It will be 2688 * written into the block number saved in the s_mmp_block field in the 2689 * superblock. Programs that check MMP should assume that if 2690 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2691 * to use the filesystem, regardless of how old the timestamp is. 2692 */ 2693 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2694 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2695 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2696 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2697 2698 struct mmp_struct { 2699 __le32 mmp_magic; /* Magic number for MMP */ 2700 __le32 mmp_seq; /* Sequence no. updated periodically */ 2701 2702 /* 2703 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2704 * purposes and do not affect the correctness of the algorithm 2705 */ 2706 __le64 mmp_time; /* Time last updated */ 2707 char mmp_nodename[64]; /* Node which last updated MMP block */ 2708 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2709 2710 /* 2711 * mmp_check_interval is used to verify if the MMP block has been 2712 * updated on the block device. The value is updated based on the 2713 * maximum time to write the MMP block during an update cycle. 2714 */ 2715 __le16 mmp_check_interval; 2716 2717 __le16 mmp_pad1; 2718 __le32 mmp_pad2[226]; 2719 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2720 }; 2721 2722 /* arguments passed to the mmp thread */ 2723 struct mmpd_data { 2724 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2725 struct super_block *sb; /* super block of the fs */ 2726 }; 2727 2728 /* 2729 * Check interval multiplier 2730 * The MMP block is written every update interval and initially checked every 2731 * update interval x the multiplier (the value is then adapted based on the 2732 * write latency). The reason is that writes can be delayed under load and we 2733 * don't want readers to incorrectly assume that the filesystem is no longer 2734 * in use. 2735 */ 2736 #define EXT4_MMP_CHECK_MULT 2UL 2737 2738 /* 2739 * Minimum interval for MMP checking in seconds. 2740 */ 2741 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2742 2743 /* 2744 * Maximum interval for MMP checking in seconds. 2745 */ 2746 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2747 2748 /* 2749 * Function prototypes 2750 */ 2751 2752 /* 2753 * Ok, these declarations are also in <linux/kernel.h> but none of the 2754 * ext4 source programs needs to include it so they are duplicated here. 2755 */ 2756 # define NORET_TYPE /**/ 2757 # define ATTRIB_NORET __attribute__((noreturn)) 2758 # define NORET_AND noreturn, 2759 2760 /* bitmap.c */ 2761 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2762 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2763 struct ext4_group_desc *gdp, 2764 struct buffer_head *bh); 2765 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2766 struct ext4_group_desc *gdp, 2767 struct buffer_head *bh); 2768 void ext4_block_bitmap_csum_set(struct super_block *sb, 2769 struct ext4_group_desc *gdp, 2770 struct buffer_head *bh); 2771 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2772 struct ext4_group_desc *gdp, 2773 struct buffer_head *bh); 2774 2775 /* balloc.c */ 2776 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2777 ext4_fsblk_t blocknr, 2778 ext4_group_t *blockgrpp, 2779 ext4_grpblk_t *offsetp); 2780 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2781 ext4_fsblk_t block); 2782 2783 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2784 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2785 ext4_group_t group); 2786 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2787 ext4_fsblk_t goal, 2788 unsigned int flags, 2789 unsigned long *count, 2790 int *errp); 2791 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2792 s64 nclusters, unsigned int flags); 2793 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2794 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2795 ext4_group_t block_group, 2796 struct buffer_head ** bh); 2797 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2798 ext4_group_t group); 2799 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2800 2801 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2802 ext4_group_t block_group, 2803 bool ignore_locked); 2804 extern int ext4_wait_block_bitmap(struct super_block *sb, 2805 ext4_group_t block_group, 2806 struct buffer_head *bh); 2807 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2808 ext4_group_t block_group); 2809 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2810 ext4_group_t block_group, 2811 struct ext4_group_desc *gdp); 2812 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2813 2814 #if IS_ENABLED(CONFIG_UNICODE) 2815 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2816 const struct qstr *iname, 2817 struct ext4_filename *fname); 2818 2819 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2820 { 2821 kfree(fname->cf_name.name); 2822 fname->cf_name.name = NULL; 2823 } 2824 #else 2825 static inline int ext4_fname_setup_ci_filename(struct inode *dir, 2826 const struct qstr *iname, 2827 struct ext4_filename *fname) 2828 { 2829 return 0; 2830 } 2831 2832 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2833 { 2834 } 2835 #endif 2836 2837 /* ext4 encryption related stuff goes here crypto.c */ 2838 #ifdef CONFIG_FS_ENCRYPTION 2839 extern const struct fscrypt_operations ext4_cryptops; 2840 2841 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2842 int lookup, struct ext4_filename *fname); 2843 2844 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2845 struct ext4_filename *fname); 2846 2847 void ext4_fname_free_filename(struct ext4_filename *fname); 2848 2849 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2850 2851 #else /* !CONFIG_FS_ENCRYPTION */ 2852 static inline int ext4_fname_setup_filename(struct inode *dir, 2853 const struct qstr *iname, 2854 int lookup, 2855 struct ext4_filename *fname) 2856 { 2857 fname->usr_fname = iname; 2858 fname->disk_name.name = (unsigned char *) iname->name; 2859 fname->disk_name.len = iname->len; 2860 2861 return ext4_fname_setup_ci_filename(dir, iname, fname); 2862 } 2863 2864 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2865 struct dentry *dentry, 2866 struct ext4_filename *fname) 2867 { 2868 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2869 } 2870 2871 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2872 { 2873 ext4_fname_free_ci_filename(fname); 2874 } 2875 2876 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2877 void __user *arg) 2878 { 2879 return -EOPNOTSUPP; 2880 } 2881 #endif /* !CONFIG_FS_ENCRYPTION */ 2882 2883 /* dir.c */ 2884 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2885 struct file *, 2886 struct ext4_dir_entry_2 *, 2887 struct buffer_head *, char *, int, 2888 unsigned int); 2889 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2890 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2891 (de), (bh), (buf), (size), (offset))) 2892 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2893 __u32 minor_hash, 2894 struct ext4_dir_entry_2 *dirent, 2895 struct fscrypt_str *ent_name); 2896 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2897 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh, 2898 void *buf, int buf_size, 2899 struct ext4_filename *fname, 2900 struct ext4_dir_entry_2 **dest_de); 2901 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2902 struct ext4_dir_entry_2 *de, 2903 int buf_size, 2904 struct ext4_filename *fname); 2905 static inline void ext4_update_dx_flag(struct inode *inode) 2906 { 2907 if (!ext4_has_feature_dir_index(inode->i_sb) && 2908 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2909 /* ext4_iget() should have caught this... */ 2910 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2911 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2912 } 2913 } 2914 static const unsigned char ext4_filetype_table[] = { 2915 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2916 }; 2917 2918 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2919 { 2920 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2921 return DT_UNKNOWN; 2922 2923 return ext4_filetype_table[filetype]; 2924 } 2925 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2926 void *buf, int buf_size); 2927 2928 /* fsync.c */ 2929 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2930 2931 /* hash.c */ 2932 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2933 struct dx_hash_info *hinfo); 2934 2935 /* ialloc.c */ 2936 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2937 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2938 struct inode *, umode_t, 2939 const struct qstr *qstr, __u32 goal, 2940 uid_t *owner, __u32 i_flags, 2941 int handle_type, unsigned int line_no, 2942 int nblocks); 2943 2944 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2945 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2946 (goal), (owner), i_flags, 0, 0, 0) 2947 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2948 type, nblocks) \ 2949 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2950 0, (type), __LINE__, (nblocks)) 2951 2952 2953 extern void ext4_free_inode(handle_t *, struct inode *); 2954 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2955 extern unsigned long ext4_count_free_inodes(struct super_block *); 2956 extern unsigned long ext4_count_dirs(struct super_block *); 2957 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2958 extern int ext4_init_inode_table(struct super_block *sb, 2959 ext4_group_t group, int barrier); 2960 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2961 2962 /* fast_commit.c */ 2963 int ext4_fc_info_show(struct seq_file *seq, void *v); 2964 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2965 void ext4_fc_init_inode(struct inode *inode); 2966 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2967 ext4_lblk_t end); 2968 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2969 struct dentry *dentry); 2970 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2971 struct dentry *dentry); 2972 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2973 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2974 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2975 struct dentry *dentry); 2976 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2977 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2978 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2979 void ext4_fc_del(struct inode *inode); 2980 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2981 void ext4_fc_replay_cleanup(struct super_block *sb); 2982 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2983 int __init ext4_fc_init_dentry_cache(void); 2984 void ext4_fc_destroy_dentry_cache(void); 2985 int ext4_fc_record_regions(struct super_block *sb, int ino, 2986 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2987 int len, int replay); 2988 2989 /* mballoc.c */ 2990 extern const struct seq_operations ext4_mb_seq_groups_ops; 2991 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2992 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2993 extern int ext4_mb_init(struct super_block *); 2994 extern void ext4_mb_release(struct super_block *); 2995 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2996 struct ext4_allocation_request *, int *); 2997 extern void ext4_discard_preallocations(struct inode *); 2998 extern int __init ext4_init_mballoc(void); 2999 extern void ext4_exit_mballoc(void); 3000 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 3001 ext4_group_t group, 3002 unsigned int nr, int *cnt); 3003 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 3004 unsigned int nr); 3005 3006 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 3007 struct buffer_head *bh, ext4_fsblk_t block, 3008 unsigned long count, int flags); 3009 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 3010 ext4_group_t ngroups); 3011 extern int ext4_mb_add_groupinfo(struct super_block *sb, 3012 ext4_group_t i, struct ext4_group_desc *desc); 3013 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 3014 ext4_fsblk_t block, unsigned long count); 3015 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 3016 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 3017 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 3018 int len, bool state); 3019 static inline bool ext4_mb_cr_expensive(enum criteria cr) 3020 { 3021 return cr >= CR_GOAL_LEN_SLOW; 3022 } 3023 3024 /* inode.c */ 3025 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 3026 struct ext4_inode_info *ei); 3027 int ext4_inode_is_fast_symlink(struct inode *inode); 3028 void ext4_check_map_extents_env(struct inode *inode); 3029 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 3030 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 3031 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 3032 bool wait, struct buffer_head **bhs); 3033 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 3034 struct buffer_head *bh_result, int create); 3035 int ext4_get_block(struct inode *inode, sector_t iblock, 3036 struct buffer_head *bh_result, int create); 3037 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 3038 struct buffer_head *bh, int create); 3039 int ext4_walk_page_buffers(handle_t *handle, 3040 struct inode *inode, 3041 struct buffer_head *head, 3042 unsigned from, 3043 unsigned to, 3044 int *partial, 3045 int (*fn)(handle_t *handle, struct inode *inode, 3046 struct buffer_head *bh)); 3047 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 3048 struct buffer_head *bh); 3049 void ext4_set_inode_mapping_order(struct inode *inode); 3050 #define FALL_BACK_TO_NONDELALLOC 1 3051 #define CONVERT_INLINE_DATA 2 3052 3053 typedef enum { 3054 EXT4_IGET_NORMAL = 0, 3055 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3056 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 3057 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 3058 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 3059 } ext4_iget_flags; 3060 3061 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3062 ext4_iget_flags flags, const char *function, 3063 unsigned int line); 3064 3065 #define ext4_iget(sb, ino, flags) \ 3066 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3067 3068 extern int ext4_write_inode(struct inode *, struct writeback_control *); 3069 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 3070 struct iattr *); 3071 extern u32 ext4_dio_alignment(struct inode *inode); 3072 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 3073 struct kstat *, u32, unsigned int); 3074 extern void ext4_evict_inode(struct inode *); 3075 extern void ext4_clear_inode(struct inode *); 3076 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 3077 struct kstat *, u32, unsigned int); 3078 extern void ext4_dirty_inode(struct inode *, int); 3079 extern int ext4_change_inode_journal_flag(struct inode *, int); 3080 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3081 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3082 struct ext4_iloc *iloc); 3083 extern int ext4_inode_attach_jinode(struct inode *inode); 3084 extern int ext4_can_truncate(struct inode *inode); 3085 extern int ext4_truncate(struct inode *); 3086 extern int ext4_break_layouts(struct inode *); 3087 extern int ext4_truncate_page_cache_block_range(struct inode *inode, 3088 loff_t start, loff_t end); 3089 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3090 extern void ext4_set_inode_flags(struct inode *, bool init); 3091 extern int ext4_alloc_da_blocks(struct inode *inode); 3092 extern void ext4_set_aops(struct inode *inode); 3093 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3094 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3095 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks); 3096 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks, 3097 int pextents); 3098 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3099 loff_t lstart, loff_t lend); 3100 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3101 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3102 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3103 extern void ext4_da_release_space(struct inode *inode, int to_free); 3104 extern void ext4_da_update_reserve_space(struct inode *inode, 3105 int used, int quota_claim); 3106 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3107 ext4_fsblk_t pblk, ext4_lblk_t len); 3108 3109 static inline bool is_special_ino(struct super_block *sb, unsigned long ino) 3110 { 3111 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 3112 3113 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) || 3114 ino == le32_to_cpu(es->s_usr_quota_inum) || 3115 ino == le32_to_cpu(es->s_grp_quota_inum) || 3116 ino == le32_to_cpu(es->s_prj_quota_inum) || 3117 ino == le32_to_cpu(es->s_orphan_file_inum); 3118 } 3119 3120 /* indirect.c */ 3121 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3122 struct ext4_map_blocks *map, int flags); 3123 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3124 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3125 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3126 ext4_lblk_t start, ext4_lblk_t end); 3127 3128 /* ioctl.c */ 3129 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3130 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3131 int ext4_fileattr_set(struct mnt_idmap *idmap, 3132 struct dentry *dentry, struct file_kattr *fa); 3133 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3134 extern void ext4_reset_inode_seed(struct inode *inode); 3135 int ext4_update_overhead(struct super_block *sb, bool force); 3136 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3137 3138 /* migrate.c */ 3139 extern int ext4_ext_migrate(struct inode *); 3140 extern int ext4_ind_migrate(struct inode *inode); 3141 3142 /* namei.c */ 3143 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3144 struct inode *inode); 3145 extern int ext4_dirblock_csum_verify(struct inode *inode, 3146 struct buffer_head *bh); 3147 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3148 __u32 start_minor_hash, __u32 *next_hash); 3149 extern int ext4_search_dir(struct buffer_head *bh, 3150 char *search_buf, 3151 int buf_size, 3152 struct inode *dir, 3153 struct ext4_filename *fname, 3154 unsigned int offset, 3155 struct ext4_dir_entry_2 **res_dir); 3156 extern int ext4_generic_delete_entry(struct inode *dir, 3157 struct ext4_dir_entry_2 *de_del, 3158 struct buffer_head *bh, 3159 void *entry_buf, 3160 int buf_size, 3161 int csum_size); 3162 extern bool ext4_empty_dir(struct inode *inode); 3163 3164 /* resize.c */ 3165 extern void ext4_kvfree_array_rcu(void *to_free); 3166 extern int ext4_group_add(struct super_block *sb, 3167 struct ext4_new_group_data *input); 3168 extern int ext4_group_extend(struct super_block *sb, 3169 struct ext4_super_block *es, 3170 ext4_fsblk_t n_blocks_count); 3171 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3172 extern unsigned int ext4_list_backups(struct super_block *sb, 3173 unsigned int *three, unsigned int *five, 3174 unsigned int *seven); 3175 3176 /* super.c */ 3177 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3178 sector_t block, blk_opf_t op_flags); 3179 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3180 sector_t block); 3181 extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb, 3182 sector_t block); 3183 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3184 bh_end_io_t *end_io, bool simu_fail); 3185 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3186 bh_end_io_t *end_io, bool simu_fail); 3187 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3188 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3189 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3190 extern int ext4_calculate_overhead(struct super_block *sb); 3191 extern __le32 ext4_superblock_csum(struct ext4_super_block *es); 3192 extern void ext4_superblock_csum_set(struct super_block *sb); 3193 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3194 ext4_group_t ngroup); 3195 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3196 char nbuf[16]); 3197 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3198 ext4_group_t block_group, 3199 unsigned int flags); 3200 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3201 ext4_group_t block_group); 3202 3203 extern __printf(7, 8) 3204 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3205 int, __u64, const char *, ...); 3206 extern __printf(6, 7) 3207 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3208 ext4_fsblk_t, int, const char *, ...); 3209 extern __printf(5, 6) 3210 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3211 const char *, ...); 3212 extern void __ext4_std_error(struct super_block *, const char *, 3213 unsigned int, int); 3214 extern __printf(4, 5) 3215 void __ext4_warning(struct super_block *, const char *, unsigned int, 3216 const char *, ...); 3217 extern __printf(4, 5) 3218 void __ext4_warning_inode(const struct inode *inode, const char *function, 3219 unsigned int line, const char *fmt, ...); 3220 extern __printf(3, 4) 3221 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3222 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3223 const char *, unsigned int, const char *); 3224 extern __printf(7, 8) 3225 void __ext4_grp_locked_error(const char *, unsigned int, 3226 struct super_block *, ext4_group_t, 3227 unsigned long, ext4_fsblk_t, 3228 const char *, ...); 3229 3230 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3231 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3232 3233 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3234 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3235 3236 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3237 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3238 (fmt), ## a) 3239 3240 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3241 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3242 3243 #define ext4_abort(sb, err, fmt, a...) \ 3244 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3245 3246 #ifdef CONFIG_PRINTK 3247 3248 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3249 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3250 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3251 __ext4_error_inode((inode), (func), (line), (block), \ 3252 (err), (fmt), ##__VA_ARGS__) 3253 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3254 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3255 #define ext4_error(sb, fmt, ...) \ 3256 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3257 ##__VA_ARGS__) 3258 #define ext4_error_err(sb, err, fmt, ...) \ 3259 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3260 ##__VA_ARGS__) 3261 #define ext4_warning(sb, fmt, ...) \ 3262 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3263 #define ext4_warning_inode(inode, fmt, ...) \ 3264 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3265 #define ext4_msg(sb, level, fmt, ...) \ 3266 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3267 #define dump_mmp_msg(sb, mmp, msg) \ 3268 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3269 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3270 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3271 fmt, ##__VA_ARGS__) 3272 3273 #else 3274 3275 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3276 do { \ 3277 no_printk(fmt, ##__VA_ARGS__); \ 3278 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3279 } while (0) 3280 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3281 do { \ 3282 no_printk(fmt, ##__VA_ARGS__); \ 3283 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3284 } while (0) 3285 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3286 do { \ 3287 no_printk(fmt, ##__VA_ARGS__); \ 3288 __ext4_error_file(file, "", 0, block, " "); \ 3289 } while (0) 3290 #define ext4_error(sb, fmt, ...) \ 3291 do { \ 3292 no_printk(fmt, ##__VA_ARGS__); \ 3293 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3294 } while (0) 3295 #define ext4_error_err(sb, err, fmt, ...) \ 3296 do { \ 3297 no_printk(fmt, ##__VA_ARGS__); \ 3298 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3299 } while (0) 3300 #define ext4_warning(sb, fmt, ...) \ 3301 do { \ 3302 no_printk(fmt, ##__VA_ARGS__); \ 3303 __ext4_warning(sb, "", 0, " "); \ 3304 } while (0) 3305 #define ext4_warning_inode(inode, fmt, ...) \ 3306 do { \ 3307 no_printk(fmt, ##__VA_ARGS__); \ 3308 __ext4_warning_inode(inode, "", 0, " "); \ 3309 } while (0) 3310 #define ext4_msg(sb, level, fmt, ...) \ 3311 do { \ 3312 no_printk(fmt, ##__VA_ARGS__); \ 3313 __ext4_msg(sb, "", " "); \ 3314 } while (0) 3315 #define dump_mmp_msg(sb, mmp, msg) \ 3316 __dump_mmp_msg(sb, mmp, "", 0, "") 3317 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3318 do { \ 3319 no_printk(fmt, ##__VA_ARGS__); \ 3320 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3321 } while (0) 3322 3323 #endif 3324 3325 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3326 struct ext4_group_desc *bg); 3327 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3328 struct ext4_group_desc *bg); 3329 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3330 struct ext4_group_desc *bg); 3331 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3332 struct ext4_group_desc *bg); 3333 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3334 struct ext4_group_desc *bg); 3335 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3336 struct ext4_group_desc *bg); 3337 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3338 struct ext4_group_desc *bg); 3339 extern void ext4_block_bitmap_set(struct super_block *sb, 3340 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3341 extern void ext4_inode_bitmap_set(struct super_block *sb, 3342 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3343 extern void ext4_inode_table_set(struct super_block *sb, 3344 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3345 extern void ext4_free_group_clusters_set(struct super_block *sb, 3346 struct ext4_group_desc *bg, 3347 __u32 count); 3348 extern void ext4_free_inodes_set(struct super_block *sb, 3349 struct ext4_group_desc *bg, __u32 count); 3350 extern void ext4_used_dirs_set(struct super_block *sb, 3351 struct ext4_group_desc *bg, __u32 count); 3352 extern void ext4_itable_unused_set(struct super_block *sb, 3353 struct ext4_group_desc *bg, __u32 count); 3354 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3355 struct ext4_group_desc *gdp); 3356 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3357 struct ext4_group_desc *gdp); 3358 extern int ext4_register_li_request(struct super_block *sb, 3359 ext4_group_t first_not_zeroed); 3360 3361 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3362 { 3363 return ext4_has_feature_gdt_csum(sb) || 3364 ext4_has_feature_metadata_csum(sb); 3365 } 3366 3367 #define ext4_read_incompat_64bit_val(es, name) \ 3368 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3369 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3370 le32_to_cpu(es->name##_lo)) 3371 3372 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3373 { 3374 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3375 } 3376 3377 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3378 { 3379 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3380 } 3381 3382 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3383 { 3384 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3385 } 3386 3387 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3388 ext4_fsblk_t blk) 3389 { 3390 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3391 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3392 } 3393 3394 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3395 ext4_fsblk_t blk) 3396 { 3397 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3398 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3399 } 3400 3401 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3402 ext4_fsblk_t blk) 3403 { 3404 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3405 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3406 } 3407 3408 static inline loff_t ext4_isize(struct super_block *sb, 3409 struct ext4_inode *raw_inode) 3410 { 3411 if (ext4_has_feature_largedir(sb) || 3412 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3413 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3414 le32_to_cpu(raw_inode->i_size_lo); 3415 3416 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3417 } 3418 3419 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3420 { 3421 raw_inode->i_size_lo = cpu_to_le32(i_size); 3422 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3423 } 3424 3425 /* 3426 * Reading s_groups_count requires using smp_rmb() afterwards. See 3427 * the locking protocol documented in the comments of ext4_group_add() 3428 * in resize.c 3429 */ 3430 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3431 { 3432 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3433 3434 smp_rmb(); 3435 return ngroups; 3436 } 3437 3438 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3439 ext4_group_t block_group) 3440 { 3441 return block_group >> sbi->s_log_groups_per_flex; 3442 } 3443 3444 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3445 { 3446 return 1 << sbi->s_log_groups_per_flex; 3447 } 3448 3449 static inline loff_t ext4_get_maxbytes(struct inode *inode) 3450 { 3451 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) 3452 return inode->i_sb->s_maxbytes; 3453 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; 3454 } 3455 3456 #define ext4_std_error(sb, errno) \ 3457 do { \ 3458 if ((errno)) \ 3459 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3460 } while (0) 3461 3462 #ifdef CONFIG_SMP 3463 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3464 * counters. So we need to make sure we have free clusters more 3465 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3466 */ 3467 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3468 #else 3469 #define EXT4_FREECLUSTERS_WATERMARK 0 3470 #endif 3471 3472 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3473 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3474 { 3475 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3476 !inode_is_locked(inode)); 3477 down_write(&EXT4_I(inode)->i_data_sem); 3478 if (newsize > EXT4_I(inode)->i_disksize) 3479 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3480 up_write(&EXT4_I(inode)->i_data_sem); 3481 } 3482 3483 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3484 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3485 { 3486 int changed = 0; 3487 3488 if (newsize > inode->i_size) { 3489 i_size_write(inode, newsize); 3490 changed = 1; 3491 } 3492 if (newsize > EXT4_I(inode)->i_disksize) { 3493 ext4_update_i_disksize(inode, newsize); 3494 changed |= 2; 3495 } 3496 return changed; 3497 } 3498 3499 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3500 loff_t len); 3501 3502 struct ext4_group_info { 3503 unsigned long bb_state; 3504 #ifdef AGGRESSIVE_CHECK 3505 unsigned long bb_check_counter; 3506 #endif 3507 struct rb_root bb_free_root; 3508 ext4_grpblk_t bb_first_free; /* first free block */ 3509 ext4_grpblk_t bb_free; /* total free blocks */ 3510 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3511 int bb_avg_fragment_size_order; /* order of average 3512 fragment in BG */ 3513 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3514 ext4_group_t bb_group; /* Group number */ 3515 struct list_head bb_prealloc_list; 3516 #ifdef DOUBLE_CHECK 3517 void *bb_bitmap; 3518 #endif 3519 struct rw_semaphore alloc_sem; 3520 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3521 * regions, index is order. 3522 * bb_counters[3] = 5 means 3523 * 5 free 8-block regions. */ 3524 }; 3525 3526 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3527 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3528 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3529 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3530 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3531 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3532 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3533 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3534 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3535 3536 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3537 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3538 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3539 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3540 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3541 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3542 3543 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3544 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3545 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3546 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3547 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3548 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3549 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3550 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3551 3552 #define EXT4_MAX_CONTENTION 8 3553 #define EXT4_CONTENTION_THRESHOLD 2 3554 3555 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3556 ext4_group_t group) 3557 { 3558 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3559 } 3560 3561 /* 3562 * Returns true if the filesystem is busy enough that attempts to 3563 * access the block group locks has run into contention. 3564 */ 3565 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3566 { 3567 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3568 } 3569 3570 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group) 3571 { 3572 if (!spin_trylock(ext4_group_lock_ptr(sb, group))) 3573 return false; 3574 /* 3575 * We're able to grab the lock right away, so drop the lock 3576 * contention counter. 3577 */ 3578 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3579 return true; 3580 } 3581 3582 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3583 { 3584 if (!ext4_try_lock_group(sb, group)) { 3585 /* 3586 * The lock is busy, so bump the contention counter, 3587 * and then wait on the spin lock. 3588 */ 3589 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3590 EXT4_MAX_CONTENTION); 3591 spin_lock(ext4_group_lock_ptr(sb, group)); 3592 } 3593 } 3594 3595 static inline void ext4_unlock_group(struct super_block *sb, 3596 ext4_group_t group) 3597 { 3598 spin_unlock(ext4_group_lock_ptr(sb, group)); 3599 } 3600 3601 #ifdef CONFIG_QUOTA 3602 static inline bool ext4_quota_capable(struct super_block *sb) 3603 { 3604 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3605 } 3606 3607 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3608 { 3609 struct ext4_sb_info *sbi = EXT4_SB(sb); 3610 3611 return (ext4_has_feature_quota(sb) || 3612 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3613 } 3614 int ext4_enable_quotas(struct super_block *sb); 3615 #endif 3616 3617 /* 3618 * Block validity checking 3619 */ 3620 #define ext4_check_indirect_blockref(inode, bh) \ 3621 ext4_check_blockref(__func__, __LINE__, inode, \ 3622 (__le32 *)(bh)->b_data, \ 3623 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3624 3625 #define ext4_ind_check_inode(inode) \ 3626 ext4_check_blockref(__func__, __LINE__, inode, \ 3627 EXT4_I(inode)->i_data, \ 3628 EXT4_NDIR_BLOCKS) 3629 3630 /* 3631 * Inodes and files operations 3632 */ 3633 3634 /* dir.c */ 3635 extern const struct file_operations ext4_dir_operations; 3636 3637 /* file.c */ 3638 extern const struct inode_operations ext4_file_inode_operations; 3639 extern const struct file_operations ext4_file_operations; 3640 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3641 3642 /* inline.c */ 3643 extern int ext4_get_max_inline_size(struct inode *inode); 3644 extern int ext4_find_inline_data_nolock(struct inode *inode); 3645 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3646 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size); 3647 3648 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3649 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3650 struct inode *inode, 3651 loff_t pos, unsigned len, 3652 struct folio **foliop); 3653 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3654 unsigned copied, struct folio *folio); 3655 extern int ext4_generic_write_inline_data(struct address_space *mapping, 3656 struct inode *inode, 3657 loff_t pos, unsigned len, 3658 struct folio **foliop, 3659 void **fsdata, bool da); 3660 extern int ext4_try_add_inline_entry(handle_t *handle, 3661 struct ext4_filename *fname, 3662 struct inode *dir, struct inode *inode); 3663 extern int ext4_try_create_inline_dir(handle_t *handle, 3664 struct inode *parent, 3665 struct inode *inode); 3666 extern int ext4_read_inline_dir(struct file *filp, 3667 struct dir_context *ctx, 3668 int *has_inline_data); 3669 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3670 struct inode *dir, ext4_lblk_t block, 3671 struct dx_hash_info *hinfo, 3672 __u32 start_hash, __u32 start_minor_hash, 3673 int *has_inline_data); 3674 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3675 struct ext4_filename *fname, 3676 struct ext4_dir_entry_2 **res_dir, 3677 int *has_inline_data); 3678 extern int ext4_delete_inline_entry(handle_t *handle, 3679 struct inode *dir, 3680 struct ext4_dir_entry_2 *de_del, 3681 struct buffer_head *bh, 3682 int *has_inline_data); 3683 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3684 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3685 struct ext4_dir_entry_2 **parent_de, 3686 int *retval); 3687 extern void *ext4_read_inline_link(struct inode *inode); 3688 3689 struct iomap; 3690 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3691 3692 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3693 3694 extern int ext4_convert_inline_data(struct inode *inode); 3695 3696 static inline int ext4_has_inline_data(struct inode *inode) 3697 { 3698 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3699 EXT4_I(inode)->i_inline_off; 3700 } 3701 3702 /* namei.c */ 3703 extern const struct inode_operations ext4_dir_inode_operations; 3704 extern const struct inode_operations ext4_special_inode_operations; 3705 extern struct dentry *ext4_get_parent(struct dentry *child); 3706 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode, 3707 struct buffer_head *dir_block, 3708 unsigned int parent_ino, void *inline_buf, 3709 int inline_size); 3710 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3711 unsigned int blocksize); 3712 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3713 struct buffer_head *bh); 3714 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3715 struct inode *inode, struct dentry *dentry); 3716 extern int __ext4_link(struct inode *dir, struct inode *inode, 3717 struct dentry *dentry); 3718 3719 #define S_SHIFT 12 3720 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3721 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3722 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3723 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3724 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3725 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3726 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3727 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3728 }; 3729 3730 static inline void ext4_set_de_type(struct super_block *sb, 3731 struct ext4_dir_entry_2 *de, 3732 umode_t mode) { 3733 if (ext4_has_feature_filetype(sb)) 3734 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3735 } 3736 3737 /* readpages.c */ 3738 extern int ext4_mpage_readpages(struct inode *inode, 3739 struct readahead_control *rac, struct folio *folio); 3740 extern int __init ext4_init_post_read_processing(void); 3741 extern void ext4_exit_post_read_processing(void); 3742 3743 /* symlink.c */ 3744 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3745 extern const struct inode_operations ext4_symlink_inode_operations; 3746 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3747 3748 /* sysfs.c */ 3749 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3750 extern int ext4_register_sysfs(struct super_block *sb); 3751 extern void ext4_unregister_sysfs(struct super_block *sb); 3752 extern int __init ext4_init_sysfs(void); 3753 extern void ext4_exit_sysfs(void); 3754 3755 /* block_validity */ 3756 extern void ext4_release_system_zone(struct super_block *sb); 3757 extern int ext4_setup_system_zone(struct super_block *sb); 3758 extern int __init ext4_init_system_zone(void); 3759 extern void ext4_exit_system_zone(void); 3760 extern int ext4_inode_block_valid(struct inode *inode, 3761 ext4_fsblk_t start_blk, 3762 unsigned int count); 3763 extern int ext4_check_blockref(const char *, unsigned int, 3764 struct inode *, __le32 *, unsigned int); 3765 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3766 ext4_fsblk_t start_blk, unsigned int count); 3767 3768 3769 /* extents.c */ 3770 struct ext4_ext_path; 3771 struct ext4_extent; 3772 3773 /* 3774 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3775 * __le32. 3776 */ 3777 #define EXT_MAX_BLOCKS 0xffffffff 3778 3779 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3780 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3781 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3782 struct ext4_map_blocks *map, int flags); 3783 extern int ext4_ext_truncate(handle_t *, struct inode *); 3784 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3785 ext4_lblk_t end); 3786 extern void ext4_ext_init(struct super_block *); 3787 extern void ext4_ext_release(struct super_block *); 3788 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3789 loff_t len); 3790 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3791 loff_t offset, ssize_t len); 3792 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle, 3793 struct inode *inode, loff_t offset, ssize_t len); 3794 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3795 ext4_io_end_t *io_end); 3796 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3797 struct ext4_map_blocks *map, int flags); 3798 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3799 int num, 3800 struct ext4_ext_path *path); 3801 extern struct ext4_ext_path *ext4_ext_insert_extent( 3802 handle_t *handle, struct inode *inode, 3803 struct ext4_ext_path *path, 3804 struct ext4_extent *newext, int gb_flags); 3805 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3806 struct ext4_ext_path *, 3807 int flags); 3808 extern void ext4_free_ext_path(struct ext4_ext_path *); 3809 extern int ext4_ext_check_inode(struct inode *inode); 3810 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3811 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3812 __u64 start, __u64 len); 3813 extern int ext4_get_es_cache(struct inode *inode, 3814 struct fiemap_extent_info *fieinfo, 3815 __u64 start, __u64 len); 3816 extern int ext4_ext_precache(struct inode *inode); 3817 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3818 struct inode *inode2, ext4_lblk_t lblk1, 3819 ext4_lblk_t lblk2, ext4_lblk_t count, 3820 int mark_unwritten,int *err); 3821 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3822 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3823 int check_cred, int restart_cred, 3824 int revoke_cred); 3825 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3826 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3827 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3828 int len, int unwritten, ext4_fsblk_t pblk); 3829 extern int ext4_ext_clear_bb(struct inode *inode); 3830 3831 3832 /* move_extent.c */ 3833 extern void ext4_double_down_write_data_sem(struct inode *first, 3834 struct inode *second); 3835 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3836 struct inode *donor_inode); 3837 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3838 __u64 start_orig, __u64 start_donor, 3839 __u64 len, __u64 *moved_len); 3840 3841 /* page-io.c */ 3842 extern int __init ext4_init_pageio(void); 3843 extern void ext4_exit_pageio(void); 3844 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3845 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3846 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3847 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3848 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3849 struct writeback_control *wbc); 3850 extern void ext4_end_io_rsv_work(struct work_struct *work); 3851 extern void ext4_io_submit(struct ext4_io_submit *io); 3852 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3853 size_t len); 3854 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3855 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3856 3857 /* mmp.c */ 3858 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3859 3860 /* mmp.c */ 3861 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3862 3863 /* verity.c */ 3864 extern const struct fsverity_operations ext4_verityops; 3865 3866 /* orphan.c */ 3867 extern int ext4_orphan_add(handle_t *, struct inode *); 3868 extern int ext4_orphan_del(handle_t *, struct inode *); 3869 extern void ext4_orphan_cleanup(struct super_block *sb, 3870 struct ext4_super_block *es); 3871 extern void ext4_release_orphan_info(struct super_block *sb); 3872 extern int ext4_init_orphan_info(struct super_block *sb); 3873 extern int ext4_orphan_file_empty(struct super_block *sb); 3874 extern void ext4_orphan_file_block_trigger( 3875 struct jbd2_buffer_trigger_type *triggers, 3876 struct buffer_head *bh, 3877 void *data, size_t size); 3878 3879 /* 3880 * Add new method to test whether block and inode bitmaps are properly 3881 * initialized. With uninit_bg reading the block from disk is not enough 3882 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3883 */ 3884 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3885 3886 static inline int bitmap_uptodate(struct buffer_head *bh) 3887 { 3888 return (buffer_uptodate(bh) && 3889 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3890 } 3891 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3892 { 3893 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3894 } 3895 3896 extern int ext4_resize_begin(struct super_block *sb); 3897 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3898 3899 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end) 3900 { 3901 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) 3902 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3903 } 3904 3905 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3906 { 3907 if (io_end->flag & EXT4_IO_END_UNWRITTEN) 3908 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3909 } 3910 3911 extern const struct iomap_ops ext4_iomap_ops; 3912 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3913 extern const struct iomap_ops ext4_iomap_report_ops; 3914 3915 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3916 { 3917 /* 3918 * If the buffer has the write error flag, we have failed 3919 * to write out data in the block. In this case, we don't 3920 * have to read the block because we may read the old data 3921 * successfully. 3922 */ 3923 if (buffer_write_io_error(bh)) 3924 set_buffer_uptodate(bh); 3925 return buffer_uptodate(bh); 3926 } 3927 3928 static inline bool ext4_inode_can_atomic_write(struct inode *inode) 3929 { 3930 3931 return S_ISREG(inode->i_mode) && 3932 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) && 3933 EXT4_SB(inode->i_sb)->s_awu_min > 0; 3934 } 3935 3936 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio, 3937 loff_t pos, unsigned len, 3938 get_block_t *get_block); 3939 #endif /* __KERNEL__ */ 3940 3941 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3942 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3943 3944 #endif /* _EXT4_H */ 3945