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/sched/signal.h> 32 #include <linux/blockgroup_lock.h> 33 #include <linux/percpu_counter.h> 34 #include <linux/ratelimit.h> 35 #include <linux/crc32c.h> 36 #include <linux/falloc.h> 37 #include <linux/percpu-rwsem.h> 38 #include <linux/fiemap.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 #include <uapi/linux/ext4.h> 43 44 #include <linux/fscrypt.h> 45 #include <linux/fsverity.h> 46 47 #include <linux/compiler.h> 48 49 /* 50 * The fourth extended filesystem constants/structures 51 */ 52 53 /* 54 * with AGGRESSIVE_CHECK allocator runs consistency checks over 55 * structures. these checks slow things down a lot 56 */ 57 #define AGGRESSIVE_CHECK__ 58 59 /* 60 * with DOUBLE_CHECK defined mballoc creates persistent in-core 61 * bitmaps, maintains and uses them to check for double allocations 62 */ 63 #define DOUBLE_CHECK__ 64 65 /* 66 * Define EXT4FS_DEBUG to produce debug messages 67 */ 68 #undef EXT4FS_DEBUG 69 70 /* 71 * Debug code 72 */ 73 #ifdef EXT4FS_DEBUG 74 #define ext4_debug(f, a...) \ 75 do { \ 76 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 77 __FILE__, __LINE__, __func__); \ 78 printk(KERN_DEBUG f, ## a); \ 79 } while (0) 80 #else 81 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 82 #endif 83 84 /* 85 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 86 */ 87 #define EXT_DEBUG__ 88 89 /* 90 * Dynamic printk for controlled extents debugging. 91 */ 92 #ifdef CONFIG_EXT4_DEBUG 93 #define ext_debug(ino, fmt, ...) \ 94 pr_debug("[%s/%d] EXT4-fs (%s): ino %llu: (%s, %d): %s:" fmt, \ 95 current->comm, task_pid_nr(current), \ 96 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 97 __func__, ##__VA_ARGS__) 98 #else 99 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 100 #endif 101 102 #define ASSERT(assert) \ 103 do { \ 104 if (unlikely(!(assert))) { \ 105 printk(KERN_EMERG \ 106 "Assertion failure in %s() at %s:%d: '%s'\n", \ 107 __func__, __FILE__, __LINE__, #assert); \ 108 BUG(); \ 109 } \ 110 } while (0) 111 112 /* data type for block offset of block group */ 113 typedef int ext4_grpblk_t; 114 115 /* data type for filesystem-wide blocks number */ 116 typedef unsigned long long ext4_fsblk_t; 117 118 /* data type for file logical block number */ 119 typedef __u32 ext4_lblk_t; 120 121 /* data type for block group number */ 122 typedef unsigned int ext4_group_t; 123 124 enum SHIFT_DIRECTION { 125 SHIFT_LEFT = 0, 126 SHIFT_RIGHT, 127 }; 128 129 /* 130 * For each criteria, mballoc has slightly different way of finding 131 * the required blocks nad usually, higher the criteria the slower the 132 * allocation. We start at lower criterias and keep falling back to 133 * higher ones if we are not able to find any blocks. Lower (earlier) 134 * criteria are faster. 135 */ 136 enum criteria { 137 /* 138 * Used when number of blocks needed is a power of 2. This 139 * doesn't trigger any disk IO except prefetch and is the 140 * fastest criteria. 141 */ 142 CR_POWER2_ALIGNED, 143 144 /* 145 * Tries to lookup in-memory data structures to find the most 146 * suitable group that satisfies goal request. No disk IO 147 * except block prefetch. 148 */ 149 CR_GOAL_LEN_FAST, 150 151 /* 152 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal 153 * length to the best available length for faster allocation. 154 */ 155 CR_BEST_AVAIL_LEN, 156 157 /* 158 * Reads each block group sequentially, performing disk IO if 159 * necessary, to find suitable block group. Tries to 160 * allocate goal length but might trim the request if nothing 161 * is found after enough tries. 162 */ 163 CR_GOAL_LEN_SLOW, 164 165 /* 166 * Finds the first free set of blocks and allocates 167 * those. This is only used in rare cases when 168 * CR_GOAL_LEN_SLOW also fails to allocate anything. 169 */ 170 CR_ANY_FREE, 171 172 /* 173 * Number of criterias defined. 174 */ 175 EXT4_MB_NUM_CRS 176 }; 177 178 /* 179 * Flags used in mballoc's allocation_context flags field. 180 * 181 * Also used to show what's going on for debugging purposes when the 182 * flag field is exported via the traceport interface 183 */ 184 185 /* prefer goal again. length */ 186 #define EXT4_MB_HINT_MERGE 0x0001 187 /* first blocks in the file */ 188 #define EXT4_MB_HINT_FIRST 0x0008 189 /* data is being allocated */ 190 #define EXT4_MB_HINT_DATA 0x0020 191 /* don't preallocate (for tails) */ 192 #define EXT4_MB_HINT_NOPREALLOC 0x0040 193 /* allocate for locality group */ 194 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 195 /* allocate goal blocks or none */ 196 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 197 /* goal is meaningful */ 198 #define EXT4_MB_HINT_TRY_GOAL 0x0200 199 /* blocks already pre-reserved by delayed allocation */ 200 #define EXT4_MB_DELALLOC_RESERVED 0x0400 201 /* We are doing stream allocation */ 202 #define EXT4_MB_STREAM_ALLOC 0x0800 203 /* Use reserved root blocks if needed */ 204 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 205 /* Use blocks from reserved pool */ 206 #define EXT4_MB_USE_RESERVED 0x2000 207 /* Do strict check for free blocks while retrying block allocation */ 208 #define EXT4_MB_STRICT_CHECK 0x4000 209 210 struct ext4_allocation_request { 211 /* target inode for block we're allocating */ 212 struct inode *inode; 213 /* how many blocks we want to allocate */ 214 unsigned int len; 215 /* logical block in target inode */ 216 ext4_lblk_t logical; 217 /* the closest logical allocated block to the left */ 218 ext4_lblk_t lleft; 219 /* the closest logical allocated block to the right */ 220 ext4_lblk_t lright; 221 /* phys. target (a hint) */ 222 ext4_fsblk_t goal; 223 /* phys. block for the closest logical allocated block to the left */ 224 ext4_fsblk_t pleft; 225 /* phys. block for the closest logical allocated block to the right */ 226 ext4_fsblk_t pright; 227 /* flags. see above EXT4_MB_HINT_* */ 228 unsigned int flags; 229 }; 230 231 /* 232 * Logical to physical block mapping, used by ext4_map_blocks() 233 * 234 * This structure is used to pass requests into ext4_map_blocks() as 235 * well as to store the information returned by ext4_map_blocks(). It 236 * takes less room on the stack than a struct buffer_head. 237 */ 238 #define EXT4_MAP_NEW BIT(BH_New) 239 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 240 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 241 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 242 #define EXT4_MAP_DELAYED BIT(BH_Delay) 243 /* 244 * This is for use in ext4_map_query_blocks() for a special case where we can 245 * have a physically and logically contiguous blocks split across two leaf 246 * nodes instead of a single extent. This is required in case of atomic writes 247 * to know whether the returned extent is last in leaf. If yes, then lookup for 248 * next in leaf block in ext4_map_query_blocks_next_in_leaf(). 249 * - This is never going to be added to any buffer head state. 250 * - We use the next available bit after BH_BITMAP_UPTODATE. 251 */ 252 #define EXT4_MAP_QUERY_LAST_IN_LEAF BIT(BH_BITMAP_UPTODATE + 1) 253 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 254 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 255 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF) 256 257 struct ext4_map_blocks { 258 ext4_fsblk_t m_pblk; 259 ext4_lblk_t m_lblk; 260 unsigned int m_len; 261 unsigned int m_flags; 262 u64 m_seq; 263 }; 264 265 /* 266 * Block validity checking, system zone rbtree. 267 */ 268 struct ext4_system_blocks { 269 struct rb_root root; 270 struct rcu_head rcu; 271 }; 272 273 /* 274 * Flags for ext4_io_end->flags 275 */ 276 #define EXT4_IO_END_UNWRITTEN 0x0001 277 #define EXT4_IO_END_FAILED 0x0002 278 279 #define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED) 280 281 struct ext4_io_end_vec { 282 struct list_head list; /* list of io_end_vec */ 283 loff_t offset; /* offset in the file */ 284 ssize_t size; /* size of the extent */ 285 }; 286 287 /* 288 * For converting unwritten extents on a work queue. 'handle' is used for 289 * buffered writeback. 290 */ 291 typedef struct ext4_io_end { 292 struct list_head list; /* per-file finished IO list */ 293 handle_t *handle; /* handle reserved for extent 294 * conversion */ 295 struct inode *inode; /* file being written to */ 296 struct bio *bio; /* Linked list of completed 297 * bios covering the extent */ 298 unsigned int flag; /* unwritten or not */ 299 refcount_t count; /* reference counter */ 300 struct list_head list_vec; /* list of ext4_io_end_vec */ 301 } ext4_io_end_t; 302 303 struct ext4_io_submit { 304 struct writeback_control *io_wbc; 305 struct bio *io_bio; 306 ext4_io_end_t *io_end; 307 sector_t io_next_block; 308 }; 309 310 /* 311 * Special inodes numbers 312 */ 313 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 314 #define EXT4_ROOT_INO 2 /* Root inode */ 315 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 316 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 317 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 318 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 319 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 320 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 321 322 /* First non-reserved inode for old ext4 filesystems */ 323 #define EXT4_GOOD_OLD_FIRST_INO 11 324 325 /* 326 * Maximal count of links to a file 327 */ 328 #define EXT4_LINK_MAX 65000 329 330 /* 331 * Macro-instructions used to manage several block sizes 332 */ 333 #define EXT4_MIN_BLOCK_SIZE 1024 334 #define EXT4_MAX_BLOCK_SIZE 65536 335 #define EXT4_MIN_BLOCK_LOG_SIZE 10 336 #define EXT4_MAX_BLOCK_LOG_SIZE 16 337 #define EXT4_MAX_CLUSTER_LOG_SIZE 28 338 #ifdef __KERNEL__ 339 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 340 #else 341 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 342 #endif 343 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 344 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 345 EXT4_SB(s)->s_cluster_bits) 346 #ifdef __KERNEL__ 347 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 348 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 349 #else 350 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 351 #endif 352 #ifdef __KERNEL__ 353 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 354 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 355 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 356 #else 357 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 358 EXT4_GOOD_OLD_INODE_SIZE : \ 359 (s)->s_inode_size) 360 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 361 EXT4_GOOD_OLD_FIRST_INO : \ 362 (s)->s_first_ino) 363 #endif 364 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 365 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 366 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 367 blkbits)) 368 #define EXT4_B_TO_LBLK(inode, offset) \ 369 (round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits) 370 #define EXT4_LBLK_TO_B(inode, lblk) ((loff_t)(lblk) << (inode)->i_blkbits) 371 372 /* Translate a block number to a page index */ 373 #define EXT4_LBLK_TO_PG(inode, lblk) (EXT4_LBLK_TO_B((inode), (lblk)) >> \ 374 PAGE_SHIFT) 375 /* Translate a page index to a block number */ 376 #define EXT4_PG_TO_LBLK(inode, pnum) (((loff_t)(pnum) << PAGE_SHIFT) >> \ 377 (inode)->i_blkbits) 378 /* Translate a block number to a cluster number */ 379 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 380 /* Translate a cluster number to a block number */ 381 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 382 /* Translate # of blks to # of clusters */ 383 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 384 (sbi)->s_cluster_bits) 385 /* Mask out the low bits to get the starting block of the cluster */ 386 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 387 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 388 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 389 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 390 /* Fill in the low bits to get the last block of the cluster */ 391 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 392 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 393 /* Get the cluster offset */ 394 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 395 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 396 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 397 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 398 399 /* 400 * Structure of a blocks group descriptor 401 */ 402 struct ext4_group_desc 403 { 404 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 405 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 406 __le32 bg_inode_table_lo; /* Inodes table block */ 407 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 408 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 409 __le16 bg_used_dirs_count_lo; /* Directories count */ 410 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 411 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 412 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 413 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 414 __le16 bg_itable_unused_lo; /* Unused inodes count */ 415 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 416 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 417 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 418 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 419 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 420 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 421 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 422 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 423 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 424 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 425 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 426 __u32 bg_reserved; 427 }; 428 429 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 430 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 431 sizeof(__le16)) 432 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 433 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 434 sizeof(__le16)) 435 436 /* 437 * Structure of a flex block group info 438 */ 439 440 struct flex_groups { 441 atomic64_t free_clusters; 442 atomic_t free_inodes; 443 atomic_t used_dirs; 444 }; 445 446 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 447 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 448 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 449 450 /* 451 * Macro-instructions used to manage group descriptors 452 */ 453 #define EXT4_MIN_DESC_SIZE 32 454 #define EXT4_MIN_DESC_SIZE_64BIT 64 455 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 456 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 457 #ifdef __KERNEL__ 458 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 459 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 460 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 461 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 462 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 463 #else 464 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 465 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 466 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 467 #endif 468 469 /* 470 * Constants relative to the data blocks 471 */ 472 #define EXT4_NDIR_BLOCKS 12 473 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 474 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 475 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 476 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 477 478 /* 479 * Inode flags 480 */ 481 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 482 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 483 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 484 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 485 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 486 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 487 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 488 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 489 /* Reserved for compression usage... */ 490 #define EXT4_DIRTY_FL 0x00000100 491 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 492 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 493 /* nb: was previously EXT2_ECOMPR_FL */ 494 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 495 /* End compression flags --- maybe not all used */ 496 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 497 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 498 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 499 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 500 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 501 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 502 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 503 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 504 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 505 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 506 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 507 508 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 509 510 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 511 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 512 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 513 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 514 515 /* User modifiable flags */ 516 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 517 EXT4_UNRM_FL | \ 518 EXT4_COMPR_FL | \ 519 EXT4_SYNC_FL | \ 520 EXT4_IMMUTABLE_FL | \ 521 EXT4_APPEND_FL | \ 522 EXT4_NODUMP_FL | \ 523 EXT4_NOATIME_FL | \ 524 EXT4_JOURNAL_DATA_FL | \ 525 EXT4_NOTAIL_FL | \ 526 EXT4_DIRSYNC_FL | \ 527 EXT4_TOPDIR_FL | \ 528 EXT4_EXTENTS_FL | \ 529 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 530 EXT4_DAX_FL | \ 531 EXT4_PROJINHERIT_FL | \ 532 EXT4_CASEFOLD_FL) 533 534 /* User visible flags */ 535 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 536 EXT4_DIRTY_FL | \ 537 EXT4_COMPRBLK_FL | \ 538 EXT4_NOCOMPR_FL | \ 539 EXT4_ENCRYPT_FL | \ 540 EXT4_INDEX_FL | \ 541 EXT4_VERITY_FL | \ 542 EXT4_INLINE_DATA_FL) 543 544 /* Flags that should be inherited by new inodes from their parent. */ 545 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 546 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 547 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 548 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 549 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 550 EXT4_DAX_FL) 551 552 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 553 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 554 EXT4_PROJINHERIT_FL)) 555 556 /* Flags that are appropriate for non-directories/regular files. */ 557 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 558 559 /* The only flags that should be swapped */ 560 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 561 562 /* Flags which are mutually exclusive to DAX */ 563 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 564 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 565 566 /* Mask out flags that are inappropriate for the given type of inode. */ 567 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 568 { 569 if (S_ISDIR(mode)) 570 return flags; 571 else if (S_ISREG(mode)) 572 return flags & EXT4_REG_FLMASK; 573 else 574 return flags & EXT4_OTHER_FLMASK; 575 } 576 577 /* 578 * Inode flags used for atomic set/get 579 */ 580 enum { 581 EXT4_INODE_SECRM = 0, /* Secure deletion */ 582 EXT4_INODE_UNRM = 1, /* Undelete */ 583 EXT4_INODE_COMPR = 2, /* Compress file */ 584 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 585 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 586 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 587 EXT4_INODE_NODUMP = 6, /* do not dump file */ 588 EXT4_INODE_NOATIME = 7, /* do not update atime */ 589 /* Reserved for compression usage... */ 590 EXT4_INODE_DIRTY = 8, 591 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 592 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 593 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 594 /* End compression flags --- maybe not all used */ 595 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 596 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 597 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 598 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 599 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 600 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 601 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 602 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 603 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 604 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 605 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 606 EXT4_INODE_DAX = 25, /* Inode is DAX */ 607 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 608 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 609 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 610 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 611 }; 612 613 /* 614 * Since it's pretty easy to mix up bit numbers and hex values, we use a 615 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 616 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 617 * any extra space in the compiled kernel image, otherwise, the build will fail. 618 * It's important that these values are the same, since we are using 619 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 620 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 621 * values found in ext2, ext3 and ext4 filesystems, and of course the values 622 * defined in e2fsprogs. 623 * 624 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 625 */ 626 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 627 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 628 629 static inline void ext4_check_flag_values(void) 630 { 631 CHECK_FLAG_VALUE(SECRM); 632 CHECK_FLAG_VALUE(UNRM); 633 CHECK_FLAG_VALUE(COMPR); 634 CHECK_FLAG_VALUE(SYNC); 635 CHECK_FLAG_VALUE(IMMUTABLE); 636 CHECK_FLAG_VALUE(APPEND); 637 CHECK_FLAG_VALUE(NODUMP); 638 CHECK_FLAG_VALUE(NOATIME); 639 CHECK_FLAG_VALUE(DIRTY); 640 CHECK_FLAG_VALUE(COMPRBLK); 641 CHECK_FLAG_VALUE(NOCOMPR); 642 CHECK_FLAG_VALUE(ENCRYPT); 643 CHECK_FLAG_VALUE(INDEX); 644 CHECK_FLAG_VALUE(IMAGIC); 645 CHECK_FLAG_VALUE(JOURNAL_DATA); 646 CHECK_FLAG_VALUE(NOTAIL); 647 CHECK_FLAG_VALUE(DIRSYNC); 648 CHECK_FLAG_VALUE(TOPDIR); 649 CHECK_FLAG_VALUE(HUGE_FILE); 650 CHECK_FLAG_VALUE(EXTENTS); 651 CHECK_FLAG_VALUE(VERITY); 652 CHECK_FLAG_VALUE(EA_INODE); 653 CHECK_FLAG_VALUE(INLINE_DATA); 654 CHECK_FLAG_VALUE(PROJINHERIT); 655 CHECK_FLAG_VALUE(CASEFOLD); 656 CHECK_FLAG_VALUE(RESERVED); 657 } 658 659 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 660 struct compat_ext4_new_group_input { 661 u32 group; 662 compat_u64 block_bitmap; 663 compat_u64 inode_bitmap; 664 compat_u64 inode_table; 665 u32 blocks_count; 666 u16 reserved_blocks; 667 u16 unused; 668 }; 669 #endif 670 671 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 672 struct ext4_new_group_data { 673 __u32 group; 674 __u64 block_bitmap; 675 __u64 inode_bitmap; 676 __u64 inode_table; 677 __u32 blocks_count; 678 __u16 reserved_blocks; 679 __u16 mdata_blocks; 680 __u32 free_clusters_count; 681 }; 682 683 /* Indexes used to index group tables in ext4_new_group_data */ 684 enum { 685 BLOCK_BITMAP = 0, /* block bitmap */ 686 INODE_BITMAP, /* inode bitmap */ 687 INODE_TABLE, /* inode tables */ 688 GROUP_TABLE_COUNT, 689 }; 690 691 /* 692 * Flags used by ext4_map_blocks() 693 */ 694 /* Allocate any needed blocks and/or convert an unwritten 695 extent to be an initialized ext4 */ 696 #define EXT4_GET_BLOCKS_CREATE 0x0001 697 /* Request the creation of an unwritten extent */ 698 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 699 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 700 EXT4_GET_BLOCKS_CREATE) 701 /* Caller is from the delayed allocation writeout path 702 * finally doing the actual allocation of delayed blocks */ 703 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 704 /* 705 * This means that we cannot merge newly allocated extents, and if we 706 * found an unwritten extent, we need to split it. 707 */ 708 #define EXT4_GET_BLOCKS_SPLIT_NOMERGE 0x0008 709 /* Convert unwritten extent to initialized. */ 710 #define EXT4_GET_BLOCKS_CONVERT 0x0010 711 /* Eventual metadata allocation (due to growing extent tree) 712 * should not fail, so try to use reserved blocks for that.*/ 713 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 714 /* Don't normalize allocation size (used for fallocate) */ 715 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 716 /* Convert written extents to unwritten */ 717 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 718 /* Write zeros to newly created written extents */ 719 #define EXT4_GET_BLOCKS_ZERO 0x0200 720 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 721 EXT4_GET_BLOCKS_ZERO) 722 /* Caller is in the context of data submission, such as writeback, 723 * fsync, etc. Especially, in the generic writeback path, caller will 724 * submit data before dropping transaction handle. This allows jbd2 725 * to avoid submitting data before commit. */ 726 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 727 /* Convert extent to initialized after IO complete */ 728 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\ 729 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\ 730 EXT4_GET_BLOCKS_IO_SUBMIT) 731 /* Caller is in the atomic contex, find extent if it has been cached */ 732 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 733 /* 734 * Atomic write caller needs this to query in the slow path of mixed mapping 735 * case, when a contiguous extent can be split across two adjacent leaf nodes. 736 * Look EXT4_MAP_QUERY_LAST_IN_LEAF. 737 */ 738 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000 739 740 /* 741 * The bit position of these flags must not overlap with any of the 742 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 743 * read_extent_tree_block(), ext4_split_extent_at(), 744 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 745 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 746 * caching the extents when reading from the extent tree while a 747 * truncate or punch hole operation is in progress. 748 */ 749 #define EXT4_EX_NOCACHE 0x40000000 750 #define EXT4_EX_FORCE_CACHE 0x20000000 751 #define EXT4_EX_NOFAIL 0x10000000 752 /* 753 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to 754 * pass while lookup/querying of on disk extent tree. 755 */ 756 #define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\ 757 EXT4_EX_NOFAIL |\ 758 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF) 759 760 /* 761 * Flags used by ext4_free_blocks 762 */ 763 #define EXT4_FREE_BLOCKS_METADATA 0x0001 764 #define EXT4_FREE_BLOCKS_FORGET 0x0002 765 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 766 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 767 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 768 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 769 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 770 771 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 772 /* 773 * ioctl commands in 32 bit emulation 774 */ 775 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 776 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 777 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 778 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 779 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 780 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 781 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 782 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 783 #endif 784 785 /* Max physical block we can address w/o extents */ 786 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 787 788 /* Max logical block we can support */ 789 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 790 791 /* 792 * Structure of an inode on the disk 793 */ 794 struct ext4_inode { 795 __le16 i_mode; /* File mode */ 796 __le16 i_uid; /* Low 16 bits of Owner Uid */ 797 __le32 i_size_lo; /* Size in bytes */ 798 __le32 i_atime; /* Access time */ 799 __le32 i_ctime; /* Inode Change time */ 800 __le32 i_mtime; /* Modification time */ 801 __le32 i_dtime; /* Deletion Time */ 802 __le16 i_gid; /* Low 16 bits of Group Id */ 803 __le16 i_links_count; /* Links count */ 804 __le32 i_blocks_lo; /* Blocks count */ 805 __le32 i_flags; /* File flags */ 806 union { 807 struct { 808 __le32 l_i_version; 809 } linux1; 810 struct { 811 __u32 h_i_translator; 812 } hurd1; 813 struct { 814 __u32 m_i_reserved1; 815 } masix1; 816 } osd1; /* OS dependent 1 */ 817 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 818 __le32 i_generation; /* File version (for NFS) */ 819 __le32 i_file_acl_lo; /* File ACL */ 820 __le32 i_size_high; 821 __le32 i_obso_faddr; /* Obsoleted fragment address */ 822 union { 823 struct { 824 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 825 __le16 l_i_file_acl_high; 826 __le16 l_i_uid_high; /* these 2 fields */ 827 __le16 l_i_gid_high; /* were reserved2[0] */ 828 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 829 __le16 l_i_reserved; 830 } linux2; 831 struct { 832 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 833 __u16 h_i_mode_high; 834 __u16 h_i_uid_high; 835 __u16 h_i_gid_high; 836 __u32 h_i_author; 837 } hurd2; 838 struct { 839 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 840 __le16 m_i_file_acl_high; 841 __u32 m_i_reserved2[2]; 842 } masix2; 843 } osd2; /* OS dependent 2 */ 844 __le16 i_extra_isize; 845 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 846 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 847 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 848 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 849 __le32 i_crtime; /* File Creation time */ 850 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 851 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 852 __le32 i_projid; /* Project ID */ 853 }; 854 855 #define EXT4_EPOCH_BITS 2 856 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 857 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 858 859 /* 860 * Extended fields will fit into an inode if the filesystem was formatted 861 * with large inodes (-I 256 or larger) and there are not currently any EAs 862 * consuming all of the available space. For new inodes we always reserve 863 * enough space for the kernel's known extended fields, but for inodes 864 * created with an old kernel this might not have been the case. None of 865 * the extended inode fields is critical for correct filesystem operation. 866 * This macro checks if a certain field fits in the inode. Note that 867 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 868 */ 869 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 870 ((offsetof(typeof(*ext4_inode), field) + \ 871 sizeof((ext4_inode)->field)) \ 872 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 873 (einode)->i_extra_isize)) \ 874 875 /* 876 * We use an encoding that preserves the times for extra epoch "00": 877 * 878 * extra msb of adjust for signed 879 * epoch 32-bit 32-bit tv_sec to 880 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 881 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 882 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 883 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 884 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 885 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 886 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 887 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 888 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 889 * 890 * Note that previous versions of the kernel on 64-bit systems would 891 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 892 * 1970. e2fsck will correct this, assuming that it is run on the 893 * affected filesystem before 2242. 894 */ 895 896 static inline __le32 ext4_encode_extra_time(struct timespec64 ts) 897 { 898 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK; 899 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS)); 900 } 901 902 static inline struct timespec64 ext4_decode_extra_time(__le32 base, 903 __le32 extra) 904 { 905 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) }; 906 907 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 908 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 909 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 910 return ts; 911 } 912 913 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \ 914 do { \ 915 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 916 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \ 917 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \ 918 } else \ 919 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \ 920 } while (0) 921 922 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \ 923 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode)) 924 925 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \ 926 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode)) 927 928 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \ 929 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode)) 930 931 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 932 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 933 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \ 934 raw_inode, (einode)->xtime) 935 936 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \ 937 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \ 938 ext4_decode_extra_time((raw_inode)->xtime, \ 939 (raw_inode)->xtime ## _extra) : \ 940 (struct timespec64) { \ 941 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \ 942 }) 943 944 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \ 945 do { \ 946 inode_set_atime_to_ts(inode, \ 947 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \ 948 } while (0) 949 950 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \ 951 do { \ 952 inode_set_mtime_to_ts(inode, \ 953 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \ 954 } while (0) 955 956 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \ 957 do { \ 958 inode_set_ctime_to_ts(inode, \ 959 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \ 960 } while (0) 961 962 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 963 do { \ 964 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 965 (einode)->xtime = \ 966 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \ 967 raw_inode); \ 968 else \ 969 (einode)->xtime = (struct timespec64){0, 0}; \ 970 } while (0) 971 972 #define i_disk_version osd1.linux1.l_i_version 973 974 #if defined(__KERNEL__) || defined(__linux__) 975 #define i_reserved1 osd1.linux1.l_i_reserved1 976 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 977 #define i_blocks_high osd2.linux2.l_i_blocks_high 978 #define i_uid_low i_uid 979 #define i_gid_low i_gid 980 #define i_uid_high osd2.linux2.l_i_uid_high 981 #define i_gid_high osd2.linux2.l_i_gid_high 982 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 983 984 #elif defined(__GNU__) 985 986 #define i_translator osd1.hurd1.h_i_translator 987 #define i_uid_high osd2.hurd2.h_i_uid_high 988 #define i_gid_high osd2.hurd2.h_i_gid_high 989 #define i_author osd2.hurd2.h_i_author 990 991 #elif defined(__masix__) 992 993 #define i_reserved1 osd1.masix1.m_i_reserved1 994 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 995 #define i_reserved2 osd2.masix2.m_i_reserved2 996 997 #endif /* defined(__KERNEL__) || defined(__linux__) */ 998 999 #include "extents_status.h" 1000 #include "fast_commit.h" 1001 1002 /* 1003 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 1004 * 1005 * These are needed to avoid lockdep false positives when we need to 1006 * allocate blocks to the quota inode during ext4_map_blocks(), while 1007 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1008 * do quota tracking for the quota inode, this avoids deadlock (as 1009 * well as infinite recursion, since it isn't turtles all the way 1010 * down...) 1011 * 1012 * I_DATA_SEM_NORMAL - Used for most inodes 1013 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1014 * where the second inode has larger inode number 1015 * than the first 1016 * I_DATA_SEM_QUOTA - Used for quota inodes only 1017 * I_DATA_SEM_EA - Used for ea_inodes only 1018 * I_DATA_SEM_JOURNAL - Used for journal inode 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 I_DATA_SEM_JOURNAL 1026 }; 1027 1028 struct ext4_fc_inode_snap; 1029 1030 /* 1031 * Snapshot failure reasons for ext4_fc_lock_updates tracepoint. 1032 * Keep these stable for tooling. 1033 */ 1034 enum ext4_fc_snap_err { 1035 EXT4_FC_SNAP_ERR_NONE = 0, 1036 EXT4_FC_SNAP_ERR_ES_MISS, 1037 EXT4_FC_SNAP_ERR_ES_DELAYED, 1038 EXT4_FC_SNAP_ERR_ES_OTHER, 1039 EXT4_FC_SNAP_ERR_INODES_CAP, 1040 EXT4_FC_SNAP_ERR_RANGES_CAP, 1041 EXT4_FC_SNAP_ERR_NOMEM, 1042 EXT4_FC_SNAP_ERR_INODE_LOC, 1043 }; 1044 1045 /* 1046 * fourth extended file system inode data in memory 1047 */ 1048 struct ext4_inode_info { 1049 __le32 i_data[15]; /* unconverted */ 1050 __u32 i_dtime; 1051 ext4_fsblk_t i_file_acl; 1052 1053 /* 1054 * i_block_group is the number of the block group which contains 1055 * this file's inode. Constant across the lifetime of the inode, 1056 * it is used for making block allocation decisions - we try to 1057 * place a file's data blocks near its inode block, and new inodes 1058 * near to their parent directory's inode. 1059 */ 1060 ext4_group_t i_block_group; 1061 ext4_lblk_t i_dir_start_lookup; 1062 #if (BITS_PER_LONG < 64) 1063 unsigned long i_state_flags; /* Dynamic state flags */ 1064 #endif 1065 unsigned long i_flags; 1066 1067 /* 1068 * Extended attributes can be read independently of the main file 1069 * data. Taking i_rwsem even when reading would cause contention 1070 * between readers of EAs and writers of regular file data, so 1071 * instead we synchronize on xattr_sem when reading or changing 1072 * EAs. 1073 * 1074 * EA inodes (EXT4_EA_INODE_FL) do not use xattr_sem; they reuse 1075 * the space for deferred iput linkage. 1076 */ 1077 union { 1078 struct rw_semaphore xattr_sem; 1079 struct llist_node i_ea_iput_node; 1080 }; 1081 1082 /* 1083 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1084 * i_orphan is used. 1085 */ 1086 union { 1087 struct list_head i_orphan; /* unlinked but open inodes */ 1088 unsigned int i_orphan_idx; /* Index in orphan file */ 1089 }; 1090 1091 /* Fast commit related info */ 1092 1093 /* For tracking dentry create updates */ 1094 struct list_head i_fc_dilist; 1095 struct list_head i_fc_list; /* 1096 * inodes that need fast commit 1097 * protected by sbi->s_fc_lock. 1098 */ 1099 1100 /* Start of lblk range that needs to be committed in this fast commit */ 1101 ext4_lblk_t i_fc_lblk_start; 1102 1103 /* End of lblk range that needs to be committed in this fast commit */ 1104 ext4_lblk_t i_fc_lblk_len; 1105 1106 /* 1107 * Commit-time fast commit snapshots. 1108 * 1109 * i_fc_snap is installed and freed under sbi->s_fc_lock. The fast 1110 * commit log writing path reads the snapshot under sbi->s_fc_lock while 1111 * serializing fast commit TLVs. 1112 * 1113 * The snapshot lifetime is bounded by EXT4_STATE_FC_COMMITTING and the 1114 * corresponding cleanup / eviction paths. 1115 * 1116 * i_fc_snap points to per-inode snapshot data for fast commit: 1117 * - a raw inode snapshot for EXT4_FC_TAG_INODE 1118 * - data range records for EXT4_FC_TAG_{ADD,DEL}_RANGE 1119 */ 1120 struct ext4_fc_inode_snap *i_fc_snap; 1121 1122 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1123 1124 /* 1125 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len 1126 * and inode's EXT4_FC_STATE_COMMITTING state bit. 1127 */ 1128 spinlock_t i_fc_lock; 1129 1130 /* 1131 * i_disksize keeps track of what the inode size is ON DISK, not 1132 * in memory. During truncate, i_size is set to the new size by 1133 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1134 * set i_disksize to 0 until the truncate is actually under way. 1135 * 1136 * The intent is that i_disksize always represents the blocks which 1137 * are used by this file. This allows recovery to restart truncate 1138 * on orphans if we crash during truncate. We actually write i_disksize 1139 * into the on-disk inode when writing inodes out, instead of i_size. 1140 * 1141 * The only time when i_disksize and i_size may be different is when 1142 * a truncate is in progress. The only things which change i_disksize 1143 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1144 */ 1145 loff_t i_disksize; 1146 1147 /* 1148 * i_data_sem is for serialising ext4_truncate() against 1149 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1150 * data tree are chopped off during truncate. We can't do that in 1151 * ext4 because whenever we perform intermediate commits during 1152 * truncate, the inode and all the metadata blocks *must* be in a 1153 * consistent state which allows truncation of the orphans to restart 1154 * during recovery. Hence we must fix the get_block-vs-truncate race 1155 * by other means, so we have i_data_sem. 1156 */ 1157 struct rw_semaphore i_data_sem; 1158 struct inode vfs_inode; 1159 struct jbd2_inode *jinode; 1160 struct mapping_metadata_bhs *i_metadata_bhs; 1161 1162 /* 1163 * File creation time. Its function is same as that of 1164 * struct timespec64 i_{a,c,m}time in the generic inode. 1165 */ 1166 struct timespec64 i_crtime; 1167 1168 /* mballoc */ 1169 atomic_t i_prealloc_active; 1170 1171 /* allocation reservation info for delalloc */ 1172 /* In case of bigalloc, this refer to clusters rather than blocks */ 1173 unsigned int i_reserved_data_blocks; 1174 struct rb_root i_prealloc_node; 1175 rwlock_t i_prealloc_lock; 1176 1177 /* extents status tree */ 1178 struct ext4_es_tree i_es_tree; 1179 rwlock_t i_es_lock; 1180 struct list_head i_es_list; 1181 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1182 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1183 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1184 extents to shrink. Protected by 1185 i_es_lock */ 1186 u64 i_es_seq; /* Change counter for extents. 1187 Protected by i_es_lock */ 1188 1189 /* ialloc */ 1190 ext4_group_t i_last_alloc_group; 1191 1192 /* pending cluster reservations for bigalloc file systems */ 1193 struct ext4_pending_tree i_pending_tree; 1194 1195 /* on-disk additional length */ 1196 __u16 i_extra_isize; 1197 1198 /* Indicate the inline data space. */ 1199 u16 i_inline_off; 1200 u16 i_inline_size; 1201 1202 #ifdef CONFIG_QUOTA 1203 /* quota space reservation, managed internally by quota code */ 1204 qsize_t i_reserved_quota; 1205 #endif 1206 spinlock_t i_block_reservation_lock; 1207 1208 /* Lock protecting lists below */ 1209 spinlock_t i_completed_io_lock; 1210 /* 1211 * Completed IOs that need unwritten extents handling and have a 1212 * transaction reserved for the buffer_head writeback path, and 1213 * also used by the iomap writeback path to queue ioends needing 1214 * unwritten extents conversion, i_disksize update, etc. 1215 */ 1216 struct list_head i_rsv_conversion_list; 1217 struct work_struct i_rsv_conversion_work; 1218 1219 /* 1220 * Transactions that contain inode's metadata needed to complete 1221 * fsync and fdatasync, respectively. 1222 */ 1223 tid_t i_sync_tid; 1224 tid_t i_datasync_tid; 1225 1226 #ifdef CONFIG_QUOTA 1227 struct dquot __rcu *i_dquot[MAXQUOTAS]; 1228 #endif 1229 1230 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1231 __u32 i_csum_seed; 1232 1233 kprojid_t i_projid; 1234 1235 #ifdef CONFIG_FS_ENCRYPTION 1236 struct fscrypt_inode_info *i_crypt_info; 1237 #endif 1238 }; 1239 1240 /* 1241 * File system states 1242 */ 1243 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1244 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1245 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1246 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1247 1248 /* 1249 * Misc. filesystem flags 1250 */ 1251 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1252 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1253 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1254 1255 /* 1256 * Mount flags set via mount options or defaults 1257 */ 1258 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1259 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1260 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1261 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1262 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1263 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1264 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1265 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1266 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1267 #ifdef CONFIG_FS_DAX 1268 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1269 #else 1270 #define EXT4_MOUNT_DAX_ALWAYS 0 1271 #endif 1272 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1273 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1274 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1275 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1276 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1277 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1278 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1279 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1280 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1281 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1282 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1283 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1284 * enable enforcement for hidden 1285 * quota files */ 1286 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1287 * enforcement for hidden quota 1288 * files */ 1289 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1290 * enforcement */ 1291 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1292 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1293 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1294 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1295 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1296 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1297 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1298 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1299 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1300 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1301 1302 /* 1303 * Mount flags set either automatically (could not be set by mount option) 1304 * based on per file system feature or property or in special cases such as 1305 * distinguishing between explicit mount option definition and default. 1306 */ 1307 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1308 specified delalloc */ 1309 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1310 size of blocksize * 8 1311 blocks */ 1312 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1313 file systems */ 1314 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1315 specified journal checksum */ 1316 1317 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1318 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1319 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1320 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1321 * scanning in mballoc 1322 */ 1323 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1324 #define EXT4_MOUNT2_BUFFERED_IOMAP 0x00000200 /* Use iomap for buffered I/O */ 1325 1326 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1327 ~EXT4_MOUNT_##opt 1328 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1329 EXT4_MOUNT_##opt 1330 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1331 EXT4_MOUNT_##opt) 1332 1333 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1334 ~EXT4_MOUNT2_##opt 1335 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1336 EXT4_MOUNT2_##opt 1337 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1338 EXT4_MOUNT2_##opt) 1339 1340 #define ext4_test_and_set_bit __test_and_set_bit_le 1341 #define ext4_set_bit __set_bit_le 1342 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1343 #define ext4_clear_bit __clear_bit_le 1344 #define ext4_test_bit test_bit_le 1345 #define ext4_find_next_zero_bit find_next_zero_bit_le 1346 #define ext4_find_next_bit find_next_bit_le 1347 1348 extern void mb_set_bits(void *bm, int cur, int len); 1349 1350 /* 1351 * Maximal mount counts between two filesystem checks 1352 */ 1353 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1354 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1355 1356 /* 1357 * Behaviour when detecting errors 1358 */ 1359 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1360 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1361 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1362 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1363 1364 /* Metadata checksum algorithm codes */ 1365 #define EXT4_CRC32C_CHKSUM 1 1366 1367 #define EXT4_LABEL_MAX 16 1368 1369 /* 1370 * Structure of the super block 1371 */ 1372 struct ext4_super_block { 1373 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1374 __le32 s_blocks_count_lo; /* Blocks count */ 1375 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1376 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1377 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1378 __le32 s_first_data_block; /* First Data Block */ 1379 __le32 s_log_block_size; /* Block size */ 1380 __le32 s_log_cluster_size; /* Allocation cluster size */ 1381 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1382 __le32 s_clusters_per_group; /* # Clusters per group */ 1383 __le32 s_inodes_per_group; /* # Inodes per group */ 1384 __le32 s_mtime; /* Mount time */ 1385 /*30*/ __le32 s_wtime; /* Write time */ 1386 __le16 s_mnt_count; /* Mount count */ 1387 __le16 s_max_mnt_count; /* Maximal mount count */ 1388 __le16 s_magic; /* Magic signature */ 1389 __le16 s_state; /* File system state */ 1390 __le16 s_errors; /* Behaviour when detecting errors */ 1391 __le16 s_minor_rev_level; /* minor revision level */ 1392 /*40*/ __le32 s_lastcheck; /* time of last check */ 1393 __le32 s_checkinterval; /* max. time between checks */ 1394 __le32 s_creator_os; /* OS */ 1395 __le32 s_rev_level; /* Revision level */ 1396 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1397 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1398 /* 1399 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1400 * 1401 * Note: the difference between the compatible feature set and 1402 * the incompatible feature set is that if there is a bit set 1403 * in the incompatible feature set that the kernel doesn't 1404 * know about, it should refuse to mount the filesystem. 1405 * 1406 * e2fsck's requirements are more strict; if it doesn't know 1407 * about a feature in either the compatible or incompatible 1408 * feature set, it must abort and not try to meddle with 1409 * things it doesn't understand... 1410 */ 1411 __le32 s_first_ino; /* First non-reserved inode */ 1412 __le16 s_inode_size; /* size of inode structure */ 1413 __le16 s_block_group_nr; /* block group # of this superblock */ 1414 __le32 s_feature_compat; /* compatible feature set */ 1415 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1416 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1417 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1418 /*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */ 1419 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1420 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1421 /* 1422 * Performance hints. Directory preallocation should only 1423 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1424 */ 1425 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1426 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1427 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1428 /* 1429 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1430 */ 1431 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1432 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1433 __le32 s_journal_dev; /* device number of journal file */ 1434 __le32 s_last_orphan; /* start of list of inodes to delete */ 1435 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1436 __u8 s_def_hash_version; /* Default hash version to use */ 1437 __u8 s_jnl_backup_type; 1438 __le16 s_desc_size; /* size of group descriptor */ 1439 /*100*/ __le32 s_default_mount_opts; 1440 __le32 s_first_meta_bg; /* First metablock block group */ 1441 __le32 s_mkfs_time; /* When the filesystem was created */ 1442 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1443 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1444 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1445 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1446 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1447 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1448 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1449 __le32 s_flags; /* Miscellaneous flags */ 1450 __le16 s_raid_stride; /* RAID stride */ 1451 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1452 __le64 s_mmp_block; /* Block for multi-mount protection */ 1453 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1454 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1455 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1456 __u8 s_encryption_level; /* versioning level for encryption */ 1457 __u8 s_reserved_pad; /* Padding to next 32bits */ 1458 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1459 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1460 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1461 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1462 snapshot's future use */ 1463 __le32 s_snapshot_list; /* inode number of the head of the 1464 on-disk snapshot list */ 1465 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1466 __le32 s_error_count; /* number of fs errors */ 1467 __le32 s_first_error_time; /* first time an error happened */ 1468 __le32 s_first_error_ino; /* inode involved in first error */ 1469 __le64 s_first_error_block; /* block involved of first error */ 1470 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1471 __le32 s_first_error_line; /* line number where error happened */ 1472 __le32 s_last_error_time; /* most recent time of an error */ 1473 __le32 s_last_error_ino; /* inode involved in last error */ 1474 __le32 s_last_error_line; /* line number where error happened */ 1475 __le64 s_last_error_block; /* block involved of last error */ 1476 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1477 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1478 __u8 s_mount_opts[64]; 1479 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1480 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1481 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1482 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1483 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1484 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1485 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1486 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1487 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1488 __u8 s_wtime_hi; 1489 __u8 s_mtime_hi; 1490 __u8 s_mkfs_time_hi; 1491 __u8 s_lastcheck_hi; 1492 __u8 s_first_error_time_hi; 1493 __u8 s_last_error_time_hi; 1494 __u8 s_first_error_errcode; 1495 __u8 s_last_error_errcode; 1496 __le16 s_encoding; /* Filename charset encoding */ 1497 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1498 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1499 __le16 s_def_resuid_hi; 1500 __le16 s_def_resgid_hi; 1501 __le32 s_reserved[93]; /* Padding to the end of the block */ 1502 __le32 s_checksum; /* crc32c(superblock) */ 1503 }; 1504 1505 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1506 1507 #ifdef __KERNEL__ 1508 1509 /* Number of quota types we support */ 1510 #define EXT4_MAXQUOTAS 3 1511 1512 #define EXT4_ENC_UTF8_12_1 1 1513 1514 /* Types of ext4 journal triggers */ 1515 enum ext4_journal_trigger_type { 1516 EXT4_JTR_ORPHAN_FILE, 1517 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1518 }; 1519 1520 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1521 1522 struct ext4_journal_trigger { 1523 struct jbd2_buffer_trigger_type tr_triggers; 1524 struct super_block *sb; 1525 }; 1526 1527 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1528 struct jbd2_buffer_trigger_type *trigger) 1529 { 1530 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1531 } 1532 1533 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1534 1535 /* Structure at the tail of orphan block */ 1536 struct ext4_orphan_block_tail { 1537 __le32 ob_magic; 1538 __le32 ob_checksum; 1539 }; 1540 1541 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1542 { 1543 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1544 sizeof(u32); 1545 } 1546 1547 struct ext4_orphan_block { 1548 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1549 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1550 }; 1551 1552 /* 1553 * Info about orphan file. 1554 */ 1555 struct ext4_orphan_info { 1556 int of_blocks; /* Number of orphan blocks in a file */ 1557 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1558 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1559 * file blocks */ 1560 }; 1561 1562 /* 1563 * Ext4 fast commit snapshot statistics. 1564 * 1565 * These are best-effort counters intended for debugging / performance 1566 * introspection; they are not exact under concurrent updates. 1567 */ 1568 struct ext4_fc_snap_stats { 1569 atomic64_t lock_updates_ns_total; 1570 atomic64_t lock_updates_ns_max; 1571 atomic64_t lock_updates_samples; 1572 1573 atomic64_t snap_inodes; 1574 atomic64_t snap_ranges; 1575 1576 atomic64_t snap_fail_es_miss; 1577 atomic64_t snap_fail_es_delayed; 1578 atomic64_t snap_fail_es_other; 1579 1580 atomic64_t snap_fail_inodes_cap; 1581 atomic64_t snap_fail_ranges_cap; 1582 atomic64_t snap_fail_nomem; 1583 atomic64_t snap_fail_inode_loc; 1584 1585 /* 1586 * Missing inode snapshots during log writing should never happen. 1587 * Keep this counter to help catch unexpected regressions. 1588 */ 1589 atomic64_t snap_fail_no_snap; 1590 }; 1591 1592 /* 1593 * fourth extended-fs super-block data in memory 1594 */ 1595 struct ext4_sb_info { 1596 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1597 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1598 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1599 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1600 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1601 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1602 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1603 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1604 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1605 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1606 unsigned long s_overhead; /* # of fs overhead clusters */ 1607 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1608 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1609 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1610 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1611 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1612 /* Array of bh's for the block group descriptors */ 1613 struct buffer_head * __rcu *s_group_desc; 1614 unsigned int s_mount_opt; 1615 unsigned int s_mount_opt2; 1616 unsigned long s_mount_flags; 1617 unsigned int s_def_mount_opt; 1618 unsigned int s_def_mount_opt2; 1619 ext4_fsblk_t s_sb_block; 1620 atomic64_t s_resv_clusters; 1621 kuid_t s_resuid; 1622 kgid_t s_resgid; 1623 unsigned short s_mount_state; 1624 unsigned short s_pad; 1625 int s_addr_per_block_bits; 1626 int s_desc_per_block_bits; 1627 int s_inode_size; 1628 int s_first_ino; 1629 unsigned int s_inode_readahead_blks; 1630 unsigned int s_inode_goal; 1631 u32 s_hash_seed[4]; 1632 int s_def_hash_version; 1633 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1634 struct percpu_counter s_freeclusters_counter; 1635 struct percpu_counter s_freeinodes_counter; 1636 struct percpu_counter s_dirs_counter; 1637 struct percpu_counter s_dirtyclusters_counter; 1638 struct percpu_counter s_sra_exceeded_retry_limit; 1639 struct blockgroup_lock *s_blockgroup_lock; 1640 struct proc_dir_entry *s_proc; 1641 struct kobject s_kobj; 1642 struct completion s_kobj_unregister; 1643 struct mutex s_error_notify_mutex; /* protects sysfs_notify vs kobject_del */ 1644 struct super_block *s_sb; 1645 struct buffer_head *s_mmp_bh; 1646 1647 /* Journaling */ 1648 struct journal_s *s_journal; 1649 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1650 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1651 struct list_head s_orphan; /* List of orphaned inodes in on disk 1652 list */ 1653 struct ext4_orphan_info s_orphan_info; 1654 unsigned long s_commit_interval; 1655 u32 s_max_batch_time; 1656 u32 s_min_batch_time; 1657 struct file *s_journal_bdev_file; 1658 #ifdef CONFIG_QUOTA 1659 /* Names of quota files with journalled quota */ 1660 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1661 int s_jquota_fmt; /* Format of quota to use */ 1662 #endif 1663 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1664 struct ext4_system_blocks __rcu *s_system_blks; 1665 1666 #ifdef EXTENTS_STATS 1667 /* ext4 extents stats */ 1668 unsigned long s_ext_min; 1669 unsigned long s_ext_max; 1670 unsigned long s_depth_max; 1671 spinlock_t s_ext_stats_lock; 1672 unsigned long s_ext_blocks; 1673 unsigned long s_ext_extents; 1674 #endif 1675 1676 /* for buddy allocator */ 1677 struct ext4_group_info ** __rcu *s_group_info; 1678 struct inode *s_buddy_cache; 1679 spinlock_t s_md_lock; 1680 unsigned short *s_mb_offsets; 1681 unsigned int *s_mb_maxs; 1682 unsigned int s_group_info_size; 1683 atomic_t s_mb_free_pending; 1684 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1685 after commit completed */ 1686 struct list_head s_discard_list; 1687 struct work_struct s_discard_work; 1688 atomic_t s_retry_alloc_pending; 1689 struct xarray *s_mb_avg_fragment_size; 1690 struct xarray *s_mb_largest_free_orders; 1691 1692 /* tunables */ 1693 unsigned long s_stripe; 1694 unsigned int s_mb_max_linear_groups; 1695 unsigned int s_mb_stream_request; 1696 unsigned int s_mb_max_to_scan; 1697 unsigned int s_mb_min_to_scan; 1698 unsigned int s_mb_stats; 1699 unsigned int s_mb_order2_reqs; 1700 unsigned int s_mb_group_prealloc; 1701 unsigned int s_max_dir_size_kb; 1702 unsigned int s_mb_prefetch; 1703 unsigned int s_mb_prefetch_limit; 1704 unsigned int s_mb_best_avail_max_trim_order; 1705 unsigned int s_sb_update_sec; 1706 unsigned int s_sb_update_kb; 1707 1708 /* where last allocation was done - for stream allocation */ 1709 ext4_group_t *s_mb_last_groups; 1710 unsigned int s_mb_nr_global_goals; 1711 1712 /* stats for buddy allocator */ 1713 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1714 atomic_t s_bal_success; /* we found long enough chunks */ 1715 atomic_t s_bal_allocated; /* in blocks */ 1716 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1717 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1718 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1719 atomic_t s_bal_goals; /* goal hits */ 1720 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */ 1721 atomic_t s_bal_len_goals; /* len goal hits */ 1722 atomic_t s_bal_breaks; /* too long searches */ 1723 atomic_t s_bal_2orders; /* 2^order hits */ 1724 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1725 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1726 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1727 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1728 atomic64_t s_mb_generation_time; 1729 atomic_t s_mb_lost_chunks; 1730 atomic_t s_mb_preallocated; 1731 atomic_t s_mb_discarded; 1732 atomic_t s_lock_busy; 1733 1734 /* locality groups */ 1735 struct ext4_locality_group __percpu *s_locality_groups; 1736 1737 /* for write statistics */ 1738 unsigned long s_sectors_written_start; 1739 u64 s_kbytes_written; 1740 1741 /* the size of zero-out chunk */ 1742 unsigned int s_extent_max_zeroout_kb; 1743 1744 unsigned int s_log_groups_per_flex; 1745 struct flex_groups * __rcu *s_flex_groups; 1746 ext4_group_t s_flex_groups_allocated; 1747 1748 /* workqueue for reserved extent conversions (buffered io) */ 1749 struct workqueue_struct *rsv_conversion_wq; 1750 1751 /* timer for periodic error stats printing */ 1752 struct timer_list s_err_report; 1753 /* timeout in seconds for s_err_report; 0 disables the timer. */ 1754 unsigned long s_err_report_sec; 1755 1756 /* Lazy inode table initialization info */ 1757 struct ext4_li_request *s_li_request; 1758 /* Wait multiplier for lazy initialization thread */ 1759 unsigned int s_li_wait_mult; 1760 1761 /* Kernel thread for multiple mount protection */ 1762 struct task_struct *s_mmp_tsk; 1763 1764 /* record the last minlen when FITRIM is called. */ 1765 unsigned long s_last_trim_minblks; 1766 1767 /* minimum folio order of a page cache allocation */ 1768 u16 s_min_folio_order; 1769 /* supported maximum folio order, 0 means not supported */ 1770 u16 s_max_folio_order; 1771 1772 /* Precomputed FS UUID checksum for seeding other checksums */ 1773 __u32 s_csum_seed; 1774 1775 /* Reclaim extents from extent status tree */ 1776 struct shrinker *s_es_shrinker; 1777 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1778 long s_es_nr_inode; 1779 struct ext4_es_stats s_es_stats; 1780 /* ext4_get_link() reads the sbi in rcu pathwalk */ 1781 struct rcu_head s_rcu; 1782 struct mb_cache *s_ea_block_cache; 1783 struct mb_cache *s_ea_inode_cache; 1784 1785 /* Deferred iput for EA inodes to avoid lock ordering issues */ 1786 struct llist_head s_ea_inode_to_free; 1787 struct delayed_work s_ea_inode_work; 1788 1789 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1790 1791 /* Journal triggers for checksum computation */ 1792 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1793 1794 /* Ratelimit ext4 messages. */ 1795 struct ratelimit_state s_err_ratelimit_state; 1796 struct ratelimit_state s_warning_ratelimit_state; 1797 struct ratelimit_state s_msg_ratelimit_state; 1798 atomic_t s_warning_count; 1799 atomic_t s_msg_count; 1800 1801 /* Encryption policy for '-o test_dummy_encryption' */ 1802 struct fscrypt_dummy_policy s_dummy_enc_policy; 1803 1804 /* 1805 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1806 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1807 * DIOREAD_NOLOCK mount options on remount. 1808 */ 1809 struct percpu_rw_semaphore s_writepages_rwsem; 1810 struct dax_device *s_daxdev; 1811 u64 s_dax_part_off; 1812 #ifdef CONFIG_EXT4_DEBUG 1813 unsigned long s_simulate_fail; 1814 #endif 1815 /* Record the errseq of the backing block device */ 1816 errseq_t s_bdev_wb_err; 1817 spinlock_t s_bdev_wb_lock; 1818 1819 /* Information about errors that happened during this mount */ 1820 spinlock_t s_error_lock; 1821 int s_add_error_count; 1822 int s_first_error_code; 1823 __u32 s_first_error_line; 1824 __u32 s_first_error_ino; 1825 __u64 s_first_error_block; 1826 const char *s_first_error_func; 1827 time64_t s_first_error_time; 1828 int s_last_error_code; 1829 __u32 s_last_error_line; 1830 __u32 s_last_error_ino; 1831 __u64 s_last_error_block; 1832 const char *s_last_error_func; 1833 time64_t s_last_error_time; 1834 /* 1835 * If we are in a context where we cannot update the on-disk 1836 * superblock, we queue the work here. This is used to update 1837 * the error information in the superblock, and for periodic 1838 * updates of the superblock called from the commit callback 1839 * function. 1840 */ 1841 struct work_struct s_sb_upd_work; 1842 1843 /* Atomic write unit values in bytes */ 1844 unsigned int s_awu_min; 1845 unsigned int s_awu_max; 1846 1847 /* Ext4 fast commit sub transaction ID */ 1848 atomic_t s_fc_subtid; 1849 1850 /* 1851 * After commit starts, the main queue gets locked, and the further 1852 * updates get added in the staging queue. 1853 */ 1854 #define FC_Q_MAIN 0 1855 #define FC_Q_STAGING 1 1856 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1857 * that have data changes in them. 1858 */ 1859 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1860 unsigned int s_fc_bytes; 1861 /* 1862 * Main fast commit lock. This lock protects accesses to the 1863 * following fields: 1864 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1865 * 1866 * s_fc_lock can be taken from reclaim context (inode eviction) and is 1867 * thus reclaim unsafe. Use ext4_fc_lock()/ext4_fc_unlock() helpers 1868 * when acquiring / releasing the lock. 1869 */ 1870 struct mutex s_fc_lock; 1871 struct buffer_head *s_fc_bh; 1872 struct ext4_fc_stats s_fc_stats; 1873 struct ext4_fc_snap_stats s_fc_snap_stats; 1874 tid_t s_fc_ineligible_tid; 1875 #ifdef CONFIG_EXT4_DEBUG 1876 int s_fc_debug_max_replay; 1877 #endif 1878 struct ext4_fc_replay_state s_fc_replay_state; 1879 }; 1880 1881 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1882 { 1883 return sb->s_fs_info; 1884 } 1885 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1886 { 1887 return container_of(inode, struct ext4_inode_info, vfs_inode); 1888 } 1889 1890 static inline int ext4_writepages_down_read(struct super_block *sb) 1891 { 1892 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1893 return memalloc_nofs_save(); 1894 } 1895 1896 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1897 { 1898 memalloc_nofs_restore(ctx); 1899 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1900 } 1901 1902 static inline int ext4_writepages_down_write(struct super_block *sb) 1903 { 1904 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1905 return memalloc_nofs_save(); 1906 } 1907 1908 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1909 { 1910 memalloc_nofs_restore(ctx); 1911 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1912 } 1913 1914 static inline int ext4_fc_lock(struct super_block *sb) 1915 { 1916 mutex_lock(&EXT4_SB(sb)->s_fc_lock); 1917 return memalloc_nofs_save(); 1918 } 1919 1920 static inline void ext4_fc_unlock(struct super_block *sb, int ctx) 1921 { 1922 memalloc_nofs_restore(ctx); 1923 mutex_unlock(&EXT4_SB(sb)->s_fc_lock); 1924 } 1925 1926 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1927 { 1928 return ino == EXT4_ROOT_INO || 1929 (ino >= EXT4_FIRST_INO(sb) && 1930 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1931 } 1932 1933 static inline int ext4_get_resuid(struct ext4_super_block *es) 1934 { 1935 return le16_to_cpu(es->s_def_resuid) | 1936 le16_to_cpu(es->s_def_resuid_hi) << 16; 1937 } 1938 1939 static inline int ext4_get_resgid(struct ext4_super_block *es) 1940 { 1941 return le16_to_cpu(es->s_def_resgid) | 1942 le16_to_cpu(es->s_def_resgid_hi) << 16; 1943 } 1944 1945 /* 1946 * Returns: sbi->field[index] 1947 * Used to access an array element from the following sbi fields which require 1948 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1949 * - s_group_desc 1950 * - s_group_info 1951 * - s_flex_group 1952 */ 1953 #define sbi_array_rcu_deref(sbi, field, index) \ 1954 ({ \ 1955 typeof(*((sbi)->field)) _v; \ 1956 rcu_read_lock(); \ 1957 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1958 rcu_read_unlock(); \ 1959 _v; \ 1960 }) 1961 1962 /* 1963 * run-time mount flags 1964 */ 1965 enum { 1966 EXT4_MF_MNTDIR_SAMPLED, 1967 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1968 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */ 1969 }; 1970 1971 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1972 { 1973 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1974 } 1975 1976 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1977 { 1978 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1979 } 1980 1981 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1982 { 1983 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1984 } 1985 1986 1987 /* 1988 * Simulate_fail codes 1989 */ 1990 #define EXT4_SIM_BBITMAP_EIO 1 1991 #define EXT4_SIM_BBITMAP_CRC 2 1992 #define EXT4_SIM_IBITMAP_EIO 3 1993 #define EXT4_SIM_IBITMAP_CRC 4 1994 #define EXT4_SIM_INODE_EIO 5 1995 #define EXT4_SIM_INODE_CRC 6 1996 #define EXT4_SIM_DIRBLOCK_EIO 7 1997 #define EXT4_SIM_DIRBLOCK_CRC 8 1998 1999 static inline bool ext4_simulate_fail(struct super_block *sb, 2000 unsigned long code) 2001 { 2002 #ifdef CONFIG_EXT4_DEBUG 2003 struct ext4_sb_info *sbi = EXT4_SB(sb); 2004 2005 if (unlikely(sbi->s_simulate_fail == code)) { 2006 sbi->s_simulate_fail = 0; 2007 return true; 2008 } 2009 #endif 2010 return false; 2011 } 2012 2013 /* 2014 * Error number codes for s_{first,last}_error_errno 2015 * 2016 * Linux errno numbers are architecture specific, so we need to translate 2017 * them into something which is architecture independent. We don't define 2018 * codes for all errno's; just the ones which are most likely to be the cause 2019 * of an ext4_error() call. 2020 */ 2021 #define EXT4_ERR_UNKNOWN 1 2022 #define EXT4_ERR_EIO 2 2023 #define EXT4_ERR_ENOMEM 3 2024 #define EXT4_ERR_EFSBADCRC 4 2025 #define EXT4_ERR_EFSCORRUPTED 5 2026 #define EXT4_ERR_ENOSPC 6 2027 #define EXT4_ERR_ENOKEY 7 2028 #define EXT4_ERR_EROFS 8 2029 #define EXT4_ERR_EFBIG 9 2030 #define EXT4_ERR_EEXIST 10 2031 #define EXT4_ERR_ERANGE 11 2032 #define EXT4_ERR_EOVERFLOW 12 2033 #define EXT4_ERR_EBUSY 13 2034 #define EXT4_ERR_ENOTDIR 14 2035 #define EXT4_ERR_ENOTEMPTY 15 2036 #define EXT4_ERR_ESHUTDOWN 16 2037 #define EXT4_ERR_EFAULT 17 2038 2039 /* 2040 * Inode dynamic state flags 2041 */ 2042 enum { 2043 EXT4_STATE_NEW, /* inode is newly created */ 2044 EXT4_STATE_XATTR, /* has in-inode xattrs */ 2045 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 2046 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 2047 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 2048 EXT4_STATE_NEWENTRY, /* File just added to dir */ 2049 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 2050 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 2051 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 2052 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 2053 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 2054 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */ 2055 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 2056 EXT4_STATE_FC_REQUEUE, /* Inode modified during fast commit */ 2057 EXT4_STATE_BUFFERED_IOMAP, /* Inode use iomap for buffered IO */ 2058 EXT4_STATE_DISKSIZE_GROW_PENDING, 2059 /* Has zeroed EOF block straddles 2060 * i_disksize awaiting writeback */ 2061 }; 2062 2063 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 2064 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 2065 { \ 2066 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 2067 } \ 2068 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 2069 { \ 2070 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 2071 } \ 2072 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 2073 { \ 2074 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 2075 } 2076 2077 /* Add these declarations here only so that these functions can be 2078 * found by name. Otherwise, they are very hard to locate. */ 2079 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 2080 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 2081 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 2082 EXT4_INODE_BIT_FNS(flag, flags, 0) 2083 2084 /* Add these declarations here only so that these functions can be 2085 * found by name. Otherwise, they are very hard to locate. */ 2086 static inline int ext4_test_inode_state(struct inode *inode, int bit); 2087 static inline void ext4_set_inode_state(struct inode *inode, int bit); 2088 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 2089 static inline unsigned long *ext4_inode_state_wait_word(struct inode *inode); 2090 static inline int ext4_inode_state_wait_bit(int bit); 2091 #if (BITS_PER_LONG < 64) 2092 EXT4_INODE_BIT_FNS(state, state_flags, 0) 2093 2094 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 2095 { 2096 (ei)->i_state_flags = 0; 2097 } 2098 #else 2099 EXT4_INODE_BIT_FNS(state, flags, 32) 2100 2101 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 2102 { 2103 /* We depend on the fact that callers will set i_flags */ 2104 } 2105 #endif 2106 2107 static inline unsigned long *ext4_inode_state_wait_word(struct inode *inode) 2108 { 2109 #if (BITS_PER_LONG < 64) 2110 return &EXT4_I(inode)->i_state_flags; 2111 #else 2112 return &EXT4_I(inode)->i_flags; 2113 #endif 2114 } 2115 2116 static inline int ext4_inode_state_wait_bit(int bit) 2117 { 2118 #if (BITS_PER_LONG < 64) 2119 return bit; 2120 #else 2121 return bit + 32; 2122 #endif 2123 } 2124 #else 2125 /* Assume that user mode programs are passing in an ext4fs superblock, not 2126 * a kernel struct super_block. This will allow us to call the feature-test 2127 * macros from user land. */ 2128 #define EXT4_SB(sb) (sb) 2129 #endif 2130 2131 static inline bool ext4_verity_in_progress(struct inode *inode) 2132 { 2133 return IS_ENABLED(CONFIG_FS_VERITY) && 2134 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 2135 } 2136 2137 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 2138 2139 /* 2140 * Check whether the inode is tracked as orphan (either in orphan file or 2141 * orphan list). 2142 */ 2143 static inline bool ext4_inode_orphan_tracked(struct inode *inode) 2144 { 2145 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) || 2146 !list_empty(&EXT4_I(inode)->i_orphan); 2147 } 2148 2149 static inline struct mapping_metadata_bhs *ext4_i_metadata_bhs( 2150 struct inode *inode) 2151 { 2152 /* 2153 * i_metadata_bhs is set in ext4_inode_attach_mmb() using cmpxchg(). 2154 * We use READ_ONCE when accessing i_metadata_bhs to make sure we get 2155 * consistent view for all accesses. 2156 */ 2157 return READ_ONCE(EXT4_I(inode)->i_metadata_bhs); 2158 } 2159 2160 /* Whether the inode pass through the iomap infrastructure for buffered I/O */ 2161 static inline bool ext4_inode_buffered_iomap(struct inode *inode) 2162 { 2163 return ext4_test_inode_state(inode, EXT4_STATE_BUFFERED_IOMAP); 2164 } 2165 2166 /* 2167 * Codes for operating systems 2168 */ 2169 #define EXT4_OS_LINUX 0 2170 #define EXT4_OS_HURD 1 2171 #define EXT4_OS_MASIX 2 2172 #define EXT4_OS_FREEBSD 3 2173 #define EXT4_OS_LITES 4 2174 2175 /* 2176 * Revision levels 2177 */ 2178 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 2179 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 2180 2181 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 2182 2183 #define EXT4_GOOD_OLD_INODE_SIZE 128 2184 2185 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 2186 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 2187 #define EXT4_TIMESTAMP_MIN S32_MIN 2188 2189 /* 2190 * Feature set definitions 2191 */ 2192 2193 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 2194 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2195 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2196 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2197 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2198 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2199 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2200 /* 2201 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2202 * incompatible only if fast commit blocks are present in the FS. Since we 2203 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2204 * incompatible. We also have a JBD2 incompat feature, which gets set when 2205 * there are fast commit blocks present in the journal. 2206 */ 2207 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2208 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2209 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2210 2211 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2212 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2213 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2214 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2215 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2216 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2217 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2218 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2219 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2220 /* 2221 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2222 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2223 * all other data structures' checksums. However, the METADATA_CSUM and 2224 * GDT_CSUM bits are mutually exclusive. 2225 */ 2226 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2227 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2228 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2229 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2230 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2231 non-empty */ 2232 2233 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2234 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2235 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2236 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2237 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2238 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2239 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2240 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2241 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2242 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2243 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2244 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2245 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2246 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2247 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2248 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2249 2250 extern void ext4_update_dynamic_rev(struct super_block *sb); 2251 2252 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2253 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2254 { \ 2255 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2256 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2257 } \ 2258 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2259 { \ 2260 ext4_update_dynamic_rev(sb); \ 2261 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2262 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2263 } \ 2264 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2265 { \ 2266 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2267 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2268 } 2269 2270 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2271 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2272 { \ 2273 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2274 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2275 } \ 2276 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2277 { \ 2278 ext4_update_dynamic_rev(sb); \ 2279 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2280 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2281 } \ 2282 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2283 { \ 2284 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2285 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2286 } 2287 2288 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2289 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2290 { \ 2291 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2292 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2293 } \ 2294 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2295 { \ 2296 ext4_update_dynamic_rev(sb); \ 2297 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2298 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2299 } \ 2300 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2301 { \ 2302 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2303 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2304 } 2305 2306 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2307 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2308 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2309 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2310 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2311 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2312 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2313 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2314 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2315 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2316 2317 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2318 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2319 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2320 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2321 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2322 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2323 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2324 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2325 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2326 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2327 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2328 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2329 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2330 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2331 2332 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2333 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2334 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2335 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2336 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2337 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2338 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2339 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2340 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2341 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2342 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2343 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2344 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2345 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2346 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2347 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2348 2349 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2350 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2351 EXT4_FEATURE_INCOMPAT_META_BG) 2352 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2353 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2354 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2355 2356 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2357 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2358 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2359 EXT4_FEATURE_INCOMPAT_META_BG) 2360 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2361 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2362 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2363 2364 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2365 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2366 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2367 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2368 EXT4_FEATURE_INCOMPAT_META_BG| \ 2369 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2370 EXT4_FEATURE_INCOMPAT_64BIT| \ 2371 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2372 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2373 EXT4_FEATURE_INCOMPAT_MMP | \ 2374 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2375 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2376 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2377 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2378 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2379 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2380 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2381 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2382 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2383 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2384 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2385 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2386 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2387 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2388 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2389 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2390 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2391 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2392 2393 #define EXTN_FEATURE_FUNCS(ver) \ 2394 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2395 { \ 2396 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2397 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2398 } \ 2399 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2400 { \ 2401 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2402 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2403 } \ 2404 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2405 { \ 2406 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2407 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2408 } 2409 2410 EXTN_FEATURE_FUNCS(2) 2411 EXTN_FEATURE_FUNCS(3) 2412 EXTN_FEATURE_FUNCS(4) 2413 2414 static inline bool ext4_has_compat_features(struct super_block *sb) 2415 { 2416 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2417 } 2418 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2419 { 2420 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2421 } 2422 static inline bool ext4_has_incompat_features(struct super_block *sb) 2423 { 2424 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2425 } 2426 2427 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2428 2429 /* 2430 * Superblock flags 2431 */ 2432 enum { 2433 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */ 2434 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */ 2435 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */ 2436 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */ 2437 }; 2438 2439 static inline int ext4_forced_shutdown(struct super_block *sb) 2440 { 2441 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2442 } 2443 2444 static inline int ext4_emergency_ro(struct super_block *sb) 2445 { 2446 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags); 2447 } 2448 2449 static inline int ext4_emergency_state(struct super_block *sb) 2450 { 2451 if (unlikely(ext4_forced_shutdown(sb))) 2452 return -EIO; 2453 if (unlikely(ext4_emergency_ro(sb))) 2454 return -EROFS; 2455 return 0; 2456 } 2457 2458 /* 2459 * Default values for user and/or group using reserved blocks 2460 */ 2461 #define EXT4_DEF_RESUID 0 2462 #define EXT4_DEF_RESGID 0 2463 2464 /* 2465 * Default project ID 2466 */ 2467 #define EXT4_DEF_PROJID 0 2468 2469 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2470 2471 /* 2472 * Default mount options 2473 */ 2474 #define EXT4_DEFM_DEBUG 0x0001 2475 #define EXT4_DEFM_BSDGROUPS 0x0002 2476 #define EXT4_DEFM_XATTR_USER 0x0004 2477 #define EXT4_DEFM_ACL 0x0008 2478 #define EXT4_DEFM_UID16 0x0010 2479 #define EXT4_DEFM_JMODE 0x0060 2480 #define EXT4_DEFM_JMODE_DATA 0x0020 2481 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2482 #define EXT4_DEFM_JMODE_WBACK 0x0060 2483 #define EXT4_DEFM_NOBARRIER 0x0100 2484 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2485 #define EXT4_DEFM_DISCARD 0x0400 2486 #define EXT4_DEFM_NODELALLOC 0x0800 2487 2488 /* 2489 * Default journal batch times and ioprio. 2490 */ 2491 #define EXT4_DEF_MIN_BATCH_TIME 0 2492 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2493 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) 2494 2495 2496 /* 2497 * Default values for superblock update 2498 */ 2499 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */ 2500 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */ 2501 2502 /* 2503 * Minimum number of groups in a flexgroup before we separate out 2504 * directories into the first block group of a flexgroup 2505 */ 2506 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2507 2508 /* 2509 * Structure of a directory entry 2510 */ 2511 #define EXT4_NAME_LEN 255 2512 /* 2513 * Base length of the ext4 directory entry excluding the name length 2514 */ 2515 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2516 2517 struct ext4_dir_entry { 2518 __le32 inode; /* Inode number */ 2519 __le16 rec_len; /* Directory entry length */ 2520 __le16 name_len; /* Name length */ 2521 char name[EXT4_NAME_LEN]; /* File name */ 2522 }; 2523 2524 2525 /* 2526 * Encrypted Casefolded entries require saving the hash on disk. This structure 2527 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2528 * boundary. 2529 */ 2530 struct ext4_dir_entry_hash { 2531 __le32 hash; 2532 __le32 minor_hash; 2533 }; 2534 2535 /* 2536 * The new version of the directory entry. Since EXT4 structures are 2537 * stored in intel byte order, and the name_len field could never be 2538 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2539 * file_type field. 2540 */ 2541 struct ext4_dir_entry_2 { 2542 __le32 inode; /* Inode number */ 2543 __le16 rec_len; /* Directory entry length */ 2544 __u8 name_len; /* Name length */ 2545 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2546 char name[EXT4_NAME_LEN]; /* File name */ 2547 }; 2548 2549 /* 2550 * Access the hashes at the end of ext4_dir_entry_2 2551 */ 2552 #define EXT4_DIRENT_HASHES(entry) \ 2553 ((struct ext4_dir_entry_hash *) \ 2554 (((void *)(entry)) + \ 2555 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2556 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash) 2557 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2558 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash) 2559 2560 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2561 { 2562 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2563 } 2564 2565 /* 2566 * This is a bogus directory entry at the end of each leaf block that 2567 * records checksums. 2568 */ 2569 struct ext4_dir_entry_tail { 2570 __le32 det_reserved_zero1; /* Pretend to be unused */ 2571 __le16 det_rec_len; /* 12 */ 2572 __u8 det_reserved_zero2; /* Zero name length */ 2573 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2574 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2575 }; 2576 2577 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2578 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2579 ((blocksize) - \ 2580 sizeof(struct ext4_dir_entry_tail)))) 2581 2582 /* 2583 * Ext4 directory file types. Only the low 3 bits are used. The 2584 * other bits are reserved for now. 2585 */ 2586 #define EXT4_FT_UNKNOWN 0 2587 #define EXT4_FT_REG_FILE 1 2588 #define EXT4_FT_DIR 2 2589 #define EXT4_FT_CHRDEV 3 2590 #define EXT4_FT_BLKDEV 4 2591 #define EXT4_FT_FIFO 5 2592 #define EXT4_FT_SOCK 6 2593 #define EXT4_FT_SYMLINK 7 2594 2595 #define EXT4_FT_MAX 8 2596 2597 #define EXT4_FT_DIR_CSUM 0xDE 2598 2599 /* 2600 * EXT4_DIR_PAD defines the directory entries boundaries 2601 * 2602 * NOTE: It must be a multiple of 4 2603 */ 2604 #define EXT4_DIR_PAD 4 2605 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2606 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2607 2608 /* 2609 * The rec_len is dependent on the type of directory. Directories that are 2610 * casefolded and encrypted need to store the hash as well, so we add room for 2611 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2612 * pass NULL for dir, as those entries do not use the extra fields. 2613 */ 2614 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2615 const struct inode *dir) 2616 { 2617 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2618 2619 if (dir && ext4_hash_in_dirent(dir)) 2620 rec_len += sizeof(struct ext4_dir_entry_hash); 2621 return (rec_len & ~EXT4_DIR_ROUND); 2622 } 2623 2624 static inline unsigned int 2625 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2626 { 2627 unsigned len = le16_to_cpu(dlen); 2628 2629 if (len == EXT4_MAX_REC_LEN || len == 0) 2630 return blocksize; 2631 return (len & 65532) | ((len & 3) << 16); 2632 } 2633 2634 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2635 { 2636 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2637 if (len < 65536) 2638 return cpu_to_le16(len); 2639 if (len == blocksize) { 2640 if (blocksize == 65536) 2641 return cpu_to_le16(EXT4_MAX_REC_LEN); 2642 else 2643 return cpu_to_le16(0); 2644 } 2645 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2646 } 2647 2648 /* 2649 * Hash Tree Directory indexing 2650 * (c) Daniel Phillips, 2001 2651 */ 2652 2653 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2654 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2655 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2656 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2657 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2658 2659 /* Legal values for the dx_root hash_version field: */ 2660 2661 #define DX_HASH_LEGACY 0 2662 #define DX_HASH_HALF_MD4 1 2663 #define DX_HASH_TEA 2 2664 #define DX_HASH_LEGACY_UNSIGNED 3 2665 #define DX_HASH_HALF_MD4_UNSIGNED 4 2666 #define DX_HASH_TEA_UNSIGNED 5 2667 #define DX_HASH_SIPHASH 6 2668 #define DX_HASH_LAST DX_HASH_SIPHASH 2669 2670 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length) 2671 { 2672 return crc32c(crc, address, length); 2673 } 2674 2675 #ifdef __KERNEL__ 2676 2677 /* hash info structure used by the directory hash */ 2678 struct dx_hash_info 2679 { 2680 u32 hash; 2681 u32 minor_hash; 2682 int hash_version; 2683 u32 *seed; 2684 }; 2685 2686 2687 /* 32 and 64 bit signed EOF for dx directories */ 2688 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2689 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2690 2691 2692 /* 2693 * Control parameters used by ext4_htree_next_block 2694 */ 2695 #define HASH_NB_ALWAYS 1 2696 2697 struct ext4_filename { 2698 const struct qstr *usr_fname; 2699 struct fscrypt_str disk_name; 2700 struct dx_hash_info hinfo; 2701 #ifdef CONFIG_FS_ENCRYPTION 2702 struct fscrypt_str crypto_buf; 2703 #endif 2704 #if IS_ENABLED(CONFIG_UNICODE) 2705 struct qstr cf_name; 2706 #endif 2707 }; 2708 2709 #define fname_name(p) ((p)->disk_name.name) 2710 #define fname_usr_name(p) ((p)->usr_fname->name) 2711 #define fname_len(p) ((p)->disk_name.len) 2712 2713 /* 2714 * Describe an inode's exact location on disk and in memory 2715 */ 2716 struct ext4_iloc 2717 { 2718 struct buffer_head *bh; 2719 unsigned long offset; 2720 ext4_group_t block_group; 2721 }; 2722 2723 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2724 { 2725 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2726 } 2727 2728 static inline bool ext4_is_quota_file(struct inode *inode) 2729 { 2730 return IS_NOQUOTA(inode) && 2731 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2732 } 2733 2734 /* 2735 * This structure is stuffed into the struct file's private_data field 2736 * for directories. It is where we put information so that we can do 2737 * readdir operations in hash tree order. 2738 */ 2739 struct dir_private_info { 2740 struct rb_root root; 2741 struct rb_node *curr_node; 2742 struct fname *extra_fname; 2743 loff_t last_pos; 2744 __u32 curr_hash; 2745 __u32 curr_minor_hash; 2746 __u32 next_hash; 2747 u64 cookie; 2748 bool initialized; 2749 }; 2750 2751 /* calculate the first block number of the group */ 2752 static inline ext4_fsblk_t 2753 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2754 { 2755 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2756 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2757 } 2758 2759 /* 2760 * Special error return code only used by dx_probe() and its callers. 2761 */ 2762 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2763 2764 /* htree levels for ext4 */ 2765 #define EXT4_HTREE_LEVEL_COMPAT 2 2766 #define EXT4_HTREE_LEVEL 3 2767 2768 static inline int ext4_dir_htree_level(struct super_block *sb) 2769 { 2770 return ext4_has_feature_largedir(sb) ? 2771 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2772 } 2773 2774 /* 2775 * Timeout and state flag for lazy initialization inode thread. 2776 */ 2777 #define EXT4_DEF_LI_WAIT_MULT 10 2778 #define EXT4_DEF_LI_MAX_START_DELAY 5 2779 #define EXT4_LAZYINIT_QUIT 0x0001 2780 #define EXT4_LAZYINIT_RUNNING 0x0002 2781 2782 /* 2783 * Lazy inode table initialization info 2784 */ 2785 struct ext4_lazy_init { 2786 unsigned long li_state; 2787 struct list_head li_request_list; 2788 struct mutex li_list_mtx; 2789 }; 2790 2791 enum ext4_li_mode { 2792 EXT4_LI_MODE_PREFETCH_BBITMAP, 2793 EXT4_LI_MODE_ITABLE, 2794 }; 2795 2796 struct ext4_li_request { 2797 struct super_block *lr_super; 2798 enum ext4_li_mode lr_mode; 2799 ext4_group_t lr_first_not_zeroed; 2800 ext4_group_t lr_next_group; 2801 struct list_head lr_request; 2802 unsigned long lr_next_sched; 2803 unsigned long lr_timeout; 2804 }; 2805 2806 struct ext4_features { 2807 struct kobject f_kobj; 2808 struct completion f_kobj_unregister; 2809 }; 2810 2811 /* 2812 * This structure will be used for multiple mount protection. It will be 2813 * written into the block number saved in the s_mmp_block field in the 2814 * superblock. Programs that check MMP should assume that if 2815 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2816 * to use the filesystem, regardless of how old the timestamp is. 2817 */ 2818 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2819 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2820 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2821 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2822 2823 struct mmp_struct { 2824 __le32 mmp_magic; /* Magic number for MMP */ 2825 __le32 mmp_seq; /* Sequence no. updated periodically */ 2826 2827 /* 2828 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2829 * purposes and do not affect the correctness of the algorithm 2830 */ 2831 __le64 mmp_time; /* Time last updated */ 2832 char mmp_nodename[64]; /* Node which last updated MMP block */ 2833 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2834 2835 /* 2836 * mmp_check_interval is used to verify if the MMP block has been 2837 * updated on the block device. The value is updated based on the 2838 * maximum time to write the MMP block during an update cycle. 2839 */ 2840 __le16 mmp_check_interval; 2841 2842 __le16 mmp_pad1; 2843 __le32 mmp_pad2[226]; 2844 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2845 }; 2846 2847 /* arguments passed to the mmp thread */ 2848 struct mmpd_data { 2849 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2850 struct super_block *sb; /* super block of the fs */ 2851 }; 2852 2853 /* 2854 * Check interval multiplier 2855 * The MMP block is written every update interval and initially checked every 2856 * update interval x the multiplier (the value is then adapted based on the 2857 * write latency). The reason is that writes can be delayed under load and we 2858 * don't want readers to incorrectly assume that the filesystem is no longer 2859 * in use. 2860 */ 2861 #define EXT4_MMP_CHECK_MULT 2UL 2862 2863 /* 2864 * Minimum interval for MMP checking in seconds. 2865 */ 2866 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2867 2868 /* 2869 * Maximum interval for MMP checking in seconds. 2870 */ 2871 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2872 2873 /* 2874 * Function prototypes 2875 */ 2876 2877 /* 2878 * Ok, these declarations are also in <linux/kernel.h> but none of the 2879 * ext4 source programs needs to include it so they are duplicated here. 2880 */ 2881 # define NORET_TYPE /**/ 2882 # define ATTRIB_NORET __attribute__((noreturn)) 2883 # define NORET_AND noreturn, 2884 2885 /* bitmap.c */ 2886 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2887 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2888 struct ext4_group_desc *gdp, 2889 struct buffer_head *bh); 2890 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2891 struct ext4_group_desc *gdp, 2892 struct buffer_head *bh); 2893 void ext4_block_bitmap_csum_set(struct super_block *sb, 2894 struct ext4_group_desc *gdp, 2895 struct buffer_head *bh); 2896 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2897 struct ext4_group_desc *gdp, 2898 struct buffer_head *bh); 2899 2900 /* balloc.c */ 2901 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2902 ext4_fsblk_t blocknr, 2903 ext4_group_t *blockgrpp, 2904 ext4_grpblk_t *offsetp); 2905 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2906 ext4_fsblk_t block); 2907 2908 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2909 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2910 ext4_group_t group); 2911 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2912 ext4_fsblk_t goal, 2913 unsigned int flags, 2914 unsigned long *count, 2915 int *errp); 2916 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2917 s64 nclusters, unsigned int flags); 2918 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2919 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2920 ext4_group_t block_group, 2921 struct buffer_head ** bh); 2922 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2923 ext4_group_t group); 2924 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2925 2926 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2927 ext4_group_t block_group, 2928 bool ignore_locked); 2929 extern int ext4_wait_block_bitmap(struct super_block *sb, 2930 ext4_group_t block_group, 2931 struct buffer_head *bh); 2932 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2933 ext4_group_t block_group); 2934 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2935 ext4_group_t block_group, 2936 struct ext4_group_desc *gdp); 2937 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2938 2939 #if IS_ENABLED(CONFIG_UNICODE) 2940 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2941 const struct qstr *iname, 2942 struct ext4_filename *fname); 2943 2944 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2945 { 2946 kfree(fname->cf_name.name); 2947 fname->cf_name.name = NULL; 2948 } 2949 #else 2950 static inline int ext4_fname_setup_ci_filename(struct inode *dir, 2951 const struct qstr *iname, 2952 struct ext4_filename *fname) 2953 { 2954 return 0; 2955 } 2956 2957 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2958 { 2959 } 2960 #endif 2961 2962 /* ext4 encryption related stuff goes here crypto.c */ 2963 #ifdef CONFIG_FS_ENCRYPTION 2964 extern const struct fscrypt_operations ext4_cryptops; 2965 2966 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2967 int lookup, struct ext4_filename *fname); 2968 2969 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2970 struct ext4_filename *fname); 2971 2972 void ext4_fname_free_filename(struct ext4_filename *fname); 2973 2974 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2975 2976 #else /* !CONFIG_FS_ENCRYPTION */ 2977 static inline int ext4_fname_setup_filename(struct inode *dir, 2978 const struct qstr *iname, 2979 int lookup, 2980 struct ext4_filename *fname) 2981 { 2982 fname->usr_fname = iname; 2983 fname->disk_name.name = (unsigned char *) iname->name; 2984 fname->disk_name.len = iname->len; 2985 2986 return ext4_fname_setup_ci_filename(dir, iname, fname); 2987 } 2988 2989 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2990 struct dentry *dentry, 2991 struct ext4_filename *fname) 2992 { 2993 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2994 } 2995 2996 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2997 { 2998 ext4_fname_free_ci_filename(fname); 2999 } 3000 3001 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 3002 void __user *arg) 3003 { 3004 return -EOPNOTSUPP; 3005 } 3006 #endif /* !CONFIG_FS_ENCRYPTION */ 3007 3008 /* dir.c */ 3009 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 3010 struct file *, 3011 struct ext4_dir_entry_2 *, 3012 struct buffer_head *, char *, int, 3013 unsigned int); 3014 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 3015 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 3016 (de), (bh), (buf), (size), (offset))) 3017 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 3018 __u32 minor_hash, 3019 struct ext4_dir_entry_2 *dirent, 3020 struct fscrypt_str *ent_name); 3021 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 3022 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh, 3023 void *buf, int buf_size, 3024 struct ext4_filename *fname, 3025 struct ext4_dir_entry_2 **dest_de); 3026 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 3027 struct ext4_dir_entry_2 *de, 3028 int buf_size, 3029 struct ext4_filename *fname); 3030 static inline void ext4_update_dx_flag(struct inode *inode) 3031 { 3032 if (!ext4_has_feature_dir_index(inode->i_sb) && 3033 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 3034 /* ext4_iget() should have caught this... */ 3035 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 3036 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 3037 } 3038 } 3039 static const unsigned char ext4_filetype_table[] = { 3040 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 3041 }; 3042 3043 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 3044 { 3045 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 3046 return DT_UNKNOWN; 3047 3048 return ext4_filetype_table[filetype]; 3049 } 3050 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 3051 void *buf, int buf_size); 3052 3053 /* fsync.c */ 3054 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 3055 3056 /* hash.c */ 3057 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 3058 struct dx_hash_info *hinfo); 3059 3060 /* ialloc.c */ 3061 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 3062 extern struct inode *__ext4_new_inode(const struct mnt_idmap *, handle_t *, 3063 struct inode *, umode_t, 3064 const struct qstr *qstr, __u32 goal, 3065 uid_t *owner, __u32 i_flags, 3066 int handle_type, unsigned int line_no, 3067 int nblocks); 3068 3069 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 3070 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 3071 (goal), (owner), i_flags, 0, 0, 0) 3072 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 3073 type, nblocks) \ 3074 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 3075 0, (type), __LINE__, (nblocks)) 3076 3077 3078 extern void ext4_free_inode(handle_t *, struct inode *); 3079 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 3080 extern unsigned long ext4_count_free_inodes(struct super_block *); 3081 extern unsigned long ext4_count_dirs(struct super_block *); 3082 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 3083 extern int ext4_init_inode_table(struct super_block *sb, 3084 ext4_group_t group, int barrier); 3085 void ext4_end_bitmap_read(struct bio *bio); 3086 3087 /* fast_commit.c */ 3088 int ext4_fc_info_show(struct seq_file *seq, void *v); 3089 void ext4_fc_init(struct super_block *sb, journal_t *journal); 3090 void ext4_fc_init_inode(struct inode *inode); 3091 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 3092 ext4_lblk_t end); 3093 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 3094 struct dentry *dentry); 3095 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 3096 struct dentry *dentry); 3097 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 3098 void ext4_fc_track_link(handle_t *handle, struct inode *inode, 3099 struct dentry *dentry); 3100 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 3101 struct dentry *dentry); 3102 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 3103 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 3104 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 3105 void ext4_fc_del(struct inode *inode); 3106 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 3107 void ext4_fc_replay_cleanup(struct super_block *sb); 3108 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 3109 int __init ext4_fc_init_dentry_cache(void); 3110 void ext4_fc_destroy_dentry_cache(void); 3111 int ext4_fc_record_regions(struct super_block *sb, int ino, 3112 ext4_lblk_t lblk, ext4_fsblk_t pblk, 3113 int len, int replay); 3114 3115 /* mballoc.c */ 3116 extern const struct seq_operations ext4_mb_seq_groups_ops; 3117 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 3118 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 3119 extern int ext4_mb_init(struct super_block *); 3120 extern void ext4_mb_release(struct super_block *); 3121 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 3122 struct ext4_allocation_request *, int *); 3123 extern void ext4_discard_preallocations(struct inode *); 3124 extern int __init ext4_init_mballoc(void); 3125 extern void ext4_exit_mballoc(void); 3126 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 3127 ext4_group_t group, 3128 unsigned int nr, int *cnt); 3129 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 3130 unsigned int nr); 3131 3132 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 3133 struct buffer_head *bh, ext4_fsblk_t block, 3134 unsigned long count, int flags); 3135 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 3136 ext4_group_t ngroups); 3137 extern int ext4_mb_add_groupinfo(struct super_block *sb, 3138 ext4_group_t i, struct ext4_group_desc *desc); 3139 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 3140 ext4_fsblk_t block, unsigned long count); 3141 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 3142 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 3143 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 3144 int len, bool state); 3145 static inline bool ext4_mb_cr_expensive(enum criteria cr) 3146 { 3147 return cr >= CR_GOAL_LEN_SLOW; 3148 } 3149 3150 /* inode.c */ 3151 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 3152 struct ext4_inode_info *ei); 3153 int ext4_inode_is_fast_symlink(struct inode *inode); 3154 void ext4_check_map_extents_env(struct inode *inode); 3155 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 3156 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 3157 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 3158 bool wait, struct buffer_head **bhs); 3159 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 3160 struct buffer_head *bh_result, int create); 3161 int ext4_get_block(struct inode *inode, sector_t iblock, 3162 struct buffer_head *bh_result, int create); 3163 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 3164 struct buffer_head *bh, int create); 3165 int ext4_walk_page_buffers(handle_t *handle, 3166 struct inode *inode, 3167 struct buffer_head *head, 3168 unsigned from, 3169 unsigned to, 3170 int *partial, 3171 int (*fn)(handle_t *handle, struct inode *inode, 3172 struct buffer_head *bh)); 3173 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 3174 struct buffer_head *bh); 3175 void ext4_set_inode_mapping_order(struct inode *inode); 3176 void ext4_enable_buffered_iomap(struct inode *inode); 3177 int ext4_nonda_switch(struct super_block *sb); 3178 #define FALL_BACK_TO_NONDELALLOC 1 3179 #define EXT4_WRITE_DATA_INLINE 2 3180 3181 typedef enum { 3182 EXT4_IGET_NORMAL = 0, 3183 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3184 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 3185 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 3186 EXT4_IGET_EA_INODE = 0x0008, /* Inode should contain an EA value */ 3187 EXT4_IGET_NOWAIT = 0x0010 /* Non-blocking lookup (skip if freeing) */ 3188 } ext4_iget_flags; 3189 3190 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3191 ext4_iget_flags flags, const char *function, 3192 unsigned int line); 3193 3194 #define ext4_iget(sb, ino, flags) \ 3195 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3196 3197 extern int ext4_write_inode(struct inode *, struct writeback_control *); 3198 extern int ext4_sync_inode_metadata(struct inode *, struct writeback_control *); 3199 extern int ext4_setattr(const struct mnt_idmap *, struct dentry *, 3200 struct iattr *); 3201 extern u32 ext4_dio_alignment(struct inode *inode); 3202 extern int ext4_getattr(const struct mnt_idmap *, const struct path *, 3203 struct kstat *, u32, unsigned int); 3204 extern void ext4_evict_inode(struct inode *); 3205 extern void ext4_clear_inode(struct inode *); 3206 extern int ext4_file_getattr(const struct mnt_idmap *, const struct path *, 3207 struct kstat *, u32, unsigned int); 3208 extern void ext4_dirty_inode(struct inode *, int); 3209 extern int ext4_change_inode_journal_flag(struct inode *, int); 3210 int ext4_get_inode_loc(struct inode *inode, struct ext4_iloc *iloc); 3211 int ext4_get_inode_loc_noio(struct inode *inode, struct ext4_iloc *iloc); 3212 int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3213 struct ext4_iloc *iloc); 3214 extern int ext4_inode_attach_jinode(struct inode *inode); 3215 extern int ext4_can_truncate(struct inode *inode); 3216 extern int ext4_truncate(struct inode *); 3217 extern int ext4_break_layouts(struct inode *); 3218 extern int ext4_truncate_page_cache_block_range(struct inode *inode, 3219 loff_t start, loff_t end); 3220 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3221 extern void ext4_set_inode_flags(struct inode *, bool init); 3222 extern int ext4_alloc_da_blocks(struct inode *inode); 3223 extern void ext4_set_aops(struct inode *inode); 3224 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3225 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3226 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks); 3227 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks, 3228 int pextents, int alloc_extents); 3229 void ext4_iomap_clear_disksize_pending(struct inode *inode); 3230 void ext4_iomap_wait_disksize_pending(struct inode *inode); 3231 unsigned int ext4_iomap_get_disksize_pending_range(struct inode *inode, 3232 loff_t *start); 3233 extern int ext4_iomap_sync_zeroed_eof(struct inode *inode, 3234 loff_t offset, loff_t end); 3235 extern int ext4_block_zero_eof(struct inode *inode, loff_t from, loff_t end); 3236 3237 #define EXT4_PARTIAL_ZERO_START 0x1 3238 #define EXT4_PARTIAL_ZERO_END 0x2 3239 extern int ext4_zero_partial_blocks(struct inode *inode, loff_t lstart, 3240 loff_t length, unsigned int *partial_zeroed); 3241 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3242 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3243 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3244 extern void ext4_da_release_space(struct inode *inode, int to_free); 3245 extern void ext4_da_update_reserve_space(struct inode *inode, 3246 int used, int quota_claim); 3247 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3248 ext4_fsblk_t pblk, ext4_lblk_t len); 3249 3250 static inline bool is_special_ino(struct super_block *sb, unsigned long ino) 3251 { 3252 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 3253 3254 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) || 3255 ino == le32_to_cpu(es->s_usr_quota_inum) || 3256 ino == le32_to_cpu(es->s_grp_quota_inum) || 3257 ino == le32_to_cpu(es->s_prj_quota_inum) || 3258 ino == le32_to_cpu(es->s_orphan_file_inum); 3259 } 3260 3261 /* indirect.c */ 3262 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3263 struct ext4_map_blocks *map, int flags); 3264 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3265 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3266 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3267 ext4_lblk_t start, ext4_lblk_t end); 3268 3269 /* ioctl.c */ 3270 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3271 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3272 int ext4_fileattr_set(const struct mnt_idmap *idmap, 3273 struct dentry *dentry, struct file_kattr *fa); 3274 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3275 extern void ext4_reset_inode_seed(struct inode *inode); 3276 int ext4_update_overhead(struct super_block *sb, bool force); 3277 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3278 3279 /* migrate.c */ 3280 extern int ext4_ext_migrate(struct inode *); 3281 extern int ext4_ind_migrate(struct inode *inode); 3282 3283 /* namei.c */ 3284 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3285 struct inode *inode); 3286 extern int ext4_dirblock_csum_verify(struct inode *inode, 3287 struct buffer_head *bh); 3288 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3289 __u32 start_minor_hash, __u32 *next_hash); 3290 extern int ext4_search_dir(struct buffer_head *bh, 3291 char *search_buf, 3292 int buf_size, 3293 struct inode *dir, 3294 struct ext4_filename *fname, 3295 unsigned int offset, 3296 struct ext4_dir_entry_2 **res_dir); 3297 extern int ext4_generic_delete_entry(struct inode *dir, 3298 struct ext4_dir_entry_2 *de_del, 3299 struct buffer_head *bh, 3300 void *entry_buf, 3301 int buf_size, 3302 int csum_size); 3303 extern bool ext4_empty_dir(struct inode *inode); 3304 3305 /* resize.c */ 3306 extern void ext4_kvfree_array_rcu(void *to_free); 3307 extern int ext4_group_add(struct super_block *sb, 3308 struct ext4_new_group_data *input); 3309 extern int ext4_group_extend(struct super_block *sb, 3310 struct ext4_super_block *es, 3311 ext4_fsblk_t n_blocks_count); 3312 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3313 extern unsigned int ext4_list_backups(struct super_block *sb, 3314 unsigned int *three, unsigned int *five, 3315 unsigned int *seven); 3316 3317 /* super.c */ 3318 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3319 sector_t block, blk_opf_t op_flags); 3320 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3321 sector_t block); 3322 extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb, 3323 sector_t block); 3324 void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3325 bio_end_io_t end_io, bool simu_fail); 3326 int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3327 bio_end_io_t end_io, bool simu_fail); 3328 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3329 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3330 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3331 extern int ext4_calculate_overhead(struct super_block *sb); 3332 extern __le32 ext4_superblock_csum(struct ext4_super_block *es); 3333 extern void ext4_superblock_csum_set(struct super_block *sb); 3334 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3335 ext4_group_t ngroup); 3336 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3337 char nbuf[16]); 3338 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3339 ext4_group_t block_group, 3340 unsigned int flags); 3341 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3342 ext4_group_t block_group); 3343 extern void print_daily_error_info(struct timer_list *t); 3344 3345 extern __printf(7, 8) 3346 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3347 int, __u64, const char *, ...); 3348 extern __printf(6, 7) 3349 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3350 ext4_fsblk_t, int, const char *, ...); 3351 extern __printf(5, 6) 3352 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3353 const char *, ...); 3354 extern void __ext4_std_error(struct super_block *, const char *, 3355 unsigned int, int); 3356 extern __printf(4, 5) 3357 void __ext4_warning(struct super_block *, const char *, unsigned int, 3358 const char *, ...); 3359 extern __printf(4, 5) 3360 void __ext4_warning_inode(const struct inode *inode, const char *function, 3361 unsigned int line, const char *fmt, ...); 3362 extern __printf(3, 4) 3363 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3364 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3365 const char *, unsigned int, const char *); 3366 extern __printf(7, 8) 3367 void __ext4_grp_locked_error(const char *, unsigned int, 3368 struct super_block *, ext4_group_t, 3369 u64, ext4_fsblk_t, 3370 const char *, ...); 3371 3372 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3373 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3374 3375 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3376 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3377 3378 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3379 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3380 (fmt), ## a) 3381 3382 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3383 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3384 3385 #define ext4_abort(sb, err, fmt, a...) \ 3386 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3387 3388 #ifdef CONFIG_PRINTK 3389 3390 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3391 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3392 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3393 __ext4_error_inode((inode), (func), (line), (block), \ 3394 (err), (fmt), ##__VA_ARGS__) 3395 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3396 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3397 #define ext4_error(sb, fmt, ...) \ 3398 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3399 ##__VA_ARGS__) 3400 #define ext4_error_err(sb, err, fmt, ...) \ 3401 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3402 ##__VA_ARGS__) 3403 #define ext4_warning(sb, fmt, ...) \ 3404 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3405 #define ext4_warning_inode(inode, fmt, ...) \ 3406 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3407 #define ext4_msg(sb, level, fmt, ...) \ 3408 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3409 #define dump_mmp_msg(sb, mmp, msg) \ 3410 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3411 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3412 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3413 fmt, ##__VA_ARGS__) 3414 3415 #else 3416 3417 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3418 do { \ 3419 no_printk(fmt, ##__VA_ARGS__); \ 3420 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3421 } while (0) 3422 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3423 do { \ 3424 no_printk(fmt, ##__VA_ARGS__); \ 3425 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3426 } while (0) 3427 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3428 do { \ 3429 no_printk(fmt, ##__VA_ARGS__); \ 3430 __ext4_error_file(file, "", 0, block, " "); \ 3431 } while (0) 3432 #define ext4_error(sb, fmt, ...) \ 3433 do { \ 3434 no_printk(fmt, ##__VA_ARGS__); \ 3435 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3436 } while (0) 3437 #define ext4_error_err(sb, err, fmt, ...) \ 3438 do { \ 3439 no_printk(fmt, ##__VA_ARGS__); \ 3440 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3441 } while (0) 3442 #define ext4_warning(sb, fmt, ...) \ 3443 do { \ 3444 no_printk(fmt, ##__VA_ARGS__); \ 3445 __ext4_warning(sb, "", 0, " "); \ 3446 } while (0) 3447 #define ext4_warning_inode(inode, fmt, ...) \ 3448 do { \ 3449 no_printk(fmt, ##__VA_ARGS__); \ 3450 __ext4_warning_inode(inode, "", 0, " "); \ 3451 } while (0) 3452 #define ext4_msg(sb, level, fmt, ...) \ 3453 do { \ 3454 no_printk(fmt, ##__VA_ARGS__); \ 3455 __ext4_msg(sb, "", " "); \ 3456 } while (0) 3457 #define dump_mmp_msg(sb, mmp, msg) \ 3458 __dump_mmp_msg(sb, mmp, "", 0, "") 3459 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3460 do { \ 3461 no_printk(fmt, ##__VA_ARGS__); \ 3462 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3463 } while (0) 3464 3465 #endif 3466 3467 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3468 struct ext4_group_desc *bg); 3469 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3470 struct ext4_group_desc *bg); 3471 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3472 struct ext4_group_desc *bg); 3473 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3474 struct ext4_group_desc *bg); 3475 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3476 struct ext4_group_desc *bg); 3477 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3478 struct ext4_group_desc *bg); 3479 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3480 struct ext4_group_desc *bg); 3481 extern void ext4_block_bitmap_set(struct super_block *sb, 3482 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3483 extern void ext4_inode_bitmap_set(struct super_block *sb, 3484 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3485 extern void ext4_inode_table_set(struct super_block *sb, 3486 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3487 extern void ext4_free_group_clusters_set(struct super_block *sb, 3488 struct ext4_group_desc *bg, 3489 __u32 count); 3490 extern void ext4_free_inodes_set(struct super_block *sb, 3491 struct ext4_group_desc *bg, __u32 count); 3492 extern void ext4_used_dirs_set(struct super_block *sb, 3493 struct ext4_group_desc *bg, __u32 count); 3494 extern void ext4_itable_unused_set(struct super_block *sb, 3495 struct ext4_group_desc *bg, __u32 count); 3496 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3497 struct ext4_group_desc *gdp); 3498 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3499 struct ext4_group_desc *gdp); 3500 extern int ext4_register_li_request(struct super_block *sb, 3501 ext4_group_t first_not_zeroed); 3502 3503 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3504 { 3505 return ext4_has_feature_gdt_csum(sb) || 3506 ext4_has_feature_metadata_csum(sb); 3507 } 3508 3509 #define ext4_read_incompat_64bit_val(es, name) \ 3510 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3511 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3512 le32_to_cpu(es->name##_lo)) 3513 3514 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3515 { 3516 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3517 } 3518 3519 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3520 { 3521 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3522 } 3523 3524 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3525 { 3526 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3527 } 3528 3529 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3530 ext4_fsblk_t blk) 3531 { 3532 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3533 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3534 } 3535 3536 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3537 ext4_fsblk_t blk) 3538 { 3539 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3540 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3541 } 3542 3543 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3544 ext4_fsblk_t blk) 3545 { 3546 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3547 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3548 } 3549 3550 static inline loff_t ext4_isize(struct super_block *sb, 3551 struct ext4_inode *raw_inode) 3552 { 3553 if (ext4_has_feature_largedir(sb) || 3554 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3555 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3556 le32_to_cpu(raw_inode->i_size_lo); 3557 3558 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3559 } 3560 3561 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3562 { 3563 raw_inode->i_size_lo = cpu_to_le32(i_size); 3564 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3565 } 3566 3567 /* 3568 * Reading s_groups_count uses acquire semantics. 3569 */ 3570 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3571 { 3572 /* Pairs with smp_store_release() in ext4_update_super() */ 3573 return smp_load_acquire(&EXT4_SB(sb)->s_groups_count); 3574 } 3575 3576 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3577 ext4_group_t block_group) 3578 { 3579 return block_group >> sbi->s_log_groups_per_flex; 3580 } 3581 3582 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3583 { 3584 return 1 << sbi->s_log_groups_per_flex; 3585 } 3586 3587 static inline loff_t ext4_get_maxbytes(struct inode *inode) 3588 { 3589 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) 3590 return inode->i_sb->s_maxbytes; 3591 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; 3592 } 3593 3594 #define ext4_std_error(sb, errno) \ 3595 do { \ 3596 if ((errno)) \ 3597 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3598 } while (0) 3599 3600 #ifdef CONFIG_SMP 3601 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3602 * counters. So we need to make sure we have free clusters more 3603 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3604 */ 3605 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3606 #else 3607 #define EXT4_FREECLUSTERS_WATERMARK 0 3608 #endif 3609 3610 /* 3611 * Update i_disksize. Requires i_rwsem to avoid races with truncate. 3612 * 3613 * In the iomap buffered I/O path, the EXT4_STATE_DISKSIZE_GROW_PENDING 3614 * inode state bit indicates that the zeroed EOF partial block which 3615 * straddles i_disksize is still waiting writeback. In that case, 3616 * i_disksize will be updated after the pending zeroed data has been 3617 * written out. 3618 */ 3619 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3620 { 3621 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3622 !inode_is_locked(inode)); 3623 down_write(&EXT4_I(inode)->i_data_sem); 3624 if (newsize > EXT4_I(inode)->i_disksize && 3625 !ext4_test_inode_state(inode, EXT4_STATE_DISKSIZE_GROW_PENDING)) 3626 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3627 up_write(&EXT4_I(inode)->i_data_sem); 3628 } 3629 3630 static inline void __ext4_set_i_disksize(struct inode *inode, loff_t newsize) 3631 { 3632 WARN_ON_ONCE(!rwsem_is_locked(&EXT4_I(inode)->i_data_sem)); 3633 3634 if (newsize < EXT4_I(inode)->i_disksize || 3635 !ext4_test_inode_state(inode, EXT4_STATE_DISKSIZE_GROW_PENDING)) 3636 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3637 } 3638 3639 /* 3640 * Update i_size and i_disksize to @newsize. Requires i_rwsem to avoid 3641 * races with truncate. 3642 * 3643 * In the iomap buffered I/O path, i_disksize is updated only if no zeroed 3644 * pending block straddles i_disksize (EXT4_STATE_DISKSIZE_GROW_PENDING 3645 * clear), otherwise the ioend worker for the pending block will advance 3646 * i_disksize once the pending block is written back. Both updates happen 3647 * under i_data_sem so that the writeback ioend worker can always see the 3648 * latest i_size under the same semaphore. 3649 * 3650 * Returns 0 if nothing changed, 1 if i_size was raised, 2 if i_disksize 3651 * was raised, or 3 if both were. 3652 */ 3653 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3654 { 3655 int changed = 0; 3656 3657 if (newsize <= inode->i_size && newsize <= EXT4_I(inode)->i_disksize) 3658 return 0; 3659 3660 down_write(&EXT4_I(inode)->i_data_sem); 3661 if (newsize > inode->i_size) { 3662 i_size_write(inode, newsize); 3663 changed = 1; 3664 } 3665 if (newsize > EXT4_I(inode)->i_disksize && 3666 !ext4_test_inode_state(inode, EXT4_STATE_DISKSIZE_GROW_PENDING)) { 3667 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3668 changed |= 2; 3669 } 3670 up_write(&EXT4_I(inode)->i_data_sem); 3671 return changed; 3672 } 3673 3674 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3675 loff_t len); 3676 3677 struct ext4_group_info { 3678 unsigned long bb_state; 3679 #ifdef AGGRESSIVE_CHECK 3680 unsigned long bb_check_counter; 3681 #endif 3682 struct rb_root bb_free_root; 3683 ext4_grpblk_t bb_first_free; /* first free block */ 3684 ext4_grpblk_t bb_free; /* total free blocks */ 3685 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3686 int bb_avg_fragment_size_order; /* order of average 3687 fragment in BG */ 3688 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3689 ext4_group_t bb_group; /* Group number */ 3690 struct list_head bb_prealloc_list; 3691 #ifdef DOUBLE_CHECK 3692 void *bb_bitmap; 3693 #endif 3694 struct rw_semaphore alloc_sem; 3695 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3696 * regions, index is order. 3697 * bb_counters[3] = 5 means 3698 * 5 free 8-block regions. */ 3699 }; 3700 3701 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3702 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3703 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3704 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3705 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3706 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3707 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3708 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3709 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3710 3711 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3712 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3713 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3714 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3715 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3716 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3717 3718 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3719 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3720 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3721 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3722 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3723 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3724 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3725 (ext4_mb_grp_test_and_set_read((grp))) 3726 3727 static inline int ext4_mb_grp_test_and_set_read(struct ext4_group_info *grp) 3728 { 3729 return (test_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &grp->bb_state) || 3730 test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &grp->bb_state)); 3731 } 3732 3733 #define EXT4_MAX_CONTENTION 8 3734 #define EXT4_CONTENTION_THRESHOLD 2 3735 3736 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3737 ext4_group_t group) 3738 { 3739 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3740 } 3741 3742 /* 3743 * Returns true if the filesystem is busy enough that attempts to 3744 * access the block group locks has run into contention. 3745 */ 3746 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3747 { 3748 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3749 } 3750 3751 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group) 3752 { 3753 if (!spin_trylock(ext4_group_lock_ptr(sb, group))) 3754 return false; 3755 /* 3756 * We're able to grab the lock right away, so drop the lock 3757 * contention counter. 3758 */ 3759 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3760 return true; 3761 } 3762 3763 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3764 { 3765 if (!ext4_try_lock_group(sb, group)) { 3766 /* 3767 * The lock is busy, so bump the contention counter, 3768 * and then wait on the spin lock. 3769 */ 3770 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3771 EXT4_MAX_CONTENTION); 3772 spin_lock(ext4_group_lock_ptr(sb, group)); 3773 } 3774 } 3775 3776 static inline void ext4_unlock_group(struct super_block *sb, 3777 ext4_group_t group) 3778 { 3779 spin_unlock(ext4_group_lock_ptr(sb, group)); 3780 } 3781 3782 #ifdef CONFIG_QUOTA 3783 static inline bool ext4_quota_capable(struct super_block *sb) 3784 { 3785 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3786 } 3787 3788 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3789 { 3790 struct ext4_sb_info *sbi = EXT4_SB(sb); 3791 3792 return (ext4_has_feature_quota(sb) || 3793 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3794 } 3795 int ext4_enable_quotas(struct super_block *sb); 3796 #endif 3797 3798 /* 3799 * Block validity checking 3800 */ 3801 #define ext4_check_indirect_blockref(inode, bh) \ 3802 ext4_check_blockref(__func__, __LINE__, inode, \ 3803 (__le32 *)(bh)->b_data, \ 3804 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3805 3806 #define ext4_ind_check_inode(inode) \ 3807 ext4_check_blockref(__func__, __LINE__, inode, \ 3808 EXT4_I(inode)->i_data, \ 3809 EXT4_NDIR_BLOCKS) 3810 3811 /* 3812 * Inodes and files operations 3813 */ 3814 3815 /* dir.c */ 3816 extern const struct file_operations ext4_dir_operations; 3817 3818 /* file.c */ 3819 extern const struct inode_operations ext4_file_inode_operations; 3820 extern const struct file_operations ext4_file_operations; 3821 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3822 3823 /* inline.c */ 3824 extern int ext4_get_max_inline_size(struct inode *inode); 3825 extern int ext4_find_inline_data_nolock(struct inode *inode); 3826 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3827 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size); 3828 3829 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3830 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3831 struct inode *inode, 3832 loff_t pos, unsigned len, 3833 struct folio **foliop); 3834 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3835 unsigned copied, struct folio *folio); 3836 extern int ext4_generic_write_inline_data(struct address_space *mapping, 3837 struct inode *inode, 3838 loff_t pos, unsigned len, 3839 struct folio **foliop, 3840 bool da); 3841 extern int ext4_try_add_inline_entry(handle_t *handle, 3842 struct ext4_filename *fname, 3843 struct inode *dir, struct inode *inode); 3844 extern int ext4_try_create_inline_dir(handle_t *handle, 3845 struct inode *parent, 3846 struct inode *inode); 3847 extern int ext4_read_inline_dir(struct file *filp, 3848 struct dir_context *ctx, 3849 int *has_inline_data); 3850 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3851 struct inode *dir, ext4_lblk_t block, 3852 struct dx_hash_info *hinfo, 3853 __u32 start_hash, __u32 start_minor_hash, 3854 int *has_inline_data); 3855 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3856 struct ext4_filename *fname, 3857 struct ext4_dir_entry_2 **res_dir, 3858 int *has_inline_data); 3859 extern int ext4_delete_inline_entry(handle_t *handle, 3860 struct inode *dir, 3861 struct ext4_dir_entry_2 *de_del, 3862 struct buffer_head *bh, 3863 int *has_inline_data); 3864 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3865 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3866 struct ext4_dir_entry_2 **parent_de, 3867 int *retval); 3868 extern void *ext4_read_inline_link(struct inode *inode); 3869 3870 struct iomap; 3871 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3872 3873 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3874 3875 extern int ext4_convert_inline_data(struct inode *inode); 3876 3877 static inline int ext4_has_inline_data(struct inode *inode) 3878 { 3879 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3880 EXT4_I(inode)->i_inline_off; 3881 } 3882 3883 /* namei.c */ 3884 extern const struct inode_operations ext4_dir_inode_operations; 3885 extern const struct inode_operations ext4_special_inode_operations; 3886 extern struct dentry *ext4_get_parent(struct dentry *child); 3887 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode, 3888 struct buffer_head *dir_block, 3889 unsigned int parent_ino, void *inline_buf, 3890 int inline_size); 3891 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3892 unsigned int blocksize); 3893 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3894 struct buffer_head *bh); 3895 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3896 struct inode *inode, struct dentry *dentry); 3897 extern int __ext4_link(struct inode *dir, struct inode *inode, 3898 const struct qstr *d_name, struct dentry *dentry); 3899 3900 #define S_SHIFT 12 3901 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3902 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3903 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3904 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3905 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3906 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3907 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3908 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3909 }; 3910 3911 static inline void ext4_set_de_type(struct super_block *sb, 3912 struct ext4_dir_entry_2 *de, 3913 umode_t mode) { 3914 if (ext4_has_feature_filetype(sb)) 3915 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3916 } 3917 3918 /* readpages.c */ 3919 int ext4_read_folio(struct file *file, struct folio *folio); 3920 void ext4_readahead(struct readahead_control *rac); 3921 int __init ext4_init_verity_caches(void); 3922 void ext4_exit_verity_caches(void); 3923 3924 /* symlink.c */ 3925 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3926 extern const struct inode_operations ext4_symlink_inode_operations; 3927 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3928 3929 /* sysfs.c */ 3930 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3931 extern int ext4_register_sysfs(struct super_block *sb); 3932 extern void ext4_unregister_sysfs(struct super_block *sb); 3933 extern int __init ext4_init_sysfs(void); 3934 extern void ext4_exit_sysfs(void); 3935 3936 /* block_validity */ 3937 extern void ext4_release_system_zone(struct super_block *sb); 3938 extern int ext4_setup_system_zone(struct super_block *sb); 3939 extern int __init ext4_init_system_zone(void); 3940 extern void ext4_exit_system_zone(void); 3941 extern int ext4_inode_block_valid(struct inode *inode, 3942 ext4_fsblk_t start_blk, 3943 unsigned int count); 3944 extern int ext4_check_blockref(const char *, unsigned int, 3945 struct inode *, __le32 *, unsigned int); 3946 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3947 ext4_fsblk_t start_blk, unsigned int count); 3948 3949 3950 /* extents.c */ 3951 struct ext4_ext_path; 3952 struct ext4_extent; 3953 3954 /* 3955 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3956 * __le32. 3957 */ 3958 #define EXT_MAX_BLOCKS 0xffffffff 3959 3960 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3961 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3962 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3963 struct ext4_map_blocks *map, int flags); 3964 ext4_lblk_t ext4_determine_insert_hole(struct inode *inode, ext4_lblk_t lblk, 3965 ext4_lblk_t hole_start, 3966 ext4_lblk_t hole_len); 3967 extern int ext4_ext_truncate(handle_t *, struct inode *); 3968 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3969 ext4_lblk_t end); 3970 extern void ext4_ext_init(struct super_block *); 3971 extern void ext4_ext_release(struct super_block *); 3972 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3973 loff_t len); 3974 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3975 loff_t offset, ssize_t len, 3976 ext4_lblk_t *converted); 3977 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle, 3978 struct inode *inode, loff_t offset, ssize_t len); 3979 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3980 ext4_io_end_t *io_end); 3981 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3982 struct ext4_map_blocks *map, int flags); 3983 extern int ext4_map_query_blocks(handle_t *handle, struct inode *inode, 3984 struct ext4_map_blocks *map, int flags); 3985 extern int ext4_map_create_blocks(handle_t *handle, struct inode *inode, 3986 struct ext4_map_blocks *map, int flags); 3987 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3988 int num, 3989 struct ext4_ext_path *path); 3990 extern struct ext4_ext_path *ext4_ext_insert_extent( 3991 handle_t *handle, struct inode *inode, 3992 struct ext4_ext_path *path, 3993 struct ext4_extent *newext, int gb_flags); 3994 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3995 struct ext4_ext_path *, 3996 int flags); 3997 extern void ext4_free_ext_path(struct ext4_ext_path *); 3998 extern int ext4_ext_check_inode(struct inode *inode); 3999 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 4000 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 4001 __u64 start, __u64 len); 4002 extern int ext4_get_es_cache(struct inode *inode, 4003 struct fiemap_extent_info *fieinfo, 4004 __u64 start, __u64 len); 4005 extern int ext4_ext_precache(struct inode *inode); 4006 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 4007 struct inode *inode2, ext4_lblk_t lblk1, 4008 ext4_lblk_t lblk2, ext4_lblk_t count, 4009 int mark_unwritten,int *err); 4010 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 4011 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 4012 int check_cred, int restart_cred, 4013 int revoke_cred); 4014 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 4015 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 4016 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 4017 int len, int unwritten, ext4_fsblk_t pblk); 4018 extern int ext4_ext_clear_bb(struct inode *inode); 4019 4020 4021 /* move_extent.c */ 4022 extern void ext4_double_down_write_data_sem(struct inode *first, 4023 struct inode *second); 4024 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 4025 struct inode *donor_inode); 4026 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 4027 __u64 start_orig, __u64 start_donor, 4028 __u64 len, __u64 *moved_len); 4029 4030 /* page-io.c */ 4031 /* 4032 * The I/O range covers the zeroed EOF block that straddles i_disksize 4033 * and will advance it upon completion. 4034 */ 4035 #define EXT4_IOMAP_IOEND_DISKSIZE_GROW_IO 1UL 4036 4037 extern int __init ext4_init_pageio(void); 4038 extern void ext4_exit_pageio(void); 4039 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 4040 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 4041 extern int ext4_put_io_end(ext4_io_end_t *io_end); 4042 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 4043 extern void ext4_io_submit_init(struct ext4_io_submit *io, 4044 struct writeback_control *wbc); 4045 extern void ext4_end_io_rsv_work(struct work_struct *work); 4046 extern void ext4_io_submit(struct ext4_io_submit *io); 4047 void ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 4048 size_t len); 4049 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 4050 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 4051 extern void ext4_iomap_end_io(struct work_struct *work); 4052 extern void ext4_iomap_end_bio(struct bio *bio); 4053 4054 /* mmp.c */ 4055 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 4056 4057 /* mmp.c */ 4058 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 4059 4060 /* verity.c */ 4061 extern const struct fsverity_operations ext4_verityops; 4062 4063 /* orphan.c */ 4064 extern int ext4_orphan_add(handle_t *, struct inode *); 4065 extern int ext4_orphan_del(handle_t *, struct inode *); 4066 extern void ext4_orphan_cleanup(struct super_block *sb, 4067 struct ext4_super_block *es); 4068 extern void ext4_release_orphan_info(struct super_block *sb); 4069 extern int ext4_init_orphan_info(struct super_block *sb); 4070 extern int ext4_orphan_file_empty(struct super_block *sb); 4071 extern void ext4_orphan_file_block_trigger( 4072 struct jbd2_buffer_trigger_type *triggers, 4073 struct buffer_head *bh, 4074 void *data, size_t size); 4075 4076 /* 4077 * Add new method to test whether block and inode bitmaps are properly 4078 * initialized. With uninit_bg reading the block from disk is not enough 4079 * to mark the bitmap uptodate. We need to also zero-out the bitmap 4080 */ 4081 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 4082 4083 static inline int bitmap_uptodate(struct buffer_head *bh) 4084 { 4085 return (buffer_uptodate(bh) && 4086 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 4087 } 4088 static inline void set_bitmap_uptodate(struct buffer_head *bh) 4089 { 4090 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 4091 } 4092 4093 extern int ext4_resize_begin(struct super_block *sb); 4094 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 4095 4096 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end) 4097 { 4098 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) 4099 io_end->flag |= EXT4_IO_END_UNWRITTEN; 4100 } 4101 4102 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 4103 { 4104 if (io_end->flag & EXT4_IO_END_UNWRITTEN) 4105 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 4106 } 4107 4108 extern const struct iomap_ops ext4_iomap_ops; 4109 extern const struct iomap_ops ext4_iomap_report_ops; 4110 extern const struct iomap_ops ext4_iomap_buffered_write_ops; 4111 extern const struct iomap_ops ext4_iomap_buffered_da_write_ops; 4112 extern const struct iomap_write_ops ext4_iomap_write_ops; 4113 4114 int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length, 4115 unsigned flags, struct iomap *iomap, struct iomap *srcmap); 4116 4117 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 4118 { 4119 /* 4120 * If the buffer has the write error flag, we have failed 4121 * to write out data in the block. In this case, we don't 4122 * have to read the block because we may read the old data 4123 * successfully. 4124 */ 4125 if (buffer_write_io_error(bh)) 4126 set_buffer_uptodate(bh); 4127 return buffer_uptodate(bh); 4128 } 4129 4130 static inline bool ext4_inode_can_atomic_write(struct inode *inode) 4131 { 4132 4133 return S_ISREG(inode->i_mode) && 4134 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) && 4135 EXT4_SB(inode->i_sb)->s_awu_min > 0; 4136 } 4137 4138 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio, 4139 loff_t pos, unsigned len, 4140 get_block_t *get_block); 4141 4142 #if IS_ENABLED(CONFIG_EXT4_KUNIT_TESTS) 4143 #define EXPORT_SYMBOL_FOR_EXT4_TEST(sym) \ 4144 EXPORT_SYMBOL_FOR_MODULES(sym, "ext4-test") 4145 #endif 4146 #endif /* __KERNEL__ */ 4147 4148 #endif /* _EXT4_H */ 4149