1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/refcount.h> 21 #include <linux/types.h> 22 #include <linux/blkdev.h> 23 #include <linux/magic.h> 24 #include <linux/jbd2.h> 25 #include <linux/quota.h> 26 #include <linux/rwsem.h> 27 #include <linux/rbtree.h> 28 #include <linux/seqlock.h> 29 #include <linux/mutex.h> 30 #include <linux/timer.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <linux/crc32c.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #include <linux/fiemap.h> 40 #ifdef __KERNEL__ 41 #include <linux/compat.h> 42 #endif 43 #include <uapi/linux/ext4.h> 44 45 #include <linux/fscrypt.h> 46 #include <linux/fsverity.h> 47 48 #include <linux/compiler.h> 49 50 /* 51 * The fourth extended filesystem constants/structures 52 */ 53 54 /* 55 * with AGGRESSIVE_CHECK allocator runs consistency checks over 56 * structures. these checks slow things down a lot 57 */ 58 #define AGGRESSIVE_CHECK__ 59 60 /* 61 * with DOUBLE_CHECK defined mballoc creates persistent in-core 62 * bitmaps, maintains and uses them to check for double allocations 63 */ 64 #define DOUBLE_CHECK__ 65 66 /* 67 * Define EXT4FS_DEBUG to produce debug messages 68 */ 69 #undef EXT4FS_DEBUG 70 71 /* 72 * Debug code 73 */ 74 #ifdef EXT4FS_DEBUG 75 #define ext4_debug(f, a...) \ 76 do { \ 77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 78 __FILE__, __LINE__, __func__); \ 79 printk(KERN_DEBUG f, ## a); \ 80 } while (0) 81 #else 82 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 83 #endif 84 85 /* 86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 87 */ 88 #define EXT_DEBUG__ 89 90 /* 91 * Dynamic printk for controlled extents debugging. 92 */ 93 #ifdef CONFIG_EXT4_DEBUG 94 #define ext_debug(ino, fmt, ...) \ 95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 96 current->comm, task_pid_nr(current), \ 97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 98 __func__, ##__VA_ARGS__) 99 #else 100 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 101 #endif 102 103 #define ASSERT(assert) \ 104 do { \ 105 if (unlikely(!(assert))) { \ 106 printk(KERN_EMERG \ 107 "Assertion failure in %s() at %s:%d: '%s'\n", \ 108 __func__, __FILE__, __LINE__, #assert); \ 109 BUG(); \ 110 } \ 111 } while (0) 112 113 /* data type for block offset of block group */ 114 typedef int ext4_grpblk_t; 115 116 /* data type for filesystem-wide blocks number */ 117 typedef unsigned long long ext4_fsblk_t; 118 119 /* data type for file logical block number */ 120 typedef __u32 ext4_lblk_t; 121 122 /* data type for block group number */ 123 typedef unsigned int ext4_group_t; 124 125 enum SHIFT_DIRECTION { 126 SHIFT_LEFT = 0, 127 SHIFT_RIGHT, 128 }; 129 130 /* 131 * For each criteria, mballoc has slightly different way of finding 132 * the required blocks nad usually, higher the criteria the slower the 133 * allocation. We start at lower criterias and keep falling back to 134 * higher ones if we are not able to find any blocks. Lower (earlier) 135 * criteria are faster. 136 */ 137 enum criteria { 138 /* 139 * Used when number of blocks needed is a power of 2. This 140 * doesn't trigger any disk IO except prefetch and is the 141 * fastest criteria. 142 */ 143 CR_POWER2_ALIGNED, 144 145 /* 146 * Tries to lookup in-memory data structures to find the most 147 * suitable group that satisfies goal request. No disk IO 148 * except block prefetch. 149 */ 150 CR_GOAL_LEN_FAST, 151 152 /* 153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal 154 * length to the best available length for faster allocation. 155 */ 156 CR_BEST_AVAIL_LEN, 157 158 /* 159 * Reads each block group sequentially, performing disk IO if 160 * necessary, to find suitable block group. Tries to 161 * allocate goal length but might trim the request if nothing 162 * is found after enough tries. 163 */ 164 CR_GOAL_LEN_SLOW, 165 166 /* 167 * Finds the first free set of blocks and allocates 168 * those. This is only used in rare cases when 169 * CR_GOAL_LEN_SLOW also fails to allocate anything. 170 */ 171 CR_ANY_FREE, 172 173 /* 174 * Number of criterias defined. 175 */ 176 EXT4_MB_NUM_CRS 177 }; 178 179 /* 180 * Flags used in mballoc's allocation_context flags field. 181 * 182 * Also used to show what's going on for debugging purposes when the 183 * flag field is exported via the traceport interface 184 */ 185 186 /* prefer goal again. length */ 187 #define EXT4_MB_HINT_MERGE 0x0001 188 /* first blocks in the file */ 189 #define EXT4_MB_HINT_FIRST 0x0008 190 /* data is being allocated */ 191 #define EXT4_MB_HINT_DATA 0x0020 192 /* don't preallocate (for tails) */ 193 #define EXT4_MB_HINT_NOPREALLOC 0x0040 194 /* allocate for locality group */ 195 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 196 /* allocate goal blocks or none */ 197 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 198 /* goal is meaningful */ 199 #define EXT4_MB_HINT_TRY_GOAL 0x0200 200 /* blocks already pre-reserved by delayed allocation */ 201 #define EXT4_MB_DELALLOC_RESERVED 0x0400 202 /* We are doing stream allocation */ 203 #define EXT4_MB_STREAM_ALLOC 0x0800 204 /* Use reserved root blocks if needed */ 205 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 206 /* Use blocks from reserved pool */ 207 #define EXT4_MB_USE_RESERVED 0x2000 208 /* Do strict check for free blocks while retrying block allocation */ 209 #define EXT4_MB_STRICT_CHECK 0x4000 210 211 struct ext4_allocation_request { 212 /* target inode for block we're allocating */ 213 struct inode *inode; 214 /* how many blocks we want to allocate */ 215 unsigned int len; 216 /* logical block in target inode */ 217 ext4_lblk_t logical; 218 /* the closest logical allocated block to the left */ 219 ext4_lblk_t lleft; 220 /* the closest logical allocated block to the right */ 221 ext4_lblk_t lright; 222 /* phys. target (a hint) */ 223 ext4_fsblk_t goal; 224 /* phys. block for the closest logical allocated block to the left */ 225 ext4_fsblk_t pleft; 226 /* phys. block for the closest logical allocated block to the right */ 227 ext4_fsblk_t pright; 228 /* flags. see above EXT4_MB_HINT_* */ 229 unsigned int flags; 230 }; 231 232 /* 233 * Logical to physical block mapping, used by ext4_map_blocks() 234 * 235 * This structure is used to pass requests into ext4_map_blocks() as 236 * well as to store the information returned by ext4_map_blocks(). It 237 * takes less room on the stack than a struct buffer_head. 238 */ 239 #define EXT4_MAP_NEW BIT(BH_New) 240 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 241 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 242 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 243 #define EXT4_MAP_DELAYED BIT(BH_Delay) 244 /* 245 * This is for use in ext4_map_query_blocks() for a special case where we can 246 * have a physically and logically contiguous blocks split across two leaf 247 * nodes instead of a single extent. This is required in case of atomic writes 248 * to know whether the returned extent is last in leaf. If yes, then lookup for 249 * next in leaf block in ext4_map_query_blocks_next_in_leaf(). 250 * - This is never going to be added to any buffer head state. 251 * - We use the next available bit after BH_BITMAP_UPTODATE. 252 */ 253 #define EXT4_MAP_QUERY_LAST_IN_LEAF BIT(BH_BITMAP_UPTODATE + 1) 254 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 255 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 256 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF) 257 258 struct ext4_map_blocks { 259 ext4_fsblk_t m_pblk; 260 ext4_lblk_t m_lblk; 261 unsigned int m_len; 262 unsigned int m_flags; 263 }; 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) 370 371 /* Translate a block number to a cluster number */ 372 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 373 /* Translate a cluster number to a block number */ 374 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 375 /* Translate # of blks to # of clusters */ 376 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 377 (sbi)->s_cluster_bits) 378 /* Mask out the low bits to get the starting block of the cluster */ 379 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 380 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 381 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 382 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 383 /* Fill in the low bits to get the last block of the cluster */ 384 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 385 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 386 /* Get the cluster offset */ 387 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 388 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 389 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 390 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 391 392 /* 393 * Structure of a blocks group descriptor 394 */ 395 struct ext4_group_desc 396 { 397 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 398 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 399 __le32 bg_inode_table_lo; /* Inodes table block */ 400 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 401 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 402 __le16 bg_used_dirs_count_lo; /* Directories count */ 403 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 404 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 405 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 406 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 407 __le16 bg_itable_unused_lo; /* Unused inodes count */ 408 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 409 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 410 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 411 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 412 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 413 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 414 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 415 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 416 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 417 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 418 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 419 __u32 bg_reserved; 420 }; 421 422 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 423 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 424 sizeof(__le16)) 425 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 426 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 427 sizeof(__le16)) 428 429 /* 430 * Structure of a flex block group info 431 */ 432 433 struct flex_groups { 434 atomic64_t free_clusters; 435 atomic_t free_inodes; 436 atomic_t used_dirs; 437 }; 438 439 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 440 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 441 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 442 443 /* 444 * Macro-instructions used to manage group descriptors 445 */ 446 #define EXT4_MIN_DESC_SIZE 32 447 #define EXT4_MIN_DESC_SIZE_64BIT 64 448 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 449 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 450 #ifdef __KERNEL__ 451 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 452 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 453 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 454 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 455 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 456 #else 457 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 458 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 459 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 460 #endif 461 462 /* 463 * Constants relative to the data blocks 464 */ 465 #define EXT4_NDIR_BLOCKS 12 466 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 467 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 468 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 469 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 470 471 /* 472 * Inode flags 473 */ 474 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 475 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 476 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 477 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 478 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 479 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 480 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 481 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 482 /* Reserved for compression usage... */ 483 #define EXT4_DIRTY_FL 0x00000100 484 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 485 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 486 /* nb: was previously EXT2_ECOMPR_FL */ 487 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 488 /* End compression flags --- maybe not all used */ 489 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 490 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 491 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 492 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 493 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 494 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 495 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 496 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 497 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 498 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 499 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 500 501 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 502 503 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 504 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 505 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 506 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 507 508 /* User modifiable flags */ 509 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 510 EXT4_UNRM_FL | \ 511 EXT4_COMPR_FL | \ 512 EXT4_SYNC_FL | \ 513 EXT4_IMMUTABLE_FL | \ 514 EXT4_APPEND_FL | \ 515 EXT4_NODUMP_FL | \ 516 EXT4_NOATIME_FL | \ 517 EXT4_JOURNAL_DATA_FL | \ 518 EXT4_NOTAIL_FL | \ 519 EXT4_DIRSYNC_FL | \ 520 EXT4_TOPDIR_FL | \ 521 EXT4_EXTENTS_FL | \ 522 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 523 EXT4_DAX_FL | \ 524 EXT4_PROJINHERIT_FL | \ 525 EXT4_CASEFOLD_FL) 526 527 /* User visible flags */ 528 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 529 EXT4_DIRTY_FL | \ 530 EXT4_COMPRBLK_FL | \ 531 EXT4_NOCOMPR_FL | \ 532 EXT4_ENCRYPT_FL | \ 533 EXT4_INDEX_FL | \ 534 EXT4_VERITY_FL | \ 535 EXT4_INLINE_DATA_FL) 536 537 /* Flags that should be inherited by new inodes from their parent. */ 538 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 539 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 540 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 541 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 542 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 543 EXT4_DAX_FL) 544 545 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 546 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 547 EXT4_PROJINHERIT_FL)) 548 549 /* Flags that are appropriate for non-directories/regular files. */ 550 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 551 552 /* The only flags that should be swapped */ 553 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 554 555 /* Flags which are mutually exclusive to DAX */ 556 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 557 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 558 559 /* Mask out flags that are inappropriate for the given type of inode. */ 560 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 561 { 562 if (S_ISDIR(mode)) 563 return flags; 564 else if (S_ISREG(mode)) 565 return flags & EXT4_REG_FLMASK; 566 else 567 return flags & EXT4_OTHER_FLMASK; 568 } 569 570 /* 571 * Inode flags used for atomic set/get 572 */ 573 enum { 574 EXT4_INODE_SECRM = 0, /* Secure deletion */ 575 EXT4_INODE_UNRM = 1, /* Undelete */ 576 EXT4_INODE_COMPR = 2, /* Compress file */ 577 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 578 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 579 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 580 EXT4_INODE_NODUMP = 6, /* do not dump file */ 581 EXT4_INODE_NOATIME = 7, /* do not update atime */ 582 /* Reserved for compression usage... */ 583 EXT4_INODE_DIRTY = 8, 584 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 585 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 586 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 587 /* End compression flags --- maybe not all used */ 588 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 589 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 590 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 591 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 592 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 593 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 594 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 595 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 596 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 597 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 598 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 599 EXT4_INODE_DAX = 25, /* Inode is DAX */ 600 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 601 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 602 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 603 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 604 }; 605 606 /* 607 * Since it's pretty easy to mix up bit numbers and hex values, we use a 608 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 609 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 610 * any extra space in the compiled kernel image, otherwise, the build will fail. 611 * It's important that these values are the same, since we are using 612 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 613 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 614 * values found in ext2, ext3 and ext4 filesystems, and of course the values 615 * defined in e2fsprogs. 616 * 617 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 618 */ 619 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 620 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 621 622 static inline void ext4_check_flag_values(void) 623 { 624 CHECK_FLAG_VALUE(SECRM); 625 CHECK_FLAG_VALUE(UNRM); 626 CHECK_FLAG_VALUE(COMPR); 627 CHECK_FLAG_VALUE(SYNC); 628 CHECK_FLAG_VALUE(IMMUTABLE); 629 CHECK_FLAG_VALUE(APPEND); 630 CHECK_FLAG_VALUE(NODUMP); 631 CHECK_FLAG_VALUE(NOATIME); 632 CHECK_FLAG_VALUE(DIRTY); 633 CHECK_FLAG_VALUE(COMPRBLK); 634 CHECK_FLAG_VALUE(NOCOMPR); 635 CHECK_FLAG_VALUE(ENCRYPT); 636 CHECK_FLAG_VALUE(INDEX); 637 CHECK_FLAG_VALUE(IMAGIC); 638 CHECK_FLAG_VALUE(JOURNAL_DATA); 639 CHECK_FLAG_VALUE(NOTAIL); 640 CHECK_FLAG_VALUE(DIRSYNC); 641 CHECK_FLAG_VALUE(TOPDIR); 642 CHECK_FLAG_VALUE(HUGE_FILE); 643 CHECK_FLAG_VALUE(EXTENTS); 644 CHECK_FLAG_VALUE(VERITY); 645 CHECK_FLAG_VALUE(EA_INODE); 646 CHECK_FLAG_VALUE(INLINE_DATA); 647 CHECK_FLAG_VALUE(PROJINHERIT); 648 CHECK_FLAG_VALUE(CASEFOLD); 649 CHECK_FLAG_VALUE(RESERVED); 650 } 651 652 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 653 struct compat_ext4_new_group_input { 654 u32 group; 655 compat_u64 block_bitmap; 656 compat_u64 inode_bitmap; 657 compat_u64 inode_table; 658 u32 blocks_count; 659 u16 reserved_blocks; 660 u16 unused; 661 }; 662 #endif 663 664 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 665 struct ext4_new_group_data { 666 __u32 group; 667 __u64 block_bitmap; 668 __u64 inode_bitmap; 669 __u64 inode_table; 670 __u32 blocks_count; 671 __u16 reserved_blocks; 672 __u16 mdata_blocks; 673 __u32 free_clusters_count; 674 }; 675 676 /* Indexes used to index group tables in ext4_new_group_data */ 677 enum { 678 BLOCK_BITMAP = 0, /* block bitmap */ 679 INODE_BITMAP, /* inode bitmap */ 680 INODE_TABLE, /* inode tables */ 681 GROUP_TABLE_COUNT, 682 }; 683 684 /* 685 * Flags used by ext4_map_blocks() 686 */ 687 /* Allocate any needed blocks and/or convert an unwritten 688 extent to be an initialized ext4 */ 689 #define EXT4_GET_BLOCKS_CREATE 0x0001 690 /* Request the creation of an unwritten extent */ 691 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 692 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 693 EXT4_GET_BLOCKS_CREATE) 694 /* Caller is from the delayed allocation writeout path 695 * finally doing the actual allocation of delayed blocks */ 696 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 697 /* caller is from the direct IO path, request to creation of an 698 unwritten extents if not allocated, split the unwritten 699 extent if blocks has been preallocated already*/ 700 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 701 #define EXT4_GET_BLOCKS_CONVERT 0x0010 702 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 703 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 704 /* Eventual metadata allocation (due to growing extent tree) 705 * should not fail, so try to use reserved blocks for that.*/ 706 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 707 /* Don't normalize allocation size (used for fallocate) */ 708 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 709 /* Convert written extents to unwritten */ 710 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 711 /* Write zeros to newly created written extents */ 712 #define EXT4_GET_BLOCKS_ZERO 0x0200 713 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 714 EXT4_GET_BLOCKS_ZERO) 715 /* Caller is in the context of data submission, such as writeback, 716 * fsync, etc. Especially, in the generic writeback path, caller will 717 * submit data before dropping transaction handle. This allows jbd2 718 * to avoid submitting data before commit. */ 719 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 720 /* Convert extent to initialized after IO complete */ 721 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\ 722 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\ 723 EXT4_GET_BLOCKS_IO_SUBMIT) 724 /* Caller is in the atomic contex, find extent if it has been cached */ 725 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 726 /* 727 * Atomic write caller needs this to query in the slow path of mixed mapping 728 * case, when a contiguous extent can be split across two adjacent leaf nodes. 729 * Look EXT4_MAP_QUERY_LAST_IN_LEAF. 730 */ 731 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000 732 733 /* 734 * The bit position of these flags must not overlap with any of the 735 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 736 * read_extent_tree_block(), ext4_split_extent_at(), 737 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 738 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 739 * caching the extents when reading from the extent tree while a 740 * truncate or punch hole operation is in progress. 741 */ 742 #define EXT4_EX_NOCACHE 0x40000000 743 #define EXT4_EX_FORCE_CACHE 0x20000000 744 #define EXT4_EX_NOFAIL 0x10000000 745 /* 746 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to 747 * pass while lookup/querying of on disk extent tree. 748 */ 749 #define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\ 750 EXT4_EX_NOFAIL |\ 751 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF) 752 753 /* 754 * Flags used by ext4_free_blocks 755 */ 756 #define EXT4_FREE_BLOCKS_METADATA 0x0001 757 #define EXT4_FREE_BLOCKS_FORGET 0x0002 758 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 759 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 760 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 761 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 762 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 763 764 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 765 /* 766 * ioctl commands in 32 bit emulation 767 */ 768 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 769 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 770 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 771 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 772 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 773 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 774 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 775 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 776 #endif 777 778 /* Max physical block we can address w/o extents */ 779 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 780 781 /* Max logical block we can support */ 782 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 783 784 /* 785 * Structure of an inode on the disk 786 */ 787 struct ext4_inode { 788 __le16 i_mode; /* File mode */ 789 __le16 i_uid; /* Low 16 bits of Owner Uid */ 790 __le32 i_size_lo; /* Size in bytes */ 791 __le32 i_atime; /* Access time */ 792 __le32 i_ctime; /* Inode Change time */ 793 __le32 i_mtime; /* Modification time */ 794 __le32 i_dtime; /* Deletion Time */ 795 __le16 i_gid; /* Low 16 bits of Group Id */ 796 __le16 i_links_count; /* Links count */ 797 __le32 i_blocks_lo; /* Blocks count */ 798 __le32 i_flags; /* File flags */ 799 union { 800 struct { 801 __le32 l_i_version; 802 } linux1; 803 struct { 804 __u32 h_i_translator; 805 } hurd1; 806 struct { 807 __u32 m_i_reserved1; 808 } masix1; 809 } osd1; /* OS dependent 1 */ 810 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 811 __le32 i_generation; /* File version (for NFS) */ 812 __le32 i_file_acl_lo; /* File ACL */ 813 __le32 i_size_high; 814 __le32 i_obso_faddr; /* Obsoleted fragment address */ 815 union { 816 struct { 817 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 818 __le16 l_i_file_acl_high; 819 __le16 l_i_uid_high; /* these 2 fields */ 820 __le16 l_i_gid_high; /* were reserved2[0] */ 821 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 822 __le16 l_i_reserved; 823 } linux2; 824 struct { 825 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 826 __u16 h_i_mode_high; 827 __u16 h_i_uid_high; 828 __u16 h_i_gid_high; 829 __u32 h_i_author; 830 } hurd2; 831 struct { 832 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 833 __le16 m_i_file_acl_high; 834 __u32 m_i_reserved2[2]; 835 } masix2; 836 } osd2; /* OS dependent 2 */ 837 __le16 i_extra_isize; 838 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 839 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 840 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 841 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 842 __le32 i_crtime; /* File Creation time */ 843 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 844 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 845 __le32 i_projid; /* Project ID */ 846 }; 847 848 #define EXT4_EPOCH_BITS 2 849 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 850 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 851 852 /* 853 * Extended fields will fit into an inode if the filesystem was formatted 854 * with large inodes (-I 256 or larger) and there are not currently any EAs 855 * consuming all of the available space. For new inodes we always reserve 856 * enough space for the kernel's known extended fields, but for inodes 857 * created with an old kernel this might not have been the case. None of 858 * the extended inode fields is critical for correct filesystem operation. 859 * This macro checks if a certain field fits in the inode. Note that 860 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 861 */ 862 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 863 ((offsetof(typeof(*ext4_inode), field) + \ 864 sizeof((ext4_inode)->field)) \ 865 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 866 (einode)->i_extra_isize)) \ 867 868 /* 869 * We use an encoding that preserves the times for extra epoch "00": 870 * 871 * extra msb of adjust for signed 872 * epoch 32-bit 32-bit tv_sec to 873 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 874 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 875 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 876 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 877 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 878 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 879 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 880 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 881 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 882 * 883 * Note that previous versions of the kernel on 64-bit systems would 884 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 885 * 1970. e2fsck will correct this, assuming that it is run on the 886 * affected filesystem before 2242. 887 */ 888 889 static inline __le32 ext4_encode_extra_time(struct timespec64 ts) 890 { 891 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK; 892 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS)); 893 } 894 895 static inline struct timespec64 ext4_decode_extra_time(__le32 base, 896 __le32 extra) 897 { 898 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) }; 899 900 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 901 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 902 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 903 return ts; 904 } 905 906 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \ 907 do { \ 908 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 909 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \ 910 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \ 911 } else \ 912 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \ 913 } while (0) 914 915 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \ 916 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode)) 917 918 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \ 919 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode)) 920 921 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \ 922 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode)) 923 924 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 925 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 926 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \ 927 raw_inode, (einode)->xtime) 928 929 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \ 930 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \ 931 ext4_decode_extra_time((raw_inode)->xtime, \ 932 (raw_inode)->xtime ## _extra) : \ 933 (struct timespec64) { \ 934 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \ 935 }) 936 937 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \ 938 do { \ 939 inode_set_atime_to_ts(inode, \ 940 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \ 941 } while (0) 942 943 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \ 944 do { \ 945 inode_set_mtime_to_ts(inode, \ 946 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \ 947 } while (0) 948 949 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \ 950 do { \ 951 inode_set_ctime_to_ts(inode, \ 952 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \ 953 } while (0) 954 955 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 956 do { \ 957 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 958 (einode)->xtime = \ 959 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \ 960 raw_inode); \ 961 else \ 962 (einode)->xtime = (struct timespec64){0, 0}; \ 963 } while (0) 964 965 #define i_disk_version osd1.linux1.l_i_version 966 967 #if defined(__KERNEL__) || defined(__linux__) 968 #define i_reserved1 osd1.linux1.l_i_reserved1 969 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 970 #define i_blocks_high osd2.linux2.l_i_blocks_high 971 #define i_uid_low i_uid 972 #define i_gid_low i_gid 973 #define i_uid_high osd2.linux2.l_i_uid_high 974 #define i_gid_high osd2.linux2.l_i_gid_high 975 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 976 977 #elif defined(__GNU__) 978 979 #define i_translator osd1.hurd1.h_i_translator 980 #define i_uid_high osd2.hurd2.h_i_uid_high 981 #define i_gid_high osd2.hurd2.h_i_gid_high 982 #define i_author osd2.hurd2.h_i_author 983 984 #elif defined(__masix__) 985 986 #define i_reserved1 osd1.masix1.m_i_reserved1 987 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 988 #define i_reserved2 osd2.masix2.m_i_reserved2 989 990 #endif /* defined(__KERNEL__) || defined(__linux__) */ 991 992 #include "extents_status.h" 993 #include "fast_commit.h" 994 995 /* 996 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 997 * 998 * These are needed to avoid lockdep false positives when we need to 999 * allocate blocks to the quota inode during ext4_map_blocks(), while 1000 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1001 * do quota tracking for the quota inode, this avoids deadlock (as 1002 * well as infinite recursion, since it isn't turtles all the way 1003 * down...) 1004 * 1005 * I_DATA_SEM_NORMAL - Used for most inodes 1006 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1007 * where the second inode has larger inode number 1008 * than the first 1009 * I_DATA_SEM_QUOTA - Used for quota inodes only 1010 * I_DATA_SEM_EA - Used for ea_inodes only 1011 */ 1012 enum { 1013 I_DATA_SEM_NORMAL = 0, 1014 I_DATA_SEM_OTHER, 1015 I_DATA_SEM_QUOTA, 1016 I_DATA_SEM_EA 1017 }; 1018 1019 1020 /* 1021 * fourth extended file system inode data in memory 1022 */ 1023 struct ext4_inode_info { 1024 __le32 i_data[15]; /* unconverted */ 1025 __u32 i_dtime; 1026 ext4_fsblk_t i_file_acl; 1027 1028 /* 1029 * i_block_group is the number of the block group which contains 1030 * this file's inode. Constant across the lifetime of the inode, 1031 * it is used for making block allocation decisions - we try to 1032 * place a file's data blocks near its inode block, and new inodes 1033 * near to their parent directory's inode. 1034 */ 1035 ext4_group_t i_block_group; 1036 ext4_lblk_t i_dir_start_lookup; 1037 #if (BITS_PER_LONG < 64) 1038 unsigned long i_state_flags; /* Dynamic state flags */ 1039 #endif 1040 unsigned long i_flags; 1041 1042 /* 1043 * Extended attributes can be read independently of the main file 1044 * data. Taking i_rwsem even when reading would cause contention 1045 * between readers of EAs and writers of regular file data, so 1046 * instead we synchronize on xattr_sem when reading or changing 1047 * EAs. 1048 */ 1049 struct rw_semaphore xattr_sem; 1050 1051 /* 1052 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1053 * i_orphan is used. 1054 */ 1055 union { 1056 struct list_head i_orphan; /* unlinked but open inodes */ 1057 unsigned int i_orphan_idx; /* Index in orphan file */ 1058 }; 1059 1060 /* Fast commit related info */ 1061 1062 /* For tracking dentry create updates */ 1063 struct list_head i_fc_dilist; 1064 struct list_head i_fc_list; /* 1065 * inodes that need fast commit 1066 * protected by sbi->s_fc_lock. 1067 */ 1068 1069 /* Start of lblk range that needs to be committed in this fast commit */ 1070 ext4_lblk_t i_fc_lblk_start; 1071 1072 /* End of lblk range that needs to be committed in this fast commit */ 1073 ext4_lblk_t i_fc_lblk_len; 1074 1075 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1076 1077 /* Fast commit wait queue for this inode */ 1078 wait_queue_head_t i_fc_wait; 1079 1080 /* 1081 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len 1082 * and inode's EXT4_FC_STATE_COMMITTING state bit. 1083 */ 1084 spinlock_t i_fc_lock; 1085 1086 /* 1087 * i_disksize keeps track of what the inode size is ON DISK, not 1088 * in memory. During truncate, i_size is set to the new size by 1089 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1090 * set i_disksize to 0 until the truncate is actually under way. 1091 * 1092 * The intent is that i_disksize always represents the blocks which 1093 * are used by this file. This allows recovery to restart truncate 1094 * on orphans if we crash during truncate. We actually write i_disksize 1095 * into the on-disk inode when writing inodes out, instead of i_size. 1096 * 1097 * The only time when i_disksize and i_size may be different is when 1098 * a truncate is in progress. The only things which change i_disksize 1099 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1100 */ 1101 loff_t i_disksize; 1102 1103 /* 1104 * i_data_sem is for serialising ext4_truncate() against 1105 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1106 * data tree are chopped off during truncate. We can't do that in 1107 * ext4 because whenever we perform intermediate commits during 1108 * truncate, the inode and all the metadata blocks *must* be in a 1109 * consistent state which allows truncation of the orphans to restart 1110 * during recovery. Hence we must fix the get_block-vs-truncate race 1111 * by other means, so we have i_data_sem. 1112 */ 1113 struct rw_semaphore i_data_sem; 1114 struct inode vfs_inode; 1115 struct jbd2_inode *jinode; 1116 1117 /* 1118 * File creation time. Its function is same as that of 1119 * struct timespec64 i_{a,c,m}time in the generic inode. 1120 */ 1121 struct timespec64 i_crtime; 1122 1123 /* mballoc */ 1124 atomic_t i_prealloc_active; 1125 1126 /* allocation reservation info for delalloc */ 1127 /* In case of bigalloc, this refer to clusters rather than blocks */ 1128 unsigned int i_reserved_data_blocks; 1129 struct rb_root i_prealloc_node; 1130 rwlock_t i_prealloc_lock; 1131 1132 /* extents status tree */ 1133 struct ext4_es_tree i_es_tree; 1134 rwlock_t i_es_lock; 1135 struct list_head i_es_list; 1136 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1137 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1138 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1139 extents to shrink. Protected by 1140 i_es_lock */ 1141 1142 /* ialloc */ 1143 ext4_group_t i_last_alloc_group; 1144 1145 /* pending cluster reservations for bigalloc file systems */ 1146 struct ext4_pending_tree i_pending_tree; 1147 1148 /* on-disk additional length */ 1149 __u16 i_extra_isize; 1150 1151 /* Indicate the inline data space. */ 1152 u16 i_inline_off; 1153 u16 i_inline_size; 1154 1155 #ifdef CONFIG_QUOTA 1156 /* quota space reservation, managed internally by quota code */ 1157 qsize_t i_reserved_quota; 1158 #endif 1159 spinlock_t i_block_reservation_lock; 1160 1161 /* Lock protecting lists below */ 1162 spinlock_t i_completed_io_lock; 1163 /* 1164 * Completed IOs that need unwritten extents handling and have 1165 * transaction reserved 1166 */ 1167 struct list_head i_rsv_conversion_list; 1168 struct work_struct i_rsv_conversion_work; 1169 1170 /* 1171 * Transactions that contain inode's metadata needed to complete 1172 * fsync and fdatasync, respectively. 1173 */ 1174 tid_t i_sync_tid; 1175 tid_t i_datasync_tid; 1176 1177 #ifdef CONFIG_QUOTA 1178 struct dquot __rcu *i_dquot[MAXQUOTAS]; 1179 #endif 1180 1181 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1182 __u32 i_csum_seed; 1183 1184 kprojid_t i_projid; 1185 }; 1186 1187 /* 1188 * File system states 1189 */ 1190 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1191 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1192 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1193 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1194 1195 /* 1196 * Misc. filesystem flags 1197 */ 1198 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1199 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1200 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1201 1202 /* 1203 * Mount flags set via mount options or defaults 1204 */ 1205 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1206 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1207 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1208 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1209 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1210 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1211 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1212 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1213 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1214 #ifdef CONFIG_FS_DAX 1215 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1216 #else 1217 #define EXT4_MOUNT_DAX_ALWAYS 0 1218 #endif 1219 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1220 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1221 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1222 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1223 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1224 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1225 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1226 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1227 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1228 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1229 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1230 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1231 * enable enforcement for hidden 1232 * quota files */ 1233 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1234 * enforcement for hidden quota 1235 * files */ 1236 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1237 * enforcement */ 1238 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1239 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1240 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1241 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1242 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1243 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1244 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1245 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1246 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1247 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1248 1249 /* 1250 * Mount flags set either automatically (could not be set by mount option) 1251 * based on per file system feature or property or in special cases such as 1252 * distinguishing between explicit mount option definition and default. 1253 */ 1254 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1255 specified delalloc */ 1256 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1257 size of blocksize * 8 1258 blocks */ 1259 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1260 file systems */ 1261 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1262 specified journal checksum */ 1263 1264 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1265 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1266 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1267 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1268 * scanning in mballoc 1269 */ 1270 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1271 1272 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1273 ~EXT4_MOUNT_##opt 1274 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1275 EXT4_MOUNT_##opt 1276 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1277 EXT4_MOUNT_##opt) 1278 1279 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1280 ~EXT4_MOUNT2_##opt 1281 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1282 EXT4_MOUNT2_##opt 1283 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1284 EXT4_MOUNT2_##opt) 1285 1286 #define ext4_test_and_set_bit __test_and_set_bit_le 1287 #define ext4_set_bit __set_bit_le 1288 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1289 #define ext4_clear_bit __clear_bit_le 1290 #define ext4_test_bit test_bit_le 1291 #define ext4_find_next_zero_bit find_next_zero_bit_le 1292 #define ext4_find_next_bit find_next_bit_le 1293 1294 extern void mb_set_bits(void *bm, int cur, int len); 1295 1296 /* 1297 * Maximal mount counts between two filesystem checks 1298 */ 1299 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1300 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1301 1302 /* 1303 * Behaviour when detecting errors 1304 */ 1305 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1306 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1307 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1308 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1309 1310 /* Metadata checksum algorithm codes */ 1311 #define EXT4_CRC32C_CHKSUM 1 1312 1313 #define EXT4_LABEL_MAX 16 1314 1315 /* 1316 * Structure of the super block 1317 */ 1318 struct ext4_super_block { 1319 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1320 __le32 s_blocks_count_lo; /* Blocks count */ 1321 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1322 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1323 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1324 __le32 s_first_data_block; /* First Data Block */ 1325 __le32 s_log_block_size; /* Block size */ 1326 __le32 s_log_cluster_size; /* Allocation cluster size */ 1327 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1328 __le32 s_clusters_per_group; /* # Clusters per group */ 1329 __le32 s_inodes_per_group; /* # Inodes per group */ 1330 __le32 s_mtime; /* Mount time */ 1331 /*30*/ __le32 s_wtime; /* Write time */ 1332 __le16 s_mnt_count; /* Mount count */ 1333 __le16 s_max_mnt_count; /* Maximal mount count */ 1334 __le16 s_magic; /* Magic signature */ 1335 __le16 s_state; /* File system state */ 1336 __le16 s_errors; /* Behaviour when detecting errors */ 1337 __le16 s_minor_rev_level; /* minor revision level */ 1338 /*40*/ __le32 s_lastcheck; /* time of last check */ 1339 __le32 s_checkinterval; /* max. time between checks */ 1340 __le32 s_creator_os; /* OS */ 1341 __le32 s_rev_level; /* Revision level */ 1342 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1343 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1344 /* 1345 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1346 * 1347 * Note: the difference between the compatible feature set and 1348 * the incompatible feature set is that if there is a bit set 1349 * in the incompatible feature set that the kernel doesn't 1350 * know about, it should refuse to mount the filesystem. 1351 * 1352 * e2fsck's requirements are more strict; if it doesn't know 1353 * about a feature in either the compatible or incompatible 1354 * feature set, it must abort and not try to meddle with 1355 * things it doesn't understand... 1356 */ 1357 __le32 s_first_ino; /* First non-reserved inode */ 1358 __le16 s_inode_size; /* size of inode structure */ 1359 __le16 s_block_group_nr; /* block group # of this superblock */ 1360 __le32 s_feature_compat; /* compatible feature set */ 1361 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1362 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1363 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1364 /*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */ 1365 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1366 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1367 /* 1368 * Performance hints. Directory preallocation should only 1369 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1370 */ 1371 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1372 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1373 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1374 /* 1375 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1376 */ 1377 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1378 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1379 __le32 s_journal_dev; /* device number of journal file */ 1380 __le32 s_last_orphan; /* start of list of inodes to delete */ 1381 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1382 __u8 s_def_hash_version; /* Default hash version to use */ 1383 __u8 s_jnl_backup_type; 1384 __le16 s_desc_size; /* size of group descriptor */ 1385 /*100*/ __le32 s_default_mount_opts; 1386 __le32 s_first_meta_bg; /* First metablock block group */ 1387 __le32 s_mkfs_time; /* When the filesystem was created */ 1388 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1389 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1390 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1391 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1392 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1393 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1394 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1395 __le32 s_flags; /* Miscellaneous flags */ 1396 __le16 s_raid_stride; /* RAID stride */ 1397 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1398 __le64 s_mmp_block; /* Block for multi-mount protection */ 1399 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1400 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1401 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1402 __u8 s_encryption_level; /* versioning level for encryption */ 1403 __u8 s_reserved_pad; /* Padding to next 32bits */ 1404 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1405 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1406 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1407 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1408 snapshot's future use */ 1409 __le32 s_snapshot_list; /* inode number of the head of the 1410 on-disk snapshot list */ 1411 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1412 __le32 s_error_count; /* number of fs errors */ 1413 __le32 s_first_error_time; /* first time an error happened */ 1414 __le32 s_first_error_ino; /* inode involved in first error */ 1415 __le64 s_first_error_block; /* block involved of first error */ 1416 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1417 __le32 s_first_error_line; /* line number where error happened */ 1418 __le32 s_last_error_time; /* most recent time of an error */ 1419 __le32 s_last_error_ino; /* inode involved in last error */ 1420 __le32 s_last_error_line; /* line number where error happened */ 1421 __le64 s_last_error_block; /* block involved of last error */ 1422 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1423 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1424 __u8 s_mount_opts[64]; 1425 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1426 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1427 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1428 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1429 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1430 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1431 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1432 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1433 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1434 __u8 s_wtime_hi; 1435 __u8 s_mtime_hi; 1436 __u8 s_mkfs_time_hi; 1437 __u8 s_lastcheck_hi; 1438 __u8 s_first_error_time_hi; 1439 __u8 s_last_error_time_hi; 1440 __u8 s_first_error_errcode; 1441 __u8 s_last_error_errcode; 1442 __le16 s_encoding; /* Filename charset encoding */ 1443 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1444 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1445 __le16 s_def_resuid_hi; 1446 __le16 s_def_resgid_hi; 1447 __le32 s_reserved[93]; /* Padding to the end of the block */ 1448 __le32 s_checksum; /* crc32c(superblock) */ 1449 }; 1450 1451 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1452 1453 #ifdef __KERNEL__ 1454 1455 /* Number of quota types we support */ 1456 #define EXT4_MAXQUOTAS 3 1457 1458 #define EXT4_ENC_UTF8_12_1 1 1459 1460 /* Types of ext4 journal triggers */ 1461 enum ext4_journal_trigger_type { 1462 EXT4_JTR_ORPHAN_FILE, 1463 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1464 }; 1465 1466 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1467 1468 struct ext4_journal_trigger { 1469 struct jbd2_buffer_trigger_type tr_triggers; 1470 struct super_block *sb; 1471 }; 1472 1473 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1474 struct jbd2_buffer_trigger_type *trigger) 1475 { 1476 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1477 } 1478 1479 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1480 1481 /* Structure at the tail of orphan block */ 1482 struct ext4_orphan_block_tail { 1483 __le32 ob_magic; 1484 __le32 ob_checksum; 1485 }; 1486 1487 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1488 { 1489 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1490 sizeof(u32); 1491 } 1492 1493 struct ext4_orphan_block { 1494 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1495 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1496 }; 1497 1498 /* 1499 * Info about orphan file. 1500 */ 1501 struct ext4_orphan_info { 1502 int of_blocks; /* Number of orphan blocks in a file */ 1503 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1504 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1505 * file blocks */ 1506 }; 1507 1508 /* 1509 * fourth extended-fs super-block data in memory 1510 */ 1511 struct ext4_sb_info { 1512 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1513 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1514 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1515 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1516 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1517 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1518 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1519 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1520 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1521 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1522 unsigned long s_overhead; /* # of fs overhead clusters */ 1523 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1524 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1525 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1526 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1527 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1528 /* Array of bh's for the block group descriptors */ 1529 struct buffer_head * __rcu *s_group_desc; 1530 unsigned int s_mount_opt; 1531 unsigned int s_mount_opt2; 1532 unsigned long s_mount_flags; 1533 unsigned int s_def_mount_opt; 1534 unsigned int s_def_mount_opt2; 1535 ext4_fsblk_t s_sb_block; 1536 atomic64_t s_resv_clusters; 1537 kuid_t s_resuid; 1538 kgid_t s_resgid; 1539 unsigned short s_mount_state; 1540 unsigned short s_pad; 1541 int s_addr_per_block_bits; 1542 int s_desc_per_block_bits; 1543 int s_inode_size; 1544 int s_first_ino; 1545 unsigned int s_inode_readahead_blks; 1546 unsigned int s_inode_goal; 1547 u32 s_hash_seed[4]; 1548 int s_def_hash_version; 1549 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1550 struct percpu_counter s_freeclusters_counter; 1551 struct percpu_counter s_freeinodes_counter; 1552 struct percpu_counter s_dirs_counter; 1553 struct percpu_counter s_dirtyclusters_counter; 1554 struct percpu_counter s_sra_exceeded_retry_limit; 1555 struct blockgroup_lock *s_blockgroup_lock; 1556 struct proc_dir_entry *s_proc; 1557 struct kobject s_kobj; 1558 struct completion s_kobj_unregister; 1559 struct super_block *s_sb; 1560 struct buffer_head *s_mmp_bh; 1561 1562 /* Journaling */ 1563 struct journal_s *s_journal; 1564 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1565 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1566 struct list_head s_orphan; /* List of orphaned inodes in on disk 1567 list */ 1568 struct ext4_orphan_info s_orphan_info; 1569 unsigned long s_commit_interval; 1570 u32 s_max_batch_time; 1571 u32 s_min_batch_time; 1572 struct file *s_journal_bdev_file; 1573 #ifdef CONFIG_QUOTA 1574 /* Names of quota files with journalled quota */ 1575 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1576 int s_jquota_fmt; /* Format of quota to use */ 1577 #endif 1578 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1579 struct ext4_system_blocks __rcu *s_system_blks; 1580 1581 #ifdef EXTENTS_STATS 1582 /* ext4 extents stats */ 1583 unsigned long s_ext_min; 1584 unsigned long s_ext_max; 1585 unsigned long s_depth_max; 1586 spinlock_t s_ext_stats_lock; 1587 unsigned long s_ext_blocks; 1588 unsigned long s_ext_extents; 1589 #endif 1590 1591 /* for buddy allocator */ 1592 struct ext4_group_info ** __rcu *s_group_info; 1593 struct inode *s_buddy_cache; 1594 spinlock_t s_md_lock; 1595 unsigned short *s_mb_offsets; 1596 unsigned int *s_mb_maxs; 1597 unsigned int s_group_info_size; 1598 atomic_t s_mb_free_pending; 1599 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1600 after commit completed */ 1601 struct list_head s_discard_list; 1602 struct work_struct s_discard_work; 1603 atomic_t s_retry_alloc_pending; 1604 struct xarray *s_mb_avg_fragment_size; 1605 struct xarray *s_mb_largest_free_orders; 1606 1607 /* tunables */ 1608 unsigned long s_stripe; 1609 unsigned int s_mb_max_linear_groups; 1610 unsigned int s_mb_stream_request; 1611 unsigned int s_mb_max_to_scan; 1612 unsigned int s_mb_min_to_scan; 1613 unsigned int s_mb_stats; 1614 unsigned int s_mb_order2_reqs; 1615 unsigned int s_mb_group_prealloc; 1616 unsigned int s_max_dir_size_kb; 1617 unsigned int s_mb_prefetch; 1618 unsigned int s_mb_prefetch_limit; 1619 unsigned int s_mb_best_avail_max_trim_order; 1620 unsigned int s_sb_update_sec; 1621 unsigned int s_sb_update_kb; 1622 1623 /* where last allocation was done - for stream allocation */ 1624 ext4_group_t *s_mb_last_groups; 1625 unsigned int s_mb_nr_global_goals; 1626 1627 /* stats for buddy allocator */ 1628 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1629 atomic_t s_bal_success; /* we found long enough chunks */ 1630 atomic_t s_bal_allocated; /* in blocks */ 1631 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1632 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1633 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1634 atomic_t s_bal_goals; /* goal hits */ 1635 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */ 1636 atomic_t s_bal_len_goals; /* len goal hits */ 1637 atomic_t s_bal_breaks; /* too long searches */ 1638 atomic_t s_bal_2orders; /* 2^order hits */ 1639 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1640 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1641 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1642 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1643 atomic64_t s_mb_generation_time; 1644 atomic_t s_mb_lost_chunks; 1645 atomic_t s_mb_preallocated; 1646 atomic_t s_mb_discarded; 1647 atomic_t s_lock_busy; 1648 1649 /* locality groups */ 1650 struct ext4_locality_group __percpu *s_locality_groups; 1651 1652 /* for write statistics */ 1653 unsigned long s_sectors_written_start; 1654 u64 s_kbytes_written; 1655 1656 /* the size of zero-out chunk */ 1657 unsigned int s_extent_max_zeroout_kb; 1658 1659 unsigned int s_log_groups_per_flex; 1660 struct flex_groups * __rcu *s_flex_groups; 1661 ext4_group_t s_flex_groups_allocated; 1662 1663 /* workqueue for reserved extent conversions (buffered io) */ 1664 struct workqueue_struct *rsv_conversion_wq; 1665 1666 /* timer for periodic error stats printing */ 1667 struct timer_list s_err_report; 1668 1669 /* Lazy inode table initialization info */ 1670 struct ext4_li_request *s_li_request; 1671 /* Wait multiplier for lazy initialization thread */ 1672 unsigned int s_li_wait_mult; 1673 1674 /* Kernel thread for multiple mount protection */ 1675 struct task_struct *s_mmp_tsk; 1676 1677 /* record the last minlen when FITRIM is called. */ 1678 unsigned long s_last_trim_minblks; 1679 1680 /* Precomputed FS UUID checksum for seeding other checksums */ 1681 __u32 s_csum_seed; 1682 1683 /* Reclaim extents from extent status tree */ 1684 struct shrinker *s_es_shrinker; 1685 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1686 long s_es_nr_inode; 1687 struct ext4_es_stats s_es_stats; 1688 struct mb_cache *s_ea_block_cache; 1689 struct mb_cache *s_ea_inode_cache; 1690 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1691 1692 /* Journal triggers for checksum computation */ 1693 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1694 1695 /* Ratelimit ext4 messages. */ 1696 struct ratelimit_state s_err_ratelimit_state; 1697 struct ratelimit_state s_warning_ratelimit_state; 1698 struct ratelimit_state s_msg_ratelimit_state; 1699 atomic_t s_warning_count; 1700 atomic_t s_msg_count; 1701 1702 /* Encryption policy for '-o test_dummy_encryption' */ 1703 struct fscrypt_dummy_policy s_dummy_enc_policy; 1704 1705 /* 1706 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1707 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1708 * DIOREAD_NOLOCK mount options on remount. 1709 */ 1710 struct percpu_rw_semaphore s_writepages_rwsem; 1711 struct dax_device *s_daxdev; 1712 u64 s_dax_part_off; 1713 #ifdef CONFIG_EXT4_DEBUG 1714 unsigned long s_simulate_fail; 1715 #endif 1716 /* Record the errseq of the backing block device */ 1717 errseq_t s_bdev_wb_err; 1718 spinlock_t s_bdev_wb_lock; 1719 1720 /* Information about errors that happened during this mount */ 1721 spinlock_t s_error_lock; 1722 int s_add_error_count; 1723 int s_first_error_code; 1724 __u32 s_first_error_line; 1725 __u32 s_first_error_ino; 1726 __u64 s_first_error_block; 1727 const char *s_first_error_func; 1728 time64_t s_first_error_time; 1729 int s_last_error_code; 1730 __u32 s_last_error_line; 1731 __u32 s_last_error_ino; 1732 __u64 s_last_error_block; 1733 const char *s_last_error_func; 1734 time64_t s_last_error_time; 1735 /* 1736 * If we are in a context where we cannot update the on-disk 1737 * superblock, we queue the work here. This is used to update 1738 * the error information in the superblock, and for periodic 1739 * updates of the superblock called from the commit callback 1740 * function. 1741 */ 1742 struct work_struct s_sb_upd_work; 1743 1744 /* Atomic write unit values in bytes */ 1745 unsigned int s_awu_min; 1746 unsigned int s_awu_max; 1747 1748 /* Ext4 fast commit sub transaction ID */ 1749 atomic_t s_fc_subtid; 1750 1751 /* 1752 * After commit starts, the main queue gets locked, and the further 1753 * updates get added in the staging queue. 1754 */ 1755 #define FC_Q_MAIN 0 1756 #define FC_Q_STAGING 1 1757 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1758 * that have data changes in them. 1759 */ 1760 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1761 unsigned int s_fc_bytes; 1762 /* 1763 * Main fast commit lock. This lock protects accesses to the 1764 * following fields: 1765 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1766 */ 1767 struct mutex s_fc_lock; 1768 struct buffer_head *s_fc_bh; 1769 struct ext4_fc_stats s_fc_stats; 1770 tid_t s_fc_ineligible_tid; 1771 #ifdef CONFIG_EXT4_DEBUG 1772 int s_fc_debug_max_replay; 1773 #endif 1774 struct ext4_fc_replay_state s_fc_replay_state; 1775 }; 1776 1777 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1778 { 1779 return sb->s_fs_info; 1780 } 1781 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1782 { 1783 return container_of(inode, struct ext4_inode_info, vfs_inode); 1784 } 1785 1786 static inline int ext4_writepages_down_read(struct super_block *sb) 1787 { 1788 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1789 return memalloc_nofs_save(); 1790 } 1791 1792 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1793 { 1794 memalloc_nofs_restore(ctx); 1795 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1796 } 1797 1798 static inline int ext4_writepages_down_write(struct super_block *sb) 1799 { 1800 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1801 return memalloc_nofs_save(); 1802 } 1803 1804 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1805 { 1806 memalloc_nofs_restore(ctx); 1807 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1808 } 1809 1810 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1811 { 1812 return ino == EXT4_ROOT_INO || 1813 (ino >= EXT4_FIRST_INO(sb) && 1814 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1815 } 1816 1817 static inline int ext4_get_resuid(struct ext4_super_block *es) 1818 { 1819 return le16_to_cpu(es->s_def_resuid) | 1820 le16_to_cpu(es->s_def_resuid_hi) << 16; 1821 } 1822 1823 static inline int ext4_get_resgid(struct ext4_super_block *es) 1824 { 1825 return le16_to_cpu(es->s_def_resgid) | 1826 le16_to_cpu(es->s_def_resgid_hi) << 16; 1827 } 1828 1829 /* 1830 * Returns: sbi->field[index] 1831 * Used to access an array element from the following sbi fields which require 1832 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1833 * - s_group_desc 1834 * - s_group_info 1835 * - s_flex_group 1836 */ 1837 #define sbi_array_rcu_deref(sbi, field, index) \ 1838 ({ \ 1839 typeof(*((sbi)->field)) _v; \ 1840 rcu_read_lock(); \ 1841 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1842 rcu_read_unlock(); \ 1843 _v; \ 1844 }) 1845 1846 /* 1847 * run-time mount flags 1848 */ 1849 enum { 1850 EXT4_MF_MNTDIR_SAMPLED, 1851 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1852 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */ 1853 }; 1854 1855 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1856 { 1857 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1858 } 1859 1860 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1861 { 1862 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1863 } 1864 1865 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1866 { 1867 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1868 } 1869 1870 1871 /* 1872 * Simulate_fail codes 1873 */ 1874 #define EXT4_SIM_BBITMAP_EIO 1 1875 #define EXT4_SIM_BBITMAP_CRC 2 1876 #define EXT4_SIM_IBITMAP_EIO 3 1877 #define EXT4_SIM_IBITMAP_CRC 4 1878 #define EXT4_SIM_INODE_EIO 5 1879 #define EXT4_SIM_INODE_CRC 6 1880 #define EXT4_SIM_DIRBLOCK_EIO 7 1881 #define EXT4_SIM_DIRBLOCK_CRC 8 1882 1883 static inline bool ext4_simulate_fail(struct super_block *sb, 1884 unsigned long code) 1885 { 1886 #ifdef CONFIG_EXT4_DEBUG 1887 struct ext4_sb_info *sbi = EXT4_SB(sb); 1888 1889 if (unlikely(sbi->s_simulate_fail == code)) { 1890 sbi->s_simulate_fail = 0; 1891 return true; 1892 } 1893 #endif 1894 return false; 1895 } 1896 1897 /* 1898 * Error number codes for s_{first,last}_error_errno 1899 * 1900 * Linux errno numbers are architecture specific, so we need to translate 1901 * them into something which is architecture independent. We don't define 1902 * codes for all errno's; just the ones which are most likely to be the cause 1903 * of an ext4_error() call. 1904 */ 1905 #define EXT4_ERR_UNKNOWN 1 1906 #define EXT4_ERR_EIO 2 1907 #define EXT4_ERR_ENOMEM 3 1908 #define EXT4_ERR_EFSBADCRC 4 1909 #define EXT4_ERR_EFSCORRUPTED 5 1910 #define EXT4_ERR_ENOSPC 6 1911 #define EXT4_ERR_ENOKEY 7 1912 #define EXT4_ERR_EROFS 8 1913 #define EXT4_ERR_EFBIG 9 1914 #define EXT4_ERR_EEXIST 10 1915 #define EXT4_ERR_ERANGE 11 1916 #define EXT4_ERR_EOVERFLOW 12 1917 #define EXT4_ERR_EBUSY 13 1918 #define EXT4_ERR_ENOTDIR 14 1919 #define EXT4_ERR_ENOTEMPTY 15 1920 #define EXT4_ERR_ESHUTDOWN 16 1921 #define EXT4_ERR_EFAULT 17 1922 1923 /* 1924 * Inode dynamic state flags 1925 */ 1926 enum { 1927 EXT4_STATE_NEW, /* inode is newly created */ 1928 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1929 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1930 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1931 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1932 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1933 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1934 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1935 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1936 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1937 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1938 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */ 1939 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1940 }; 1941 1942 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1943 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1944 { \ 1945 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1946 } \ 1947 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1948 { \ 1949 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1950 } \ 1951 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1952 { \ 1953 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1954 } 1955 1956 /* Add these declarations here only so that these functions can be 1957 * found by name. Otherwise, they are very hard to locate. */ 1958 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1959 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1960 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1961 EXT4_INODE_BIT_FNS(flag, flags, 0) 1962 1963 /* Add these declarations here only so that these functions can be 1964 * found by name. Otherwise, they are very hard to locate. */ 1965 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1966 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1967 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1968 #if (BITS_PER_LONG < 64) 1969 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1970 1971 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1972 { 1973 (ei)->i_state_flags = 0; 1974 } 1975 #else 1976 EXT4_INODE_BIT_FNS(state, flags, 32) 1977 1978 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1979 { 1980 /* We depend on the fact that callers will set i_flags */ 1981 } 1982 #endif 1983 #else 1984 /* Assume that user mode programs are passing in an ext4fs superblock, not 1985 * a kernel struct super_block. This will allow us to call the feature-test 1986 * macros from user land. */ 1987 #define EXT4_SB(sb) (sb) 1988 #endif 1989 1990 static inline bool ext4_verity_in_progress(struct inode *inode) 1991 { 1992 return IS_ENABLED(CONFIG_FS_VERITY) && 1993 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1994 } 1995 1996 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1997 1998 /* 1999 * Check whether the inode is tracked as orphan (either in orphan file or 2000 * orphan list). 2001 */ 2002 static inline bool ext4_inode_orphan_tracked(struct inode *inode) 2003 { 2004 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) || 2005 !list_empty(&EXT4_I(inode)->i_orphan); 2006 } 2007 2008 /* 2009 * Codes for operating systems 2010 */ 2011 #define EXT4_OS_LINUX 0 2012 #define EXT4_OS_HURD 1 2013 #define EXT4_OS_MASIX 2 2014 #define EXT4_OS_FREEBSD 3 2015 #define EXT4_OS_LITES 4 2016 2017 /* 2018 * Revision levels 2019 */ 2020 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 2021 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 2022 2023 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 2024 2025 #define EXT4_GOOD_OLD_INODE_SIZE 128 2026 2027 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 2028 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 2029 #define EXT4_TIMESTAMP_MIN S32_MIN 2030 2031 /* 2032 * Feature set definitions 2033 */ 2034 2035 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 2036 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2037 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2038 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2039 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2040 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2041 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2042 /* 2043 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2044 * incompatible only if fast commit blocks are present in the FS. Since we 2045 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2046 * incompatible. We also have a JBD2 incompat feature, which gets set when 2047 * there are fast commit blocks present in the journal. 2048 */ 2049 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2050 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2051 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2052 2053 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2054 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2055 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2056 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2057 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2058 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2059 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2060 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2061 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2062 /* 2063 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2064 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2065 * all other data structures' checksums. However, the METADATA_CSUM and 2066 * GDT_CSUM bits are mutually exclusive. 2067 */ 2068 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2069 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2070 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2071 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2072 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2073 non-empty */ 2074 2075 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2076 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2077 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2078 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2079 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2080 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2081 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2082 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2083 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2084 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2085 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2086 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2087 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2088 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2089 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2090 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2091 2092 extern void ext4_update_dynamic_rev(struct super_block *sb); 2093 2094 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2095 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2096 { \ 2097 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2098 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2099 } \ 2100 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2101 { \ 2102 ext4_update_dynamic_rev(sb); \ 2103 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2104 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2105 } \ 2106 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2107 { \ 2108 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2109 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2110 } 2111 2112 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2113 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2114 { \ 2115 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2116 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2117 } \ 2118 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2119 { \ 2120 ext4_update_dynamic_rev(sb); \ 2121 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2122 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2123 } \ 2124 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2125 { \ 2126 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2127 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2128 } 2129 2130 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2131 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2132 { \ 2133 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2134 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2135 } \ 2136 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2137 { \ 2138 ext4_update_dynamic_rev(sb); \ 2139 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2140 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2141 } \ 2142 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2143 { \ 2144 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2145 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2146 } 2147 2148 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2149 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2150 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2151 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2152 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2153 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2154 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2155 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2156 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2157 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2158 2159 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2160 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2161 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2162 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2163 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2164 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2165 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2166 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2167 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2168 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2169 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2170 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2171 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2172 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2173 2174 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2175 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2176 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2177 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2178 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2179 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2180 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2181 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2182 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2183 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2184 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2185 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2186 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2187 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2188 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2189 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2190 2191 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2192 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2193 EXT4_FEATURE_INCOMPAT_META_BG) 2194 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2195 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2196 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2197 2198 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2199 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2200 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2201 EXT4_FEATURE_INCOMPAT_META_BG) 2202 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2203 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2204 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2205 2206 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2207 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2208 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2209 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2210 EXT4_FEATURE_INCOMPAT_META_BG| \ 2211 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2212 EXT4_FEATURE_INCOMPAT_64BIT| \ 2213 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2214 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2215 EXT4_FEATURE_INCOMPAT_MMP | \ 2216 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2217 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2218 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2219 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2220 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2221 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2222 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2223 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2224 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2225 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2226 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2227 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2228 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2229 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2230 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2231 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2232 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2233 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2234 2235 #define EXTN_FEATURE_FUNCS(ver) \ 2236 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2237 { \ 2238 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2239 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2240 } \ 2241 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2242 { \ 2243 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2244 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2245 } \ 2246 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2247 { \ 2248 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2249 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2250 } 2251 2252 EXTN_FEATURE_FUNCS(2) 2253 EXTN_FEATURE_FUNCS(3) 2254 EXTN_FEATURE_FUNCS(4) 2255 2256 static inline bool ext4_has_compat_features(struct super_block *sb) 2257 { 2258 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2259 } 2260 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2261 { 2262 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2263 } 2264 static inline bool ext4_has_incompat_features(struct super_block *sb) 2265 { 2266 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2267 } 2268 2269 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2270 2271 /* 2272 * Superblock flags 2273 */ 2274 enum { 2275 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */ 2276 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */ 2277 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */ 2278 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */ 2279 }; 2280 2281 static inline int ext4_forced_shutdown(struct super_block *sb) 2282 { 2283 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2284 } 2285 2286 static inline int ext4_emergency_ro(struct super_block *sb) 2287 { 2288 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags); 2289 } 2290 2291 static inline int ext4_emergency_state(struct super_block *sb) 2292 { 2293 if (unlikely(ext4_forced_shutdown(sb))) 2294 return -EIO; 2295 if (unlikely(ext4_emergency_ro(sb))) 2296 return -EROFS; 2297 return 0; 2298 } 2299 2300 /* 2301 * Default values for user and/or group using reserved blocks 2302 */ 2303 #define EXT4_DEF_RESUID 0 2304 #define EXT4_DEF_RESGID 0 2305 2306 /* 2307 * Default project ID 2308 */ 2309 #define EXT4_DEF_PROJID 0 2310 2311 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2312 2313 /* 2314 * Default mount options 2315 */ 2316 #define EXT4_DEFM_DEBUG 0x0001 2317 #define EXT4_DEFM_BSDGROUPS 0x0002 2318 #define EXT4_DEFM_XATTR_USER 0x0004 2319 #define EXT4_DEFM_ACL 0x0008 2320 #define EXT4_DEFM_UID16 0x0010 2321 #define EXT4_DEFM_JMODE 0x0060 2322 #define EXT4_DEFM_JMODE_DATA 0x0020 2323 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2324 #define EXT4_DEFM_JMODE_WBACK 0x0060 2325 #define EXT4_DEFM_NOBARRIER 0x0100 2326 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2327 #define EXT4_DEFM_DISCARD 0x0400 2328 #define EXT4_DEFM_NODELALLOC 0x0800 2329 2330 /* 2331 * Default journal batch times and ioprio. 2332 */ 2333 #define EXT4_DEF_MIN_BATCH_TIME 0 2334 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2335 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) 2336 2337 2338 /* 2339 * Default values for superblock update 2340 */ 2341 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */ 2342 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */ 2343 2344 2345 /* 2346 * Minimum number of groups in a flexgroup before we separate out 2347 * directories into the first block group of a flexgroup 2348 */ 2349 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2350 2351 /* 2352 * Structure of a directory entry 2353 */ 2354 #define EXT4_NAME_LEN 255 2355 /* 2356 * Base length of the ext4 directory entry excluding the name length 2357 */ 2358 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2359 2360 struct ext4_dir_entry { 2361 __le32 inode; /* Inode number */ 2362 __le16 rec_len; /* Directory entry length */ 2363 __le16 name_len; /* Name length */ 2364 char name[EXT4_NAME_LEN]; /* File name */ 2365 }; 2366 2367 2368 /* 2369 * Encrypted Casefolded entries require saving the hash on disk. This structure 2370 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2371 * boundary. 2372 */ 2373 struct ext4_dir_entry_hash { 2374 __le32 hash; 2375 __le32 minor_hash; 2376 }; 2377 2378 /* 2379 * The new version of the directory entry. Since EXT4 structures are 2380 * stored in intel byte order, and the name_len field could never be 2381 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2382 * file_type field. 2383 */ 2384 struct ext4_dir_entry_2 { 2385 __le32 inode; /* Inode number */ 2386 __le16 rec_len; /* Directory entry length */ 2387 __u8 name_len; /* Name length */ 2388 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2389 char name[EXT4_NAME_LEN]; /* File name */ 2390 }; 2391 2392 /* 2393 * Access the hashes at the end of ext4_dir_entry_2 2394 */ 2395 #define EXT4_DIRENT_HASHES(entry) \ 2396 ((struct ext4_dir_entry_hash *) \ 2397 (((void *)(entry)) + \ 2398 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2399 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash) 2400 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2401 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash) 2402 2403 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2404 { 2405 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2406 } 2407 2408 /* 2409 * This is a bogus directory entry at the end of each leaf block that 2410 * records checksums. 2411 */ 2412 struct ext4_dir_entry_tail { 2413 __le32 det_reserved_zero1; /* Pretend to be unused */ 2414 __le16 det_rec_len; /* 12 */ 2415 __u8 det_reserved_zero2; /* Zero name length */ 2416 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2417 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2418 }; 2419 2420 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2421 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2422 ((blocksize) - \ 2423 sizeof(struct ext4_dir_entry_tail)))) 2424 2425 /* 2426 * Ext4 directory file types. Only the low 3 bits are used. The 2427 * other bits are reserved for now. 2428 */ 2429 #define EXT4_FT_UNKNOWN 0 2430 #define EXT4_FT_REG_FILE 1 2431 #define EXT4_FT_DIR 2 2432 #define EXT4_FT_CHRDEV 3 2433 #define EXT4_FT_BLKDEV 4 2434 #define EXT4_FT_FIFO 5 2435 #define EXT4_FT_SOCK 6 2436 #define EXT4_FT_SYMLINK 7 2437 2438 #define EXT4_FT_MAX 8 2439 2440 #define EXT4_FT_DIR_CSUM 0xDE 2441 2442 /* 2443 * EXT4_DIR_PAD defines the directory entries boundaries 2444 * 2445 * NOTE: It must be a multiple of 4 2446 */ 2447 #define EXT4_DIR_PAD 4 2448 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2449 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2450 2451 /* 2452 * The rec_len is dependent on the type of directory. Directories that are 2453 * casefolded and encrypted need to store the hash as well, so we add room for 2454 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2455 * pass NULL for dir, as those entries do not use the extra fields. 2456 */ 2457 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2458 const struct inode *dir) 2459 { 2460 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2461 2462 if (dir && ext4_hash_in_dirent(dir)) 2463 rec_len += sizeof(struct ext4_dir_entry_hash); 2464 return (rec_len & ~EXT4_DIR_ROUND); 2465 } 2466 2467 /* 2468 * If we ever get support for fs block sizes > page_size, we'll need 2469 * to remove the #if statements in the next two functions... 2470 */ 2471 static inline unsigned int 2472 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2473 { 2474 unsigned len = le16_to_cpu(dlen); 2475 2476 #if (PAGE_SIZE >= 65536) 2477 if (len == EXT4_MAX_REC_LEN || len == 0) 2478 return blocksize; 2479 return (len & 65532) | ((len & 3) << 16); 2480 #else 2481 return len; 2482 #endif 2483 } 2484 2485 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2486 { 2487 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2488 #if (PAGE_SIZE >= 65536) 2489 if (len < 65536) 2490 return cpu_to_le16(len); 2491 if (len == blocksize) { 2492 if (blocksize == 65536) 2493 return cpu_to_le16(EXT4_MAX_REC_LEN); 2494 else 2495 return cpu_to_le16(0); 2496 } 2497 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2498 #else 2499 return cpu_to_le16(len); 2500 #endif 2501 } 2502 2503 /* 2504 * Hash Tree Directory indexing 2505 * (c) Daniel Phillips, 2001 2506 */ 2507 2508 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2509 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2510 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2511 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2512 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2513 2514 /* Legal values for the dx_root hash_version field: */ 2515 2516 #define DX_HASH_LEGACY 0 2517 #define DX_HASH_HALF_MD4 1 2518 #define DX_HASH_TEA 2 2519 #define DX_HASH_LEGACY_UNSIGNED 3 2520 #define DX_HASH_HALF_MD4_UNSIGNED 4 2521 #define DX_HASH_TEA_UNSIGNED 5 2522 #define DX_HASH_SIPHASH 6 2523 #define DX_HASH_LAST DX_HASH_SIPHASH 2524 2525 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length) 2526 { 2527 return crc32c(crc, address, length); 2528 } 2529 2530 #ifdef __KERNEL__ 2531 2532 /* hash info structure used by the directory hash */ 2533 struct dx_hash_info 2534 { 2535 u32 hash; 2536 u32 minor_hash; 2537 int hash_version; 2538 u32 *seed; 2539 }; 2540 2541 2542 /* 32 and 64 bit signed EOF for dx directories */ 2543 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2544 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2545 2546 2547 /* 2548 * Control parameters used by ext4_htree_next_block 2549 */ 2550 #define HASH_NB_ALWAYS 1 2551 2552 struct ext4_filename { 2553 const struct qstr *usr_fname; 2554 struct fscrypt_str disk_name; 2555 struct dx_hash_info hinfo; 2556 #ifdef CONFIG_FS_ENCRYPTION 2557 struct fscrypt_str crypto_buf; 2558 #endif 2559 #if IS_ENABLED(CONFIG_UNICODE) 2560 struct qstr cf_name; 2561 #endif 2562 }; 2563 2564 #define fname_name(p) ((p)->disk_name.name) 2565 #define fname_usr_name(p) ((p)->usr_fname->name) 2566 #define fname_len(p) ((p)->disk_name.len) 2567 2568 /* 2569 * Describe an inode's exact location on disk and in memory 2570 */ 2571 struct ext4_iloc 2572 { 2573 struct buffer_head *bh; 2574 unsigned long offset; 2575 ext4_group_t block_group; 2576 }; 2577 2578 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2579 { 2580 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2581 } 2582 2583 static inline bool ext4_is_quota_file(struct inode *inode) 2584 { 2585 return IS_NOQUOTA(inode) && 2586 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2587 } 2588 2589 /* 2590 * This structure is stuffed into the struct file's private_data field 2591 * for directories. It is where we put information so that we can do 2592 * readdir operations in hash tree order. 2593 */ 2594 struct dir_private_info { 2595 struct rb_root root; 2596 struct rb_node *curr_node; 2597 struct fname *extra_fname; 2598 loff_t last_pos; 2599 __u32 curr_hash; 2600 __u32 curr_minor_hash; 2601 __u32 next_hash; 2602 u64 cookie; 2603 bool initialized; 2604 }; 2605 2606 /* calculate the first block number of the group */ 2607 static inline ext4_fsblk_t 2608 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2609 { 2610 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2611 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2612 } 2613 2614 /* 2615 * Special error return code only used by dx_probe() and its callers. 2616 */ 2617 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2618 2619 /* htree levels for ext4 */ 2620 #define EXT4_HTREE_LEVEL_COMPAT 2 2621 #define EXT4_HTREE_LEVEL 3 2622 2623 static inline int ext4_dir_htree_level(struct super_block *sb) 2624 { 2625 return ext4_has_feature_largedir(sb) ? 2626 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2627 } 2628 2629 /* 2630 * Timeout and state flag for lazy initialization inode thread. 2631 */ 2632 #define EXT4_DEF_LI_WAIT_MULT 10 2633 #define EXT4_DEF_LI_MAX_START_DELAY 5 2634 #define EXT4_LAZYINIT_QUIT 0x0001 2635 #define EXT4_LAZYINIT_RUNNING 0x0002 2636 2637 /* 2638 * Lazy inode table initialization info 2639 */ 2640 struct ext4_lazy_init { 2641 unsigned long li_state; 2642 struct list_head li_request_list; 2643 struct mutex li_list_mtx; 2644 }; 2645 2646 enum ext4_li_mode { 2647 EXT4_LI_MODE_PREFETCH_BBITMAP, 2648 EXT4_LI_MODE_ITABLE, 2649 }; 2650 2651 struct ext4_li_request { 2652 struct super_block *lr_super; 2653 enum ext4_li_mode lr_mode; 2654 ext4_group_t lr_first_not_zeroed; 2655 ext4_group_t lr_next_group; 2656 struct list_head lr_request; 2657 unsigned long lr_next_sched; 2658 unsigned long lr_timeout; 2659 }; 2660 2661 struct ext4_features { 2662 struct kobject f_kobj; 2663 struct completion f_kobj_unregister; 2664 }; 2665 2666 /* 2667 * This structure will be used for multiple mount protection. It will be 2668 * written into the block number saved in the s_mmp_block field in the 2669 * superblock. Programs that check MMP should assume that if 2670 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2671 * to use the filesystem, regardless of how old the timestamp is. 2672 */ 2673 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2674 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2675 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2676 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2677 2678 struct mmp_struct { 2679 __le32 mmp_magic; /* Magic number for MMP */ 2680 __le32 mmp_seq; /* Sequence no. updated periodically */ 2681 2682 /* 2683 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2684 * purposes and do not affect the correctness of the algorithm 2685 */ 2686 __le64 mmp_time; /* Time last updated */ 2687 char mmp_nodename[64]; /* Node which last updated MMP block */ 2688 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2689 2690 /* 2691 * mmp_check_interval is used to verify if the MMP block has been 2692 * updated on the block device. The value is updated based on the 2693 * maximum time to write the MMP block during an update cycle. 2694 */ 2695 __le16 mmp_check_interval; 2696 2697 __le16 mmp_pad1; 2698 __le32 mmp_pad2[226]; 2699 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2700 }; 2701 2702 /* arguments passed to the mmp thread */ 2703 struct mmpd_data { 2704 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2705 struct super_block *sb; /* super block of the fs */ 2706 }; 2707 2708 /* 2709 * Check interval multiplier 2710 * The MMP block is written every update interval and initially checked every 2711 * update interval x the multiplier (the value is then adapted based on the 2712 * write latency). The reason is that writes can be delayed under load and we 2713 * don't want readers to incorrectly assume that the filesystem is no longer 2714 * in use. 2715 */ 2716 #define EXT4_MMP_CHECK_MULT 2UL 2717 2718 /* 2719 * Minimum interval for MMP checking in seconds. 2720 */ 2721 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2722 2723 /* 2724 * Maximum interval for MMP checking in seconds. 2725 */ 2726 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2727 2728 /* 2729 * Function prototypes 2730 */ 2731 2732 /* 2733 * Ok, these declarations are also in <linux/kernel.h> but none of the 2734 * ext4 source programs needs to include it so they are duplicated here. 2735 */ 2736 # define NORET_TYPE /**/ 2737 # define ATTRIB_NORET __attribute__((noreturn)) 2738 # define NORET_AND noreturn, 2739 2740 /* bitmap.c */ 2741 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2742 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2743 struct ext4_group_desc *gdp, 2744 struct buffer_head *bh); 2745 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2746 struct ext4_group_desc *gdp, 2747 struct buffer_head *bh); 2748 void ext4_block_bitmap_csum_set(struct super_block *sb, 2749 struct ext4_group_desc *gdp, 2750 struct buffer_head *bh); 2751 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2752 struct ext4_group_desc *gdp, 2753 struct buffer_head *bh); 2754 2755 /* balloc.c */ 2756 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2757 ext4_fsblk_t blocknr, 2758 ext4_group_t *blockgrpp, 2759 ext4_grpblk_t *offsetp); 2760 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2761 ext4_fsblk_t block); 2762 2763 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2764 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2765 ext4_group_t group); 2766 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2767 ext4_fsblk_t goal, 2768 unsigned int flags, 2769 unsigned long *count, 2770 int *errp); 2771 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2772 s64 nclusters, unsigned int flags); 2773 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2774 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2775 ext4_group_t block_group, 2776 struct buffer_head ** bh); 2777 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2778 ext4_group_t group); 2779 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2780 2781 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2782 ext4_group_t block_group, 2783 bool ignore_locked); 2784 extern int ext4_wait_block_bitmap(struct super_block *sb, 2785 ext4_group_t block_group, 2786 struct buffer_head *bh); 2787 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2788 ext4_group_t block_group); 2789 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2790 ext4_group_t block_group, 2791 struct ext4_group_desc *gdp); 2792 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2793 2794 #if IS_ENABLED(CONFIG_UNICODE) 2795 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2796 const struct qstr *iname, 2797 struct ext4_filename *fname); 2798 2799 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2800 { 2801 kfree(fname->cf_name.name); 2802 fname->cf_name.name = NULL; 2803 } 2804 #else 2805 static inline int ext4_fname_setup_ci_filename(struct inode *dir, 2806 const struct qstr *iname, 2807 struct ext4_filename *fname) 2808 { 2809 return 0; 2810 } 2811 2812 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2813 { 2814 } 2815 #endif 2816 2817 /* ext4 encryption related stuff goes here crypto.c */ 2818 #ifdef CONFIG_FS_ENCRYPTION 2819 extern const struct fscrypt_operations ext4_cryptops; 2820 2821 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2822 int lookup, struct ext4_filename *fname); 2823 2824 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2825 struct ext4_filename *fname); 2826 2827 void ext4_fname_free_filename(struct ext4_filename *fname); 2828 2829 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2830 2831 #else /* !CONFIG_FS_ENCRYPTION */ 2832 static inline int ext4_fname_setup_filename(struct inode *dir, 2833 const struct qstr *iname, 2834 int lookup, 2835 struct ext4_filename *fname) 2836 { 2837 fname->usr_fname = iname; 2838 fname->disk_name.name = (unsigned char *) iname->name; 2839 fname->disk_name.len = iname->len; 2840 2841 return ext4_fname_setup_ci_filename(dir, iname, fname); 2842 } 2843 2844 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2845 struct dentry *dentry, 2846 struct ext4_filename *fname) 2847 { 2848 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2849 } 2850 2851 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2852 { 2853 ext4_fname_free_ci_filename(fname); 2854 } 2855 2856 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2857 void __user *arg) 2858 { 2859 return -EOPNOTSUPP; 2860 } 2861 #endif /* !CONFIG_FS_ENCRYPTION */ 2862 2863 /* dir.c */ 2864 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2865 struct file *, 2866 struct ext4_dir_entry_2 *, 2867 struct buffer_head *, char *, int, 2868 unsigned int); 2869 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2870 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2871 (de), (bh), (buf), (size), (offset))) 2872 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2873 __u32 minor_hash, 2874 struct ext4_dir_entry_2 *dirent, 2875 struct fscrypt_str *ent_name); 2876 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2877 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh, 2878 void *buf, int buf_size, 2879 struct ext4_filename *fname, 2880 struct ext4_dir_entry_2 **dest_de); 2881 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2882 struct ext4_dir_entry_2 *de, 2883 int buf_size, 2884 struct ext4_filename *fname); 2885 static inline void ext4_update_dx_flag(struct inode *inode) 2886 { 2887 if (!ext4_has_feature_dir_index(inode->i_sb) && 2888 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2889 /* ext4_iget() should have caught this... */ 2890 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2891 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2892 } 2893 } 2894 static const unsigned char ext4_filetype_table[] = { 2895 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2896 }; 2897 2898 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2899 { 2900 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2901 return DT_UNKNOWN; 2902 2903 return ext4_filetype_table[filetype]; 2904 } 2905 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2906 void *buf, int buf_size); 2907 2908 /* fsync.c */ 2909 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2910 2911 /* hash.c */ 2912 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2913 struct dx_hash_info *hinfo); 2914 2915 /* ialloc.c */ 2916 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2917 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2918 struct inode *, umode_t, 2919 const struct qstr *qstr, __u32 goal, 2920 uid_t *owner, __u32 i_flags, 2921 int handle_type, unsigned int line_no, 2922 int nblocks); 2923 2924 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2925 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2926 (goal), (owner), i_flags, 0, 0, 0) 2927 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2928 type, nblocks) \ 2929 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2930 0, (type), __LINE__, (nblocks)) 2931 2932 2933 extern void ext4_free_inode(handle_t *, struct inode *); 2934 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2935 extern unsigned long ext4_count_free_inodes(struct super_block *); 2936 extern unsigned long ext4_count_dirs(struct super_block *); 2937 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2938 extern int ext4_init_inode_table(struct super_block *sb, 2939 ext4_group_t group, int barrier); 2940 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2941 2942 /* fast_commit.c */ 2943 int ext4_fc_info_show(struct seq_file *seq, void *v); 2944 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2945 void ext4_fc_init_inode(struct inode *inode); 2946 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2947 ext4_lblk_t end); 2948 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2949 struct dentry *dentry); 2950 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2951 struct dentry *dentry); 2952 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2953 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2954 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2955 struct dentry *dentry); 2956 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2957 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2958 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2959 void ext4_fc_del(struct inode *inode); 2960 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2961 void ext4_fc_replay_cleanup(struct super_block *sb); 2962 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2963 int __init ext4_fc_init_dentry_cache(void); 2964 void ext4_fc_destroy_dentry_cache(void); 2965 int ext4_fc_record_regions(struct super_block *sb, int ino, 2966 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2967 int len, int replay); 2968 2969 /* mballoc.c */ 2970 extern const struct seq_operations ext4_mb_seq_groups_ops; 2971 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2972 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2973 extern int ext4_mb_init(struct super_block *); 2974 extern void ext4_mb_release(struct super_block *); 2975 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2976 struct ext4_allocation_request *, int *); 2977 extern void ext4_discard_preallocations(struct inode *); 2978 extern int __init ext4_init_mballoc(void); 2979 extern void ext4_exit_mballoc(void); 2980 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2981 ext4_group_t group, 2982 unsigned int nr, int *cnt); 2983 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2984 unsigned int nr); 2985 2986 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2987 struct buffer_head *bh, ext4_fsblk_t block, 2988 unsigned long count, int flags); 2989 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2990 ext4_group_t ngroups); 2991 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2992 ext4_group_t i, struct ext4_group_desc *desc); 2993 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2994 ext4_fsblk_t block, unsigned long count); 2995 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2996 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2997 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2998 int len, bool state); 2999 static inline bool ext4_mb_cr_expensive(enum criteria cr) 3000 { 3001 return cr >= CR_GOAL_LEN_SLOW; 3002 } 3003 3004 /* inode.c */ 3005 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 3006 struct ext4_inode_info *ei); 3007 int ext4_inode_is_fast_symlink(struct inode *inode); 3008 void ext4_check_map_extents_env(struct inode *inode); 3009 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 3010 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 3011 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 3012 bool wait, struct buffer_head **bhs); 3013 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 3014 struct buffer_head *bh_result, int create); 3015 int ext4_get_block(struct inode *inode, sector_t iblock, 3016 struct buffer_head *bh_result, int create); 3017 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 3018 struct buffer_head *bh, int create); 3019 int ext4_walk_page_buffers(handle_t *handle, 3020 struct inode *inode, 3021 struct buffer_head *head, 3022 unsigned from, 3023 unsigned to, 3024 int *partial, 3025 int (*fn)(handle_t *handle, struct inode *inode, 3026 struct buffer_head *bh)); 3027 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 3028 struct buffer_head *bh); 3029 void ext4_set_inode_mapping_order(struct inode *inode); 3030 #define FALL_BACK_TO_NONDELALLOC 1 3031 #define CONVERT_INLINE_DATA 2 3032 3033 typedef enum { 3034 EXT4_IGET_NORMAL = 0, 3035 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3036 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 3037 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 3038 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 3039 } ext4_iget_flags; 3040 3041 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3042 ext4_iget_flags flags, const char *function, 3043 unsigned int line); 3044 3045 #define ext4_iget(sb, ino, flags) \ 3046 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3047 3048 extern int ext4_write_inode(struct inode *, struct writeback_control *); 3049 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 3050 struct iattr *); 3051 extern u32 ext4_dio_alignment(struct inode *inode); 3052 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 3053 struct kstat *, u32, unsigned int); 3054 extern void ext4_evict_inode(struct inode *); 3055 extern void ext4_clear_inode(struct inode *); 3056 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 3057 struct kstat *, u32, unsigned int); 3058 extern void ext4_dirty_inode(struct inode *, int); 3059 extern int ext4_change_inode_journal_flag(struct inode *, int); 3060 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3061 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3062 struct ext4_iloc *iloc); 3063 extern int ext4_inode_attach_jinode(struct inode *inode); 3064 extern int ext4_can_truncate(struct inode *inode); 3065 extern int ext4_truncate(struct inode *); 3066 extern int ext4_break_layouts(struct inode *); 3067 extern int ext4_truncate_page_cache_block_range(struct inode *inode, 3068 loff_t start, loff_t end); 3069 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3070 extern void ext4_set_inode_flags(struct inode *, bool init); 3071 extern int ext4_alloc_da_blocks(struct inode *inode); 3072 extern void ext4_set_aops(struct inode *inode); 3073 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3074 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3075 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks); 3076 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks, 3077 int pextents); 3078 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3079 loff_t lstart, loff_t lend); 3080 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3081 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3082 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3083 extern void ext4_da_release_space(struct inode *inode, int to_free); 3084 extern void ext4_da_update_reserve_space(struct inode *inode, 3085 int used, int quota_claim); 3086 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3087 ext4_fsblk_t pblk, ext4_lblk_t len); 3088 3089 static inline bool is_special_ino(struct super_block *sb, unsigned long ino) 3090 { 3091 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 3092 3093 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) || 3094 ino == le32_to_cpu(es->s_usr_quota_inum) || 3095 ino == le32_to_cpu(es->s_grp_quota_inum) || 3096 ino == le32_to_cpu(es->s_prj_quota_inum) || 3097 ino == le32_to_cpu(es->s_orphan_file_inum); 3098 } 3099 3100 /* indirect.c */ 3101 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3102 struct ext4_map_blocks *map, int flags); 3103 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3104 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3105 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3106 ext4_lblk_t start, ext4_lblk_t end); 3107 3108 /* ioctl.c */ 3109 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3110 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3111 int ext4_fileattr_set(struct mnt_idmap *idmap, 3112 struct dentry *dentry, struct file_kattr *fa); 3113 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3114 extern void ext4_reset_inode_seed(struct inode *inode); 3115 int ext4_update_overhead(struct super_block *sb, bool force); 3116 int ext4_force_shutdown(struct super_block *sb, u32 flags); 3117 3118 /* migrate.c */ 3119 extern int ext4_ext_migrate(struct inode *); 3120 extern int ext4_ind_migrate(struct inode *inode); 3121 3122 /* namei.c */ 3123 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3124 struct inode *inode); 3125 extern int ext4_dirblock_csum_verify(struct inode *inode, 3126 struct buffer_head *bh); 3127 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3128 __u32 start_minor_hash, __u32 *next_hash); 3129 extern int ext4_search_dir(struct buffer_head *bh, 3130 char *search_buf, 3131 int buf_size, 3132 struct inode *dir, 3133 struct ext4_filename *fname, 3134 unsigned int offset, 3135 struct ext4_dir_entry_2 **res_dir); 3136 extern int ext4_generic_delete_entry(struct inode *dir, 3137 struct ext4_dir_entry_2 *de_del, 3138 struct buffer_head *bh, 3139 void *entry_buf, 3140 int buf_size, 3141 int csum_size); 3142 extern bool ext4_empty_dir(struct inode *inode); 3143 3144 /* resize.c */ 3145 extern void ext4_kvfree_array_rcu(void *to_free); 3146 extern int ext4_group_add(struct super_block *sb, 3147 struct ext4_new_group_data *input); 3148 extern int ext4_group_extend(struct super_block *sb, 3149 struct ext4_super_block *es, 3150 ext4_fsblk_t n_blocks_count); 3151 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3152 extern unsigned int ext4_list_backups(struct super_block *sb, 3153 unsigned int *three, unsigned int *five, 3154 unsigned int *seven); 3155 3156 /* super.c */ 3157 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3158 sector_t block, blk_opf_t op_flags); 3159 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3160 sector_t block); 3161 extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb, 3162 sector_t block); 3163 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3164 bh_end_io_t *end_io, bool simu_fail); 3165 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3166 bh_end_io_t *end_io, bool simu_fail); 3167 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3168 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3169 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3170 extern int ext4_calculate_overhead(struct super_block *sb); 3171 extern __le32 ext4_superblock_csum(struct ext4_super_block *es); 3172 extern void ext4_superblock_csum_set(struct super_block *sb); 3173 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3174 ext4_group_t ngroup); 3175 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3176 char nbuf[16]); 3177 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3178 ext4_group_t block_group, 3179 unsigned int flags); 3180 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3181 ext4_group_t block_group); 3182 3183 extern __printf(7, 8) 3184 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3185 int, __u64, const char *, ...); 3186 extern __printf(6, 7) 3187 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3188 ext4_fsblk_t, int, const char *, ...); 3189 extern __printf(5, 6) 3190 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3191 const char *, ...); 3192 extern void __ext4_std_error(struct super_block *, const char *, 3193 unsigned int, int); 3194 extern __printf(4, 5) 3195 void __ext4_warning(struct super_block *, const char *, unsigned int, 3196 const char *, ...); 3197 extern __printf(4, 5) 3198 void __ext4_warning_inode(const struct inode *inode, const char *function, 3199 unsigned int line, const char *fmt, ...); 3200 extern __printf(3, 4) 3201 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3202 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3203 const char *, unsigned int, const char *); 3204 extern __printf(7, 8) 3205 void __ext4_grp_locked_error(const char *, unsigned int, 3206 struct super_block *, ext4_group_t, 3207 unsigned long, ext4_fsblk_t, 3208 const char *, ...); 3209 3210 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3211 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3212 3213 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3214 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3215 3216 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3217 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3218 (fmt), ## a) 3219 3220 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3221 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3222 3223 #define ext4_abort(sb, err, fmt, a...) \ 3224 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3225 3226 #ifdef CONFIG_PRINTK 3227 3228 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3229 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3230 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3231 __ext4_error_inode((inode), (func), (line), (block), \ 3232 (err), (fmt), ##__VA_ARGS__) 3233 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3234 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3235 #define ext4_error(sb, fmt, ...) \ 3236 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3237 ##__VA_ARGS__) 3238 #define ext4_error_err(sb, err, fmt, ...) \ 3239 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3240 ##__VA_ARGS__) 3241 #define ext4_warning(sb, fmt, ...) \ 3242 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3243 #define ext4_warning_inode(inode, fmt, ...) \ 3244 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3245 #define ext4_msg(sb, level, fmt, ...) \ 3246 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3247 #define dump_mmp_msg(sb, mmp, msg) \ 3248 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3249 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3250 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3251 fmt, ##__VA_ARGS__) 3252 3253 #else 3254 3255 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3256 do { \ 3257 no_printk(fmt, ##__VA_ARGS__); \ 3258 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3259 } while (0) 3260 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3261 do { \ 3262 no_printk(fmt, ##__VA_ARGS__); \ 3263 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3264 } while (0) 3265 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3266 do { \ 3267 no_printk(fmt, ##__VA_ARGS__); \ 3268 __ext4_error_file(file, "", 0, block, " "); \ 3269 } while (0) 3270 #define ext4_error(sb, fmt, ...) \ 3271 do { \ 3272 no_printk(fmt, ##__VA_ARGS__); \ 3273 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3274 } while (0) 3275 #define ext4_error_err(sb, err, fmt, ...) \ 3276 do { \ 3277 no_printk(fmt, ##__VA_ARGS__); \ 3278 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3279 } while (0) 3280 #define ext4_warning(sb, fmt, ...) \ 3281 do { \ 3282 no_printk(fmt, ##__VA_ARGS__); \ 3283 __ext4_warning(sb, "", 0, " "); \ 3284 } while (0) 3285 #define ext4_warning_inode(inode, fmt, ...) \ 3286 do { \ 3287 no_printk(fmt, ##__VA_ARGS__); \ 3288 __ext4_warning_inode(inode, "", 0, " "); \ 3289 } while (0) 3290 #define ext4_msg(sb, level, fmt, ...) \ 3291 do { \ 3292 no_printk(fmt, ##__VA_ARGS__); \ 3293 __ext4_msg(sb, "", " "); \ 3294 } while (0) 3295 #define dump_mmp_msg(sb, mmp, msg) \ 3296 __dump_mmp_msg(sb, mmp, "", 0, "") 3297 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3298 do { \ 3299 no_printk(fmt, ##__VA_ARGS__); \ 3300 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3301 } while (0) 3302 3303 #endif 3304 3305 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3306 struct ext4_group_desc *bg); 3307 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3308 struct ext4_group_desc *bg); 3309 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3310 struct ext4_group_desc *bg); 3311 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3312 struct ext4_group_desc *bg); 3313 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3314 struct ext4_group_desc *bg); 3315 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3316 struct ext4_group_desc *bg); 3317 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3318 struct ext4_group_desc *bg); 3319 extern void ext4_block_bitmap_set(struct super_block *sb, 3320 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3321 extern void ext4_inode_bitmap_set(struct super_block *sb, 3322 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3323 extern void ext4_inode_table_set(struct super_block *sb, 3324 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3325 extern void ext4_free_group_clusters_set(struct super_block *sb, 3326 struct ext4_group_desc *bg, 3327 __u32 count); 3328 extern void ext4_free_inodes_set(struct super_block *sb, 3329 struct ext4_group_desc *bg, __u32 count); 3330 extern void ext4_used_dirs_set(struct super_block *sb, 3331 struct ext4_group_desc *bg, __u32 count); 3332 extern void ext4_itable_unused_set(struct super_block *sb, 3333 struct ext4_group_desc *bg, __u32 count); 3334 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3335 struct ext4_group_desc *gdp); 3336 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3337 struct ext4_group_desc *gdp); 3338 extern int ext4_register_li_request(struct super_block *sb, 3339 ext4_group_t first_not_zeroed); 3340 3341 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3342 { 3343 return ext4_has_feature_gdt_csum(sb) || 3344 ext4_has_feature_metadata_csum(sb); 3345 } 3346 3347 #define ext4_read_incompat_64bit_val(es, name) \ 3348 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3349 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3350 le32_to_cpu(es->name##_lo)) 3351 3352 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3353 { 3354 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3355 } 3356 3357 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3358 { 3359 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3360 } 3361 3362 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3363 { 3364 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3365 } 3366 3367 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3368 ext4_fsblk_t blk) 3369 { 3370 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3371 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3372 } 3373 3374 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3375 ext4_fsblk_t blk) 3376 { 3377 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3378 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3379 } 3380 3381 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3382 ext4_fsblk_t blk) 3383 { 3384 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3385 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3386 } 3387 3388 static inline loff_t ext4_isize(struct super_block *sb, 3389 struct ext4_inode *raw_inode) 3390 { 3391 if (ext4_has_feature_largedir(sb) || 3392 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3393 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3394 le32_to_cpu(raw_inode->i_size_lo); 3395 3396 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3397 } 3398 3399 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3400 { 3401 raw_inode->i_size_lo = cpu_to_le32(i_size); 3402 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3403 } 3404 3405 /* 3406 * Reading s_groups_count requires using smp_rmb() afterwards. See 3407 * the locking protocol documented in the comments of ext4_group_add() 3408 * in resize.c 3409 */ 3410 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3411 { 3412 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3413 3414 smp_rmb(); 3415 return ngroups; 3416 } 3417 3418 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3419 ext4_group_t block_group) 3420 { 3421 return block_group >> sbi->s_log_groups_per_flex; 3422 } 3423 3424 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3425 { 3426 return 1 << sbi->s_log_groups_per_flex; 3427 } 3428 3429 static inline loff_t ext4_get_maxbytes(struct inode *inode) 3430 { 3431 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) 3432 return inode->i_sb->s_maxbytes; 3433 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; 3434 } 3435 3436 #define ext4_std_error(sb, errno) \ 3437 do { \ 3438 if ((errno)) \ 3439 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3440 } while (0) 3441 3442 #ifdef CONFIG_SMP 3443 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3444 * counters. So we need to make sure we have free clusters more 3445 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3446 */ 3447 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3448 #else 3449 #define EXT4_FREECLUSTERS_WATERMARK 0 3450 #endif 3451 3452 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3453 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3454 { 3455 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3456 !inode_is_locked(inode)); 3457 down_write(&EXT4_I(inode)->i_data_sem); 3458 if (newsize > EXT4_I(inode)->i_disksize) 3459 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3460 up_write(&EXT4_I(inode)->i_data_sem); 3461 } 3462 3463 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3464 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3465 { 3466 int changed = 0; 3467 3468 if (newsize > inode->i_size) { 3469 i_size_write(inode, newsize); 3470 changed = 1; 3471 } 3472 if (newsize > EXT4_I(inode)->i_disksize) { 3473 ext4_update_i_disksize(inode, newsize); 3474 changed |= 2; 3475 } 3476 return changed; 3477 } 3478 3479 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3480 loff_t len); 3481 3482 struct ext4_group_info { 3483 unsigned long bb_state; 3484 #ifdef AGGRESSIVE_CHECK 3485 unsigned long bb_check_counter; 3486 #endif 3487 struct rb_root bb_free_root; 3488 ext4_grpblk_t bb_first_free; /* first free block */ 3489 ext4_grpblk_t bb_free; /* total free blocks */ 3490 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3491 int bb_avg_fragment_size_order; /* order of average 3492 fragment in BG */ 3493 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3494 ext4_group_t bb_group; /* Group number */ 3495 struct list_head bb_prealloc_list; 3496 #ifdef DOUBLE_CHECK 3497 void *bb_bitmap; 3498 #endif 3499 struct rw_semaphore alloc_sem; 3500 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3501 * regions, index is order. 3502 * bb_counters[3] = 5 means 3503 * 5 free 8-block regions. */ 3504 }; 3505 3506 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3507 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3508 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3509 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3510 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3511 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3512 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3513 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3514 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3515 3516 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3517 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3518 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3519 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3520 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3521 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3522 3523 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3524 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3525 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3526 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3527 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3528 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3529 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3530 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3531 3532 #define EXT4_MAX_CONTENTION 8 3533 #define EXT4_CONTENTION_THRESHOLD 2 3534 3535 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3536 ext4_group_t group) 3537 { 3538 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3539 } 3540 3541 /* 3542 * Returns true if the filesystem is busy enough that attempts to 3543 * access the block group locks has run into contention. 3544 */ 3545 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3546 { 3547 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3548 } 3549 3550 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group) 3551 { 3552 if (!spin_trylock(ext4_group_lock_ptr(sb, group))) 3553 return false; 3554 /* 3555 * We're able to grab the lock right away, so drop the lock 3556 * contention counter. 3557 */ 3558 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3559 return true; 3560 } 3561 3562 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3563 { 3564 if (!ext4_try_lock_group(sb, group)) { 3565 /* 3566 * The lock is busy, so bump the contention counter, 3567 * and then wait on the spin lock. 3568 */ 3569 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3570 EXT4_MAX_CONTENTION); 3571 spin_lock(ext4_group_lock_ptr(sb, group)); 3572 } 3573 } 3574 3575 static inline void ext4_unlock_group(struct super_block *sb, 3576 ext4_group_t group) 3577 { 3578 spin_unlock(ext4_group_lock_ptr(sb, group)); 3579 } 3580 3581 #ifdef CONFIG_QUOTA 3582 static inline bool ext4_quota_capable(struct super_block *sb) 3583 { 3584 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3585 } 3586 3587 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3588 { 3589 struct ext4_sb_info *sbi = EXT4_SB(sb); 3590 3591 return (ext4_has_feature_quota(sb) || 3592 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3593 } 3594 int ext4_enable_quotas(struct super_block *sb); 3595 #endif 3596 3597 /* 3598 * Block validity checking 3599 */ 3600 #define ext4_check_indirect_blockref(inode, bh) \ 3601 ext4_check_blockref(__func__, __LINE__, inode, \ 3602 (__le32 *)(bh)->b_data, \ 3603 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3604 3605 #define ext4_ind_check_inode(inode) \ 3606 ext4_check_blockref(__func__, __LINE__, inode, \ 3607 EXT4_I(inode)->i_data, \ 3608 EXT4_NDIR_BLOCKS) 3609 3610 /* 3611 * Inodes and files operations 3612 */ 3613 3614 /* dir.c */ 3615 extern const struct file_operations ext4_dir_operations; 3616 3617 /* file.c */ 3618 extern const struct inode_operations ext4_file_inode_operations; 3619 extern const struct file_operations ext4_file_operations; 3620 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3621 3622 /* inline.c */ 3623 extern int ext4_get_max_inline_size(struct inode *inode); 3624 extern int ext4_find_inline_data_nolock(struct inode *inode); 3625 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3626 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size); 3627 3628 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3629 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3630 struct inode *inode, 3631 loff_t pos, unsigned len, 3632 struct folio **foliop); 3633 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3634 unsigned copied, struct folio *folio); 3635 extern int ext4_generic_write_inline_data(struct address_space *mapping, 3636 struct inode *inode, 3637 loff_t pos, unsigned len, 3638 struct folio **foliop, 3639 void **fsdata, bool da); 3640 extern int ext4_try_add_inline_entry(handle_t *handle, 3641 struct ext4_filename *fname, 3642 struct inode *dir, struct inode *inode); 3643 extern int ext4_try_create_inline_dir(handle_t *handle, 3644 struct inode *parent, 3645 struct inode *inode); 3646 extern int ext4_read_inline_dir(struct file *filp, 3647 struct dir_context *ctx, 3648 int *has_inline_data); 3649 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3650 struct inode *dir, ext4_lblk_t block, 3651 struct dx_hash_info *hinfo, 3652 __u32 start_hash, __u32 start_minor_hash, 3653 int *has_inline_data); 3654 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3655 struct ext4_filename *fname, 3656 struct ext4_dir_entry_2 **res_dir, 3657 int *has_inline_data); 3658 extern int ext4_delete_inline_entry(handle_t *handle, 3659 struct inode *dir, 3660 struct ext4_dir_entry_2 *de_del, 3661 struct buffer_head *bh, 3662 int *has_inline_data); 3663 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3664 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3665 struct ext4_dir_entry_2 **parent_de, 3666 int *retval); 3667 extern void *ext4_read_inline_link(struct inode *inode); 3668 3669 struct iomap; 3670 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3671 3672 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3673 3674 extern int ext4_convert_inline_data(struct inode *inode); 3675 3676 static inline int ext4_has_inline_data(struct inode *inode) 3677 { 3678 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3679 EXT4_I(inode)->i_inline_off; 3680 } 3681 3682 /* namei.c */ 3683 extern const struct inode_operations ext4_dir_inode_operations; 3684 extern const struct inode_operations ext4_special_inode_operations; 3685 extern struct dentry *ext4_get_parent(struct dentry *child); 3686 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode, 3687 struct buffer_head *dir_block, 3688 unsigned int parent_ino, void *inline_buf, 3689 int inline_size); 3690 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3691 unsigned int blocksize); 3692 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3693 struct buffer_head *bh); 3694 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3695 struct inode *inode, struct dentry *dentry); 3696 extern int __ext4_link(struct inode *dir, struct inode *inode, 3697 struct dentry *dentry); 3698 3699 #define S_SHIFT 12 3700 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3701 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3702 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3703 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3704 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3705 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3706 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3707 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3708 }; 3709 3710 static inline void ext4_set_de_type(struct super_block *sb, 3711 struct ext4_dir_entry_2 *de, 3712 umode_t mode) { 3713 if (ext4_has_feature_filetype(sb)) 3714 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3715 } 3716 3717 /* readpages.c */ 3718 extern int ext4_mpage_readpages(struct inode *inode, 3719 struct readahead_control *rac, struct folio *folio); 3720 extern int __init ext4_init_post_read_processing(void); 3721 extern void ext4_exit_post_read_processing(void); 3722 3723 /* symlink.c */ 3724 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3725 extern const struct inode_operations ext4_symlink_inode_operations; 3726 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3727 3728 /* sysfs.c */ 3729 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3730 extern int ext4_register_sysfs(struct super_block *sb); 3731 extern void ext4_unregister_sysfs(struct super_block *sb); 3732 extern int __init ext4_init_sysfs(void); 3733 extern void ext4_exit_sysfs(void); 3734 3735 /* block_validity */ 3736 extern void ext4_release_system_zone(struct super_block *sb); 3737 extern int ext4_setup_system_zone(struct super_block *sb); 3738 extern int __init ext4_init_system_zone(void); 3739 extern void ext4_exit_system_zone(void); 3740 extern int ext4_inode_block_valid(struct inode *inode, 3741 ext4_fsblk_t start_blk, 3742 unsigned int count); 3743 extern int ext4_check_blockref(const char *, unsigned int, 3744 struct inode *, __le32 *, unsigned int); 3745 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3746 ext4_fsblk_t start_blk, unsigned int count); 3747 3748 3749 /* extents.c */ 3750 struct ext4_ext_path; 3751 struct ext4_extent; 3752 3753 /* 3754 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3755 * __le32. 3756 */ 3757 #define EXT_MAX_BLOCKS 0xffffffff 3758 3759 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3760 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3761 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3762 struct ext4_map_blocks *map, int flags); 3763 extern int ext4_ext_truncate(handle_t *, struct inode *); 3764 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3765 ext4_lblk_t end); 3766 extern void ext4_ext_init(struct super_block *); 3767 extern void ext4_ext_release(struct super_block *); 3768 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3769 loff_t len); 3770 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3771 loff_t offset, ssize_t len); 3772 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle, 3773 struct inode *inode, loff_t offset, ssize_t len); 3774 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3775 ext4_io_end_t *io_end); 3776 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3777 struct ext4_map_blocks *map, int flags); 3778 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3779 int num, 3780 struct ext4_ext_path *path); 3781 extern struct ext4_ext_path *ext4_ext_insert_extent( 3782 handle_t *handle, struct inode *inode, 3783 struct ext4_ext_path *path, 3784 struct ext4_extent *newext, int gb_flags); 3785 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3786 struct ext4_ext_path *, 3787 int flags); 3788 extern void ext4_free_ext_path(struct ext4_ext_path *); 3789 extern int ext4_ext_check_inode(struct inode *inode); 3790 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3791 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3792 __u64 start, __u64 len); 3793 extern int ext4_get_es_cache(struct inode *inode, 3794 struct fiemap_extent_info *fieinfo, 3795 __u64 start, __u64 len); 3796 extern int ext4_ext_precache(struct inode *inode); 3797 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3798 struct inode *inode2, ext4_lblk_t lblk1, 3799 ext4_lblk_t lblk2, ext4_lblk_t count, 3800 int mark_unwritten,int *err); 3801 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3802 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3803 int check_cred, int restart_cred, 3804 int revoke_cred); 3805 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3806 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3807 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3808 int len, int unwritten, ext4_fsblk_t pblk); 3809 extern int ext4_ext_clear_bb(struct inode *inode); 3810 3811 3812 /* move_extent.c */ 3813 extern void ext4_double_down_write_data_sem(struct inode *first, 3814 struct inode *second); 3815 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3816 struct inode *donor_inode); 3817 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3818 __u64 start_orig, __u64 start_donor, 3819 __u64 len, __u64 *moved_len); 3820 3821 /* page-io.c */ 3822 extern int __init ext4_init_pageio(void); 3823 extern void ext4_exit_pageio(void); 3824 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3825 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3826 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3827 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3828 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3829 struct writeback_control *wbc); 3830 extern void ext4_end_io_rsv_work(struct work_struct *work); 3831 extern void ext4_io_submit(struct ext4_io_submit *io); 3832 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3833 size_t len); 3834 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3835 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3836 3837 /* mmp.c */ 3838 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3839 3840 /* mmp.c */ 3841 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3842 3843 /* verity.c */ 3844 extern const struct fsverity_operations ext4_verityops; 3845 3846 /* orphan.c */ 3847 extern int ext4_orphan_add(handle_t *, struct inode *); 3848 extern int ext4_orphan_del(handle_t *, struct inode *); 3849 extern void ext4_orphan_cleanup(struct super_block *sb, 3850 struct ext4_super_block *es); 3851 extern void ext4_release_orphan_info(struct super_block *sb); 3852 extern int ext4_init_orphan_info(struct super_block *sb); 3853 extern int ext4_orphan_file_empty(struct super_block *sb); 3854 extern void ext4_orphan_file_block_trigger( 3855 struct jbd2_buffer_trigger_type *triggers, 3856 struct buffer_head *bh, 3857 void *data, size_t size); 3858 3859 /* 3860 * Add new method to test whether block and inode bitmaps are properly 3861 * initialized. With uninit_bg reading the block from disk is not enough 3862 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3863 */ 3864 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3865 3866 static inline int bitmap_uptodate(struct buffer_head *bh) 3867 { 3868 return (buffer_uptodate(bh) && 3869 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3870 } 3871 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3872 { 3873 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3874 } 3875 3876 extern int ext4_resize_begin(struct super_block *sb); 3877 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3878 3879 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end) 3880 { 3881 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) 3882 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3883 } 3884 3885 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3886 { 3887 if (io_end->flag & EXT4_IO_END_UNWRITTEN) 3888 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3889 } 3890 3891 extern const struct iomap_ops ext4_iomap_ops; 3892 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3893 extern const struct iomap_ops ext4_iomap_report_ops; 3894 3895 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3896 { 3897 /* 3898 * If the buffer has the write error flag, we have failed 3899 * to write out data in the block. In this case, we don't 3900 * have to read the block because we may read the old data 3901 * successfully. 3902 */ 3903 if (buffer_write_io_error(bh)) 3904 set_buffer_uptodate(bh); 3905 return buffer_uptodate(bh); 3906 } 3907 3908 static inline bool ext4_inode_can_atomic_write(struct inode *inode) 3909 { 3910 3911 return S_ISREG(inode->i_mode) && 3912 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) && 3913 EXT4_SB(inode->i_sb)->s_awu_min > 0; 3914 } 3915 3916 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio, 3917 loff_t pos, unsigned len, 3918 get_block_t *get_block); 3919 #endif /* __KERNEL__ */ 3920 3921 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3922 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3923 3924 #endif /* _EXT4_H */ 3925