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 %llu: (%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 #define EXT4_LBLK_TO_B(inode, lblk) ((loff_t)(lblk) << (inode)->i_blkbits) 372 373 /* Translate a block number to a page index */ 374 #define EXT4_LBLK_TO_PG(inode, lblk) (EXT4_LBLK_TO_B((inode), (lblk)) >> \ 375 PAGE_SHIFT) 376 /* Translate a page index to a block number */ 377 #define EXT4_PG_TO_LBLK(inode, pnum) (((loff_t)(pnum) << PAGE_SHIFT) >> \ 378 (inode)->i_blkbits) 379 /* Translate a block number to a cluster number */ 380 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 381 /* Translate a cluster number to a block number */ 382 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 383 /* Translate # of blks to # of clusters */ 384 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 385 (sbi)->s_cluster_bits) 386 /* Mask out the low bits to get the starting block of the cluster */ 387 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 388 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 389 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 390 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 391 /* Fill in the low bits to get the last block of the cluster */ 392 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 393 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 394 /* Get the cluster offset */ 395 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 396 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 397 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 398 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 399 400 /* 401 * Structure of a blocks group descriptor 402 */ 403 struct ext4_group_desc 404 { 405 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 406 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 407 __le32 bg_inode_table_lo; /* Inodes table block */ 408 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 409 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 410 __le16 bg_used_dirs_count_lo; /* Directories count */ 411 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 412 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 413 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 414 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 415 __le16 bg_itable_unused_lo; /* Unused inodes count */ 416 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 417 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 418 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 419 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 420 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 421 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 422 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 423 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 424 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 425 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 426 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 427 __u32 bg_reserved; 428 }; 429 430 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 431 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 432 sizeof(__le16)) 433 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 434 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 435 sizeof(__le16)) 436 437 /* 438 * Structure of a flex block group info 439 */ 440 441 struct flex_groups { 442 atomic64_t free_clusters; 443 atomic_t free_inodes; 444 atomic_t used_dirs; 445 }; 446 447 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 448 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 449 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 450 451 /* 452 * Macro-instructions used to manage group descriptors 453 */ 454 #define EXT4_MIN_DESC_SIZE 32 455 #define EXT4_MIN_DESC_SIZE_64BIT 64 456 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 457 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 458 #ifdef __KERNEL__ 459 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 460 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 461 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 462 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 463 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 464 #else 465 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 466 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 467 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 468 #endif 469 470 /* 471 * Constants relative to the data blocks 472 */ 473 #define EXT4_NDIR_BLOCKS 12 474 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 475 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 476 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 477 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 478 479 /* 480 * Inode flags 481 */ 482 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 483 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 484 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 485 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 486 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 487 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 488 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 489 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 490 /* Reserved for compression usage... */ 491 #define EXT4_DIRTY_FL 0x00000100 492 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 493 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 494 /* nb: was previously EXT2_ECOMPR_FL */ 495 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 496 /* End compression flags --- maybe not all used */ 497 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 498 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 499 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 500 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 501 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 502 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 503 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 504 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 505 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 506 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 507 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 508 509 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 510 511 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 512 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 513 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 514 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 515 516 /* User modifiable flags */ 517 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 518 EXT4_UNRM_FL | \ 519 EXT4_COMPR_FL | \ 520 EXT4_SYNC_FL | \ 521 EXT4_IMMUTABLE_FL | \ 522 EXT4_APPEND_FL | \ 523 EXT4_NODUMP_FL | \ 524 EXT4_NOATIME_FL | \ 525 EXT4_JOURNAL_DATA_FL | \ 526 EXT4_NOTAIL_FL | \ 527 EXT4_DIRSYNC_FL | \ 528 EXT4_TOPDIR_FL | \ 529 EXT4_EXTENTS_FL | \ 530 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 531 EXT4_DAX_FL | \ 532 EXT4_PROJINHERIT_FL | \ 533 EXT4_CASEFOLD_FL) 534 535 /* User visible flags */ 536 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 537 EXT4_DIRTY_FL | \ 538 EXT4_COMPRBLK_FL | \ 539 EXT4_NOCOMPR_FL | \ 540 EXT4_ENCRYPT_FL | \ 541 EXT4_INDEX_FL | \ 542 EXT4_VERITY_FL | \ 543 EXT4_INLINE_DATA_FL) 544 545 /* Flags that should be inherited by new inodes from their parent. */ 546 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 547 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 548 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 549 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 550 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 551 EXT4_DAX_FL) 552 553 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 554 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 555 EXT4_PROJINHERIT_FL)) 556 557 /* Flags that are appropriate for non-directories/regular files. */ 558 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 559 560 /* The only flags that should be swapped */ 561 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 562 563 /* Flags which are mutually exclusive to DAX */ 564 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 565 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 566 567 /* Mask out flags that are inappropriate for the given type of inode. */ 568 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 569 { 570 if (S_ISDIR(mode)) 571 return flags; 572 else if (S_ISREG(mode)) 573 return flags & EXT4_REG_FLMASK; 574 else 575 return flags & EXT4_OTHER_FLMASK; 576 } 577 578 /* 579 * Inode flags used for atomic set/get 580 */ 581 enum { 582 EXT4_INODE_SECRM = 0, /* Secure deletion */ 583 EXT4_INODE_UNRM = 1, /* Undelete */ 584 EXT4_INODE_COMPR = 2, /* Compress file */ 585 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 586 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 587 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 588 EXT4_INODE_NODUMP = 6, /* do not dump file */ 589 EXT4_INODE_NOATIME = 7, /* do not update atime */ 590 /* Reserved for compression usage... */ 591 EXT4_INODE_DIRTY = 8, 592 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 593 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 594 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 595 /* End compression flags --- maybe not all used */ 596 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 597 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 598 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 599 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 600 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 601 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 602 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 603 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 604 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 605 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 606 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 607 EXT4_INODE_DAX = 25, /* Inode is DAX */ 608 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 609 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 610 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 611 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 612 }; 613 614 /* 615 * Since it's pretty easy to mix up bit numbers and hex values, we use a 616 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 617 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 618 * any extra space in the compiled kernel image, otherwise, the build will fail. 619 * It's important that these values are the same, since we are using 620 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 621 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 622 * values found in ext2, ext3 and ext4 filesystems, and of course the values 623 * defined in e2fsprogs. 624 * 625 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 626 */ 627 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 628 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 629 630 static inline void ext4_check_flag_values(void) 631 { 632 CHECK_FLAG_VALUE(SECRM); 633 CHECK_FLAG_VALUE(UNRM); 634 CHECK_FLAG_VALUE(COMPR); 635 CHECK_FLAG_VALUE(SYNC); 636 CHECK_FLAG_VALUE(IMMUTABLE); 637 CHECK_FLAG_VALUE(APPEND); 638 CHECK_FLAG_VALUE(NODUMP); 639 CHECK_FLAG_VALUE(NOATIME); 640 CHECK_FLAG_VALUE(DIRTY); 641 CHECK_FLAG_VALUE(COMPRBLK); 642 CHECK_FLAG_VALUE(NOCOMPR); 643 CHECK_FLAG_VALUE(ENCRYPT); 644 CHECK_FLAG_VALUE(INDEX); 645 CHECK_FLAG_VALUE(IMAGIC); 646 CHECK_FLAG_VALUE(JOURNAL_DATA); 647 CHECK_FLAG_VALUE(NOTAIL); 648 CHECK_FLAG_VALUE(DIRSYNC); 649 CHECK_FLAG_VALUE(TOPDIR); 650 CHECK_FLAG_VALUE(HUGE_FILE); 651 CHECK_FLAG_VALUE(EXTENTS); 652 CHECK_FLAG_VALUE(VERITY); 653 CHECK_FLAG_VALUE(EA_INODE); 654 CHECK_FLAG_VALUE(INLINE_DATA); 655 CHECK_FLAG_VALUE(PROJINHERIT); 656 CHECK_FLAG_VALUE(CASEFOLD); 657 CHECK_FLAG_VALUE(RESERVED); 658 } 659 660 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 661 struct compat_ext4_new_group_input { 662 u32 group; 663 compat_u64 block_bitmap; 664 compat_u64 inode_bitmap; 665 compat_u64 inode_table; 666 u32 blocks_count; 667 u16 reserved_blocks; 668 u16 unused; 669 }; 670 #endif 671 672 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 673 struct ext4_new_group_data { 674 __u32 group; 675 __u64 block_bitmap; 676 __u64 inode_bitmap; 677 __u64 inode_table; 678 __u32 blocks_count; 679 __u16 reserved_blocks; 680 __u16 mdata_blocks; 681 __u32 free_clusters_count; 682 }; 683 684 /* Indexes used to index group tables in ext4_new_group_data */ 685 enum { 686 BLOCK_BITMAP = 0, /* block bitmap */ 687 INODE_BITMAP, /* inode bitmap */ 688 INODE_TABLE, /* inode tables */ 689 GROUP_TABLE_COUNT, 690 }; 691 692 /* 693 * Flags used by ext4_map_blocks() 694 */ 695 /* Allocate any needed blocks and/or convert an unwritten 696 extent to be an initialized ext4 */ 697 #define EXT4_GET_BLOCKS_CREATE 0x0001 698 /* Request the creation of an unwritten extent */ 699 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 700 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 701 EXT4_GET_BLOCKS_CREATE) 702 /* Caller is from the delayed allocation writeout path 703 * finally doing the actual allocation of delayed blocks */ 704 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 705 /* 706 * This means that we cannot merge newly allocated extents, and if we 707 * found an unwritten extent, we need to split it. 708 */ 709 #define EXT4_GET_BLOCKS_SPLIT_NOMERGE 0x0008 710 /* Convert unwritten extent to initialized. */ 711 #define EXT4_GET_BLOCKS_CONVERT 0x0010 712 /* Eventual metadata allocation (due to growing extent tree) 713 * should not fail, so try to use reserved blocks for that.*/ 714 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 715 /* Don't normalize allocation size (used for fallocate) */ 716 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 717 /* Convert written extents to unwritten */ 718 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 719 /* Write zeros to newly created written extents */ 720 #define EXT4_GET_BLOCKS_ZERO 0x0200 721 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 722 EXT4_GET_BLOCKS_ZERO) 723 /* Caller is in the context of data submission, such as writeback, 724 * fsync, etc. Especially, in the generic writeback path, caller will 725 * submit data before dropping transaction handle. This allows jbd2 726 * to avoid submitting data before commit. */ 727 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 728 /* Convert extent to initialized after IO complete */ 729 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\ 730 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\ 731 EXT4_GET_BLOCKS_IO_SUBMIT) 732 /* Caller is in the atomic contex, find extent if it has been cached */ 733 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 734 /* 735 * Atomic write caller needs this to query in the slow path of mixed mapping 736 * case, when a contiguous extent can be split across two adjacent leaf nodes. 737 * Look EXT4_MAP_QUERY_LAST_IN_LEAF. 738 */ 739 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000 740 741 /* 742 * The bit position of these flags must not overlap with any of the 743 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 744 * read_extent_tree_block(), ext4_split_extent_at(), 745 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 746 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 747 * caching the extents when reading from the extent tree while a 748 * truncate or punch hole operation is in progress. 749 */ 750 #define EXT4_EX_NOCACHE 0x40000000 751 #define EXT4_EX_FORCE_CACHE 0x20000000 752 #define EXT4_EX_NOFAIL 0x10000000 753 /* 754 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to 755 * pass while lookup/querying of on disk extent tree. 756 */ 757 #define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\ 758 EXT4_EX_NOFAIL |\ 759 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF) 760 761 /* 762 * Flags used by ext4_free_blocks 763 */ 764 #define EXT4_FREE_BLOCKS_METADATA 0x0001 765 #define EXT4_FREE_BLOCKS_FORGET 0x0002 766 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 767 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 768 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 769 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 770 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 771 772 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 773 /* 774 * ioctl commands in 32 bit emulation 775 */ 776 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 777 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 778 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 779 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 780 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 781 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 782 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 783 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 784 #endif 785 786 /* Max physical block we can address w/o extents */ 787 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 788 789 /* Max logical block we can support */ 790 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 791 792 /* 793 * Structure of an inode on the disk 794 */ 795 struct ext4_inode { 796 __le16 i_mode; /* File mode */ 797 __le16 i_uid; /* Low 16 bits of Owner Uid */ 798 __le32 i_size_lo; /* Size in bytes */ 799 __le32 i_atime; /* Access time */ 800 __le32 i_ctime; /* Inode Change time */ 801 __le32 i_mtime; /* Modification time */ 802 __le32 i_dtime; /* Deletion Time */ 803 __le16 i_gid; /* Low 16 bits of Group Id */ 804 __le16 i_links_count; /* Links count */ 805 __le32 i_blocks_lo; /* Blocks count */ 806 __le32 i_flags; /* File flags */ 807 union { 808 struct { 809 __le32 l_i_version; 810 } linux1; 811 struct { 812 __u32 h_i_translator; 813 } hurd1; 814 struct { 815 __u32 m_i_reserved1; 816 } masix1; 817 } osd1; /* OS dependent 1 */ 818 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 819 __le32 i_generation; /* File version (for NFS) */ 820 __le32 i_file_acl_lo; /* File ACL */ 821 __le32 i_size_high; 822 __le32 i_obso_faddr; /* Obsoleted fragment address */ 823 union { 824 struct { 825 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 826 __le16 l_i_file_acl_high; 827 __le16 l_i_uid_high; /* these 2 fields */ 828 __le16 l_i_gid_high; /* were reserved2[0] */ 829 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 830 __le16 l_i_reserved; 831 } linux2; 832 struct { 833 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 834 __u16 h_i_mode_high; 835 __u16 h_i_uid_high; 836 __u16 h_i_gid_high; 837 __u32 h_i_author; 838 } hurd2; 839 struct { 840 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 841 __le16 m_i_file_acl_high; 842 __u32 m_i_reserved2[2]; 843 } masix2; 844 } osd2; /* OS dependent 2 */ 845 __le16 i_extra_isize; 846 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 847 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 848 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 849 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 850 __le32 i_crtime; /* File Creation time */ 851 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 852 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 853 __le32 i_projid; /* Project ID */ 854 }; 855 856 #define EXT4_EPOCH_BITS 2 857 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 858 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 859 860 /* 861 * Extended fields will fit into an inode if the filesystem was formatted 862 * with large inodes (-I 256 or larger) and there are not currently any EAs 863 * consuming all of the available space. For new inodes we always reserve 864 * enough space for the kernel's known extended fields, but for inodes 865 * created with an old kernel this might not have been the case. None of 866 * the extended inode fields is critical for correct filesystem operation. 867 * This macro checks if a certain field fits in the inode. Note that 868 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 869 */ 870 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 871 ((offsetof(typeof(*ext4_inode), field) + \ 872 sizeof((ext4_inode)->field)) \ 873 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 874 (einode)->i_extra_isize)) \ 875 876 /* 877 * We use an encoding that preserves the times for extra epoch "00": 878 * 879 * extra msb of adjust for signed 880 * epoch 32-bit 32-bit tv_sec to 881 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 882 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 883 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 884 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 885 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 886 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 887 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 888 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 889 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 890 * 891 * Note that previous versions of the kernel on 64-bit systems would 892 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 893 * 1970. e2fsck will correct this, assuming that it is run on the 894 * affected filesystem before 2242. 895 */ 896 897 static inline __le32 ext4_encode_extra_time(struct timespec64 ts) 898 { 899 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK; 900 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS)); 901 } 902 903 static inline struct timespec64 ext4_decode_extra_time(__le32 base, 904 __le32 extra) 905 { 906 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) }; 907 908 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 909 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 910 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 911 return ts; 912 } 913 914 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \ 915 do { \ 916 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 917 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \ 918 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \ 919 } else \ 920 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \ 921 } while (0) 922 923 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \ 924 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode)) 925 926 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \ 927 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode)) 928 929 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \ 930 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode)) 931 932 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 933 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 934 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \ 935 raw_inode, (einode)->xtime) 936 937 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \ 938 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \ 939 ext4_decode_extra_time((raw_inode)->xtime, \ 940 (raw_inode)->xtime ## _extra) : \ 941 (struct timespec64) { \ 942 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \ 943 }) 944 945 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \ 946 do { \ 947 inode_set_atime_to_ts(inode, \ 948 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \ 949 } while (0) 950 951 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \ 952 do { \ 953 inode_set_mtime_to_ts(inode, \ 954 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \ 955 } while (0) 956 957 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \ 958 do { \ 959 inode_set_ctime_to_ts(inode, \ 960 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \ 961 } while (0) 962 963 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 964 do { \ 965 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 966 (einode)->xtime = \ 967 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \ 968 raw_inode); \ 969 else \ 970 (einode)->xtime = (struct timespec64){0, 0}; \ 971 } while (0) 972 973 #define i_disk_version osd1.linux1.l_i_version 974 975 #if defined(__KERNEL__) || defined(__linux__) 976 #define i_reserved1 osd1.linux1.l_i_reserved1 977 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 978 #define i_blocks_high osd2.linux2.l_i_blocks_high 979 #define i_uid_low i_uid 980 #define i_gid_low i_gid 981 #define i_uid_high osd2.linux2.l_i_uid_high 982 #define i_gid_high osd2.linux2.l_i_gid_high 983 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 984 985 #elif defined(__GNU__) 986 987 #define i_translator osd1.hurd1.h_i_translator 988 #define i_uid_high osd2.hurd2.h_i_uid_high 989 #define i_gid_high osd2.hurd2.h_i_gid_high 990 #define i_author osd2.hurd2.h_i_author 991 992 #elif defined(__masix__) 993 994 #define i_reserved1 osd1.masix1.m_i_reserved1 995 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 996 #define i_reserved2 osd2.masix2.m_i_reserved2 997 998 #endif /* defined(__KERNEL__) || defined(__linux__) */ 999 1000 #include "extents_status.h" 1001 #include "fast_commit.h" 1002 1003 /* 1004 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 1005 * 1006 * These are needed to avoid lockdep false positives when we need to 1007 * allocate blocks to the quota inode during ext4_map_blocks(), while 1008 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1009 * do quota tracking for the quota inode, this avoids deadlock (as 1010 * well as infinite recursion, since it isn't turtles all the way 1011 * down...) 1012 * 1013 * I_DATA_SEM_NORMAL - Used for most inodes 1014 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1015 * where the second inode has larger inode number 1016 * than the first 1017 * I_DATA_SEM_QUOTA - Used for quota inodes only 1018 * I_DATA_SEM_EA - Used for ea_inodes only 1019 */ 1020 enum { 1021 I_DATA_SEM_NORMAL = 0, 1022 I_DATA_SEM_OTHER, 1023 I_DATA_SEM_QUOTA, 1024 I_DATA_SEM_EA 1025 }; 1026 1027 1028 /* 1029 * fourth extended file system inode data in memory 1030 */ 1031 struct ext4_inode_info { 1032 __le32 i_data[15]; /* unconverted */ 1033 __u32 i_dtime; 1034 ext4_fsblk_t i_file_acl; 1035 1036 /* 1037 * i_block_group is the number of the block group which contains 1038 * this file's inode. Constant across the lifetime of the inode, 1039 * it is used for making block allocation decisions - we try to 1040 * place a file's data blocks near its inode block, and new inodes 1041 * near to their parent directory's inode. 1042 */ 1043 ext4_group_t i_block_group; 1044 ext4_lblk_t i_dir_start_lookup; 1045 #if (BITS_PER_LONG < 64) 1046 unsigned long i_state_flags; /* Dynamic state flags */ 1047 #endif 1048 unsigned long i_flags; 1049 1050 /* 1051 * Extended attributes can be read independently of the main file 1052 * data. Taking i_rwsem even when reading would cause contention 1053 * between readers of EAs and writers of regular file data, so 1054 * instead we synchronize on xattr_sem when reading or changing 1055 * EAs. 1056 */ 1057 struct rw_semaphore xattr_sem; 1058 1059 /* 1060 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1061 * i_orphan is used. 1062 */ 1063 union { 1064 struct list_head i_orphan; /* unlinked but open inodes */ 1065 unsigned int i_orphan_idx; /* Index in orphan file */ 1066 }; 1067 1068 /* Fast commit related info */ 1069 1070 /* For tracking dentry create updates */ 1071 struct list_head i_fc_dilist; 1072 struct list_head i_fc_list; /* 1073 * inodes that need fast commit 1074 * protected by sbi->s_fc_lock. 1075 */ 1076 1077 /* Start of lblk range that needs to be committed in this fast commit */ 1078 ext4_lblk_t i_fc_lblk_start; 1079 1080 /* End of lblk range that needs to be committed in this fast commit */ 1081 ext4_lblk_t i_fc_lblk_len; 1082 1083 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1084 1085 /* Fast commit wait queue for this inode */ 1086 wait_queue_head_t i_fc_wait; 1087 1088 /* 1089 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len 1090 * and inode's EXT4_FC_STATE_COMMITTING state bit. 1091 */ 1092 spinlock_t i_fc_lock; 1093 1094 /* 1095 * i_disksize keeps track of what the inode size is ON DISK, not 1096 * in memory. During truncate, i_size is set to the new size by 1097 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1098 * set i_disksize to 0 until the truncate is actually under way. 1099 * 1100 * The intent is that i_disksize always represents the blocks which 1101 * are used by this file. This allows recovery to restart truncate 1102 * on orphans if we crash during truncate. We actually write i_disksize 1103 * into the on-disk inode when writing inodes out, instead of i_size. 1104 * 1105 * The only time when i_disksize and i_size may be different is when 1106 * a truncate is in progress. The only things which change i_disksize 1107 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1108 */ 1109 loff_t i_disksize; 1110 1111 /* 1112 * i_data_sem is for serialising ext4_truncate() against 1113 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1114 * data tree are chopped off during truncate. We can't do that in 1115 * ext4 because whenever we perform intermediate commits during 1116 * truncate, the inode and all the metadata blocks *must* be in a 1117 * consistent state which allows truncation of the orphans to restart 1118 * during recovery. Hence we must fix the get_block-vs-truncate race 1119 * by other means, so we have i_data_sem. 1120 */ 1121 struct rw_semaphore i_data_sem; 1122 struct inode vfs_inode; 1123 struct jbd2_inode *jinode; 1124 struct mapping_metadata_bhs i_metadata_bhs; 1125 1126 /* 1127 * File creation time. Its function is same as that of 1128 * struct timespec64 i_{a,c,m}time in the generic inode. 1129 */ 1130 struct timespec64 i_crtime; 1131 1132 /* mballoc */ 1133 atomic_t i_prealloc_active; 1134 1135 /* allocation reservation info for delalloc */ 1136 /* In case of bigalloc, this refer to clusters rather than blocks */ 1137 unsigned int i_reserved_data_blocks; 1138 struct rb_root i_prealloc_node; 1139 rwlock_t i_prealloc_lock; 1140 1141 /* extents status tree */ 1142 struct ext4_es_tree i_es_tree; 1143 rwlock_t i_es_lock; 1144 struct list_head i_es_list; 1145 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1146 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1147 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1148 extents to shrink. Protected by 1149 i_es_lock */ 1150 u64 i_es_seq; /* Change counter for extents. 1151 Protected by i_es_lock */ 1152 1153 /* ialloc */ 1154 ext4_group_t i_last_alloc_group; 1155 1156 /* pending cluster reservations for bigalloc file systems */ 1157 struct ext4_pending_tree i_pending_tree; 1158 1159 /* on-disk additional length */ 1160 __u16 i_extra_isize; 1161 1162 /* Indicate the inline data space. */ 1163 u16 i_inline_off; 1164 u16 i_inline_size; 1165 1166 #ifdef CONFIG_QUOTA 1167 /* quota space reservation, managed internally by quota code */ 1168 qsize_t i_reserved_quota; 1169 #endif 1170 spinlock_t i_block_reservation_lock; 1171 1172 /* Lock protecting lists below */ 1173 spinlock_t i_completed_io_lock; 1174 /* 1175 * Completed IOs that need unwritten extents handling and have 1176 * transaction reserved 1177 */ 1178 struct list_head i_rsv_conversion_list; 1179 struct work_struct i_rsv_conversion_work; 1180 1181 /* 1182 * Transactions that contain inode's metadata needed to complete 1183 * fsync and fdatasync, respectively. 1184 */ 1185 tid_t i_sync_tid; 1186 tid_t i_datasync_tid; 1187 1188 #ifdef CONFIG_QUOTA 1189 struct dquot __rcu *i_dquot[MAXQUOTAS]; 1190 #endif 1191 1192 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1193 __u32 i_csum_seed; 1194 1195 kprojid_t i_projid; 1196 1197 #ifdef CONFIG_FS_ENCRYPTION 1198 struct fscrypt_inode_info *i_crypt_info; 1199 #endif 1200 }; 1201 1202 /* 1203 * File system states 1204 */ 1205 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1206 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1207 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1208 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1209 1210 /* 1211 * Misc. filesystem flags 1212 */ 1213 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1214 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1215 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1216 1217 /* 1218 * Mount flags set via mount options or defaults 1219 */ 1220 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1221 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1222 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1223 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1224 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1225 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1226 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1227 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1228 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1229 #ifdef CONFIG_FS_DAX 1230 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1231 #else 1232 #define EXT4_MOUNT_DAX_ALWAYS 0 1233 #endif 1234 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1235 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1236 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1237 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1238 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1239 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1240 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1241 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1242 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1243 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1244 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1245 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1246 * enable enforcement for hidden 1247 * quota files */ 1248 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1249 * enforcement for hidden quota 1250 * files */ 1251 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1252 * enforcement */ 1253 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1254 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1255 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1256 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1257 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1258 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1259 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1260 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1261 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1262 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1263 1264 /* 1265 * Mount flags set either automatically (could not be set by mount option) 1266 * based on per file system feature or property or in special cases such as 1267 * distinguishing between explicit mount option definition and default. 1268 */ 1269 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1270 specified delalloc */ 1271 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1272 size of blocksize * 8 1273 blocks */ 1274 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1275 file systems */ 1276 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1277 specified journal checksum */ 1278 1279 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1280 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1281 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1282 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1283 * scanning in mballoc 1284 */ 1285 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1286 1287 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1288 ~EXT4_MOUNT_##opt 1289 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1290 EXT4_MOUNT_##opt 1291 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1292 EXT4_MOUNT_##opt) 1293 1294 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1295 ~EXT4_MOUNT2_##opt 1296 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1297 EXT4_MOUNT2_##opt 1298 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1299 EXT4_MOUNT2_##opt) 1300 1301 #define ext4_test_and_set_bit __test_and_set_bit_le 1302 #define ext4_set_bit __set_bit_le 1303 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1304 #define ext4_clear_bit __clear_bit_le 1305 #define ext4_test_bit test_bit_le 1306 #define ext4_find_next_zero_bit find_next_zero_bit_le 1307 #define ext4_find_next_bit find_next_bit_le 1308 1309 extern void mb_set_bits(void *bm, int cur, int len); 1310 1311 /* 1312 * Maximal mount counts between two filesystem checks 1313 */ 1314 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1315 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1316 1317 /* 1318 * Behaviour when detecting errors 1319 */ 1320 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1321 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1322 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1323 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1324 1325 /* Metadata checksum algorithm codes */ 1326 #define EXT4_CRC32C_CHKSUM 1 1327 1328 #define EXT4_LABEL_MAX 16 1329 1330 /* 1331 * Structure of the super block 1332 */ 1333 struct ext4_super_block { 1334 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1335 __le32 s_blocks_count_lo; /* Blocks count */ 1336 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1337 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1338 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1339 __le32 s_first_data_block; /* First Data Block */ 1340 __le32 s_log_block_size; /* Block size */ 1341 __le32 s_log_cluster_size; /* Allocation cluster size */ 1342 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1343 __le32 s_clusters_per_group; /* # Clusters per group */ 1344 __le32 s_inodes_per_group; /* # Inodes per group */ 1345 __le32 s_mtime; /* Mount time */ 1346 /*30*/ __le32 s_wtime; /* Write time */ 1347 __le16 s_mnt_count; /* Mount count */ 1348 __le16 s_max_mnt_count; /* Maximal mount count */ 1349 __le16 s_magic; /* Magic signature */ 1350 __le16 s_state; /* File system state */ 1351 __le16 s_errors; /* Behaviour when detecting errors */ 1352 __le16 s_minor_rev_level; /* minor revision level */ 1353 /*40*/ __le32 s_lastcheck; /* time of last check */ 1354 __le32 s_checkinterval; /* max. time between checks */ 1355 __le32 s_creator_os; /* OS */ 1356 __le32 s_rev_level; /* Revision level */ 1357 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1358 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1359 /* 1360 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1361 * 1362 * Note: the difference between the compatible feature set and 1363 * the incompatible feature set is that if there is a bit set 1364 * in the incompatible feature set that the kernel doesn't 1365 * know about, it should refuse to mount the filesystem. 1366 * 1367 * e2fsck's requirements are more strict; if it doesn't know 1368 * about a feature in either the compatible or incompatible 1369 * feature set, it must abort and not try to meddle with 1370 * things it doesn't understand... 1371 */ 1372 __le32 s_first_ino; /* First non-reserved inode */ 1373 __le16 s_inode_size; /* size of inode structure */ 1374 __le16 s_block_group_nr; /* block group # of this superblock */ 1375 __le32 s_feature_compat; /* compatible feature set */ 1376 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1377 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1378 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1379 /*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */ 1380 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1381 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1382 /* 1383 * Performance hints. Directory preallocation should only 1384 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1385 */ 1386 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1387 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1388 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1389 /* 1390 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1391 */ 1392 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1393 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1394 __le32 s_journal_dev; /* device number of journal file */ 1395 __le32 s_last_orphan; /* start of list of inodes to delete */ 1396 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1397 __u8 s_def_hash_version; /* Default hash version to use */ 1398 __u8 s_jnl_backup_type; 1399 __le16 s_desc_size; /* size of group descriptor */ 1400 /*100*/ __le32 s_default_mount_opts; 1401 __le32 s_first_meta_bg; /* First metablock block group */ 1402 __le32 s_mkfs_time; /* When the filesystem was created */ 1403 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1404 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1405 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1406 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1407 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1408 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1409 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1410 __le32 s_flags; /* Miscellaneous flags */ 1411 __le16 s_raid_stride; /* RAID stride */ 1412 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1413 __le64 s_mmp_block; /* Block for multi-mount protection */ 1414 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1415 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1416 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1417 __u8 s_encryption_level; /* versioning level for encryption */ 1418 __u8 s_reserved_pad; /* Padding to next 32bits */ 1419 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1420 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1421 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1422 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1423 snapshot's future use */ 1424 __le32 s_snapshot_list; /* inode number of the head of the 1425 on-disk snapshot list */ 1426 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1427 __le32 s_error_count; /* number of fs errors */ 1428 __le32 s_first_error_time; /* first time an error happened */ 1429 __le32 s_first_error_ino; /* inode involved in first error */ 1430 __le64 s_first_error_block; /* block involved of first error */ 1431 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1432 __le32 s_first_error_line; /* line number where error happened */ 1433 __le32 s_last_error_time; /* most recent time of an error */ 1434 __le32 s_last_error_ino; /* inode involved in last error */ 1435 __le32 s_last_error_line; /* line number where error happened */ 1436 __le64 s_last_error_block; /* block involved of last error */ 1437 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1438 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1439 __u8 s_mount_opts[64]; 1440 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1441 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1442 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1443 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1444 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1445 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1446 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1447 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1448 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1449 __u8 s_wtime_hi; 1450 __u8 s_mtime_hi; 1451 __u8 s_mkfs_time_hi; 1452 __u8 s_lastcheck_hi; 1453 __u8 s_first_error_time_hi; 1454 __u8 s_last_error_time_hi; 1455 __u8 s_first_error_errcode; 1456 __u8 s_last_error_errcode; 1457 __le16 s_encoding; /* Filename charset encoding */ 1458 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1459 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1460 __le16 s_def_resuid_hi; 1461 __le16 s_def_resgid_hi; 1462 __le32 s_reserved[93]; /* Padding to the end of the block */ 1463 __le32 s_checksum; /* crc32c(superblock) */ 1464 }; 1465 1466 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1467 1468 #ifdef __KERNEL__ 1469 1470 /* Number of quota types we support */ 1471 #define EXT4_MAXQUOTAS 3 1472 1473 #define EXT4_ENC_UTF8_12_1 1 1474 1475 /* Types of ext4 journal triggers */ 1476 enum ext4_journal_trigger_type { 1477 EXT4_JTR_ORPHAN_FILE, 1478 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1479 }; 1480 1481 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1482 1483 struct ext4_journal_trigger { 1484 struct jbd2_buffer_trigger_type tr_triggers; 1485 struct super_block *sb; 1486 }; 1487 1488 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1489 struct jbd2_buffer_trigger_type *trigger) 1490 { 1491 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1492 } 1493 1494 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1495 1496 /* Structure at the tail of orphan block */ 1497 struct ext4_orphan_block_tail { 1498 __le32 ob_magic; 1499 __le32 ob_checksum; 1500 }; 1501 1502 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1503 { 1504 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1505 sizeof(u32); 1506 } 1507 1508 struct ext4_orphan_block { 1509 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1510 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1511 }; 1512 1513 /* 1514 * Info about orphan file. 1515 */ 1516 struct ext4_orphan_info { 1517 int of_blocks; /* Number of orphan blocks in a file */ 1518 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1519 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1520 * file blocks */ 1521 }; 1522 1523 /* 1524 * fourth extended-fs super-block data in memory 1525 */ 1526 struct ext4_sb_info { 1527 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1528 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1529 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1530 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1531 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1532 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1533 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1534 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1535 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1536 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1537 unsigned long s_overhead; /* # of fs overhead clusters */ 1538 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1539 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1540 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1541 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1542 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1543 /* Array of bh's for the block group descriptors */ 1544 struct buffer_head * __rcu *s_group_desc; 1545 unsigned int s_mount_opt; 1546 unsigned int s_mount_opt2; 1547 unsigned long s_mount_flags; 1548 unsigned int s_def_mount_opt; 1549 unsigned int s_def_mount_opt2; 1550 ext4_fsblk_t s_sb_block; 1551 atomic64_t s_resv_clusters; 1552 kuid_t s_resuid; 1553 kgid_t s_resgid; 1554 unsigned short s_mount_state; 1555 unsigned short s_pad; 1556 int s_addr_per_block_bits; 1557 int s_desc_per_block_bits; 1558 int s_inode_size; 1559 int s_first_ino; 1560 unsigned int s_inode_readahead_blks; 1561 unsigned int s_inode_goal; 1562 u32 s_hash_seed[4]; 1563 int s_def_hash_version; 1564 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1565 struct percpu_counter s_freeclusters_counter; 1566 struct percpu_counter s_freeinodes_counter; 1567 struct percpu_counter s_dirs_counter; 1568 struct percpu_counter s_dirtyclusters_counter; 1569 struct percpu_counter s_sra_exceeded_retry_limit; 1570 struct blockgroup_lock *s_blockgroup_lock; 1571 struct proc_dir_entry *s_proc; 1572 struct kobject s_kobj; 1573 struct completion s_kobj_unregister; 1574 struct mutex s_error_notify_mutex; /* protects sysfs_notify vs kobject_del */ 1575 struct super_block *s_sb; 1576 struct buffer_head *s_mmp_bh; 1577 1578 /* Journaling */ 1579 struct journal_s *s_journal; 1580 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1581 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1582 struct list_head s_orphan; /* List of orphaned inodes in on disk 1583 list */ 1584 struct ext4_orphan_info s_orphan_info; 1585 unsigned long s_commit_interval; 1586 u32 s_max_batch_time; 1587 u32 s_min_batch_time; 1588 struct file *s_journal_bdev_file; 1589 #ifdef CONFIG_QUOTA 1590 /* Names of quota files with journalled quota */ 1591 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1592 int s_jquota_fmt; /* Format of quota to use */ 1593 #endif 1594 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1595 struct ext4_system_blocks __rcu *s_system_blks; 1596 1597 #ifdef EXTENTS_STATS 1598 /* ext4 extents stats */ 1599 unsigned long s_ext_min; 1600 unsigned long s_ext_max; 1601 unsigned long s_depth_max; 1602 spinlock_t s_ext_stats_lock; 1603 unsigned long s_ext_blocks; 1604 unsigned long s_ext_extents; 1605 #endif 1606 1607 /* for buddy allocator */ 1608 struct ext4_group_info ** __rcu *s_group_info; 1609 struct inode *s_buddy_cache; 1610 spinlock_t s_md_lock; 1611 unsigned short *s_mb_offsets; 1612 unsigned int *s_mb_maxs; 1613 unsigned int s_group_info_size; 1614 atomic_t s_mb_free_pending; 1615 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1616 after commit completed */ 1617 struct list_head s_discard_list; 1618 struct work_struct s_discard_work; 1619 atomic_t s_retry_alloc_pending; 1620 struct xarray *s_mb_avg_fragment_size; 1621 struct xarray *s_mb_largest_free_orders; 1622 1623 /* tunables */ 1624 unsigned long s_stripe; 1625 unsigned int s_mb_max_linear_groups; 1626 unsigned int s_mb_stream_request; 1627 unsigned int s_mb_max_to_scan; 1628 unsigned int s_mb_min_to_scan; 1629 unsigned int s_mb_stats; 1630 unsigned int s_mb_order2_reqs; 1631 unsigned int s_mb_group_prealloc; 1632 unsigned int s_max_dir_size_kb; 1633 unsigned int s_mb_prefetch; 1634 unsigned int s_mb_prefetch_limit; 1635 unsigned int s_mb_best_avail_max_trim_order; 1636 unsigned int s_sb_update_sec; 1637 unsigned int s_sb_update_kb; 1638 1639 /* where last allocation was done - for stream allocation */ 1640 ext4_group_t *s_mb_last_groups; 1641 unsigned int s_mb_nr_global_goals; 1642 1643 /* stats for buddy allocator */ 1644 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1645 atomic_t s_bal_success; /* we found long enough chunks */ 1646 atomic_t s_bal_allocated; /* in blocks */ 1647 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1648 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1649 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1650 atomic_t s_bal_goals; /* goal hits */ 1651 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */ 1652 atomic_t s_bal_len_goals; /* len goal hits */ 1653 atomic_t s_bal_breaks; /* too long searches */ 1654 atomic_t s_bal_2orders; /* 2^order hits */ 1655 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1656 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1657 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1658 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1659 atomic64_t s_mb_generation_time; 1660 atomic_t s_mb_lost_chunks; 1661 atomic_t s_mb_preallocated; 1662 atomic_t s_mb_discarded; 1663 atomic_t s_lock_busy; 1664 1665 /* locality groups */ 1666 struct ext4_locality_group __percpu *s_locality_groups; 1667 1668 /* for write statistics */ 1669 unsigned long s_sectors_written_start; 1670 u64 s_kbytes_written; 1671 1672 /* the size of zero-out chunk */ 1673 unsigned int s_extent_max_zeroout_kb; 1674 1675 unsigned int s_log_groups_per_flex; 1676 struct flex_groups * __rcu *s_flex_groups; 1677 ext4_group_t s_flex_groups_allocated; 1678 1679 /* workqueue for reserved extent conversions (buffered io) */ 1680 struct workqueue_struct *rsv_conversion_wq; 1681 1682 /* timer for periodic error stats printing */ 1683 struct timer_list s_err_report; 1684 /* timeout in seconds for s_err_report; 0 disables the timer. */ 1685 unsigned long s_err_report_sec; 1686 1687 /* Lazy inode table initialization info */ 1688 struct ext4_li_request *s_li_request; 1689 /* Wait multiplier for lazy initialization thread */ 1690 unsigned int s_li_wait_mult; 1691 1692 /* Kernel thread for multiple mount protection */ 1693 struct task_struct *s_mmp_tsk; 1694 1695 /* record the last minlen when FITRIM is called. */ 1696 unsigned long s_last_trim_minblks; 1697 1698 /* minimum folio order of a page cache allocation */ 1699 u16 s_min_folio_order; 1700 /* supported maximum folio order, 0 means not supported */ 1701 u16 s_max_folio_order; 1702 1703 /* Precomputed FS UUID checksum for seeding other checksums */ 1704 __u32 s_csum_seed; 1705 1706 /* Reclaim extents from extent status tree */ 1707 struct shrinker *s_es_shrinker; 1708 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1709 long s_es_nr_inode; 1710 struct ext4_es_stats s_es_stats; 1711 struct mb_cache *s_ea_block_cache; 1712 struct mb_cache *s_ea_inode_cache; 1713 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1714 1715 /* Journal triggers for checksum computation */ 1716 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1717 1718 /* Ratelimit ext4 messages. */ 1719 struct ratelimit_state s_err_ratelimit_state; 1720 struct ratelimit_state s_warning_ratelimit_state; 1721 struct ratelimit_state s_msg_ratelimit_state; 1722 atomic_t s_warning_count; 1723 atomic_t s_msg_count; 1724 1725 /* Encryption policy for '-o test_dummy_encryption' */ 1726 struct fscrypt_dummy_policy s_dummy_enc_policy; 1727 1728 /* 1729 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1730 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1731 * DIOREAD_NOLOCK mount options on remount. 1732 */ 1733 struct percpu_rw_semaphore s_writepages_rwsem; 1734 struct dax_device *s_daxdev; 1735 u64 s_dax_part_off; 1736 #ifdef CONFIG_EXT4_DEBUG 1737 unsigned long s_simulate_fail; 1738 #endif 1739 /* Record the errseq of the backing block device */ 1740 errseq_t s_bdev_wb_err; 1741 spinlock_t s_bdev_wb_lock; 1742 1743 /* Information about errors that happened during this mount */ 1744 spinlock_t s_error_lock; 1745 int s_add_error_count; 1746 int s_first_error_code; 1747 __u32 s_first_error_line; 1748 __u32 s_first_error_ino; 1749 __u64 s_first_error_block; 1750 const char *s_first_error_func; 1751 time64_t s_first_error_time; 1752 int s_last_error_code; 1753 __u32 s_last_error_line; 1754 __u32 s_last_error_ino; 1755 __u64 s_last_error_block; 1756 const char *s_last_error_func; 1757 time64_t s_last_error_time; 1758 /* 1759 * If we are in a context where we cannot update the on-disk 1760 * superblock, we queue the work here. This is used to update 1761 * the error information in the superblock, and for periodic 1762 * updates of the superblock called from the commit callback 1763 * function. 1764 */ 1765 struct work_struct s_sb_upd_work; 1766 1767 /* Atomic write unit values in bytes */ 1768 unsigned int s_awu_min; 1769 unsigned int s_awu_max; 1770 1771 /* Ext4 fast commit sub transaction ID */ 1772 atomic_t s_fc_subtid; 1773 1774 /* 1775 * After commit starts, the main queue gets locked, and the further 1776 * updates get added in the staging queue. 1777 */ 1778 #define FC_Q_MAIN 0 1779 #define FC_Q_STAGING 1 1780 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1781 * that have data changes in them. 1782 */ 1783 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1784 unsigned int s_fc_bytes; 1785 /* 1786 * Main fast commit lock. This lock protects accesses to the 1787 * following fields: 1788 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1789 * 1790 * s_fc_lock can be taken from reclaim context (inode eviction) and is 1791 * thus reclaim unsafe. Use ext4_fc_lock()/ext4_fc_unlock() helpers 1792 * when acquiring / releasing the lock. 1793 */ 1794 struct mutex s_fc_lock; 1795 struct buffer_head *s_fc_bh; 1796 struct ext4_fc_stats s_fc_stats; 1797 tid_t s_fc_ineligible_tid; 1798 #ifdef CONFIG_EXT4_DEBUG 1799 int s_fc_debug_max_replay; 1800 #endif 1801 struct ext4_fc_replay_state s_fc_replay_state; 1802 }; 1803 1804 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1805 { 1806 return sb->s_fs_info; 1807 } 1808 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1809 { 1810 return container_of(inode, struct ext4_inode_info, vfs_inode); 1811 } 1812 1813 static inline int ext4_writepages_down_read(struct super_block *sb) 1814 { 1815 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1816 return memalloc_nofs_save(); 1817 } 1818 1819 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1820 { 1821 memalloc_nofs_restore(ctx); 1822 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1823 } 1824 1825 static inline int ext4_writepages_down_write(struct super_block *sb) 1826 { 1827 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1828 return memalloc_nofs_save(); 1829 } 1830 1831 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1832 { 1833 memalloc_nofs_restore(ctx); 1834 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1835 } 1836 1837 static inline int ext4_fc_lock(struct super_block *sb) 1838 { 1839 mutex_lock(&EXT4_SB(sb)->s_fc_lock); 1840 return memalloc_nofs_save(); 1841 } 1842 1843 static inline void ext4_fc_unlock(struct super_block *sb, int ctx) 1844 { 1845 memalloc_nofs_restore(ctx); 1846 mutex_unlock(&EXT4_SB(sb)->s_fc_lock); 1847 } 1848 1849 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1850 { 1851 return ino == EXT4_ROOT_INO || 1852 (ino >= EXT4_FIRST_INO(sb) && 1853 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1854 } 1855 1856 static inline int ext4_get_resuid(struct ext4_super_block *es) 1857 { 1858 return le16_to_cpu(es->s_def_resuid) | 1859 le16_to_cpu(es->s_def_resuid_hi) << 16; 1860 } 1861 1862 static inline int ext4_get_resgid(struct ext4_super_block *es) 1863 { 1864 return le16_to_cpu(es->s_def_resgid) | 1865 le16_to_cpu(es->s_def_resgid_hi) << 16; 1866 } 1867 1868 /* 1869 * Returns: sbi->field[index] 1870 * Used to access an array element from the following sbi fields which require 1871 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1872 * - s_group_desc 1873 * - s_group_info 1874 * - s_flex_group 1875 */ 1876 #define sbi_array_rcu_deref(sbi, field, index) \ 1877 ({ \ 1878 typeof(*((sbi)->field)) _v; \ 1879 rcu_read_lock(); \ 1880 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1881 rcu_read_unlock(); \ 1882 _v; \ 1883 }) 1884 1885 /* 1886 * run-time mount flags 1887 */ 1888 enum { 1889 EXT4_MF_MNTDIR_SAMPLED, 1890 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1891 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */ 1892 }; 1893 1894 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1895 { 1896 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1897 } 1898 1899 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1900 { 1901 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1902 } 1903 1904 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1905 { 1906 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1907 } 1908 1909 1910 /* 1911 * Simulate_fail codes 1912 */ 1913 #define EXT4_SIM_BBITMAP_EIO 1 1914 #define EXT4_SIM_BBITMAP_CRC 2 1915 #define EXT4_SIM_IBITMAP_EIO 3 1916 #define EXT4_SIM_IBITMAP_CRC 4 1917 #define EXT4_SIM_INODE_EIO 5 1918 #define EXT4_SIM_INODE_CRC 6 1919 #define EXT4_SIM_DIRBLOCK_EIO 7 1920 #define EXT4_SIM_DIRBLOCK_CRC 8 1921 1922 static inline bool ext4_simulate_fail(struct super_block *sb, 1923 unsigned long code) 1924 { 1925 #ifdef CONFIG_EXT4_DEBUG 1926 struct ext4_sb_info *sbi = EXT4_SB(sb); 1927 1928 if (unlikely(sbi->s_simulate_fail == code)) { 1929 sbi->s_simulate_fail = 0; 1930 return true; 1931 } 1932 #endif 1933 return false; 1934 } 1935 1936 /* 1937 * Error number codes for s_{first,last}_error_errno 1938 * 1939 * Linux errno numbers are architecture specific, so we need to translate 1940 * them into something which is architecture independent. We don't define 1941 * codes for all errno's; just the ones which are most likely to be the cause 1942 * of an ext4_error() call. 1943 */ 1944 #define EXT4_ERR_UNKNOWN 1 1945 #define EXT4_ERR_EIO 2 1946 #define EXT4_ERR_ENOMEM 3 1947 #define EXT4_ERR_EFSBADCRC 4 1948 #define EXT4_ERR_EFSCORRUPTED 5 1949 #define EXT4_ERR_ENOSPC 6 1950 #define EXT4_ERR_ENOKEY 7 1951 #define EXT4_ERR_EROFS 8 1952 #define EXT4_ERR_EFBIG 9 1953 #define EXT4_ERR_EEXIST 10 1954 #define EXT4_ERR_ERANGE 11 1955 #define EXT4_ERR_EOVERFLOW 12 1956 #define EXT4_ERR_EBUSY 13 1957 #define EXT4_ERR_ENOTDIR 14 1958 #define EXT4_ERR_ENOTEMPTY 15 1959 #define EXT4_ERR_ESHUTDOWN 16 1960 #define EXT4_ERR_EFAULT 17 1961 1962 /* 1963 * Inode dynamic state flags 1964 */ 1965 enum { 1966 EXT4_STATE_NEW, /* inode is newly created */ 1967 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1968 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1969 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1970 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1971 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1972 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1973 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1974 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1975 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1976 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1977 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */ 1978 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1979 }; 1980 1981 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1982 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1983 { \ 1984 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1985 } \ 1986 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1987 { \ 1988 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1989 } \ 1990 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1991 { \ 1992 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1993 } 1994 1995 /* Add these declarations here only so that these functions can be 1996 * found by name. Otherwise, they are very hard to locate. */ 1997 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1998 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1999 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 2000 EXT4_INODE_BIT_FNS(flag, flags, 0) 2001 2002 /* Add these declarations here only so that these functions can be 2003 * found by name. Otherwise, they are very hard to locate. */ 2004 static inline int ext4_test_inode_state(struct inode *inode, int bit); 2005 static inline void ext4_set_inode_state(struct inode *inode, int bit); 2006 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 2007 #if (BITS_PER_LONG < 64) 2008 EXT4_INODE_BIT_FNS(state, state_flags, 0) 2009 2010 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 2011 { 2012 (ei)->i_state_flags = 0; 2013 } 2014 #else 2015 EXT4_INODE_BIT_FNS(state, flags, 32) 2016 2017 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 2018 { 2019 /* We depend on the fact that callers will set i_flags */ 2020 } 2021 #endif 2022 #else 2023 /* Assume that user mode programs are passing in an ext4fs superblock, not 2024 * a kernel struct super_block. This will allow us to call the feature-test 2025 * macros from user land. */ 2026 #define EXT4_SB(sb) (sb) 2027 #endif 2028 2029 static inline bool ext4_verity_in_progress(struct inode *inode) 2030 { 2031 return IS_ENABLED(CONFIG_FS_VERITY) && 2032 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 2033 } 2034 2035 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 2036 2037 /* 2038 * Check whether the inode is tracked as orphan (either in orphan file or 2039 * orphan list). 2040 */ 2041 static inline bool ext4_inode_orphan_tracked(struct inode *inode) 2042 { 2043 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) || 2044 !list_empty(&EXT4_I(inode)->i_orphan); 2045 } 2046 2047 /* 2048 * Codes for operating systems 2049 */ 2050 #define EXT4_OS_LINUX 0 2051 #define EXT4_OS_HURD 1 2052 #define EXT4_OS_MASIX 2 2053 #define EXT4_OS_FREEBSD 3 2054 #define EXT4_OS_LITES 4 2055 2056 /* 2057 * Revision levels 2058 */ 2059 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 2060 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 2061 2062 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 2063 2064 #define EXT4_GOOD_OLD_INODE_SIZE 128 2065 2066 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 2067 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 2068 #define EXT4_TIMESTAMP_MIN S32_MIN 2069 2070 /* 2071 * Feature set definitions 2072 */ 2073 2074 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 2075 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2076 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2077 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2078 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2079 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2080 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2081 /* 2082 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2083 * incompatible only if fast commit blocks are present in the FS. Since we 2084 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2085 * incompatible. We also have a JBD2 incompat feature, which gets set when 2086 * there are fast commit blocks present in the journal. 2087 */ 2088 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2089 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2090 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2091 2092 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2093 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2094 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2095 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2096 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2097 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2098 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2099 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2100 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2101 /* 2102 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2103 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2104 * all other data structures' checksums. However, the METADATA_CSUM and 2105 * GDT_CSUM bits are mutually exclusive. 2106 */ 2107 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2108 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2109 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2110 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2111 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2112 non-empty */ 2113 2114 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2115 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2116 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2117 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2118 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2119 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2120 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2121 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2122 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2123 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2124 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2125 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2126 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2127 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2128 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2129 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2130 2131 extern void ext4_update_dynamic_rev(struct super_block *sb); 2132 2133 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2134 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2135 { \ 2136 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2137 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2138 } \ 2139 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2140 { \ 2141 ext4_update_dynamic_rev(sb); \ 2142 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2143 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2144 } \ 2145 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2146 { \ 2147 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2148 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2149 } 2150 2151 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2152 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2153 { \ 2154 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2155 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2156 } \ 2157 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2158 { \ 2159 ext4_update_dynamic_rev(sb); \ 2160 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2161 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2162 } \ 2163 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2164 { \ 2165 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2166 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2167 } 2168 2169 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2170 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2171 { \ 2172 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2173 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2174 } \ 2175 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2176 { \ 2177 ext4_update_dynamic_rev(sb); \ 2178 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2179 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2180 } \ 2181 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2182 { \ 2183 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2184 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2185 } 2186 2187 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2188 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2189 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2190 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2191 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2192 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2193 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2194 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2195 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2196 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2197 2198 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2199 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2200 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2201 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2202 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2203 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2204 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2205 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2206 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2207 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2208 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2209 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2210 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2211 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2212 2213 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2214 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2215 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2216 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2217 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2218 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2219 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2220 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2221 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2222 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2223 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2224 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2225 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2226 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2227 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2228 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2229 2230 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2231 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2232 EXT4_FEATURE_INCOMPAT_META_BG) 2233 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2234 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2235 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2236 2237 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2238 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2239 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2240 EXT4_FEATURE_INCOMPAT_META_BG) 2241 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2242 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2243 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2244 2245 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2246 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2247 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2248 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2249 EXT4_FEATURE_INCOMPAT_META_BG| \ 2250 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2251 EXT4_FEATURE_INCOMPAT_64BIT| \ 2252 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2253 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2254 EXT4_FEATURE_INCOMPAT_MMP | \ 2255 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2256 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2257 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2258 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2259 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2260 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2261 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2262 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2263 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2264 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2265 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2266 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2267 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2268 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2269 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2270 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2271 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2272 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2273 2274 #define EXTN_FEATURE_FUNCS(ver) \ 2275 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2276 { \ 2277 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2278 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2279 } \ 2280 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2281 { \ 2282 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2283 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2284 } \ 2285 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2286 { \ 2287 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2288 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2289 } 2290 2291 EXTN_FEATURE_FUNCS(2) 2292 EXTN_FEATURE_FUNCS(3) 2293 EXTN_FEATURE_FUNCS(4) 2294 2295 static inline bool ext4_has_compat_features(struct super_block *sb) 2296 { 2297 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2298 } 2299 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2300 { 2301 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2302 } 2303 static inline bool ext4_has_incompat_features(struct super_block *sb) 2304 { 2305 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2306 } 2307 2308 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2309 2310 /* 2311 * Superblock flags 2312 */ 2313 enum { 2314 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */ 2315 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */ 2316 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */ 2317 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */ 2318 }; 2319 2320 static inline int ext4_forced_shutdown(struct super_block *sb) 2321 { 2322 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2323 } 2324 2325 static inline int ext4_emergency_ro(struct super_block *sb) 2326 { 2327 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags); 2328 } 2329 2330 static inline int ext4_emergency_state(struct super_block *sb) 2331 { 2332 if (unlikely(ext4_forced_shutdown(sb))) 2333 return -EIO; 2334 if (unlikely(ext4_emergency_ro(sb))) 2335 return -EROFS; 2336 return 0; 2337 } 2338 2339 /* 2340 * Default values for user and/or group using reserved blocks 2341 */ 2342 #define EXT4_DEF_RESUID 0 2343 #define EXT4_DEF_RESGID 0 2344 2345 /* 2346 * Default project ID 2347 */ 2348 #define EXT4_DEF_PROJID 0 2349 2350 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2351 2352 /* 2353 * Default mount options 2354 */ 2355 #define EXT4_DEFM_DEBUG 0x0001 2356 #define EXT4_DEFM_BSDGROUPS 0x0002 2357 #define EXT4_DEFM_XATTR_USER 0x0004 2358 #define EXT4_DEFM_ACL 0x0008 2359 #define EXT4_DEFM_UID16 0x0010 2360 #define EXT4_DEFM_JMODE 0x0060 2361 #define EXT4_DEFM_JMODE_DATA 0x0020 2362 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2363 #define EXT4_DEFM_JMODE_WBACK 0x0060 2364 #define EXT4_DEFM_NOBARRIER 0x0100 2365 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2366 #define EXT4_DEFM_DISCARD 0x0400 2367 #define EXT4_DEFM_NODELALLOC 0x0800 2368 2369 /* 2370 * Default journal batch times and ioprio. 2371 */ 2372 #define EXT4_DEF_MIN_BATCH_TIME 0 2373 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2374 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) 2375 2376 2377 /* 2378 * Default values for superblock update 2379 */ 2380 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */ 2381 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */ 2382 2383 /* 2384 * Minimum number of groups in a flexgroup before we separate out 2385 * directories into the first block group of a flexgroup 2386 */ 2387 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2388 2389 /* 2390 * Structure of a directory entry 2391 */ 2392 #define EXT4_NAME_LEN 255 2393 /* 2394 * Base length of the ext4 directory entry excluding the name length 2395 */ 2396 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2397 2398 struct ext4_dir_entry { 2399 __le32 inode; /* Inode number */ 2400 __le16 rec_len; /* Directory entry length */ 2401 __le16 name_len; /* Name length */ 2402 char name[EXT4_NAME_LEN]; /* File name */ 2403 }; 2404 2405 2406 /* 2407 * Encrypted Casefolded entries require saving the hash on disk. This structure 2408 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2409 * boundary. 2410 */ 2411 struct ext4_dir_entry_hash { 2412 __le32 hash; 2413 __le32 minor_hash; 2414 }; 2415 2416 /* 2417 * The new version of the directory entry. Since EXT4 structures are 2418 * stored in intel byte order, and the name_len field could never be 2419 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2420 * file_type field. 2421 */ 2422 struct ext4_dir_entry_2 { 2423 __le32 inode; /* Inode number */ 2424 __le16 rec_len; /* Directory entry length */ 2425 __u8 name_len; /* Name length */ 2426 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2427 char name[EXT4_NAME_LEN]; /* File name */ 2428 }; 2429 2430 /* 2431 * Access the hashes at the end of ext4_dir_entry_2 2432 */ 2433 #define EXT4_DIRENT_HASHES(entry) \ 2434 ((struct ext4_dir_entry_hash *) \ 2435 (((void *)(entry)) + \ 2436 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2437 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash) 2438 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2439 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash) 2440 2441 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2442 { 2443 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2444 } 2445 2446 /* 2447 * This is a bogus directory entry at the end of each leaf block that 2448 * records checksums. 2449 */ 2450 struct ext4_dir_entry_tail { 2451 __le32 det_reserved_zero1; /* Pretend to be unused */ 2452 __le16 det_rec_len; /* 12 */ 2453 __u8 det_reserved_zero2; /* Zero name length */ 2454 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2455 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2456 }; 2457 2458 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2459 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2460 ((blocksize) - \ 2461 sizeof(struct ext4_dir_entry_tail)))) 2462 2463 /* 2464 * Ext4 directory file types. Only the low 3 bits are used. The 2465 * other bits are reserved for now. 2466 */ 2467 #define EXT4_FT_UNKNOWN 0 2468 #define EXT4_FT_REG_FILE 1 2469 #define EXT4_FT_DIR 2 2470 #define EXT4_FT_CHRDEV 3 2471 #define EXT4_FT_BLKDEV 4 2472 #define EXT4_FT_FIFO 5 2473 #define EXT4_FT_SOCK 6 2474 #define EXT4_FT_SYMLINK 7 2475 2476 #define EXT4_FT_MAX 8 2477 2478 #define EXT4_FT_DIR_CSUM 0xDE 2479 2480 /* 2481 * EXT4_DIR_PAD defines the directory entries boundaries 2482 * 2483 * NOTE: It must be a multiple of 4 2484 */ 2485 #define EXT4_DIR_PAD 4 2486 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2487 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2488 2489 /* 2490 * The rec_len is dependent on the type of directory. Directories that are 2491 * casefolded and encrypted need to store the hash as well, so we add room for 2492 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2493 * pass NULL for dir, as those entries do not use the extra fields. 2494 */ 2495 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2496 const struct inode *dir) 2497 { 2498 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2499 2500 if (dir && ext4_hash_in_dirent(dir)) 2501 rec_len += sizeof(struct ext4_dir_entry_hash); 2502 return (rec_len & ~EXT4_DIR_ROUND); 2503 } 2504 2505 static inline unsigned int 2506 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2507 { 2508 unsigned len = le16_to_cpu(dlen); 2509 2510 if (len == EXT4_MAX_REC_LEN || len == 0) 2511 return blocksize; 2512 return (len & 65532) | ((len & 3) << 16); 2513 } 2514 2515 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2516 { 2517 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2518 if (len < 65536) 2519 return cpu_to_le16(len); 2520 if (len == blocksize) { 2521 if (blocksize == 65536) 2522 return cpu_to_le16(EXT4_MAX_REC_LEN); 2523 else 2524 return cpu_to_le16(0); 2525 } 2526 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2527 } 2528 2529 /* 2530 * Hash Tree Directory indexing 2531 * (c) Daniel Phillips, 2001 2532 */ 2533 2534 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2535 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2536 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2537 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2538 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2539 2540 /* Legal values for the dx_root hash_version field: */ 2541 2542 #define DX_HASH_LEGACY 0 2543 #define DX_HASH_HALF_MD4 1 2544 #define DX_HASH_TEA 2 2545 #define DX_HASH_LEGACY_UNSIGNED 3 2546 #define DX_HASH_HALF_MD4_UNSIGNED 4 2547 #define DX_HASH_TEA_UNSIGNED 5 2548 #define DX_HASH_SIPHASH 6 2549 #define DX_HASH_LAST DX_HASH_SIPHASH 2550 2551 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length) 2552 { 2553 return crc32c(crc, address, length); 2554 } 2555 2556 #ifdef __KERNEL__ 2557 2558 /* hash info structure used by the directory hash */ 2559 struct dx_hash_info 2560 { 2561 u32 hash; 2562 u32 minor_hash; 2563 int hash_version; 2564 u32 *seed; 2565 }; 2566 2567 2568 /* 32 and 64 bit signed EOF for dx directories */ 2569 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2570 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2571 2572 2573 /* 2574 * Control parameters used by ext4_htree_next_block 2575 */ 2576 #define HASH_NB_ALWAYS 1 2577 2578 struct ext4_filename { 2579 const struct qstr *usr_fname; 2580 struct fscrypt_str disk_name; 2581 struct dx_hash_info hinfo; 2582 #ifdef CONFIG_FS_ENCRYPTION 2583 struct fscrypt_str crypto_buf; 2584 #endif 2585 #if IS_ENABLED(CONFIG_UNICODE) 2586 struct qstr cf_name; 2587 #endif 2588 }; 2589 2590 #define fname_name(p) ((p)->disk_name.name) 2591 #define fname_usr_name(p) ((p)->usr_fname->name) 2592 #define fname_len(p) ((p)->disk_name.len) 2593 2594 /* 2595 * Describe an inode's exact location on disk and in memory 2596 */ 2597 struct ext4_iloc 2598 { 2599 struct buffer_head *bh; 2600 unsigned long offset; 2601 ext4_group_t block_group; 2602 }; 2603 2604 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2605 { 2606 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2607 } 2608 2609 static inline bool ext4_is_quota_file(struct inode *inode) 2610 { 2611 return IS_NOQUOTA(inode) && 2612 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2613 } 2614 2615 /* 2616 * This structure is stuffed into the struct file's private_data field 2617 * for directories. It is where we put information so that we can do 2618 * readdir operations in hash tree order. 2619 */ 2620 struct dir_private_info { 2621 struct rb_root root; 2622 struct rb_node *curr_node; 2623 struct fname *extra_fname; 2624 loff_t last_pos; 2625 __u32 curr_hash; 2626 __u32 curr_minor_hash; 2627 __u32 next_hash; 2628 u64 cookie; 2629 bool initialized; 2630 }; 2631 2632 /* calculate the first block number of the group */ 2633 static inline ext4_fsblk_t 2634 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2635 { 2636 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2637 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2638 } 2639 2640 /* 2641 * Special error return code only used by dx_probe() and its callers. 2642 */ 2643 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2644 2645 /* htree levels for ext4 */ 2646 #define EXT4_HTREE_LEVEL_COMPAT 2 2647 #define EXT4_HTREE_LEVEL 3 2648 2649 static inline int ext4_dir_htree_level(struct super_block *sb) 2650 { 2651 return ext4_has_feature_largedir(sb) ? 2652 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2653 } 2654 2655 /* 2656 * Timeout and state flag for lazy initialization inode thread. 2657 */ 2658 #define EXT4_DEF_LI_WAIT_MULT 10 2659 #define EXT4_DEF_LI_MAX_START_DELAY 5 2660 #define EXT4_LAZYINIT_QUIT 0x0001 2661 #define EXT4_LAZYINIT_RUNNING 0x0002 2662 2663 /* 2664 * Lazy inode table initialization info 2665 */ 2666 struct ext4_lazy_init { 2667 unsigned long li_state; 2668 struct list_head li_request_list; 2669 struct mutex li_list_mtx; 2670 }; 2671 2672 enum ext4_li_mode { 2673 EXT4_LI_MODE_PREFETCH_BBITMAP, 2674 EXT4_LI_MODE_ITABLE, 2675 }; 2676 2677 struct ext4_li_request { 2678 struct super_block *lr_super; 2679 enum ext4_li_mode lr_mode; 2680 ext4_group_t lr_first_not_zeroed; 2681 ext4_group_t lr_next_group; 2682 struct list_head lr_request; 2683 unsigned long lr_next_sched; 2684 unsigned long lr_timeout; 2685 }; 2686 2687 struct ext4_features { 2688 struct kobject f_kobj; 2689 struct completion f_kobj_unregister; 2690 }; 2691 2692 /* 2693 * This structure will be used for multiple mount protection. It will be 2694 * written into the block number saved in the s_mmp_block field in the 2695 * superblock. Programs that check MMP should assume that if 2696 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2697 * to use the filesystem, regardless of how old the timestamp is. 2698 */ 2699 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2700 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2701 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2702 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2703 2704 struct mmp_struct { 2705 __le32 mmp_magic; /* Magic number for MMP */ 2706 __le32 mmp_seq; /* Sequence no. updated periodically */ 2707 2708 /* 2709 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2710 * purposes and do not affect the correctness of the algorithm 2711 */ 2712 __le64 mmp_time; /* Time last updated */ 2713 char mmp_nodename[64]; /* Node which last updated MMP block */ 2714 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2715 2716 /* 2717 * mmp_check_interval is used to verify if the MMP block has been 2718 * updated on the block device. The value is updated based on the 2719 * maximum time to write the MMP block during an update cycle. 2720 */ 2721 __le16 mmp_check_interval; 2722 2723 __le16 mmp_pad1; 2724 __le32 mmp_pad2[226]; 2725 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2726 }; 2727 2728 /* arguments passed to the mmp thread */ 2729 struct mmpd_data { 2730 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2731 struct super_block *sb; /* super block of the fs */ 2732 }; 2733 2734 /* 2735 * Check interval multiplier 2736 * The MMP block is written every update interval and initially checked every 2737 * update interval x the multiplier (the value is then adapted based on the 2738 * write latency). The reason is that writes can be delayed under load and we 2739 * don't want readers to incorrectly assume that the filesystem is no longer 2740 * in use. 2741 */ 2742 #define EXT4_MMP_CHECK_MULT 2UL 2743 2744 /* 2745 * Minimum interval for MMP checking in seconds. 2746 */ 2747 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2748 2749 /* 2750 * Maximum interval for MMP checking in seconds. 2751 */ 2752 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2753 2754 /* 2755 * Function prototypes 2756 */ 2757 2758 /* 2759 * Ok, these declarations are also in <linux/kernel.h> but none of the 2760 * ext4 source programs needs to include it so they are duplicated here. 2761 */ 2762 # define NORET_TYPE /**/ 2763 # define ATTRIB_NORET __attribute__((noreturn)) 2764 # define NORET_AND noreturn, 2765 2766 /* bitmap.c */ 2767 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2768 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2769 struct ext4_group_desc *gdp, 2770 struct buffer_head *bh); 2771 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2772 struct ext4_group_desc *gdp, 2773 struct buffer_head *bh); 2774 void ext4_block_bitmap_csum_set(struct super_block *sb, 2775 struct ext4_group_desc *gdp, 2776 struct buffer_head *bh); 2777 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2778 struct ext4_group_desc *gdp, 2779 struct buffer_head *bh); 2780 2781 /* balloc.c */ 2782 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2783 ext4_fsblk_t blocknr, 2784 ext4_group_t *blockgrpp, 2785 ext4_grpblk_t *offsetp); 2786 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2787 ext4_fsblk_t block); 2788 2789 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2790 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2791 ext4_group_t group); 2792 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2793 ext4_fsblk_t goal, 2794 unsigned int flags, 2795 unsigned long *count, 2796 int *errp); 2797 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2798 s64 nclusters, unsigned int flags); 2799 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2800 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2801 ext4_group_t block_group, 2802 struct buffer_head ** bh); 2803 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2804 ext4_group_t group); 2805 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2806 2807 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2808 ext4_group_t block_group, 2809 bool ignore_locked); 2810 extern int ext4_wait_block_bitmap(struct super_block *sb, 2811 ext4_group_t block_group, 2812 struct buffer_head *bh); 2813 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2814 ext4_group_t block_group); 2815 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2816 ext4_group_t block_group, 2817 struct ext4_group_desc *gdp); 2818 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2819 2820 #if IS_ENABLED(CONFIG_UNICODE) 2821 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2822 const struct qstr *iname, 2823 struct ext4_filename *fname); 2824 2825 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2826 { 2827 kfree(fname->cf_name.name); 2828 fname->cf_name.name = NULL; 2829 } 2830 #else 2831 static inline int ext4_fname_setup_ci_filename(struct inode *dir, 2832 const struct qstr *iname, 2833 struct ext4_filename *fname) 2834 { 2835 return 0; 2836 } 2837 2838 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2839 { 2840 } 2841 #endif 2842 2843 /* ext4 encryption related stuff goes here crypto.c */ 2844 #ifdef CONFIG_FS_ENCRYPTION 2845 extern const struct fscrypt_operations ext4_cryptops; 2846 2847 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2848 int lookup, struct ext4_filename *fname); 2849 2850 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2851 struct ext4_filename *fname); 2852 2853 void ext4_fname_free_filename(struct ext4_filename *fname); 2854 2855 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2856 2857 #else /* !CONFIG_FS_ENCRYPTION */ 2858 static inline int ext4_fname_setup_filename(struct inode *dir, 2859 const struct qstr *iname, 2860 int lookup, 2861 struct ext4_filename *fname) 2862 { 2863 fname->usr_fname = iname; 2864 fname->disk_name.name = (unsigned char *) iname->name; 2865 fname->disk_name.len = iname->len; 2866 2867 return ext4_fname_setup_ci_filename(dir, iname, fname); 2868 } 2869 2870 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2871 struct dentry *dentry, 2872 struct ext4_filename *fname) 2873 { 2874 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2875 } 2876 2877 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2878 { 2879 ext4_fname_free_ci_filename(fname); 2880 } 2881 2882 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2883 void __user *arg) 2884 { 2885 return -EOPNOTSUPP; 2886 } 2887 #endif /* !CONFIG_FS_ENCRYPTION */ 2888 2889 /* dir.c */ 2890 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2891 struct file *, 2892 struct ext4_dir_entry_2 *, 2893 struct buffer_head *, char *, int, 2894 unsigned int); 2895 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2896 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2897 (de), (bh), (buf), (size), (offset))) 2898 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2899 __u32 minor_hash, 2900 struct ext4_dir_entry_2 *dirent, 2901 struct fscrypt_str *ent_name); 2902 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2903 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh, 2904 void *buf, int buf_size, 2905 struct ext4_filename *fname, 2906 struct ext4_dir_entry_2 **dest_de); 2907 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2908 struct ext4_dir_entry_2 *de, 2909 int buf_size, 2910 struct ext4_filename *fname); 2911 static inline void ext4_update_dx_flag(struct inode *inode) 2912 { 2913 if (!ext4_has_feature_dir_index(inode->i_sb) && 2914 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2915 /* ext4_iget() should have caught this... */ 2916 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2917 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2918 } 2919 } 2920 static const unsigned char ext4_filetype_table[] = { 2921 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2922 }; 2923 2924 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2925 { 2926 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2927 return DT_UNKNOWN; 2928 2929 return ext4_filetype_table[filetype]; 2930 } 2931 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2932 void *buf, int buf_size); 2933 2934 /* fsync.c */ 2935 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2936 2937 /* hash.c */ 2938 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2939 struct dx_hash_info *hinfo); 2940 2941 /* ialloc.c */ 2942 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2943 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2944 struct inode *, umode_t, 2945 const struct qstr *qstr, __u32 goal, 2946 uid_t *owner, __u32 i_flags, 2947 int handle_type, unsigned int line_no, 2948 int nblocks); 2949 2950 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2951 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2952 (goal), (owner), i_flags, 0, 0, 0) 2953 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2954 type, nblocks) \ 2955 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2956 0, (type), __LINE__, (nblocks)) 2957 2958 2959 extern void ext4_free_inode(handle_t *, struct inode *); 2960 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2961 extern unsigned long ext4_count_free_inodes(struct super_block *); 2962 extern unsigned long ext4_count_dirs(struct super_block *); 2963 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2964 extern int ext4_init_inode_table(struct super_block *sb, 2965 ext4_group_t group, int barrier); 2966 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2967 2968 /* fast_commit.c */ 2969 int ext4_fc_info_show(struct seq_file *seq, void *v); 2970 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2971 void ext4_fc_init_inode(struct inode *inode); 2972 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2973 ext4_lblk_t end); 2974 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2975 struct dentry *dentry); 2976 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2977 struct dentry *dentry); 2978 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2979 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2980 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2981 struct dentry *dentry); 2982 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2983 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2984 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2985 void ext4_fc_del(struct inode *inode); 2986 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2987 void ext4_fc_replay_cleanup(struct super_block *sb); 2988 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2989 int __init ext4_fc_init_dentry_cache(void); 2990 void ext4_fc_destroy_dentry_cache(void); 2991 int ext4_fc_record_regions(struct super_block *sb, int ino, 2992 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2993 int len, int replay); 2994 2995 /* mballoc.c */ 2996 extern const struct seq_operations ext4_mb_seq_groups_ops; 2997 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2998 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2999 extern int ext4_mb_init(struct super_block *); 3000 extern void ext4_mb_release(struct super_block *); 3001 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 3002 struct ext4_allocation_request *, int *); 3003 extern void ext4_discard_preallocations(struct inode *); 3004 extern int __init ext4_init_mballoc(void); 3005 extern void ext4_exit_mballoc(void); 3006 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 3007 ext4_group_t group, 3008 unsigned int nr, int *cnt); 3009 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 3010 unsigned int nr); 3011 3012 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 3013 struct buffer_head *bh, ext4_fsblk_t block, 3014 unsigned long count, int flags); 3015 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 3016 ext4_group_t ngroups); 3017 extern int ext4_mb_add_groupinfo(struct super_block *sb, 3018 ext4_group_t i, struct ext4_group_desc *desc); 3019 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 3020 ext4_fsblk_t block, unsigned long count); 3021 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 3022 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 3023 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 3024 int len, bool state); 3025 static inline bool ext4_mb_cr_expensive(enum criteria cr) 3026 { 3027 return cr >= CR_GOAL_LEN_SLOW; 3028 } 3029 3030 /* inode.c */ 3031 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 3032 struct ext4_inode_info *ei); 3033 int ext4_inode_is_fast_symlink(struct inode *inode); 3034 void ext4_check_map_extents_env(struct inode *inode); 3035 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 3036 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 3037 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 3038 bool wait, struct buffer_head **bhs); 3039 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 3040 struct buffer_head *bh_result, int create); 3041 int ext4_get_block(struct inode *inode, sector_t iblock, 3042 struct buffer_head *bh_result, int create); 3043 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 3044 struct buffer_head *bh, int create); 3045 int ext4_walk_page_buffers(handle_t *handle, 3046 struct inode *inode, 3047 struct buffer_head *head, 3048 unsigned from, 3049 unsigned to, 3050 int *partial, 3051 int (*fn)(handle_t *handle, struct inode *inode, 3052 struct buffer_head *bh)); 3053 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 3054 struct buffer_head *bh); 3055 void ext4_set_inode_mapping_order(struct inode *inode); 3056 #define FALL_BACK_TO_NONDELALLOC 1 3057 #define CONVERT_INLINE_DATA 2 3058 3059 typedef enum { 3060 EXT4_IGET_NORMAL = 0, 3061 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3062 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 3063 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 3064 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 3065 } ext4_iget_flags; 3066 3067 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3068 ext4_iget_flags flags, const char *function, 3069 unsigned int line); 3070 3071 #define ext4_iget(sb, ino, flags) \ 3072 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3073 3074 extern int ext4_write_inode(struct inode *, struct writeback_control *); 3075 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 3076 struct iattr *); 3077 extern u32 ext4_dio_alignment(struct inode *inode); 3078 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 3079 struct kstat *, u32, unsigned int); 3080 extern void ext4_evict_inode(struct inode *); 3081 extern void ext4_clear_inode(struct inode *); 3082 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 3083 struct kstat *, u32, unsigned int); 3084 extern void ext4_dirty_inode(struct inode *, int); 3085 extern int ext4_change_inode_journal_flag(struct inode *, int); 3086 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3087 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3088 struct ext4_iloc *iloc); 3089 extern int ext4_inode_attach_jinode(struct inode *inode); 3090 extern int ext4_can_truncate(struct inode *inode); 3091 extern int ext4_truncate(struct inode *); 3092 extern int ext4_break_layouts(struct inode *); 3093 extern int ext4_truncate_page_cache_block_range(struct inode *inode, 3094 loff_t start, loff_t end); 3095 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3096 extern void ext4_set_inode_flags(struct inode *, bool init); 3097 extern int ext4_alloc_da_blocks(struct inode *inode); 3098 extern void ext4_set_aops(struct inode *inode); 3099 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3100 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3101 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks); 3102 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks, 3103 int pextents); 3104 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3105 loff_t lstart, loff_t lend); 3106 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3107 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3108 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3109 extern void ext4_da_release_space(struct inode *inode, int to_free); 3110 extern void ext4_da_update_reserve_space(struct inode *inode, 3111 int used, int quota_claim); 3112 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3113 ext4_fsblk_t pblk, ext4_lblk_t len); 3114 3115 static inline bool is_special_ino(struct super_block *sb, unsigned long ino) 3116 { 3117 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 3118 3119 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) || 3120 ino == le32_to_cpu(es->s_usr_quota_inum) || 3121 ino == le32_to_cpu(es->s_grp_quota_inum) || 3122 ino == le32_to_cpu(es->s_prj_quota_inum) || 3123 ino == le32_to_cpu(es->s_orphan_file_inum); 3124 } 3125 3126 /* indirect.c */ 3127 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3128 struct ext4_map_blocks *map, int flags); 3129 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3130 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3131 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3132 ext4_lblk_t start, ext4_lblk_t end); 3133 3134 /* ioctl.c */ 3135 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3136 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3137 int ext4_fileattr_set(struct mnt_idmap *idmap, 3138 struct dentry *dentry, struct file_kattr *fa); 3139 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3140 extern void ext4_reset_inode_seed(struct inode *inode); 3141 int ext4_update_overhead(struct super_block *sb, bool force); 3142 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3143 3144 /* migrate.c */ 3145 extern int ext4_ext_migrate(struct inode *); 3146 extern int ext4_ind_migrate(struct inode *inode); 3147 3148 /* namei.c */ 3149 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3150 struct inode *inode); 3151 extern int ext4_dirblock_csum_verify(struct inode *inode, 3152 struct buffer_head *bh); 3153 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3154 __u32 start_minor_hash, __u32 *next_hash); 3155 extern int ext4_search_dir(struct buffer_head *bh, 3156 char *search_buf, 3157 int buf_size, 3158 struct inode *dir, 3159 struct ext4_filename *fname, 3160 unsigned int offset, 3161 struct ext4_dir_entry_2 **res_dir); 3162 extern int ext4_generic_delete_entry(struct inode *dir, 3163 struct ext4_dir_entry_2 *de_del, 3164 struct buffer_head *bh, 3165 void *entry_buf, 3166 int buf_size, 3167 int csum_size); 3168 extern bool ext4_empty_dir(struct inode *inode); 3169 3170 /* resize.c */ 3171 extern void ext4_kvfree_array_rcu(void *to_free); 3172 extern int ext4_group_add(struct super_block *sb, 3173 struct ext4_new_group_data *input); 3174 extern int ext4_group_extend(struct super_block *sb, 3175 struct ext4_super_block *es, 3176 ext4_fsblk_t n_blocks_count); 3177 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3178 extern unsigned int ext4_list_backups(struct super_block *sb, 3179 unsigned int *three, unsigned int *five, 3180 unsigned int *seven); 3181 3182 /* super.c */ 3183 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3184 sector_t block, blk_opf_t op_flags); 3185 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3186 sector_t block); 3187 extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb, 3188 sector_t block); 3189 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3190 bh_end_io_t *end_io, bool simu_fail); 3191 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3192 bh_end_io_t *end_io, bool simu_fail); 3193 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3194 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3195 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3196 extern int ext4_calculate_overhead(struct super_block *sb); 3197 extern __le32 ext4_superblock_csum(struct ext4_super_block *es); 3198 extern void ext4_superblock_csum_set(struct super_block *sb); 3199 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3200 ext4_group_t ngroup); 3201 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3202 char nbuf[16]); 3203 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3204 ext4_group_t block_group, 3205 unsigned int flags); 3206 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3207 ext4_group_t block_group); 3208 extern void print_daily_error_info(struct timer_list *t); 3209 3210 extern __printf(7, 8) 3211 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3212 int, __u64, const char *, ...); 3213 extern __printf(6, 7) 3214 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3215 ext4_fsblk_t, int, const char *, ...); 3216 extern __printf(5, 6) 3217 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3218 const char *, ...); 3219 extern void __ext4_std_error(struct super_block *, const char *, 3220 unsigned int, int); 3221 extern __printf(4, 5) 3222 void __ext4_warning(struct super_block *, const char *, unsigned int, 3223 const char *, ...); 3224 extern __printf(4, 5) 3225 void __ext4_warning_inode(const struct inode *inode, const char *function, 3226 unsigned int line, const char *fmt, ...); 3227 extern __printf(3, 4) 3228 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3229 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3230 const char *, unsigned int, const char *); 3231 extern __printf(7, 8) 3232 void __ext4_grp_locked_error(const char *, unsigned int, 3233 struct super_block *, ext4_group_t, 3234 u64, ext4_fsblk_t, 3235 const char *, ...); 3236 3237 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3238 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3239 3240 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3241 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3242 3243 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3244 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3245 (fmt), ## a) 3246 3247 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3248 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3249 3250 #define ext4_abort(sb, err, fmt, a...) \ 3251 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3252 3253 #ifdef CONFIG_PRINTK 3254 3255 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3256 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3257 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3258 __ext4_error_inode((inode), (func), (line), (block), \ 3259 (err), (fmt), ##__VA_ARGS__) 3260 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3261 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3262 #define ext4_error(sb, fmt, ...) \ 3263 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3264 ##__VA_ARGS__) 3265 #define ext4_error_err(sb, err, fmt, ...) \ 3266 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3267 ##__VA_ARGS__) 3268 #define ext4_warning(sb, fmt, ...) \ 3269 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3270 #define ext4_warning_inode(inode, fmt, ...) \ 3271 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3272 #define ext4_msg(sb, level, fmt, ...) \ 3273 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3274 #define dump_mmp_msg(sb, mmp, msg) \ 3275 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3276 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3277 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3278 fmt, ##__VA_ARGS__) 3279 3280 #else 3281 3282 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3283 do { \ 3284 no_printk(fmt, ##__VA_ARGS__); \ 3285 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3286 } while (0) 3287 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3288 do { \ 3289 no_printk(fmt, ##__VA_ARGS__); \ 3290 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3291 } while (0) 3292 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3293 do { \ 3294 no_printk(fmt, ##__VA_ARGS__); \ 3295 __ext4_error_file(file, "", 0, block, " "); \ 3296 } while (0) 3297 #define ext4_error(sb, fmt, ...) \ 3298 do { \ 3299 no_printk(fmt, ##__VA_ARGS__); \ 3300 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3301 } while (0) 3302 #define ext4_error_err(sb, err, fmt, ...) \ 3303 do { \ 3304 no_printk(fmt, ##__VA_ARGS__); \ 3305 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3306 } while (0) 3307 #define ext4_warning(sb, fmt, ...) \ 3308 do { \ 3309 no_printk(fmt, ##__VA_ARGS__); \ 3310 __ext4_warning(sb, "", 0, " "); \ 3311 } while (0) 3312 #define ext4_warning_inode(inode, fmt, ...) \ 3313 do { \ 3314 no_printk(fmt, ##__VA_ARGS__); \ 3315 __ext4_warning_inode(inode, "", 0, " "); \ 3316 } while (0) 3317 #define ext4_msg(sb, level, fmt, ...) \ 3318 do { \ 3319 no_printk(fmt, ##__VA_ARGS__); \ 3320 __ext4_msg(sb, "", " "); \ 3321 } while (0) 3322 #define dump_mmp_msg(sb, mmp, msg) \ 3323 __dump_mmp_msg(sb, mmp, "", 0, "") 3324 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3325 do { \ 3326 no_printk(fmt, ##__VA_ARGS__); \ 3327 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3328 } while (0) 3329 3330 #endif 3331 3332 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3333 struct ext4_group_desc *bg); 3334 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3335 struct ext4_group_desc *bg); 3336 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3337 struct ext4_group_desc *bg); 3338 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3339 struct ext4_group_desc *bg); 3340 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3341 struct ext4_group_desc *bg); 3342 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3343 struct ext4_group_desc *bg); 3344 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3345 struct ext4_group_desc *bg); 3346 extern void ext4_block_bitmap_set(struct super_block *sb, 3347 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3348 extern void ext4_inode_bitmap_set(struct super_block *sb, 3349 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3350 extern void ext4_inode_table_set(struct super_block *sb, 3351 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3352 extern void ext4_free_group_clusters_set(struct super_block *sb, 3353 struct ext4_group_desc *bg, 3354 __u32 count); 3355 extern void ext4_free_inodes_set(struct super_block *sb, 3356 struct ext4_group_desc *bg, __u32 count); 3357 extern void ext4_used_dirs_set(struct super_block *sb, 3358 struct ext4_group_desc *bg, __u32 count); 3359 extern void ext4_itable_unused_set(struct super_block *sb, 3360 struct ext4_group_desc *bg, __u32 count); 3361 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3362 struct ext4_group_desc *gdp); 3363 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3364 struct ext4_group_desc *gdp); 3365 extern int ext4_register_li_request(struct super_block *sb, 3366 ext4_group_t first_not_zeroed); 3367 3368 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3369 { 3370 return ext4_has_feature_gdt_csum(sb) || 3371 ext4_has_feature_metadata_csum(sb); 3372 } 3373 3374 #define ext4_read_incompat_64bit_val(es, name) \ 3375 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3376 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3377 le32_to_cpu(es->name##_lo)) 3378 3379 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3380 { 3381 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3382 } 3383 3384 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3385 { 3386 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3387 } 3388 3389 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3390 { 3391 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3392 } 3393 3394 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3395 ext4_fsblk_t blk) 3396 { 3397 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3398 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3399 } 3400 3401 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3402 ext4_fsblk_t blk) 3403 { 3404 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3405 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3406 } 3407 3408 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3409 ext4_fsblk_t blk) 3410 { 3411 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3412 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3413 } 3414 3415 static inline loff_t ext4_isize(struct super_block *sb, 3416 struct ext4_inode *raw_inode) 3417 { 3418 if (ext4_has_feature_largedir(sb) || 3419 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3420 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3421 le32_to_cpu(raw_inode->i_size_lo); 3422 3423 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3424 } 3425 3426 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3427 { 3428 raw_inode->i_size_lo = cpu_to_le32(i_size); 3429 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3430 } 3431 3432 /* 3433 * Reading s_groups_count requires using smp_rmb() afterwards. See 3434 * the locking protocol documented in the comments of ext4_group_add() 3435 * in resize.c 3436 */ 3437 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3438 { 3439 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3440 3441 smp_rmb(); 3442 return ngroups; 3443 } 3444 3445 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3446 ext4_group_t block_group) 3447 { 3448 return block_group >> sbi->s_log_groups_per_flex; 3449 } 3450 3451 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3452 { 3453 return 1 << sbi->s_log_groups_per_flex; 3454 } 3455 3456 static inline loff_t ext4_get_maxbytes(struct inode *inode) 3457 { 3458 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) 3459 return inode->i_sb->s_maxbytes; 3460 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; 3461 } 3462 3463 #define ext4_std_error(sb, errno) \ 3464 do { \ 3465 if ((errno)) \ 3466 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3467 } while (0) 3468 3469 #ifdef CONFIG_SMP 3470 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3471 * counters. So we need to make sure we have free clusters more 3472 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3473 */ 3474 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3475 #else 3476 #define EXT4_FREECLUSTERS_WATERMARK 0 3477 #endif 3478 3479 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3480 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3481 { 3482 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3483 !inode_is_locked(inode)); 3484 down_write(&EXT4_I(inode)->i_data_sem); 3485 if (newsize > EXT4_I(inode)->i_disksize) 3486 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3487 up_write(&EXT4_I(inode)->i_data_sem); 3488 } 3489 3490 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3491 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3492 { 3493 int changed = 0; 3494 3495 if (newsize > inode->i_size) { 3496 i_size_write(inode, newsize); 3497 changed = 1; 3498 } 3499 if (newsize > EXT4_I(inode)->i_disksize) { 3500 ext4_update_i_disksize(inode, newsize); 3501 changed |= 2; 3502 } 3503 return changed; 3504 } 3505 3506 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3507 loff_t len); 3508 3509 struct ext4_group_info { 3510 unsigned long bb_state; 3511 #ifdef AGGRESSIVE_CHECK 3512 unsigned long bb_check_counter; 3513 #endif 3514 struct rb_root bb_free_root; 3515 ext4_grpblk_t bb_first_free; /* first free block */ 3516 ext4_grpblk_t bb_free; /* total free blocks */ 3517 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3518 int bb_avg_fragment_size_order; /* order of average 3519 fragment in BG */ 3520 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3521 ext4_group_t bb_group; /* Group number */ 3522 struct list_head bb_prealloc_list; 3523 #ifdef DOUBLE_CHECK 3524 void *bb_bitmap; 3525 #endif 3526 struct rw_semaphore alloc_sem; 3527 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3528 * regions, index is order. 3529 * bb_counters[3] = 5 means 3530 * 5 free 8-block regions. */ 3531 }; 3532 3533 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3534 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3535 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3536 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3537 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3538 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3539 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3540 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3541 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3542 3543 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3544 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3545 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3546 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3547 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3548 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3549 3550 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3551 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3552 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3553 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3554 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3555 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3556 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3557 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3558 3559 #define EXT4_MAX_CONTENTION 8 3560 #define EXT4_CONTENTION_THRESHOLD 2 3561 3562 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3563 ext4_group_t group) 3564 { 3565 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3566 } 3567 3568 /* 3569 * Returns true if the filesystem is busy enough that attempts to 3570 * access the block group locks has run into contention. 3571 */ 3572 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3573 { 3574 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3575 } 3576 3577 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group) 3578 { 3579 if (!spin_trylock(ext4_group_lock_ptr(sb, group))) 3580 return false; 3581 /* 3582 * We're able to grab the lock right away, so drop the lock 3583 * contention counter. 3584 */ 3585 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3586 return true; 3587 } 3588 3589 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3590 { 3591 if (!ext4_try_lock_group(sb, group)) { 3592 /* 3593 * The lock is busy, so bump the contention counter, 3594 * and then wait on the spin lock. 3595 */ 3596 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3597 EXT4_MAX_CONTENTION); 3598 spin_lock(ext4_group_lock_ptr(sb, group)); 3599 } 3600 } 3601 3602 static inline void ext4_unlock_group(struct super_block *sb, 3603 ext4_group_t group) 3604 { 3605 spin_unlock(ext4_group_lock_ptr(sb, group)); 3606 } 3607 3608 #ifdef CONFIG_QUOTA 3609 static inline bool ext4_quota_capable(struct super_block *sb) 3610 { 3611 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3612 } 3613 3614 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3615 { 3616 struct ext4_sb_info *sbi = EXT4_SB(sb); 3617 3618 return (ext4_has_feature_quota(sb) || 3619 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3620 } 3621 int ext4_enable_quotas(struct super_block *sb); 3622 #endif 3623 3624 /* 3625 * Block validity checking 3626 */ 3627 #define ext4_check_indirect_blockref(inode, bh) \ 3628 ext4_check_blockref(__func__, __LINE__, inode, \ 3629 (__le32 *)(bh)->b_data, \ 3630 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3631 3632 #define ext4_ind_check_inode(inode) \ 3633 ext4_check_blockref(__func__, __LINE__, inode, \ 3634 EXT4_I(inode)->i_data, \ 3635 EXT4_NDIR_BLOCKS) 3636 3637 /* 3638 * Inodes and files operations 3639 */ 3640 3641 /* dir.c */ 3642 extern const struct file_operations ext4_dir_operations; 3643 3644 /* file.c */ 3645 extern const struct inode_operations ext4_file_inode_operations; 3646 extern const struct file_operations ext4_file_operations; 3647 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3648 3649 /* inline.c */ 3650 extern int ext4_get_max_inline_size(struct inode *inode); 3651 extern int ext4_find_inline_data_nolock(struct inode *inode); 3652 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3653 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size); 3654 3655 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3656 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3657 struct inode *inode, 3658 loff_t pos, unsigned len, 3659 struct folio **foliop); 3660 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3661 unsigned copied, struct folio *folio); 3662 extern int ext4_generic_write_inline_data(struct address_space *mapping, 3663 struct inode *inode, 3664 loff_t pos, unsigned len, 3665 struct folio **foliop, 3666 void **fsdata, bool da); 3667 extern int ext4_try_add_inline_entry(handle_t *handle, 3668 struct ext4_filename *fname, 3669 struct inode *dir, struct inode *inode); 3670 extern int ext4_try_create_inline_dir(handle_t *handle, 3671 struct inode *parent, 3672 struct inode *inode); 3673 extern int ext4_read_inline_dir(struct file *filp, 3674 struct dir_context *ctx, 3675 int *has_inline_data); 3676 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3677 struct inode *dir, ext4_lblk_t block, 3678 struct dx_hash_info *hinfo, 3679 __u32 start_hash, __u32 start_minor_hash, 3680 int *has_inline_data); 3681 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3682 struct ext4_filename *fname, 3683 struct ext4_dir_entry_2 **res_dir, 3684 int *has_inline_data); 3685 extern int ext4_delete_inline_entry(handle_t *handle, 3686 struct inode *dir, 3687 struct ext4_dir_entry_2 *de_del, 3688 struct buffer_head *bh, 3689 int *has_inline_data); 3690 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3691 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3692 struct ext4_dir_entry_2 **parent_de, 3693 int *retval); 3694 extern void *ext4_read_inline_link(struct inode *inode); 3695 3696 struct iomap; 3697 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3698 3699 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3700 3701 extern int ext4_convert_inline_data(struct inode *inode); 3702 3703 static inline int ext4_has_inline_data(struct inode *inode) 3704 { 3705 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3706 EXT4_I(inode)->i_inline_off; 3707 } 3708 3709 /* namei.c */ 3710 extern const struct inode_operations ext4_dir_inode_operations; 3711 extern const struct inode_operations ext4_special_inode_operations; 3712 extern struct dentry *ext4_get_parent(struct dentry *child); 3713 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode, 3714 struct buffer_head *dir_block, 3715 unsigned int parent_ino, void *inline_buf, 3716 int inline_size); 3717 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3718 unsigned int blocksize); 3719 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3720 struct buffer_head *bh); 3721 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3722 struct inode *inode, struct dentry *dentry); 3723 extern int __ext4_link(struct inode *dir, struct inode *inode, 3724 struct dentry *dentry); 3725 3726 #define S_SHIFT 12 3727 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3728 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3729 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3730 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3731 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3732 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3733 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3734 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3735 }; 3736 3737 static inline void ext4_set_de_type(struct super_block *sb, 3738 struct ext4_dir_entry_2 *de, 3739 umode_t mode) { 3740 if (ext4_has_feature_filetype(sb)) 3741 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3742 } 3743 3744 /* readpages.c */ 3745 int ext4_read_folio(struct file *file, struct folio *folio); 3746 void ext4_readahead(struct readahead_control *rac); 3747 extern int __init ext4_init_post_read_processing(void); 3748 extern void ext4_exit_post_read_processing(void); 3749 3750 /* symlink.c */ 3751 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3752 extern const struct inode_operations ext4_symlink_inode_operations; 3753 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3754 3755 /* sysfs.c */ 3756 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3757 extern int ext4_register_sysfs(struct super_block *sb); 3758 extern void ext4_unregister_sysfs(struct super_block *sb); 3759 extern int __init ext4_init_sysfs(void); 3760 extern void ext4_exit_sysfs(void); 3761 3762 /* block_validity */ 3763 extern void ext4_release_system_zone(struct super_block *sb); 3764 extern int ext4_setup_system_zone(struct super_block *sb); 3765 extern int __init ext4_init_system_zone(void); 3766 extern void ext4_exit_system_zone(void); 3767 extern int ext4_inode_block_valid(struct inode *inode, 3768 ext4_fsblk_t start_blk, 3769 unsigned int count); 3770 extern int ext4_check_blockref(const char *, unsigned int, 3771 struct inode *, __le32 *, unsigned int); 3772 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3773 ext4_fsblk_t start_blk, unsigned int count); 3774 3775 3776 /* extents.c */ 3777 struct ext4_ext_path; 3778 struct ext4_extent; 3779 3780 /* 3781 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3782 * __le32. 3783 */ 3784 #define EXT_MAX_BLOCKS 0xffffffff 3785 3786 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3787 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3788 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3789 struct ext4_map_blocks *map, int flags); 3790 extern int ext4_ext_truncate(handle_t *, struct inode *); 3791 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3792 ext4_lblk_t end); 3793 extern void ext4_ext_init(struct super_block *); 3794 extern void ext4_ext_release(struct super_block *); 3795 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3796 loff_t len); 3797 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3798 loff_t offset, ssize_t len); 3799 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle, 3800 struct inode *inode, loff_t offset, ssize_t len); 3801 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3802 ext4_io_end_t *io_end); 3803 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3804 struct ext4_map_blocks *map, int flags); 3805 extern int ext4_map_query_blocks(handle_t *handle, struct inode *inode, 3806 struct ext4_map_blocks *map, int flags); 3807 extern int ext4_map_create_blocks(handle_t *handle, struct inode *inode, 3808 struct ext4_map_blocks *map, int flags); 3809 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3810 int num, 3811 struct ext4_ext_path *path); 3812 extern struct ext4_ext_path *ext4_ext_insert_extent( 3813 handle_t *handle, struct inode *inode, 3814 struct ext4_ext_path *path, 3815 struct ext4_extent *newext, int gb_flags); 3816 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3817 struct ext4_ext_path *, 3818 int flags); 3819 extern void ext4_free_ext_path(struct ext4_ext_path *); 3820 extern int ext4_ext_check_inode(struct inode *inode); 3821 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3822 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3823 __u64 start, __u64 len); 3824 extern int ext4_get_es_cache(struct inode *inode, 3825 struct fiemap_extent_info *fieinfo, 3826 __u64 start, __u64 len); 3827 extern int ext4_ext_precache(struct inode *inode); 3828 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3829 struct inode *inode2, ext4_lblk_t lblk1, 3830 ext4_lblk_t lblk2, ext4_lblk_t count, 3831 int mark_unwritten,int *err); 3832 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3833 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3834 int check_cred, int restart_cred, 3835 int revoke_cred); 3836 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3837 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3838 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3839 int len, int unwritten, ext4_fsblk_t pblk); 3840 extern int ext4_ext_clear_bb(struct inode *inode); 3841 3842 3843 /* move_extent.c */ 3844 extern void ext4_double_down_write_data_sem(struct inode *first, 3845 struct inode *second); 3846 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3847 struct inode *donor_inode); 3848 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3849 __u64 start_orig, __u64 start_donor, 3850 __u64 len, __u64 *moved_len); 3851 3852 /* page-io.c */ 3853 extern int __init ext4_init_pageio(void); 3854 extern void ext4_exit_pageio(void); 3855 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3856 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3857 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3858 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3859 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3860 struct writeback_control *wbc); 3861 extern void ext4_end_io_rsv_work(struct work_struct *work); 3862 extern void ext4_io_submit(struct ext4_io_submit *io); 3863 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3864 size_t len); 3865 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3866 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3867 3868 /* mmp.c */ 3869 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3870 3871 /* mmp.c */ 3872 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3873 3874 /* verity.c */ 3875 extern const struct fsverity_operations ext4_verityops; 3876 3877 /* orphan.c */ 3878 extern int ext4_orphan_add(handle_t *, struct inode *); 3879 extern int ext4_orphan_del(handle_t *, struct inode *); 3880 extern void ext4_orphan_cleanup(struct super_block *sb, 3881 struct ext4_super_block *es); 3882 extern void ext4_release_orphan_info(struct super_block *sb); 3883 extern int ext4_init_orphan_info(struct super_block *sb); 3884 extern int ext4_orphan_file_empty(struct super_block *sb); 3885 extern void ext4_orphan_file_block_trigger( 3886 struct jbd2_buffer_trigger_type *triggers, 3887 struct buffer_head *bh, 3888 void *data, size_t size); 3889 3890 /* 3891 * Add new method to test whether block and inode bitmaps are properly 3892 * initialized. With uninit_bg reading the block from disk is not enough 3893 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3894 */ 3895 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3896 3897 static inline int bitmap_uptodate(struct buffer_head *bh) 3898 { 3899 return (buffer_uptodate(bh) && 3900 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3901 } 3902 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3903 { 3904 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3905 } 3906 3907 extern int ext4_resize_begin(struct super_block *sb); 3908 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3909 3910 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end) 3911 { 3912 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) 3913 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3914 } 3915 3916 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3917 { 3918 if (io_end->flag & EXT4_IO_END_UNWRITTEN) 3919 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3920 } 3921 3922 extern const struct iomap_ops ext4_iomap_ops; 3923 extern const struct iomap_ops ext4_iomap_report_ops; 3924 3925 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3926 { 3927 /* 3928 * If the buffer has the write error flag, we have failed 3929 * to write out data in the block. In this case, we don't 3930 * have to read the block because we may read the old data 3931 * successfully. 3932 */ 3933 if (buffer_write_io_error(bh)) 3934 set_buffer_uptodate(bh); 3935 return buffer_uptodate(bh); 3936 } 3937 3938 static inline bool ext4_inode_can_atomic_write(struct inode *inode) 3939 { 3940 3941 return S_ISREG(inode->i_mode) && 3942 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) && 3943 EXT4_SB(inode->i_sb)->s_awu_min > 0; 3944 } 3945 3946 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio, 3947 loff_t pos, unsigned len, 3948 get_block_t *get_block); 3949 3950 #if IS_ENABLED(CONFIG_EXT4_KUNIT_TESTS) 3951 #define EXPORT_SYMBOL_FOR_EXT4_TEST(sym) \ 3952 EXPORT_SYMBOL_FOR_MODULES(sym, "ext4-test") 3953 #endif 3954 #endif /* __KERNEL__ */ 3955 3956 #endif /* _EXT4_H */ 3957