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 30 338 #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 */ 1019 enum { 1020 I_DATA_SEM_NORMAL = 0, 1021 I_DATA_SEM_OTHER, 1022 I_DATA_SEM_QUOTA, 1023 I_DATA_SEM_EA 1024 }; 1025 1026 1027 /* 1028 * fourth extended file system inode data in memory 1029 */ 1030 struct ext4_inode_info { 1031 __le32 i_data[15]; /* unconverted */ 1032 __u32 i_dtime; 1033 ext4_fsblk_t i_file_acl; 1034 1035 /* 1036 * i_block_group is the number of the block group which contains 1037 * this file's inode. Constant across the lifetime of the inode, 1038 * it is used for making block allocation decisions - we try to 1039 * place a file's data blocks near its inode block, and new inodes 1040 * near to their parent directory's inode. 1041 */ 1042 ext4_group_t i_block_group; 1043 ext4_lblk_t i_dir_start_lookup; 1044 #if (BITS_PER_LONG < 64) 1045 unsigned long i_state_flags; /* Dynamic state flags */ 1046 #endif 1047 unsigned long i_flags; 1048 1049 /* 1050 * Extended attributes can be read independently of the main file 1051 * data. Taking i_rwsem even when reading would cause contention 1052 * between readers of EAs and writers of regular file data, so 1053 * instead we synchronize on xattr_sem when reading or changing 1054 * EAs. 1055 */ 1056 struct rw_semaphore xattr_sem; 1057 1058 /* 1059 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1060 * i_orphan is used. 1061 */ 1062 union { 1063 struct list_head i_orphan; /* unlinked but open inodes */ 1064 unsigned int i_orphan_idx; /* Index in orphan file */ 1065 }; 1066 1067 /* Fast commit related info */ 1068 1069 /* For tracking dentry create updates */ 1070 struct list_head i_fc_dilist; 1071 struct list_head i_fc_list; /* 1072 * inodes that need fast commit 1073 * protected by sbi->s_fc_lock. 1074 */ 1075 1076 /* Start of lblk range that needs to be committed in this fast commit */ 1077 ext4_lblk_t i_fc_lblk_start; 1078 1079 /* End of lblk range that needs to be committed in this fast commit */ 1080 ext4_lblk_t i_fc_lblk_len; 1081 1082 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1083 1084 /* 1085 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len 1086 * and inode's EXT4_FC_STATE_COMMITTING state bit. 1087 */ 1088 spinlock_t i_fc_lock; 1089 1090 /* 1091 * i_disksize keeps track of what the inode size is ON DISK, not 1092 * in memory. During truncate, i_size is set to the new size by 1093 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1094 * set i_disksize to 0 until the truncate is actually under way. 1095 * 1096 * The intent is that i_disksize always represents the blocks which 1097 * are used by this file. This allows recovery to restart truncate 1098 * on orphans if we crash during truncate. We actually write i_disksize 1099 * into the on-disk inode when writing inodes out, instead of i_size. 1100 * 1101 * The only time when i_disksize and i_size may be different is when 1102 * a truncate is in progress. The only things which change i_disksize 1103 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1104 */ 1105 loff_t i_disksize; 1106 1107 /* 1108 * i_data_sem is for serialising ext4_truncate() against 1109 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1110 * data tree are chopped off during truncate. We can't do that in 1111 * ext4 because whenever we perform intermediate commits during 1112 * truncate, the inode and all the metadata blocks *must* be in a 1113 * consistent state which allows truncation of the orphans to restart 1114 * during recovery. Hence we must fix the get_block-vs-truncate race 1115 * by other means, so we have i_data_sem. 1116 */ 1117 struct rw_semaphore i_data_sem; 1118 struct inode vfs_inode; 1119 struct jbd2_inode *jinode; 1120 struct mapping_metadata_bhs i_metadata_bhs; 1121 1122 /* 1123 * File creation time. Its function is same as that of 1124 * struct timespec64 i_{a,c,m}time in the generic inode. 1125 */ 1126 struct timespec64 i_crtime; 1127 1128 /* mballoc */ 1129 atomic_t i_prealloc_active; 1130 1131 /* allocation reservation info for delalloc */ 1132 /* In case of bigalloc, this refer to clusters rather than blocks */ 1133 unsigned int i_reserved_data_blocks; 1134 struct rb_root i_prealloc_node; 1135 rwlock_t i_prealloc_lock; 1136 1137 /* extents status tree */ 1138 struct ext4_es_tree i_es_tree; 1139 rwlock_t i_es_lock; 1140 struct list_head i_es_list; 1141 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1142 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1143 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1144 extents to shrink. Protected by 1145 i_es_lock */ 1146 u64 i_es_seq; /* Change counter for extents. 1147 Protected by i_es_lock */ 1148 1149 /* ialloc */ 1150 ext4_group_t i_last_alloc_group; 1151 1152 /* pending cluster reservations for bigalloc file systems */ 1153 struct ext4_pending_tree i_pending_tree; 1154 1155 /* on-disk additional length */ 1156 __u16 i_extra_isize; 1157 1158 /* Indicate the inline data space. */ 1159 u16 i_inline_off; 1160 u16 i_inline_size; 1161 1162 #ifdef CONFIG_QUOTA 1163 /* quota space reservation, managed internally by quota code */ 1164 qsize_t i_reserved_quota; 1165 #endif 1166 spinlock_t i_block_reservation_lock; 1167 1168 /* Lock protecting lists below */ 1169 spinlock_t i_completed_io_lock; 1170 /* 1171 * Completed IOs that need unwritten extents handling and have 1172 * transaction reserved 1173 */ 1174 struct list_head i_rsv_conversion_list; 1175 struct work_struct i_rsv_conversion_work; 1176 1177 /* 1178 * Transactions that contain inode's metadata needed to complete 1179 * fsync and fdatasync, respectively. 1180 */ 1181 tid_t i_sync_tid; 1182 tid_t i_datasync_tid; 1183 1184 #ifdef CONFIG_QUOTA 1185 struct dquot __rcu *i_dquot[MAXQUOTAS]; 1186 #endif 1187 1188 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1189 __u32 i_csum_seed; 1190 1191 kprojid_t i_projid; 1192 1193 #ifdef CONFIG_FS_ENCRYPTION 1194 struct fscrypt_inode_info *i_crypt_info; 1195 #endif 1196 }; 1197 1198 /* 1199 * File system states 1200 */ 1201 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1202 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1203 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1204 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1205 1206 /* 1207 * Misc. filesystem flags 1208 */ 1209 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1210 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1211 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1212 1213 /* 1214 * Mount flags set via mount options or defaults 1215 */ 1216 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1217 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1218 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1219 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1220 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1221 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1222 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1223 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1224 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1225 #ifdef CONFIG_FS_DAX 1226 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1227 #else 1228 #define EXT4_MOUNT_DAX_ALWAYS 0 1229 #endif 1230 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1231 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1232 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1233 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1234 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1235 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1236 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1237 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1238 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1239 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1240 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1241 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1242 * enable enforcement for hidden 1243 * quota files */ 1244 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1245 * enforcement for hidden quota 1246 * files */ 1247 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1248 * enforcement */ 1249 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1250 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1251 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1252 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1253 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1254 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1255 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1256 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1257 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1258 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1259 1260 /* 1261 * Mount flags set either automatically (could not be set by mount option) 1262 * based on per file system feature or property or in special cases such as 1263 * distinguishing between explicit mount option definition and default. 1264 */ 1265 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1266 specified delalloc */ 1267 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1268 size of blocksize * 8 1269 blocks */ 1270 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1271 file systems */ 1272 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1273 specified journal checksum */ 1274 1275 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1276 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1277 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1278 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1279 * scanning in mballoc 1280 */ 1281 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1282 1283 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1284 ~EXT4_MOUNT_##opt 1285 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1286 EXT4_MOUNT_##opt 1287 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1288 EXT4_MOUNT_##opt) 1289 1290 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1291 ~EXT4_MOUNT2_##opt 1292 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1293 EXT4_MOUNT2_##opt 1294 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1295 EXT4_MOUNT2_##opt) 1296 1297 #define ext4_test_and_set_bit __test_and_set_bit_le 1298 #define ext4_set_bit __set_bit_le 1299 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1300 #define ext4_clear_bit __clear_bit_le 1301 #define ext4_test_bit test_bit_le 1302 #define ext4_find_next_zero_bit find_next_zero_bit_le 1303 #define ext4_find_next_bit find_next_bit_le 1304 1305 extern void mb_set_bits(void *bm, int cur, int len); 1306 1307 /* 1308 * Maximal mount counts between two filesystem checks 1309 */ 1310 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1311 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1312 1313 /* 1314 * Behaviour when detecting errors 1315 */ 1316 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1317 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1318 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1319 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1320 1321 /* Metadata checksum algorithm codes */ 1322 #define EXT4_CRC32C_CHKSUM 1 1323 1324 #define EXT4_LABEL_MAX 16 1325 1326 /* 1327 * Structure of the super block 1328 */ 1329 struct ext4_super_block { 1330 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1331 __le32 s_blocks_count_lo; /* Blocks count */ 1332 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1333 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1334 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1335 __le32 s_first_data_block; /* First Data Block */ 1336 __le32 s_log_block_size; /* Block size */ 1337 __le32 s_log_cluster_size; /* Allocation cluster size */ 1338 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1339 __le32 s_clusters_per_group; /* # Clusters per group */ 1340 __le32 s_inodes_per_group; /* # Inodes per group */ 1341 __le32 s_mtime; /* Mount time */ 1342 /*30*/ __le32 s_wtime; /* Write time */ 1343 __le16 s_mnt_count; /* Mount count */ 1344 __le16 s_max_mnt_count; /* Maximal mount count */ 1345 __le16 s_magic; /* Magic signature */ 1346 __le16 s_state; /* File system state */ 1347 __le16 s_errors; /* Behaviour when detecting errors */ 1348 __le16 s_minor_rev_level; /* minor revision level */ 1349 /*40*/ __le32 s_lastcheck; /* time of last check */ 1350 __le32 s_checkinterval; /* max. time between checks */ 1351 __le32 s_creator_os; /* OS */ 1352 __le32 s_rev_level; /* Revision level */ 1353 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1354 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1355 /* 1356 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1357 * 1358 * Note: the difference between the compatible feature set and 1359 * the incompatible feature set is that if there is a bit set 1360 * in the incompatible feature set that the kernel doesn't 1361 * know about, it should refuse to mount the filesystem. 1362 * 1363 * e2fsck's requirements are more strict; if it doesn't know 1364 * about a feature in either the compatible or incompatible 1365 * feature set, it must abort and not try to meddle with 1366 * things it doesn't understand... 1367 */ 1368 __le32 s_first_ino; /* First non-reserved inode */ 1369 __le16 s_inode_size; /* size of inode structure */ 1370 __le16 s_block_group_nr; /* block group # of this superblock */ 1371 __le32 s_feature_compat; /* compatible feature set */ 1372 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1373 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1374 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1375 /*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */ 1376 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1377 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1378 /* 1379 * Performance hints. Directory preallocation should only 1380 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1381 */ 1382 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1383 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1384 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1385 /* 1386 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1387 */ 1388 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1389 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1390 __le32 s_journal_dev; /* device number of journal file */ 1391 __le32 s_last_orphan; /* start of list of inodes to delete */ 1392 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1393 __u8 s_def_hash_version; /* Default hash version to use */ 1394 __u8 s_jnl_backup_type; 1395 __le16 s_desc_size; /* size of group descriptor */ 1396 /*100*/ __le32 s_default_mount_opts; 1397 __le32 s_first_meta_bg; /* First metablock block group */ 1398 __le32 s_mkfs_time; /* When the filesystem was created */ 1399 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1400 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1401 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1402 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1403 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1404 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1405 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1406 __le32 s_flags; /* Miscellaneous flags */ 1407 __le16 s_raid_stride; /* RAID stride */ 1408 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1409 __le64 s_mmp_block; /* Block for multi-mount protection */ 1410 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1411 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1412 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1413 __u8 s_encryption_level; /* versioning level for encryption */ 1414 __u8 s_reserved_pad; /* Padding to next 32bits */ 1415 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1416 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1417 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1418 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1419 snapshot's future use */ 1420 __le32 s_snapshot_list; /* inode number of the head of the 1421 on-disk snapshot list */ 1422 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1423 __le32 s_error_count; /* number of fs errors */ 1424 __le32 s_first_error_time; /* first time an error happened */ 1425 __le32 s_first_error_ino; /* inode involved in first error */ 1426 __le64 s_first_error_block; /* block involved of first error */ 1427 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1428 __le32 s_first_error_line; /* line number where error happened */ 1429 __le32 s_last_error_time; /* most recent time of an error */ 1430 __le32 s_last_error_ino; /* inode involved in last error */ 1431 __le32 s_last_error_line; /* line number where error happened */ 1432 __le64 s_last_error_block; /* block involved of last error */ 1433 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1434 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1435 __u8 s_mount_opts[64]; 1436 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1437 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1438 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1439 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1440 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1441 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1442 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1443 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1444 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1445 __u8 s_wtime_hi; 1446 __u8 s_mtime_hi; 1447 __u8 s_mkfs_time_hi; 1448 __u8 s_lastcheck_hi; 1449 __u8 s_first_error_time_hi; 1450 __u8 s_last_error_time_hi; 1451 __u8 s_first_error_errcode; 1452 __u8 s_last_error_errcode; 1453 __le16 s_encoding; /* Filename charset encoding */ 1454 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1455 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1456 __le16 s_def_resuid_hi; 1457 __le16 s_def_resgid_hi; 1458 __le32 s_reserved[93]; /* Padding to the end of the block */ 1459 __le32 s_checksum; /* crc32c(superblock) */ 1460 }; 1461 1462 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1463 1464 #ifdef __KERNEL__ 1465 1466 /* Number of quota types we support */ 1467 #define EXT4_MAXQUOTAS 3 1468 1469 #define EXT4_ENC_UTF8_12_1 1 1470 1471 /* Types of ext4 journal triggers */ 1472 enum ext4_journal_trigger_type { 1473 EXT4_JTR_ORPHAN_FILE, 1474 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1475 }; 1476 1477 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1478 1479 struct ext4_journal_trigger { 1480 struct jbd2_buffer_trigger_type tr_triggers; 1481 struct super_block *sb; 1482 }; 1483 1484 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1485 struct jbd2_buffer_trigger_type *trigger) 1486 { 1487 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1488 } 1489 1490 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1491 1492 /* Structure at the tail of orphan block */ 1493 struct ext4_orphan_block_tail { 1494 __le32 ob_magic; 1495 __le32 ob_checksum; 1496 }; 1497 1498 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1499 { 1500 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1501 sizeof(u32); 1502 } 1503 1504 struct ext4_orphan_block { 1505 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1506 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1507 }; 1508 1509 /* 1510 * Info about orphan file. 1511 */ 1512 struct ext4_orphan_info { 1513 int of_blocks; /* Number of orphan blocks in a file */ 1514 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1515 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1516 * file blocks */ 1517 }; 1518 1519 /* 1520 * fourth extended-fs super-block data in memory 1521 */ 1522 struct ext4_sb_info { 1523 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1524 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1525 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1526 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1527 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1528 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1529 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1530 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1531 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1532 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1533 unsigned long s_overhead; /* # of fs overhead clusters */ 1534 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1535 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1536 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1537 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1538 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1539 /* Array of bh's for the block group descriptors */ 1540 struct buffer_head * __rcu *s_group_desc; 1541 unsigned int s_mount_opt; 1542 unsigned int s_mount_opt2; 1543 unsigned long s_mount_flags; 1544 unsigned int s_def_mount_opt; 1545 unsigned int s_def_mount_opt2; 1546 ext4_fsblk_t s_sb_block; 1547 atomic64_t s_resv_clusters; 1548 kuid_t s_resuid; 1549 kgid_t s_resgid; 1550 unsigned short s_mount_state; 1551 unsigned short s_pad; 1552 int s_addr_per_block_bits; 1553 int s_desc_per_block_bits; 1554 int s_inode_size; 1555 int s_first_ino; 1556 unsigned int s_inode_readahead_blks; 1557 unsigned int s_inode_goal; 1558 u32 s_hash_seed[4]; 1559 int s_def_hash_version; 1560 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1561 struct percpu_counter s_freeclusters_counter; 1562 struct percpu_counter s_freeinodes_counter; 1563 struct percpu_counter s_dirs_counter; 1564 struct percpu_counter s_dirtyclusters_counter; 1565 struct percpu_counter s_sra_exceeded_retry_limit; 1566 struct blockgroup_lock *s_blockgroup_lock; 1567 struct proc_dir_entry *s_proc; 1568 struct kobject s_kobj; 1569 struct completion s_kobj_unregister; 1570 struct mutex s_error_notify_mutex; /* protects sysfs_notify vs kobject_del */ 1571 struct super_block *s_sb; 1572 struct buffer_head *s_mmp_bh; 1573 1574 /* Journaling */ 1575 struct journal_s *s_journal; 1576 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1577 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1578 struct list_head s_orphan; /* List of orphaned inodes in on disk 1579 list */ 1580 struct ext4_orphan_info s_orphan_info; 1581 unsigned long s_commit_interval; 1582 u32 s_max_batch_time; 1583 u32 s_min_batch_time; 1584 struct file *s_journal_bdev_file; 1585 #ifdef CONFIG_QUOTA 1586 /* Names of quota files with journalled quota */ 1587 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1588 int s_jquota_fmt; /* Format of quota to use */ 1589 #endif 1590 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1591 struct ext4_system_blocks __rcu *s_system_blks; 1592 1593 #ifdef EXTENTS_STATS 1594 /* ext4 extents stats */ 1595 unsigned long s_ext_min; 1596 unsigned long s_ext_max; 1597 unsigned long s_depth_max; 1598 spinlock_t s_ext_stats_lock; 1599 unsigned long s_ext_blocks; 1600 unsigned long s_ext_extents; 1601 #endif 1602 1603 /* for buddy allocator */ 1604 struct ext4_group_info ** __rcu *s_group_info; 1605 struct inode *s_buddy_cache; 1606 spinlock_t s_md_lock; 1607 unsigned short *s_mb_offsets; 1608 unsigned int *s_mb_maxs; 1609 unsigned int s_group_info_size; 1610 atomic_t s_mb_free_pending; 1611 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1612 after commit completed */ 1613 struct list_head s_discard_list; 1614 struct work_struct s_discard_work; 1615 atomic_t s_retry_alloc_pending; 1616 struct xarray *s_mb_avg_fragment_size; 1617 struct xarray *s_mb_largest_free_orders; 1618 1619 /* tunables */ 1620 unsigned long s_stripe; 1621 unsigned int s_mb_max_linear_groups; 1622 unsigned int s_mb_stream_request; 1623 unsigned int s_mb_max_to_scan; 1624 unsigned int s_mb_min_to_scan; 1625 unsigned int s_mb_stats; 1626 unsigned int s_mb_order2_reqs; 1627 unsigned int s_mb_group_prealloc; 1628 unsigned int s_max_dir_size_kb; 1629 unsigned int s_mb_prefetch; 1630 unsigned int s_mb_prefetch_limit; 1631 unsigned int s_mb_best_avail_max_trim_order; 1632 unsigned int s_sb_update_sec; 1633 unsigned int s_sb_update_kb; 1634 1635 /* where last allocation was done - for stream allocation */ 1636 ext4_group_t *s_mb_last_groups; 1637 unsigned int s_mb_nr_global_goals; 1638 1639 /* stats for buddy allocator */ 1640 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1641 atomic_t s_bal_success; /* we found long enough chunks */ 1642 atomic_t s_bal_allocated; /* in blocks */ 1643 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1644 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1645 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1646 atomic_t s_bal_goals; /* goal hits */ 1647 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */ 1648 atomic_t s_bal_len_goals; /* len goal hits */ 1649 atomic_t s_bal_breaks; /* too long searches */ 1650 atomic_t s_bal_2orders; /* 2^order hits */ 1651 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1652 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1653 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1654 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1655 atomic64_t s_mb_generation_time; 1656 atomic_t s_mb_lost_chunks; 1657 atomic_t s_mb_preallocated; 1658 atomic_t s_mb_discarded; 1659 atomic_t s_lock_busy; 1660 1661 /* locality groups */ 1662 struct ext4_locality_group __percpu *s_locality_groups; 1663 1664 /* for write statistics */ 1665 unsigned long s_sectors_written_start; 1666 u64 s_kbytes_written; 1667 1668 /* the size of zero-out chunk */ 1669 unsigned int s_extent_max_zeroout_kb; 1670 1671 unsigned int s_log_groups_per_flex; 1672 struct flex_groups * __rcu *s_flex_groups; 1673 ext4_group_t s_flex_groups_allocated; 1674 1675 /* workqueue for reserved extent conversions (buffered io) */ 1676 struct workqueue_struct *rsv_conversion_wq; 1677 1678 /* timer for periodic error stats printing */ 1679 struct timer_list s_err_report; 1680 /* timeout in seconds for s_err_report; 0 disables the timer. */ 1681 unsigned long s_err_report_sec; 1682 1683 /* Lazy inode table initialization info */ 1684 struct ext4_li_request *s_li_request; 1685 /* Wait multiplier for lazy initialization thread */ 1686 unsigned int s_li_wait_mult; 1687 1688 /* Kernel thread for multiple mount protection */ 1689 struct task_struct *s_mmp_tsk; 1690 1691 /* record the last minlen when FITRIM is called. */ 1692 unsigned long s_last_trim_minblks; 1693 1694 /* minimum folio order of a page cache allocation */ 1695 u16 s_min_folio_order; 1696 /* supported maximum folio order, 0 means not supported */ 1697 u16 s_max_folio_order; 1698 1699 /* Precomputed FS UUID checksum for seeding other checksums */ 1700 __u32 s_csum_seed; 1701 1702 /* Reclaim extents from extent status tree */ 1703 struct shrinker *s_es_shrinker; 1704 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1705 long s_es_nr_inode; 1706 struct ext4_es_stats s_es_stats; 1707 struct mb_cache *s_ea_block_cache; 1708 struct mb_cache *s_ea_inode_cache; 1709 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1710 1711 /* Journal triggers for checksum computation */ 1712 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1713 1714 /* Ratelimit ext4 messages. */ 1715 struct ratelimit_state s_err_ratelimit_state; 1716 struct ratelimit_state s_warning_ratelimit_state; 1717 struct ratelimit_state s_msg_ratelimit_state; 1718 atomic_t s_warning_count; 1719 atomic_t s_msg_count; 1720 1721 /* Encryption policy for '-o test_dummy_encryption' */ 1722 struct fscrypt_dummy_policy s_dummy_enc_policy; 1723 1724 /* 1725 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1726 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1727 * DIOREAD_NOLOCK mount options on remount. 1728 */ 1729 struct percpu_rw_semaphore s_writepages_rwsem; 1730 struct dax_device *s_daxdev; 1731 u64 s_dax_part_off; 1732 #ifdef CONFIG_EXT4_DEBUG 1733 unsigned long s_simulate_fail; 1734 #endif 1735 /* Record the errseq of the backing block device */ 1736 errseq_t s_bdev_wb_err; 1737 spinlock_t s_bdev_wb_lock; 1738 1739 /* Information about errors that happened during this mount */ 1740 spinlock_t s_error_lock; 1741 int s_add_error_count; 1742 int s_first_error_code; 1743 __u32 s_first_error_line; 1744 __u32 s_first_error_ino; 1745 __u64 s_first_error_block; 1746 const char *s_first_error_func; 1747 time64_t s_first_error_time; 1748 int s_last_error_code; 1749 __u32 s_last_error_line; 1750 __u32 s_last_error_ino; 1751 __u64 s_last_error_block; 1752 const char *s_last_error_func; 1753 time64_t s_last_error_time; 1754 /* 1755 * If we are in a context where we cannot update the on-disk 1756 * superblock, we queue the work here. This is used to update 1757 * the error information in the superblock, and for periodic 1758 * updates of the superblock called from the commit callback 1759 * function. 1760 */ 1761 struct work_struct s_sb_upd_work; 1762 1763 /* Atomic write unit values in bytes */ 1764 unsigned int s_awu_min; 1765 unsigned int s_awu_max; 1766 1767 /* Ext4 fast commit sub transaction ID */ 1768 atomic_t s_fc_subtid; 1769 1770 /* 1771 * After commit starts, the main queue gets locked, and the further 1772 * updates get added in the staging queue. 1773 */ 1774 #define FC_Q_MAIN 0 1775 #define FC_Q_STAGING 1 1776 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1777 * that have data changes in them. 1778 */ 1779 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1780 unsigned int s_fc_bytes; 1781 /* 1782 * Main fast commit lock. This lock protects accesses to the 1783 * following fields: 1784 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1785 * 1786 * s_fc_lock can be taken from reclaim context (inode eviction) and is 1787 * thus reclaim unsafe. Use ext4_fc_lock()/ext4_fc_unlock() helpers 1788 * when acquiring / releasing the lock. 1789 */ 1790 struct mutex s_fc_lock; 1791 struct buffer_head *s_fc_bh; 1792 struct ext4_fc_stats s_fc_stats; 1793 tid_t s_fc_ineligible_tid; 1794 #ifdef CONFIG_EXT4_DEBUG 1795 int s_fc_debug_max_replay; 1796 #endif 1797 struct ext4_fc_replay_state s_fc_replay_state; 1798 }; 1799 1800 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1801 { 1802 return sb->s_fs_info; 1803 } 1804 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1805 { 1806 return container_of(inode, struct ext4_inode_info, vfs_inode); 1807 } 1808 1809 static inline int ext4_writepages_down_read(struct super_block *sb) 1810 { 1811 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1812 return memalloc_nofs_save(); 1813 } 1814 1815 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1816 { 1817 memalloc_nofs_restore(ctx); 1818 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1819 } 1820 1821 static inline int ext4_writepages_down_write(struct super_block *sb) 1822 { 1823 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1824 return memalloc_nofs_save(); 1825 } 1826 1827 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1828 { 1829 memalloc_nofs_restore(ctx); 1830 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1831 } 1832 1833 static inline int ext4_fc_lock(struct super_block *sb) 1834 { 1835 mutex_lock(&EXT4_SB(sb)->s_fc_lock); 1836 return memalloc_nofs_save(); 1837 } 1838 1839 static inline void ext4_fc_unlock(struct super_block *sb, int ctx) 1840 { 1841 memalloc_nofs_restore(ctx); 1842 mutex_unlock(&EXT4_SB(sb)->s_fc_lock); 1843 } 1844 1845 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1846 { 1847 return ino == EXT4_ROOT_INO || 1848 (ino >= EXT4_FIRST_INO(sb) && 1849 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1850 } 1851 1852 static inline int ext4_get_resuid(struct ext4_super_block *es) 1853 { 1854 return le16_to_cpu(es->s_def_resuid) | 1855 le16_to_cpu(es->s_def_resuid_hi) << 16; 1856 } 1857 1858 static inline int ext4_get_resgid(struct ext4_super_block *es) 1859 { 1860 return le16_to_cpu(es->s_def_resgid) | 1861 le16_to_cpu(es->s_def_resgid_hi) << 16; 1862 } 1863 1864 /* 1865 * Returns: sbi->field[index] 1866 * Used to access an array element from the following sbi fields which require 1867 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1868 * - s_group_desc 1869 * - s_group_info 1870 * - s_flex_group 1871 */ 1872 #define sbi_array_rcu_deref(sbi, field, index) \ 1873 ({ \ 1874 typeof(*((sbi)->field)) _v; \ 1875 rcu_read_lock(); \ 1876 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1877 rcu_read_unlock(); \ 1878 _v; \ 1879 }) 1880 1881 /* 1882 * run-time mount flags 1883 */ 1884 enum { 1885 EXT4_MF_MNTDIR_SAMPLED, 1886 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1887 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */ 1888 }; 1889 1890 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1891 { 1892 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1893 } 1894 1895 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1896 { 1897 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1898 } 1899 1900 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1901 { 1902 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1903 } 1904 1905 1906 /* 1907 * Simulate_fail codes 1908 */ 1909 #define EXT4_SIM_BBITMAP_EIO 1 1910 #define EXT4_SIM_BBITMAP_CRC 2 1911 #define EXT4_SIM_IBITMAP_EIO 3 1912 #define EXT4_SIM_IBITMAP_CRC 4 1913 #define EXT4_SIM_INODE_EIO 5 1914 #define EXT4_SIM_INODE_CRC 6 1915 #define EXT4_SIM_DIRBLOCK_EIO 7 1916 #define EXT4_SIM_DIRBLOCK_CRC 8 1917 1918 static inline bool ext4_simulate_fail(struct super_block *sb, 1919 unsigned long code) 1920 { 1921 #ifdef CONFIG_EXT4_DEBUG 1922 struct ext4_sb_info *sbi = EXT4_SB(sb); 1923 1924 if (unlikely(sbi->s_simulate_fail == code)) { 1925 sbi->s_simulate_fail = 0; 1926 return true; 1927 } 1928 #endif 1929 return false; 1930 } 1931 1932 /* 1933 * Error number codes for s_{first,last}_error_errno 1934 * 1935 * Linux errno numbers are architecture specific, so we need to translate 1936 * them into something which is architecture independent. We don't define 1937 * codes for all errno's; just the ones which are most likely to be the cause 1938 * of an ext4_error() call. 1939 */ 1940 #define EXT4_ERR_UNKNOWN 1 1941 #define EXT4_ERR_EIO 2 1942 #define EXT4_ERR_ENOMEM 3 1943 #define EXT4_ERR_EFSBADCRC 4 1944 #define EXT4_ERR_EFSCORRUPTED 5 1945 #define EXT4_ERR_ENOSPC 6 1946 #define EXT4_ERR_ENOKEY 7 1947 #define EXT4_ERR_EROFS 8 1948 #define EXT4_ERR_EFBIG 9 1949 #define EXT4_ERR_EEXIST 10 1950 #define EXT4_ERR_ERANGE 11 1951 #define EXT4_ERR_EOVERFLOW 12 1952 #define EXT4_ERR_EBUSY 13 1953 #define EXT4_ERR_ENOTDIR 14 1954 #define EXT4_ERR_ENOTEMPTY 15 1955 #define EXT4_ERR_ESHUTDOWN 16 1956 #define EXT4_ERR_EFAULT 17 1957 1958 /* 1959 * Inode dynamic state flags 1960 */ 1961 enum { 1962 EXT4_STATE_NEW, /* inode is newly created */ 1963 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1964 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1965 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1966 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1967 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1968 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1969 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1970 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1971 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1972 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1973 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */ 1974 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1975 }; 1976 1977 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1978 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1979 { \ 1980 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1981 } \ 1982 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1983 { \ 1984 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1985 } \ 1986 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1987 { \ 1988 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1989 } 1990 1991 /* Add these declarations here only so that these functions can be 1992 * found by name. Otherwise, they are very hard to locate. */ 1993 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1994 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1995 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1996 EXT4_INODE_BIT_FNS(flag, flags, 0) 1997 1998 /* Add these declarations here only so that these functions can be 1999 * found by name. Otherwise, they are very hard to locate. */ 2000 static inline int ext4_test_inode_state(struct inode *inode, int bit); 2001 static inline void ext4_set_inode_state(struct inode *inode, int bit); 2002 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 2003 #if (BITS_PER_LONG < 64) 2004 EXT4_INODE_BIT_FNS(state, state_flags, 0) 2005 2006 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 2007 { 2008 (ei)->i_state_flags = 0; 2009 } 2010 #else 2011 EXT4_INODE_BIT_FNS(state, flags, 32) 2012 2013 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 2014 { 2015 /* We depend on the fact that callers will set i_flags */ 2016 } 2017 #endif 2018 #else 2019 /* Assume that user mode programs are passing in an ext4fs superblock, not 2020 * a kernel struct super_block. This will allow us to call the feature-test 2021 * macros from user land. */ 2022 #define EXT4_SB(sb) (sb) 2023 #endif 2024 2025 static inline bool ext4_verity_in_progress(struct inode *inode) 2026 { 2027 return IS_ENABLED(CONFIG_FS_VERITY) && 2028 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 2029 } 2030 2031 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 2032 2033 /* 2034 * Check whether the inode is tracked as orphan (either in orphan file or 2035 * orphan list). 2036 */ 2037 static inline bool ext4_inode_orphan_tracked(struct inode *inode) 2038 { 2039 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) || 2040 !list_empty(&EXT4_I(inode)->i_orphan); 2041 } 2042 2043 /* 2044 * Codes for operating systems 2045 */ 2046 #define EXT4_OS_LINUX 0 2047 #define EXT4_OS_HURD 1 2048 #define EXT4_OS_MASIX 2 2049 #define EXT4_OS_FREEBSD 3 2050 #define EXT4_OS_LITES 4 2051 2052 /* 2053 * Revision levels 2054 */ 2055 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 2056 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 2057 2058 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 2059 2060 #define EXT4_GOOD_OLD_INODE_SIZE 128 2061 2062 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 2063 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 2064 #define EXT4_TIMESTAMP_MIN S32_MIN 2065 2066 /* 2067 * Feature set definitions 2068 */ 2069 2070 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 2071 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2072 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2073 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2074 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2075 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2076 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2077 /* 2078 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2079 * incompatible only if fast commit blocks are present in the FS. Since we 2080 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2081 * incompatible. We also have a JBD2 incompat feature, which gets set when 2082 * there are fast commit blocks present in the journal. 2083 */ 2084 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2085 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2086 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2087 2088 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2089 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2090 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2091 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2092 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2093 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2094 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2095 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2096 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2097 /* 2098 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2099 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2100 * all other data structures' checksums. However, the METADATA_CSUM and 2101 * GDT_CSUM bits are mutually exclusive. 2102 */ 2103 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2104 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2105 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2106 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2107 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2108 non-empty */ 2109 2110 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2111 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2112 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2113 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2114 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2115 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2116 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2117 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2118 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2119 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2120 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2121 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2122 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2123 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2124 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2125 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2126 2127 extern void ext4_update_dynamic_rev(struct super_block *sb); 2128 2129 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2130 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2131 { \ 2132 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2133 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2134 } \ 2135 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2136 { \ 2137 ext4_update_dynamic_rev(sb); \ 2138 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2139 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2140 } \ 2141 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2142 { \ 2143 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2144 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2145 } 2146 2147 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2148 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2149 { \ 2150 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2151 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2152 } \ 2153 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2154 { \ 2155 ext4_update_dynamic_rev(sb); \ 2156 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2157 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2158 } \ 2159 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2160 { \ 2161 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2162 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2163 } 2164 2165 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2166 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2167 { \ 2168 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2169 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2170 } \ 2171 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2172 { \ 2173 ext4_update_dynamic_rev(sb); \ 2174 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2175 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2176 } \ 2177 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2178 { \ 2179 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2180 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2181 } 2182 2183 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2184 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2185 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2186 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2187 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2188 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2189 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2190 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2191 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2192 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2193 2194 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2195 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2196 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2197 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2198 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2199 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2200 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2201 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2202 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2203 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2204 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2205 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2206 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2207 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2208 2209 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2210 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2211 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2212 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2213 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2214 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2215 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2216 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2217 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2218 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2219 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2220 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2221 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2222 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2223 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2224 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2225 2226 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2227 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2228 EXT4_FEATURE_INCOMPAT_META_BG) 2229 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2230 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2231 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2232 2233 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2234 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2235 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2236 EXT4_FEATURE_INCOMPAT_META_BG) 2237 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2238 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2239 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2240 2241 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2242 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2243 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2244 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2245 EXT4_FEATURE_INCOMPAT_META_BG| \ 2246 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2247 EXT4_FEATURE_INCOMPAT_64BIT| \ 2248 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2249 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2250 EXT4_FEATURE_INCOMPAT_MMP | \ 2251 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2252 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2253 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2254 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2255 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2256 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2257 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2258 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2259 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2260 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2261 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2262 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2263 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2264 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2265 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2266 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2267 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2268 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2269 2270 #define EXTN_FEATURE_FUNCS(ver) \ 2271 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2272 { \ 2273 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2274 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2275 } \ 2276 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2277 { \ 2278 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2279 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2280 } \ 2281 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2282 { \ 2283 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2284 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2285 } 2286 2287 EXTN_FEATURE_FUNCS(2) 2288 EXTN_FEATURE_FUNCS(3) 2289 EXTN_FEATURE_FUNCS(4) 2290 2291 static inline bool ext4_has_compat_features(struct super_block *sb) 2292 { 2293 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2294 } 2295 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2296 { 2297 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2298 } 2299 static inline bool ext4_has_incompat_features(struct super_block *sb) 2300 { 2301 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2302 } 2303 2304 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2305 2306 /* 2307 * Superblock flags 2308 */ 2309 enum { 2310 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */ 2311 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */ 2312 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */ 2313 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */ 2314 }; 2315 2316 static inline int ext4_forced_shutdown(struct super_block *sb) 2317 { 2318 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2319 } 2320 2321 static inline int ext4_emergency_ro(struct super_block *sb) 2322 { 2323 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags); 2324 } 2325 2326 static inline int ext4_emergency_state(struct super_block *sb) 2327 { 2328 if (unlikely(ext4_forced_shutdown(sb))) 2329 return -EIO; 2330 if (unlikely(ext4_emergency_ro(sb))) 2331 return -EROFS; 2332 return 0; 2333 } 2334 2335 /* 2336 * Default values for user and/or group using reserved blocks 2337 */ 2338 #define EXT4_DEF_RESUID 0 2339 #define EXT4_DEF_RESGID 0 2340 2341 /* 2342 * Default project ID 2343 */ 2344 #define EXT4_DEF_PROJID 0 2345 2346 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2347 2348 /* 2349 * Default mount options 2350 */ 2351 #define EXT4_DEFM_DEBUG 0x0001 2352 #define EXT4_DEFM_BSDGROUPS 0x0002 2353 #define EXT4_DEFM_XATTR_USER 0x0004 2354 #define EXT4_DEFM_ACL 0x0008 2355 #define EXT4_DEFM_UID16 0x0010 2356 #define EXT4_DEFM_JMODE 0x0060 2357 #define EXT4_DEFM_JMODE_DATA 0x0020 2358 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2359 #define EXT4_DEFM_JMODE_WBACK 0x0060 2360 #define EXT4_DEFM_NOBARRIER 0x0100 2361 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2362 #define EXT4_DEFM_DISCARD 0x0400 2363 #define EXT4_DEFM_NODELALLOC 0x0800 2364 2365 /* 2366 * Default journal batch times and ioprio. 2367 */ 2368 #define EXT4_DEF_MIN_BATCH_TIME 0 2369 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2370 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) 2371 2372 2373 /* 2374 * Default values for superblock update 2375 */ 2376 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */ 2377 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */ 2378 2379 /* 2380 * Minimum number of groups in a flexgroup before we separate out 2381 * directories into the first block group of a flexgroup 2382 */ 2383 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2384 2385 /* 2386 * Structure of a directory entry 2387 */ 2388 #define EXT4_NAME_LEN 255 2389 /* 2390 * Base length of the ext4 directory entry excluding the name length 2391 */ 2392 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2393 2394 struct ext4_dir_entry { 2395 __le32 inode; /* Inode number */ 2396 __le16 rec_len; /* Directory entry length */ 2397 __le16 name_len; /* Name length */ 2398 char name[EXT4_NAME_LEN]; /* File name */ 2399 }; 2400 2401 2402 /* 2403 * Encrypted Casefolded entries require saving the hash on disk. This structure 2404 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2405 * boundary. 2406 */ 2407 struct ext4_dir_entry_hash { 2408 __le32 hash; 2409 __le32 minor_hash; 2410 }; 2411 2412 /* 2413 * The new version of the directory entry. Since EXT4 structures are 2414 * stored in intel byte order, and the name_len field could never be 2415 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2416 * file_type field. 2417 */ 2418 struct ext4_dir_entry_2 { 2419 __le32 inode; /* Inode number */ 2420 __le16 rec_len; /* Directory entry length */ 2421 __u8 name_len; /* Name length */ 2422 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2423 char name[EXT4_NAME_LEN]; /* File name */ 2424 }; 2425 2426 /* 2427 * Access the hashes at the end of ext4_dir_entry_2 2428 */ 2429 #define EXT4_DIRENT_HASHES(entry) \ 2430 ((struct ext4_dir_entry_hash *) \ 2431 (((void *)(entry)) + \ 2432 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2433 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash) 2434 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2435 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash) 2436 2437 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2438 { 2439 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2440 } 2441 2442 /* 2443 * This is a bogus directory entry at the end of each leaf block that 2444 * records checksums. 2445 */ 2446 struct ext4_dir_entry_tail { 2447 __le32 det_reserved_zero1; /* Pretend to be unused */ 2448 __le16 det_rec_len; /* 12 */ 2449 __u8 det_reserved_zero2; /* Zero name length */ 2450 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2451 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2452 }; 2453 2454 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2455 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2456 ((blocksize) - \ 2457 sizeof(struct ext4_dir_entry_tail)))) 2458 2459 /* 2460 * Ext4 directory file types. Only the low 3 bits are used. The 2461 * other bits are reserved for now. 2462 */ 2463 #define EXT4_FT_UNKNOWN 0 2464 #define EXT4_FT_REG_FILE 1 2465 #define EXT4_FT_DIR 2 2466 #define EXT4_FT_CHRDEV 3 2467 #define EXT4_FT_BLKDEV 4 2468 #define EXT4_FT_FIFO 5 2469 #define EXT4_FT_SOCK 6 2470 #define EXT4_FT_SYMLINK 7 2471 2472 #define EXT4_FT_MAX 8 2473 2474 #define EXT4_FT_DIR_CSUM 0xDE 2475 2476 /* 2477 * EXT4_DIR_PAD defines the directory entries boundaries 2478 * 2479 * NOTE: It must be a multiple of 4 2480 */ 2481 #define EXT4_DIR_PAD 4 2482 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2483 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2484 2485 /* 2486 * The rec_len is dependent on the type of directory. Directories that are 2487 * casefolded and encrypted need to store the hash as well, so we add room for 2488 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2489 * pass NULL for dir, as those entries do not use the extra fields. 2490 */ 2491 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2492 const struct inode *dir) 2493 { 2494 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2495 2496 if (dir && ext4_hash_in_dirent(dir)) 2497 rec_len += sizeof(struct ext4_dir_entry_hash); 2498 return (rec_len & ~EXT4_DIR_ROUND); 2499 } 2500 2501 static inline unsigned int 2502 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2503 { 2504 unsigned len = le16_to_cpu(dlen); 2505 2506 if (len == EXT4_MAX_REC_LEN || len == 0) 2507 return blocksize; 2508 return (len & 65532) | ((len & 3) << 16); 2509 } 2510 2511 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2512 { 2513 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2514 if (len < 65536) 2515 return cpu_to_le16(len); 2516 if (len == blocksize) { 2517 if (blocksize == 65536) 2518 return cpu_to_le16(EXT4_MAX_REC_LEN); 2519 else 2520 return cpu_to_le16(0); 2521 } 2522 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2523 } 2524 2525 /* 2526 * Hash Tree Directory indexing 2527 * (c) Daniel Phillips, 2001 2528 */ 2529 2530 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2531 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2532 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2533 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2534 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2535 2536 /* Legal values for the dx_root hash_version field: */ 2537 2538 #define DX_HASH_LEGACY 0 2539 #define DX_HASH_HALF_MD4 1 2540 #define DX_HASH_TEA 2 2541 #define DX_HASH_LEGACY_UNSIGNED 3 2542 #define DX_HASH_HALF_MD4_UNSIGNED 4 2543 #define DX_HASH_TEA_UNSIGNED 5 2544 #define DX_HASH_SIPHASH 6 2545 #define DX_HASH_LAST DX_HASH_SIPHASH 2546 2547 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length) 2548 { 2549 return crc32c(crc, address, length); 2550 } 2551 2552 #ifdef __KERNEL__ 2553 2554 /* hash info structure used by the directory hash */ 2555 struct dx_hash_info 2556 { 2557 u32 hash; 2558 u32 minor_hash; 2559 int hash_version; 2560 u32 *seed; 2561 }; 2562 2563 2564 /* 32 and 64 bit signed EOF for dx directories */ 2565 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2566 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2567 2568 2569 /* 2570 * Control parameters used by ext4_htree_next_block 2571 */ 2572 #define HASH_NB_ALWAYS 1 2573 2574 struct ext4_filename { 2575 const struct qstr *usr_fname; 2576 struct fscrypt_str disk_name; 2577 struct dx_hash_info hinfo; 2578 #ifdef CONFIG_FS_ENCRYPTION 2579 struct fscrypt_str crypto_buf; 2580 #endif 2581 #if IS_ENABLED(CONFIG_UNICODE) 2582 struct qstr cf_name; 2583 #endif 2584 }; 2585 2586 #define fname_name(p) ((p)->disk_name.name) 2587 #define fname_usr_name(p) ((p)->usr_fname->name) 2588 #define fname_len(p) ((p)->disk_name.len) 2589 2590 /* 2591 * Describe an inode's exact location on disk and in memory 2592 */ 2593 struct ext4_iloc 2594 { 2595 struct buffer_head *bh; 2596 unsigned long offset; 2597 ext4_group_t block_group; 2598 }; 2599 2600 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2601 { 2602 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2603 } 2604 2605 static inline bool ext4_is_quota_file(struct inode *inode) 2606 { 2607 return IS_NOQUOTA(inode) && 2608 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2609 } 2610 2611 /* 2612 * This structure is stuffed into the struct file's private_data field 2613 * for directories. It is where we put information so that we can do 2614 * readdir operations in hash tree order. 2615 */ 2616 struct dir_private_info { 2617 struct rb_root root; 2618 struct rb_node *curr_node; 2619 struct fname *extra_fname; 2620 loff_t last_pos; 2621 __u32 curr_hash; 2622 __u32 curr_minor_hash; 2623 __u32 next_hash; 2624 u64 cookie; 2625 bool initialized; 2626 }; 2627 2628 /* calculate the first block number of the group */ 2629 static inline ext4_fsblk_t 2630 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2631 { 2632 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2633 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2634 } 2635 2636 /* 2637 * Special error return code only used by dx_probe() and its callers. 2638 */ 2639 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2640 2641 /* htree levels for ext4 */ 2642 #define EXT4_HTREE_LEVEL_COMPAT 2 2643 #define EXT4_HTREE_LEVEL 3 2644 2645 static inline int ext4_dir_htree_level(struct super_block *sb) 2646 { 2647 return ext4_has_feature_largedir(sb) ? 2648 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2649 } 2650 2651 /* 2652 * Timeout and state flag for lazy initialization inode thread. 2653 */ 2654 #define EXT4_DEF_LI_WAIT_MULT 10 2655 #define EXT4_DEF_LI_MAX_START_DELAY 5 2656 #define EXT4_LAZYINIT_QUIT 0x0001 2657 #define EXT4_LAZYINIT_RUNNING 0x0002 2658 2659 /* 2660 * Lazy inode table initialization info 2661 */ 2662 struct ext4_lazy_init { 2663 unsigned long li_state; 2664 struct list_head li_request_list; 2665 struct mutex li_list_mtx; 2666 }; 2667 2668 enum ext4_li_mode { 2669 EXT4_LI_MODE_PREFETCH_BBITMAP, 2670 EXT4_LI_MODE_ITABLE, 2671 }; 2672 2673 struct ext4_li_request { 2674 struct super_block *lr_super; 2675 enum ext4_li_mode lr_mode; 2676 ext4_group_t lr_first_not_zeroed; 2677 ext4_group_t lr_next_group; 2678 struct list_head lr_request; 2679 unsigned long lr_next_sched; 2680 unsigned long lr_timeout; 2681 }; 2682 2683 struct ext4_features { 2684 struct kobject f_kobj; 2685 struct completion f_kobj_unregister; 2686 }; 2687 2688 /* 2689 * This structure will be used for multiple mount protection. It will be 2690 * written into the block number saved in the s_mmp_block field in the 2691 * superblock. Programs that check MMP should assume that if 2692 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2693 * to use the filesystem, regardless of how old the timestamp is. 2694 */ 2695 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2696 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2697 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2698 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2699 2700 struct mmp_struct { 2701 __le32 mmp_magic; /* Magic number for MMP */ 2702 __le32 mmp_seq; /* Sequence no. updated periodically */ 2703 2704 /* 2705 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2706 * purposes and do not affect the correctness of the algorithm 2707 */ 2708 __le64 mmp_time; /* Time last updated */ 2709 char mmp_nodename[64]; /* Node which last updated MMP block */ 2710 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2711 2712 /* 2713 * mmp_check_interval is used to verify if the MMP block has been 2714 * updated on the block device. The value is updated based on the 2715 * maximum time to write the MMP block during an update cycle. 2716 */ 2717 __le16 mmp_check_interval; 2718 2719 __le16 mmp_pad1; 2720 __le32 mmp_pad2[226]; 2721 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2722 }; 2723 2724 /* arguments passed to the mmp thread */ 2725 struct mmpd_data { 2726 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2727 struct super_block *sb; /* super block of the fs */ 2728 }; 2729 2730 /* 2731 * Check interval multiplier 2732 * The MMP block is written every update interval and initially checked every 2733 * update interval x the multiplier (the value is then adapted based on the 2734 * write latency). The reason is that writes can be delayed under load and we 2735 * don't want readers to incorrectly assume that the filesystem is no longer 2736 * in use. 2737 */ 2738 #define EXT4_MMP_CHECK_MULT 2UL 2739 2740 /* 2741 * Minimum interval for MMP checking in seconds. 2742 */ 2743 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2744 2745 /* 2746 * Maximum interval for MMP checking in seconds. 2747 */ 2748 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2749 2750 /* 2751 * Function prototypes 2752 */ 2753 2754 /* 2755 * Ok, these declarations are also in <linux/kernel.h> but none of the 2756 * ext4 source programs needs to include it so they are duplicated here. 2757 */ 2758 # define NORET_TYPE /**/ 2759 # define ATTRIB_NORET __attribute__((noreturn)) 2760 # define NORET_AND noreturn, 2761 2762 /* bitmap.c */ 2763 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2764 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2765 struct ext4_group_desc *gdp, 2766 struct buffer_head *bh); 2767 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2768 struct ext4_group_desc *gdp, 2769 struct buffer_head *bh); 2770 void ext4_block_bitmap_csum_set(struct super_block *sb, 2771 struct ext4_group_desc *gdp, 2772 struct buffer_head *bh); 2773 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2774 struct ext4_group_desc *gdp, 2775 struct buffer_head *bh); 2776 2777 /* balloc.c */ 2778 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2779 ext4_fsblk_t blocknr, 2780 ext4_group_t *blockgrpp, 2781 ext4_grpblk_t *offsetp); 2782 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2783 ext4_fsblk_t block); 2784 2785 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2786 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2787 ext4_group_t group); 2788 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2789 ext4_fsblk_t goal, 2790 unsigned int flags, 2791 unsigned long *count, 2792 int *errp); 2793 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2794 s64 nclusters, unsigned int flags); 2795 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2796 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2797 ext4_group_t block_group, 2798 struct buffer_head ** bh); 2799 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2800 ext4_group_t group); 2801 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2802 2803 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2804 ext4_group_t block_group, 2805 bool ignore_locked); 2806 extern int ext4_wait_block_bitmap(struct super_block *sb, 2807 ext4_group_t block_group, 2808 struct buffer_head *bh); 2809 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2810 ext4_group_t block_group); 2811 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2812 ext4_group_t block_group, 2813 struct ext4_group_desc *gdp); 2814 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2815 2816 #if IS_ENABLED(CONFIG_UNICODE) 2817 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2818 const struct qstr *iname, 2819 struct ext4_filename *fname); 2820 2821 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2822 { 2823 kfree(fname->cf_name.name); 2824 fname->cf_name.name = NULL; 2825 } 2826 #else 2827 static inline int ext4_fname_setup_ci_filename(struct inode *dir, 2828 const struct qstr *iname, 2829 struct ext4_filename *fname) 2830 { 2831 return 0; 2832 } 2833 2834 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2835 { 2836 } 2837 #endif 2838 2839 /* ext4 encryption related stuff goes here crypto.c */ 2840 #ifdef CONFIG_FS_ENCRYPTION 2841 extern const struct fscrypt_operations ext4_cryptops; 2842 2843 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2844 int lookup, struct ext4_filename *fname); 2845 2846 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2847 struct ext4_filename *fname); 2848 2849 void ext4_fname_free_filename(struct ext4_filename *fname); 2850 2851 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2852 2853 #else /* !CONFIG_FS_ENCRYPTION */ 2854 static inline int ext4_fname_setup_filename(struct inode *dir, 2855 const struct qstr *iname, 2856 int lookup, 2857 struct ext4_filename *fname) 2858 { 2859 fname->usr_fname = iname; 2860 fname->disk_name.name = (unsigned char *) iname->name; 2861 fname->disk_name.len = iname->len; 2862 2863 return ext4_fname_setup_ci_filename(dir, iname, fname); 2864 } 2865 2866 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2867 struct dentry *dentry, 2868 struct ext4_filename *fname) 2869 { 2870 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2871 } 2872 2873 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2874 { 2875 ext4_fname_free_ci_filename(fname); 2876 } 2877 2878 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2879 void __user *arg) 2880 { 2881 return -EOPNOTSUPP; 2882 } 2883 #endif /* !CONFIG_FS_ENCRYPTION */ 2884 2885 /* dir.c */ 2886 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2887 struct file *, 2888 struct ext4_dir_entry_2 *, 2889 struct buffer_head *, char *, int, 2890 unsigned int); 2891 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2892 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2893 (de), (bh), (buf), (size), (offset))) 2894 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2895 __u32 minor_hash, 2896 struct ext4_dir_entry_2 *dirent, 2897 struct fscrypt_str *ent_name); 2898 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2899 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh, 2900 void *buf, int buf_size, 2901 struct ext4_filename *fname, 2902 struct ext4_dir_entry_2 **dest_de); 2903 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2904 struct ext4_dir_entry_2 *de, 2905 int buf_size, 2906 struct ext4_filename *fname); 2907 static inline void ext4_update_dx_flag(struct inode *inode) 2908 { 2909 if (!ext4_has_feature_dir_index(inode->i_sb) && 2910 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2911 /* ext4_iget() should have caught this... */ 2912 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2913 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2914 } 2915 } 2916 static const unsigned char ext4_filetype_table[] = { 2917 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2918 }; 2919 2920 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2921 { 2922 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2923 return DT_UNKNOWN; 2924 2925 return ext4_filetype_table[filetype]; 2926 } 2927 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2928 void *buf, int buf_size); 2929 2930 /* fsync.c */ 2931 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2932 2933 /* hash.c */ 2934 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2935 struct dx_hash_info *hinfo); 2936 2937 /* ialloc.c */ 2938 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2939 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2940 struct inode *, umode_t, 2941 const struct qstr *qstr, __u32 goal, 2942 uid_t *owner, __u32 i_flags, 2943 int handle_type, unsigned int line_no, 2944 int nblocks); 2945 2946 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2947 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2948 (goal), (owner), i_flags, 0, 0, 0) 2949 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2950 type, nblocks) \ 2951 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2952 0, (type), __LINE__, (nblocks)) 2953 2954 2955 extern void ext4_free_inode(handle_t *, struct inode *); 2956 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2957 extern unsigned long ext4_count_free_inodes(struct super_block *); 2958 extern unsigned long ext4_count_dirs(struct super_block *); 2959 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2960 extern int ext4_init_inode_table(struct super_block *sb, 2961 ext4_group_t group, int barrier); 2962 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2963 2964 /* fast_commit.c */ 2965 int ext4_fc_info_show(struct seq_file *seq, void *v); 2966 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2967 void ext4_fc_init_inode(struct inode *inode); 2968 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2969 ext4_lblk_t end); 2970 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2971 struct dentry *dentry); 2972 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2973 struct dentry *dentry); 2974 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2975 void ext4_fc_track_link(handle_t *handle, struct inode *inode, 2976 struct dentry *dentry); 2977 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2978 struct dentry *dentry); 2979 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2980 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2981 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2982 void ext4_fc_del(struct inode *inode); 2983 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2984 void ext4_fc_replay_cleanup(struct super_block *sb); 2985 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2986 int __init ext4_fc_init_dentry_cache(void); 2987 void ext4_fc_destroy_dentry_cache(void); 2988 int ext4_fc_record_regions(struct super_block *sb, int ino, 2989 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2990 int len, int replay); 2991 2992 /* mballoc.c */ 2993 extern const struct seq_operations ext4_mb_seq_groups_ops; 2994 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2995 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2996 extern int ext4_mb_init(struct super_block *); 2997 extern void ext4_mb_release(struct super_block *); 2998 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2999 struct ext4_allocation_request *, int *); 3000 extern void ext4_discard_preallocations(struct inode *); 3001 extern int __init ext4_init_mballoc(void); 3002 extern void ext4_exit_mballoc(void); 3003 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 3004 ext4_group_t group, 3005 unsigned int nr, int *cnt); 3006 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 3007 unsigned int nr); 3008 3009 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 3010 struct buffer_head *bh, ext4_fsblk_t block, 3011 unsigned long count, int flags); 3012 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 3013 ext4_group_t ngroups); 3014 extern int ext4_mb_add_groupinfo(struct super_block *sb, 3015 ext4_group_t i, struct ext4_group_desc *desc); 3016 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 3017 ext4_fsblk_t block, unsigned long count); 3018 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 3019 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 3020 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 3021 int len, bool state); 3022 static inline bool ext4_mb_cr_expensive(enum criteria cr) 3023 { 3024 return cr >= CR_GOAL_LEN_SLOW; 3025 } 3026 3027 /* inode.c */ 3028 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 3029 struct ext4_inode_info *ei); 3030 int ext4_inode_is_fast_symlink(struct inode *inode); 3031 void ext4_check_map_extents_env(struct inode *inode); 3032 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 3033 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 3034 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 3035 bool wait, struct buffer_head **bhs); 3036 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 3037 struct buffer_head *bh_result, int create); 3038 int ext4_get_block(struct inode *inode, sector_t iblock, 3039 struct buffer_head *bh_result, int create); 3040 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 3041 struct buffer_head *bh, int create); 3042 int ext4_walk_page_buffers(handle_t *handle, 3043 struct inode *inode, 3044 struct buffer_head *head, 3045 unsigned from, 3046 unsigned to, 3047 int *partial, 3048 int (*fn)(handle_t *handle, struct inode *inode, 3049 struct buffer_head *bh)); 3050 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 3051 struct buffer_head *bh); 3052 void ext4_set_inode_mapping_order(struct inode *inode); 3053 #define FALL_BACK_TO_NONDELALLOC 1 3054 #define CONVERT_INLINE_DATA 2 3055 3056 typedef enum { 3057 EXT4_IGET_NORMAL = 0, 3058 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3059 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 3060 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 3061 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 3062 } ext4_iget_flags; 3063 3064 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3065 ext4_iget_flags flags, const char *function, 3066 unsigned int line); 3067 3068 #define ext4_iget(sb, ino, flags) \ 3069 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3070 3071 extern int ext4_write_inode(struct inode *, struct writeback_control *); 3072 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 3073 struct iattr *); 3074 extern u32 ext4_dio_alignment(struct inode *inode); 3075 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 3076 struct kstat *, u32, unsigned int); 3077 extern void ext4_evict_inode(struct inode *); 3078 extern void ext4_clear_inode(struct inode *); 3079 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 3080 struct kstat *, u32, unsigned int); 3081 extern void ext4_dirty_inode(struct inode *, int); 3082 extern int ext4_change_inode_journal_flag(struct inode *, int); 3083 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3084 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3085 struct ext4_iloc *iloc); 3086 extern int ext4_inode_attach_jinode(struct inode *inode); 3087 extern int ext4_can_truncate(struct inode *inode); 3088 extern int ext4_truncate(struct inode *); 3089 extern int ext4_break_layouts(struct inode *); 3090 extern int ext4_truncate_page_cache_block_range(struct inode *inode, 3091 loff_t start, loff_t end); 3092 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3093 extern void ext4_set_inode_flags(struct inode *, bool init); 3094 extern int ext4_alloc_da_blocks(struct inode *inode); 3095 extern void ext4_set_aops(struct inode *inode); 3096 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3097 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3098 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks); 3099 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks, 3100 int pextents); 3101 extern int ext4_block_zero_eof(struct inode *inode, loff_t from, loff_t end); 3102 extern int ext4_zero_partial_blocks(struct inode *inode, loff_t lstart, 3103 loff_t length, bool *did_zero); 3104 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3105 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3106 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3107 extern void ext4_da_release_space(struct inode *inode, int to_free); 3108 extern void ext4_da_update_reserve_space(struct inode *inode, 3109 int used, int quota_claim); 3110 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3111 ext4_fsblk_t pblk, ext4_lblk_t len); 3112 3113 static inline bool is_special_ino(struct super_block *sb, unsigned long ino) 3114 { 3115 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 3116 3117 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) || 3118 ino == le32_to_cpu(es->s_usr_quota_inum) || 3119 ino == le32_to_cpu(es->s_grp_quota_inum) || 3120 ino == le32_to_cpu(es->s_prj_quota_inum) || 3121 ino == le32_to_cpu(es->s_orphan_file_inum); 3122 } 3123 3124 /* indirect.c */ 3125 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3126 struct ext4_map_blocks *map, int flags); 3127 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3128 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3129 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3130 ext4_lblk_t start, ext4_lblk_t end); 3131 3132 /* ioctl.c */ 3133 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3134 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3135 int ext4_fileattr_set(struct mnt_idmap *idmap, 3136 struct dentry *dentry, struct file_kattr *fa); 3137 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3138 extern void ext4_reset_inode_seed(struct inode *inode); 3139 int ext4_update_overhead(struct super_block *sb, bool force); 3140 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3141 3142 /* migrate.c */ 3143 extern int ext4_ext_migrate(struct inode *); 3144 extern int ext4_ind_migrate(struct inode *inode); 3145 3146 /* namei.c */ 3147 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3148 struct inode *inode); 3149 extern int ext4_dirblock_csum_verify(struct inode *inode, 3150 struct buffer_head *bh); 3151 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3152 __u32 start_minor_hash, __u32 *next_hash); 3153 extern int ext4_search_dir(struct buffer_head *bh, 3154 char *search_buf, 3155 int buf_size, 3156 struct inode *dir, 3157 struct ext4_filename *fname, 3158 unsigned int offset, 3159 struct ext4_dir_entry_2 **res_dir); 3160 extern int ext4_generic_delete_entry(struct inode *dir, 3161 struct ext4_dir_entry_2 *de_del, 3162 struct buffer_head *bh, 3163 void *entry_buf, 3164 int buf_size, 3165 int csum_size); 3166 extern bool ext4_empty_dir(struct inode *inode); 3167 3168 /* resize.c */ 3169 extern void ext4_kvfree_array_rcu(void *to_free); 3170 extern int ext4_group_add(struct super_block *sb, 3171 struct ext4_new_group_data *input); 3172 extern int ext4_group_extend(struct super_block *sb, 3173 struct ext4_super_block *es, 3174 ext4_fsblk_t n_blocks_count); 3175 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3176 extern unsigned int ext4_list_backups(struct super_block *sb, 3177 unsigned int *three, unsigned int *five, 3178 unsigned int *seven); 3179 3180 /* super.c */ 3181 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3182 sector_t block, blk_opf_t op_flags); 3183 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3184 sector_t block); 3185 extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb, 3186 sector_t block); 3187 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3188 bh_end_io_t *end_io, bool simu_fail); 3189 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3190 bh_end_io_t *end_io, bool simu_fail); 3191 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3192 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3193 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3194 extern int ext4_calculate_overhead(struct super_block *sb); 3195 extern __le32 ext4_superblock_csum(struct ext4_super_block *es); 3196 extern void ext4_superblock_csum_set(struct super_block *sb); 3197 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3198 ext4_group_t ngroup); 3199 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3200 char nbuf[16]); 3201 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3202 ext4_group_t block_group, 3203 unsigned int flags); 3204 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3205 ext4_group_t block_group); 3206 extern void print_daily_error_info(struct timer_list *t); 3207 3208 extern __printf(7, 8) 3209 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3210 int, __u64, const char *, ...); 3211 extern __printf(6, 7) 3212 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3213 ext4_fsblk_t, int, const char *, ...); 3214 extern __printf(5, 6) 3215 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3216 const char *, ...); 3217 extern void __ext4_std_error(struct super_block *, const char *, 3218 unsigned int, int); 3219 extern __printf(4, 5) 3220 void __ext4_warning(struct super_block *, const char *, unsigned int, 3221 const char *, ...); 3222 extern __printf(4, 5) 3223 void __ext4_warning_inode(const struct inode *inode, const char *function, 3224 unsigned int line, const char *fmt, ...); 3225 extern __printf(3, 4) 3226 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3227 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3228 const char *, unsigned int, const char *); 3229 extern __printf(7, 8) 3230 void __ext4_grp_locked_error(const char *, unsigned int, 3231 struct super_block *, ext4_group_t, 3232 u64, ext4_fsblk_t, 3233 const char *, ...); 3234 3235 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3236 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3237 3238 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3239 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3240 3241 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3242 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3243 (fmt), ## a) 3244 3245 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3246 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3247 3248 #define ext4_abort(sb, err, fmt, a...) \ 3249 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3250 3251 #ifdef CONFIG_PRINTK 3252 3253 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3254 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3255 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3256 __ext4_error_inode((inode), (func), (line), (block), \ 3257 (err), (fmt), ##__VA_ARGS__) 3258 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3259 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3260 #define ext4_error(sb, fmt, ...) \ 3261 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3262 ##__VA_ARGS__) 3263 #define ext4_error_err(sb, err, fmt, ...) \ 3264 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3265 ##__VA_ARGS__) 3266 #define ext4_warning(sb, fmt, ...) \ 3267 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3268 #define ext4_warning_inode(inode, fmt, ...) \ 3269 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3270 #define ext4_msg(sb, level, fmt, ...) \ 3271 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3272 #define dump_mmp_msg(sb, mmp, msg) \ 3273 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3274 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3275 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3276 fmt, ##__VA_ARGS__) 3277 3278 #else 3279 3280 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3281 do { \ 3282 no_printk(fmt, ##__VA_ARGS__); \ 3283 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3284 } while (0) 3285 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3286 do { \ 3287 no_printk(fmt, ##__VA_ARGS__); \ 3288 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3289 } while (0) 3290 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3291 do { \ 3292 no_printk(fmt, ##__VA_ARGS__); \ 3293 __ext4_error_file(file, "", 0, block, " "); \ 3294 } while (0) 3295 #define ext4_error(sb, fmt, ...) \ 3296 do { \ 3297 no_printk(fmt, ##__VA_ARGS__); \ 3298 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3299 } while (0) 3300 #define ext4_error_err(sb, err, fmt, ...) \ 3301 do { \ 3302 no_printk(fmt, ##__VA_ARGS__); \ 3303 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3304 } while (0) 3305 #define ext4_warning(sb, fmt, ...) \ 3306 do { \ 3307 no_printk(fmt, ##__VA_ARGS__); \ 3308 __ext4_warning(sb, "", 0, " "); \ 3309 } while (0) 3310 #define ext4_warning_inode(inode, fmt, ...) \ 3311 do { \ 3312 no_printk(fmt, ##__VA_ARGS__); \ 3313 __ext4_warning_inode(inode, "", 0, " "); \ 3314 } while (0) 3315 #define ext4_msg(sb, level, fmt, ...) \ 3316 do { \ 3317 no_printk(fmt, ##__VA_ARGS__); \ 3318 __ext4_msg(sb, "", " "); \ 3319 } while (0) 3320 #define dump_mmp_msg(sb, mmp, msg) \ 3321 __dump_mmp_msg(sb, mmp, "", 0, "") 3322 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3323 do { \ 3324 no_printk(fmt, ##__VA_ARGS__); \ 3325 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3326 } while (0) 3327 3328 #endif 3329 3330 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3331 struct ext4_group_desc *bg); 3332 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3333 struct ext4_group_desc *bg); 3334 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3335 struct ext4_group_desc *bg); 3336 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3337 struct ext4_group_desc *bg); 3338 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3339 struct ext4_group_desc *bg); 3340 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3341 struct ext4_group_desc *bg); 3342 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3343 struct ext4_group_desc *bg); 3344 extern void ext4_block_bitmap_set(struct super_block *sb, 3345 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3346 extern void ext4_inode_bitmap_set(struct super_block *sb, 3347 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3348 extern void ext4_inode_table_set(struct super_block *sb, 3349 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3350 extern void ext4_free_group_clusters_set(struct super_block *sb, 3351 struct ext4_group_desc *bg, 3352 __u32 count); 3353 extern void ext4_free_inodes_set(struct super_block *sb, 3354 struct ext4_group_desc *bg, __u32 count); 3355 extern void ext4_used_dirs_set(struct super_block *sb, 3356 struct ext4_group_desc *bg, __u32 count); 3357 extern void ext4_itable_unused_set(struct super_block *sb, 3358 struct ext4_group_desc *bg, __u32 count); 3359 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3360 struct ext4_group_desc *gdp); 3361 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3362 struct ext4_group_desc *gdp); 3363 extern int ext4_register_li_request(struct super_block *sb, 3364 ext4_group_t first_not_zeroed); 3365 3366 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3367 { 3368 return ext4_has_feature_gdt_csum(sb) || 3369 ext4_has_feature_metadata_csum(sb); 3370 } 3371 3372 #define ext4_read_incompat_64bit_val(es, name) \ 3373 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3374 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3375 le32_to_cpu(es->name##_lo)) 3376 3377 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3378 { 3379 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3380 } 3381 3382 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3383 { 3384 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3385 } 3386 3387 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3388 { 3389 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3390 } 3391 3392 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3393 ext4_fsblk_t blk) 3394 { 3395 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3396 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3397 } 3398 3399 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3400 ext4_fsblk_t blk) 3401 { 3402 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3403 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3404 } 3405 3406 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3407 ext4_fsblk_t blk) 3408 { 3409 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3410 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3411 } 3412 3413 static inline loff_t ext4_isize(struct super_block *sb, 3414 struct ext4_inode *raw_inode) 3415 { 3416 if (ext4_has_feature_largedir(sb) || 3417 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3418 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3419 le32_to_cpu(raw_inode->i_size_lo); 3420 3421 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3422 } 3423 3424 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3425 { 3426 raw_inode->i_size_lo = cpu_to_le32(i_size); 3427 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3428 } 3429 3430 /* 3431 * Reading s_groups_count requires using smp_rmb() afterwards. See 3432 * the locking protocol documented in the comments of ext4_group_add() 3433 * in resize.c 3434 */ 3435 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3436 { 3437 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3438 3439 smp_rmb(); 3440 return ngroups; 3441 } 3442 3443 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3444 ext4_group_t block_group) 3445 { 3446 return block_group >> sbi->s_log_groups_per_flex; 3447 } 3448 3449 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3450 { 3451 return 1 << sbi->s_log_groups_per_flex; 3452 } 3453 3454 static inline loff_t ext4_get_maxbytes(struct inode *inode) 3455 { 3456 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) 3457 return inode->i_sb->s_maxbytes; 3458 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; 3459 } 3460 3461 #define ext4_std_error(sb, errno) \ 3462 do { \ 3463 if ((errno)) \ 3464 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3465 } while (0) 3466 3467 #ifdef CONFIG_SMP 3468 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3469 * counters. So we need to make sure we have free clusters more 3470 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3471 */ 3472 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3473 #else 3474 #define EXT4_FREECLUSTERS_WATERMARK 0 3475 #endif 3476 3477 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3478 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3479 { 3480 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3481 !inode_is_locked(inode)); 3482 down_write(&EXT4_I(inode)->i_data_sem); 3483 if (newsize > EXT4_I(inode)->i_disksize) 3484 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3485 up_write(&EXT4_I(inode)->i_data_sem); 3486 } 3487 3488 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3489 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3490 { 3491 int changed = 0; 3492 3493 if (newsize > inode->i_size) { 3494 i_size_write(inode, newsize); 3495 changed = 1; 3496 } 3497 if (newsize > EXT4_I(inode)->i_disksize) { 3498 ext4_update_i_disksize(inode, newsize); 3499 changed |= 2; 3500 } 3501 return changed; 3502 } 3503 3504 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3505 loff_t len); 3506 3507 struct ext4_group_info { 3508 unsigned long bb_state; 3509 #ifdef AGGRESSIVE_CHECK 3510 unsigned long bb_check_counter; 3511 #endif 3512 struct rb_root bb_free_root; 3513 ext4_grpblk_t bb_first_free; /* first free block */ 3514 ext4_grpblk_t bb_free; /* total free blocks */ 3515 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3516 int bb_avg_fragment_size_order; /* order of average 3517 fragment in BG */ 3518 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3519 ext4_group_t bb_group; /* Group number */ 3520 struct list_head bb_prealloc_list; 3521 #ifdef DOUBLE_CHECK 3522 void *bb_bitmap; 3523 #endif 3524 struct rw_semaphore alloc_sem; 3525 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3526 * regions, index is order. 3527 * bb_counters[3] = 5 means 3528 * 5 free 8-block regions. */ 3529 }; 3530 3531 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3532 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3533 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3534 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3535 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3536 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3537 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3538 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3539 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3540 3541 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3542 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3543 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3544 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3545 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3546 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3547 3548 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3549 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3550 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3551 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3552 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3553 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3554 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3555 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3556 3557 #define EXT4_MAX_CONTENTION 8 3558 #define EXT4_CONTENTION_THRESHOLD 2 3559 3560 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3561 ext4_group_t group) 3562 { 3563 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3564 } 3565 3566 /* 3567 * Returns true if the filesystem is busy enough that attempts to 3568 * access the block group locks has run into contention. 3569 */ 3570 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3571 { 3572 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3573 } 3574 3575 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group) 3576 { 3577 if (!spin_trylock(ext4_group_lock_ptr(sb, group))) 3578 return false; 3579 /* 3580 * We're able to grab the lock right away, so drop the lock 3581 * contention counter. 3582 */ 3583 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3584 return true; 3585 } 3586 3587 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3588 { 3589 if (!ext4_try_lock_group(sb, group)) { 3590 /* 3591 * The lock is busy, so bump the contention counter, 3592 * and then wait on the spin lock. 3593 */ 3594 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3595 EXT4_MAX_CONTENTION); 3596 spin_lock(ext4_group_lock_ptr(sb, group)); 3597 } 3598 } 3599 3600 static inline void ext4_unlock_group(struct super_block *sb, 3601 ext4_group_t group) 3602 { 3603 spin_unlock(ext4_group_lock_ptr(sb, group)); 3604 } 3605 3606 #ifdef CONFIG_QUOTA 3607 static inline bool ext4_quota_capable(struct super_block *sb) 3608 { 3609 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3610 } 3611 3612 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3613 { 3614 struct ext4_sb_info *sbi = EXT4_SB(sb); 3615 3616 return (ext4_has_feature_quota(sb) || 3617 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3618 } 3619 int ext4_enable_quotas(struct super_block *sb); 3620 #endif 3621 3622 /* 3623 * Block validity checking 3624 */ 3625 #define ext4_check_indirect_blockref(inode, bh) \ 3626 ext4_check_blockref(__func__, __LINE__, inode, \ 3627 (__le32 *)(bh)->b_data, \ 3628 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3629 3630 #define ext4_ind_check_inode(inode) \ 3631 ext4_check_blockref(__func__, __LINE__, inode, \ 3632 EXT4_I(inode)->i_data, \ 3633 EXT4_NDIR_BLOCKS) 3634 3635 /* 3636 * Inodes and files operations 3637 */ 3638 3639 /* dir.c */ 3640 extern const struct file_operations ext4_dir_operations; 3641 3642 /* file.c */ 3643 extern const struct inode_operations ext4_file_inode_operations; 3644 extern const struct file_operations ext4_file_operations; 3645 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3646 3647 /* inline.c */ 3648 extern int ext4_get_max_inline_size(struct inode *inode); 3649 extern int ext4_find_inline_data_nolock(struct inode *inode); 3650 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3651 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size); 3652 3653 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3654 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3655 struct inode *inode, 3656 loff_t pos, unsigned len, 3657 struct folio **foliop); 3658 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3659 unsigned copied, struct folio *folio); 3660 extern int ext4_generic_write_inline_data(struct address_space *mapping, 3661 struct inode *inode, 3662 loff_t pos, unsigned len, 3663 struct folio **foliop, 3664 void **fsdata, bool da); 3665 extern int ext4_try_add_inline_entry(handle_t *handle, 3666 struct ext4_filename *fname, 3667 struct inode *dir, struct inode *inode); 3668 extern int ext4_try_create_inline_dir(handle_t *handle, 3669 struct inode *parent, 3670 struct inode *inode); 3671 extern int ext4_read_inline_dir(struct file *filp, 3672 struct dir_context *ctx, 3673 int *has_inline_data); 3674 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3675 struct inode *dir, ext4_lblk_t block, 3676 struct dx_hash_info *hinfo, 3677 __u32 start_hash, __u32 start_minor_hash, 3678 int *has_inline_data); 3679 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3680 struct ext4_filename *fname, 3681 struct ext4_dir_entry_2 **res_dir, 3682 int *has_inline_data); 3683 extern int ext4_delete_inline_entry(handle_t *handle, 3684 struct inode *dir, 3685 struct ext4_dir_entry_2 *de_del, 3686 struct buffer_head *bh, 3687 int *has_inline_data); 3688 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3689 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3690 struct ext4_dir_entry_2 **parent_de, 3691 int *retval); 3692 extern void *ext4_read_inline_link(struct inode *inode); 3693 3694 struct iomap; 3695 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3696 3697 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3698 3699 extern int ext4_convert_inline_data(struct inode *inode); 3700 3701 static inline int ext4_has_inline_data(struct inode *inode) 3702 { 3703 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3704 EXT4_I(inode)->i_inline_off; 3705 } 3706 3707 /* namei.c */ 3708 extern const struct inode_operations ext4_dir_inode_operations; 3709 extern const struct inode_operations ext4_special_inode_operations; 3710 extern struct dentry *ext4_get_parent(struct dentry *child); 3711 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode, 3712 struct buffer_head *dir_block, 3713 unsigned int parent_ino, void *inline_buf, 3714 int inline_size); 3715 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3716 unsigned int blocksize); 3717 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3718 struct buffer_head *bh); 3719 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3720 struct inode *inode, struct dentry *dentry); 3721 extern int __ext4_link(struct inode *dir, struct inode *inode, 3722 const struct qstr *d_name, struct dentry *dentry); 3723 3724 #define S_SHIFT 12 3725 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3726 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3727 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3728 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3729 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3730 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3731 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3732 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3733 }; 3734 3735 static inline void ext4_set_de_type(struct super_block *sb, 3736 struct ext4_dir_entry_2 *de, 3737 umode_t mode) { 3738 if (ext4_has_feature_filetype(sb)) 3739 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3740 } 3741 3742 /* readpages.c */ 3743 int ext4_read_folio(struct file *file, struct folio *folio); 3744 void ext4_readahead(struct readahead_control *rac); 3745 extern int __init ext4_init_post_read_processing(void); 3746 extern void ext4_exit_post_read_processing(void); 3747 3748 /* symlink.c */ 3749 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3750 extern const struct inode_operations ext4_symlink_inode_operations; 3751 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3752 3753 /* sysfs.c */ 3754 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3755 extern int ext4_register_sysfs(struct super_block *sb); 3756 extern void ext4_unregister_sysfs(struct super_block *sb); 3757 extern int __init ext4_init_sysfs(void); 3758 extern void ext4_exit_sysfs(void); 3759 3760 /* block_validity */ 3761 extern void ext4_release_system_zone(struct super_block *sb); 3762 extern int ext4_setup_system_zone(struct super_block *sb); 3763 extern int __init ext4_init_system_zone(void); 3764 extern void ext4_exit_system_zone(void); 3765 extern int ext4_inode_block_valid(struct inode *inode, 3766 ext4_fsblk_t start_blk, 3767 unsigned int count); 3768 extern int ext4_check_blockref(const char *, unsigned int, 3769 struct inode *, __le32 *, unsigned int); 3770 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3771 ext4_fsblk_t start_blk, unsigned int count); 3772 3773 3774 /* extents.c */ 3775 struct ext4_ext_path; 3776 struct ext4_extent; 3777 3778 /* 3779 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3780 * __le32. 3781 */ 3782 #define EXT_MAX_BLOCKS 0xffffffff 3783 3784 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3785 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3786 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3787 struct ext4_map_blocks *map, int flags); 3788 extern int ext4_ext_truncate(handle_t *, struct inode *); 3789 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3790 ext4_lblk_t end); 3791 extern void ext4_ext_init(struct super_block *); 3792 extern void ext4_ext_release(struct super_block *); 3793 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3794 loff_t len); 3795 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3796 loff_t offset, ssize_t len); 3797 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle, 3798 struct inode *inode, loff_t offset, ssize_t len); 3799 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3800 ext4_io_end_t *io_end); 3801 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3802 struct ext4_map_blocks *map, int flags); 3803 extern int ext4_map_query_blocks(handle_t *handle, struct inode *inode, 3804 struct ext4_map_blocks *map, int flags); 3805 extern int ext4_map_create_blocks(handle_t *handle, struct inode *inode, 3806 struct ext4_map_blocks *map, int flags); 3807 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3808 int num, 3809 struct ext4_ext_path *path); 3810 extern struct ext4_ext_path *ext4_ext_insert_extent( 3811 handle_t *handle, struct inode *inode, 3812 struct ext4_ext_path *path, 3813 struct ext4_extent *newext, int gb_flags); 3814 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3815 struct ext4_ext_path *, 3816 int flags); 3817 extern void ext4_free_ext_path(struct ext4_ext_path *); 3818 extern int ext4_ext_check_inode(struct inode *inode); 3819 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3820 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3821 __u64 start, __u64 len); 3822 extern int ext4_get_es_cache(struct inode *inode, 3823 struct fiemap_extent_info *fieinfo, 3824 __u64 start, __u64 len); 3825 extern int ext4_ext_precache(struct inode *inode); 3826 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3827 struct inode *inode2, ext4_lblk_t lblk1, 3828 ext4_lblk_t lblk2, ext4_lblk_t count, 3829 int mark_unwritten,int *err); 3830 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3831 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3832 int check_cred, int restart_cred, 3833 int revoke_cred); 3834 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3835 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3836 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3837 int len, int unwritten, ext4_fsblk_t pblk); 3838 extern int ext4_ext_clear_bb(struct inode *inode); 3839 3840 3841 /* move_extent.c */ 3842 extern void ext4_double_down_write_data_sem(struct inode *first, 3843 struct inode *second); 3844 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3845 struct inode *donor_inode); 3846 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3847 __u64 start_orig, __u64 start_donor, 3848 __u64 len, __u64 *moved_len); 3849 3850 /* page-io.c */ 3851 extern int __init ext4_init_pageio(void); 3852 extern void ext4_exit_pageio(void); 3853 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3854 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3855 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3856 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3857 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3858 struct writeback_control *wbc); 3859 extern void ext4_end_io_rsv_work(struct work_struct *work); 3860 extern void ext4_io_submit(struct ext4_io_submit *io); 3861 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3862 size_t len); 3863 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3864 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3865 3866 /* mmp.c */ 3867 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3868 3869 /* mmp.c */ 3870 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3871 3872 /* verity.c */ 3873 extern const struct fsverity_operations ext4_verityops; 3874 3875 /* orphan.c */ 3876 extern int ext4_orphan_add(handle_t *, struct inode *); 3877 extern int ext4_orphan_del(handle_t *, struct inode *); 3878 extern void ext4_orphan_cleanup(struct super_block *sb, 3879 struct ext4_super_block *es); 3880 extern void ext4_release_orphan_info(struct super_block *sb); 3881 extern int ext4_init_orphan_info(struct super_block *sb); 3882 extern int ext4_orphan_file_empty(struct super_block *sb); 3883 extern void ext4_orphan_file_block_trigger( 3884 struct jbd2_buffer_trigger_type *triggers, 3885 struct buffer_head *bh, 3886 void *data, size_t size); 3887 3888 /* 3889 * Add new method to test whether block and inode bitmaps are properly 3890 * initialized. With uninit_bg reading the block from disk is not enough 3891 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3892 */ 3893 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3894 3895 static inline int bitmap_uptodate(struct buffer_head *bh) 3896 { 3897 return (buffer_uptodate(bh) && 3898 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3899 } 3900 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3901 { 3902 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3903 } 3904 3905 extern int ext4_resize_begin(struct super_block *sb); 3906 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3907 3908 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end) 3909 { 3910 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) 3911 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3912 } 3913 3914 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3915 { 3916 if (io_end->flag & EXT4_IO_END_UNWRITTEN) 3917 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3918 } 3919 3920 extern const struct iomap_ops ext4_iomap_ops; 3921 extern const struct iomap_ops ext4_iomap_report_ops; 3922 3923 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3924 { 3925 /* 3926 * If the buffer has the write error flag, we have failed 3927 * to write out data in the block. In this case, we don't 3928 * have to read the block because we may read the old data 3929 * successfully. 3930 */ 3931 if (buffer_write_io_error(bh)) 3932 set_buffer_uptodate(bh); 3933 return buffer_uptodate(bh); 3934 } 3935 3936 static inline bool ext4_inode_can_atomic_write(struct inode *inode) 3937 { 3938 3939 return S_ISREG(inode->i_mode) && 3940 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) && 3941 EXT4_SB(inode->i_sb)->s_awu_min > 0; 3942 } 3943 3944 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio, 3945 loff_t pos, unsigned len, 3946 get_block_t *get_block); 3947 3948 #if IS_ENABLED(CONFIG_EXT4_KUNIT_TESTS) 3949 #define EXPORT_SYMBOL_FOR_EXT4_TEST(sym) \ 3950 EXPORT_SYMBOL_FOR_MODULES(sym, "ext4-test") 3951 #endif 3952 #endif /* __KERNEL__ */ 3953 3954 #endif /* _EXT4_H */ 3955