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/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/version.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 <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #include <linux/fscrypt.h> 44 #include <linux/fsverity.h> 45 46 #include <linux/compiler.h> 47 48 /* 49 * The fourth extended filesystem constants/structures 50 */ 51 52 /* 53 * with AGGRESSIVE_CHECK allocator runs consistency checks over 54 * structures. these checks slow things down a lot 55 */ 56 #define AGGRESSIVE_CHECK__ 57 58 /* 59 * with DOUBLE_CHECK defined mballoc creates persistent in-core 60 * bitmaps, maintains and uses them to check for double allocations 61 */ 62 #define DOUBLE_CHECK__ 63 64 /* 65 * Define EXT4FS_DEBUG to produce debug messages 66 */ 67 #undef EXT4FS_DEBUG 68 69 /* 70 * Debug code 71 */ 72 #ifdef EXT4FS_DEBUG 73 #define ext4_debug(f, a...) \ 74 do { \ 75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 76 __FILE__, __LINE__, __func__); \ 77 printk(KERN_DEBUG f, ## a); \ 78 } while (0) 79 #else 80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 81 #endif 82 83 /* 84 * Turn on EXT_DEBUG to get lots of info about extents operations. 85 */ 86 #define EXT_DEBUG__ 87 #ifdef EXT_DEBUG 88 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 89 #else 90 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 91 #endif 92 93 /* data type for block offset of block group */ 94 typedef int ext4_grpblk_t; 95 96 /* data type for filesystem-wide blocks number */ 97 typedef unsigned long long ext4_fsblk_t; 98 99 /* data type for file logical block number */ 100 typedef __u32 ext4_lblk_t; 101 102 /* data type for block group number */ 103 typedef unsigned int ext4_group_t; 104 105 enum SHIFT_DIRECTION { 106 SHIFT_LEFT = 0, 107 SHIFT_RIGHT, 108 }; 109 110 /* 111 * Flags used in mballoc's allocation_context flags field. 112 * 113 * Also used to show what's going on for debugging purposes when the 114 * flag field is exported via the traceport interface 115 */ 116 117 /* prefer goal again. length */ 118 #define EXT4_MB_HINT_MERGE 0x0001 119 /* blocks already reserved */ 120 #define EXT4_MB_HINT_RESERVED 0x0002 121 /* metadata is being allocated */ 122 #define EXT4_MB_HINT_METADATA 0x0004 123 /* first blocks in the file */ 124 #define EXT4_MB_HINT_FIRST 0x0008 125 /* search for the best chunk */ 126 #define EXT4_MB_HINT_BEST 0x0010 127 /* data is being allocated */ 128 #define EXT4_MB_HINT_DATA 0x0020 129 /* don't preallocate (for tails) */ 130 #define EXT4_MB_HINT_NOPREALLOC 0x0040 131 /* allocate for locality group */ 132 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 133 /* allocate goal blocks or none */ 134 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 135 /* goal is meaningful */ 136 #define EXT4_MB_HINT_TRY_GOAL 0x0200 137 /* blocks already pre-reserved by delayed allocation */ 138 #define EXT4_MB_DELALLOC_RESERVED 0x0400 139 /* We are doing stream allocation */ 140 #define EXT4_MB_STREAM_ALLOC 0x0800 141 /* Use reserved root blocks if needed */ 142 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 143 /* Use blocks from reserved pool */ 144 #define EXT4_MB_USE_RESERVED 0x2000 145 146 struct ext4_allocation_request { 147 /* target inode for block we're allocating */ 148 struct inode *inode; 149 /* how many blocks we want to allocate */ 150 unsigned int len; 151 /* logical block in target inode */ 152 ext4_lblk_t logical; 153 /* the closest logical allocated block to the left */ 154 ext4_lblk_t lleft; 155 /* the closest logical allocated block to the right */ 156 ext4_lblk_t lright; 157 /* phys. target (a hint) */ 158 ext4_fsblk_t goal; 159 /* phys. block for the closest logical allocated block to the left */ 160 ext4_fsblk_t pleft; 161 /* phys. block for the closest logical allocated block to the right */ 162 ext4_fsblk_t pright; 163 /* flags. see above EXT4_MB_HINT_* */ 164 unsigned int flags; 165 }; 166 167 /* 168 * Logical to physical block mapping, used by ext4_map_blocks() 169 * 170 * This structure is used to pass requests into ext4_map_blocks() as 171 * well as to store the information returned by ext4_map_blocks(). It 172 * takes less room on the stack than a struct buffer_head. 173 */ 174 #define EXT4_MAP_NEW (1 << BH_New) 175 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 176 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 177 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 178 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 180 181 struct ext4_map_blocks { 182 ext4_fsblk_t m_pblk; 183 ext4_lblk_t m_lblk; 184 unsigned int m_len; 185 unsigned int m_flags; 186 }; 187 188 /* 189 * Flags for ext4_io_end->flags 190 */ 191 #define EXT4_IO_END_UNWRITTEN 0x0001 192 193 /* 194 * For converting unwritten extents on a work queue. 'handle' is used for 195 * buffered writeback. 196 */ 197 typedef struct ext4_io_end { 198 struct list_head list; /* per-file finished IO list */ 199 handle_t *handle; /* handle reserved for extent 200 * conversion */ 201 struct inode *inode; /* file being written to */ 202 struct bio *bio; /* Linked list of completed 203 * bios covering the extent */ 204 unsigned int flag; /* unwritten or not */ 205 atomic_t count; /* reference counter */ 206 loff_t offset; /* offset in the file */ 207 ssize_t size; /* size of the extent */ 208 } ext4_io_end_t; 209 210 struct ext4_io_submit { 211 struct writeback_control *io_wbc; 212 struct bio *io_bio; 213 ext4_io_end_t *io_end; 214 sector_t io_next_block; 215 }; 216 217 /* 218 * Special inodes numbers 219 */ 220 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 221 #define EXT4_ROOT_INO 2 /* Root inode */ 222 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 223 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 224 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 225 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 226 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 227 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 228 229 /* First non-reserved inode for old ext4 filesystems */ 230 #define EXT4_GOOD_OLD_FIRST_INO 11 231 232 /* 233 * Maximal count of links to a file 234 */ 235 #define EXT4_LINK_MAX 65000 236 237 /* 238 * Macro-instructions used to manage several block sizes 239 */ 240 #define EXT4_MIN_BLOCK_SIZE 1024 241 #define EXT4_MAX_BLOCK_SIZE 65536 242 #define EXT4_MIN_BLOCK_LOG_SIZE 10 243 #define EXT4_MAX_BLOCK_LOG_SIZE 16 244 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 245 #ifdef __KERNEL__ 246 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 247 #else 248 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 249 #endif 250 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 251 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 252 EXT4_SB(s)->s_cluster_bits) 253 #ifdef __KERNEL__ 254 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 255 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 256 #else 257 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 258 #endif 259 #ifdef __KERNEL__ 260 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 261 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 262 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 263 #else 264 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 265 EXT4_GOOD_OLD_INODE_SIZE : \ 266 (s)->s_inode_size) 267 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 268 EXT4_GOOD_OLD_FIRST_INO : \ 269 (s)->s_first_ino) 270 #endif 271 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 272 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 273 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 274 blkbits)) 275 276 /* Translate a block number to a cluster number */ 277 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 278 /* Translate a cluster number to a block number */ 279 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 280 /* Translate # of blks to # of clusters */ 281 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 282 (sbi)->s_cluster_bits) 283 /* Mask out the low bits to get the starting block of the cluster */ 284 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 285 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 286 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 287 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 288 /* Get the cluster offset */ 289 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 290 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 291 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 292 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 293 294 /* 295 * Structure of a blocks group descriptor 296 */ 297 struct ext4_group_desc 298 { 299 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 300 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 301 __le32 bg_inode_table_lo; /* Inodes table block */ 302 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 303 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 304 __le16 bg_used_dirs_count_lo; /* Directories count */ 305 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 306 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 307 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 308 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 309 __le16 bg_itable_unused_lo; /* Unused inodes count */ 310 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 311 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 312 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 313 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 314 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 315 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 316 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 317 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 318 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 319 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 320 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 321 __u32 bg_reserved; 322 }; 323 324 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 325 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 326 sizeof(__le16)) 327 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 328 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 329 sizeof(__le16)) 330 331 /* 332 * Structure of a flex block group info 333 */ 334 335 struct flex_groups { 336 atomic64_t free_clusters; 337 atomic_t free_inodes; 338 atomic_t used_dirs; 339 }; 340 341 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 342 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 343 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 344 345 /* 346 * Macro-instructions used to manage group descriptors 347 */ 348 #define EXT4_MIN_DESC_SIZE 32 349 #define EXT4_MIN_DESC_SIZE_64BIT 64 350 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 351 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 352 #ifdef __KERNEL__ 353 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 354 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 356 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 357 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 358 #else 359 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 361 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 362 #endif 363 364 /* 365 * Constants relative to the data blocks 366 */ 367 #define EXT4_NDIR_BLOCKS 12 368 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 369 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 370 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 371 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 372 373 /* 374 * Inode flags 375 */ 376 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 377 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 378 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 379 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 380 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 381 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 382 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 383 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 384 /* Reserved for compression usage... */ 385 #define EXT4_DIRTY_FL 0x00000100 386 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 387 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 388 /* nb: was previously EXT2_ECOMPR_FL */ 389 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 390 /* End compression flags --- maybe not all used */ 391 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 392 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 393 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 394 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 395 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 396 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 397 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 398 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 399 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 400 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 401 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 402 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 403 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 404 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */ 405 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 406 407 #define EXT4_FL_USER_VISIBLE 0x705BDFFF /* User visible flags */ 408 #define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */ 409 410 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 411 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 412 EXT4_IMMUTABLE_FL | \ 413 EXT4_APPEND_FL | \ 414 EXT4_NODUMP_FL | \ 415 EXT4_NOATIME_FL | \ 416 EXT4_PROJINHERIT_FL) 417 418 /* Flags that should be inherited by new inodes from their parent. */ 419 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 420 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 421 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 422 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 423 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL) 424 425 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 426 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 427 EXT4_PROJINHERIT_FL)) 428 429 /* Flags that are appropriate for non-directories/regular files. */ 430 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 431 432 /* The only flags that should be swapped */ 433 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 434 435 /* Mask out flags that are inappropriate for the given type of inode. */ 436 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 437 { 438 if (S_ISDIR(mode)) 439 return flags; 440 else if (S_ISREG(mode)) 441 return flags & EXT4_REG_FLMASK; 442 else 443 return flags & EXT4_OTHER_FLMASK; 444 } 445 446 /* 447 * Inode flags used for atomic set/get 448 */ 449 enum { 450 EXT4_INODE_SECRM = 0, /* Secure deletion */ 451 EXT4_INODE_UNRM = 1, /* Undelete */ 452 EXT4_INODE_COMPR = 2, /* Compress file */ 453 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 454 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 455 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 456 EXT4_INODE_NODUMP = 6, /* do not dump file */ 457 EXT4_INODE_NOATIME = 7, /* do not update atime */ 458 /* Reserved for compression usage... */ 459 EXT4_INODE_DIRTY = 8, 460 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 461 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 462 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 463 /* End compression flags --- maybe not all used */ 464 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 465 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 466 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 467 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 468 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 469 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 470 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 471 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 472 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 473 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 474 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 475 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 476 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 477 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 478 }; 479 480 /* 481 * Since it's pretty easy to mix up bit numbers and hex values, we use a 482 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 483 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 484 * any extra space in the compiled kernel image, otherwise, the build will fail. 485 * It's important that these values are the same, since we are using 486 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 487 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 488 * values found in ext2, ext3 and ext4 filesystems, and of course the values 489 * defined in e2fsprogs. 490 * 491 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 492 */ 493 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 494 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 495 496 static inline void ext4_check_flag_values(void) 497 { 498 CHECK_FLAG_VALUE(SECRM); 499 CHECK_FLAG_VALUE(UNRM); 500 CHECK_FLAG_VALUE(COMPR); 501 CHECK_FLAG_VALUE(SYNC); 502 CHECK_FLAG_VALUE(IMMUTABLE); 503 CHECK_FLAG_VALUE(APPEND); 504 CHECK_FLAG_VALUE(NODUMP); 505 CHECK_FLAG_VALUE(NOATIME); 506 CHECK_FLAG_VALUE(DIRTY); 507 CHECK_FLAG_VALUE(COMPRBLK); 508 CHECK_FLAG_VALUE(NOCOMPR); 509 CHECK_FLAG_VALUE(ENCRYPT); 510 CHECK_FLAG_VALUE(INDEX); 511 CHECK_FLAG_VALUE(IMAGIC); 512 CHECK_FLAG_VALUE(JOURNAL_DATA); 513 CHECK_FLAG_VALUE(NOTAIL); 514 CHECK_FLAG_VALUE(DIRSYNC); 515 CHECK_FLAG_VALUE(TOPDIR); 516 CHECK_FLAG_VALUE(HUGE_FILE); 517 CHECK_FLAG_VALUE(EXTENTS); 518 CHECK_FLAG_VALUE(VERITY); 519 CHECK_FLAG_VALUE(EA_INODE); 520 CHECK_FLAG_VALUE(EOFBLOCKS); 521 CHECK_FLAG_VALUE(INLINE_DATA); 522 CHECK_FLAG_VALUE(PROJINHERIT); 523 CHECK_FLAG_VALUE(RESERVED); 524 } 525 526 /* Used to pass group descriptor data when online resize is done */ 527 struct ext4_new_group_input { 528 __u32 group; /* Group number for this data */ 529 __u64 block_bitmap; /* Absolute block number of block bitmap */ 530 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 531 __u64 inode_table; /* Absolute block number of inode table start */ 532 __u32 blocks_count; /* Total number of blocks in this group */ 533 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 534 __u16 unused; 535 }; 536 537 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 538 struct compat_ext4_new_group_input { 539 u32 group; 540 compat_u64 block_bitmap; 541 compat_u64 inode_bitmap; 542 compat_u64 inode_table; 543 u32 blocks_count; 544 u16 reserved_blocks; 545 u16 unused; 546 }; 547 #endif 548 549 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 550 struct ext4_new_group_data { 551 __u32 group; 552 __u64 block_bitmap; 553 __u64 inode_bitmap; 554 __u64 inode_table; 555 __u32 blocks_count; 556 __u16 reserved_blocks; 557 __u16 mdata_blocks; 558 __u32 free_clusters_count; 559 }; 560 561 /* Indexes used to index group tables in ext4_new_group_data */ 562 enum { 563 BLOCK_BITMAP = 0, /* block bitmap */ 564 INODE_BITMAP, /* inode bitmap */ 565 INODE_TABLE, /* inode tables */ 566 GROUP_TABLE_COUNT, 567 }; 568 569 /* 570 * Flags used by ext4_map_blocks() 571 */ 572 /* Allocate any needed blocks and/or convert an unwritten 573 extent to be an initialized ext4 */ 574 #define EXT4_GET_BLOCKS_CREATE 0x0001 575 /* Request the creation of an unwritten extent */ 576 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 577 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 578 EXT4_GET_BLOCKS_CREATE) 579 /* Caller is from the delayed allocation writeout path 580 * finally doing the actual allocation of delayed blocks */ 581 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 582 /* caller is from the direct IO path, request to creation of an 583 unwritten extents if not allocated, split the unwritten 584 extent if blocks has been preallocated already*/ 585 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 586 #define EXT4_GET_BLOCKS_CONVERT 0x0010 587 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 588 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 589 /* Convert extent to initialized after IO complete */ 590 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 591 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 592 /* Eventual metadata allocation (due to growing extent tree) 593 * should not fail, so try to use reserved blocks for that.*/ 594 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 595 /* Don't normalize allocation size (used for fallocate) */ 596 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 597 /* Request will not result in inode size update (user for fallocate) */ 598 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 599 /* Convert written extents to unwritten */ 600 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 601 /* Write zeros to newly created written extents */ 602 #define EXT4_GET_BLOCKS_ZERO 0x0200 603 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 604 EXT4_GET_BLOCKS_ZERO) 605 /* Caller will submit data before dropping transaction handle. This 606 * allows jbd2 to avoid submitting data before commit. */ 607 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 608 609 /* 610 * The bit position of these flags must not overlap with any of the 611 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 612 * read_extent_tree_block(), ext4_split_extent_at(), 613 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 614 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 615 * caching the extents when reading from the extent tree while a 616 * truncate or punch hole operation is in progress. 617 */ 618 #define EXT4_EX_NOCACHE 0x40000000 619 #define EXT4_EX_FORCE_CACHE 0x20000000 620 621 /* 622 * Flags used by ext4_free_blocks 623 */ 624 #define EXT4_FREE_BLOCKS_METADATA 0x0001 625 #define EXT4_FREE_BLOCKS_FORGET 0x0002 626 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 627 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 628 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 629 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 630 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 631 632 /* 633 * ioctl commands 634 */ 635 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 636 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 637 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 638 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 639 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 640 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 641 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 642 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 643 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 644 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 645 #define EXT4_IOC_MIGRATE _IO('f', 9) 646 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 647 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 648 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 649 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 650 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 651 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 652 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 653 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 654 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 655 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 656 657 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 658 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 659 660 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 661 662 /* 663 * Flags for going down operation 664 */ 665 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 666 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 667 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 668 669 670 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 671 /* 672 * ioctl commands in 32 bit emulation 673 */ 674 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 675 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 676 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 677 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 678 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 679 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 680 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 681 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 682 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 683 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 684 #endif 685 686 /* Max physical block we can address w/o extents */ 687 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 688 689 /* Max logical block we can support */ 690 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF 691 692 /* 693 * Structure of an inode on the disk 694 */ 695 struct ext4_inode { 696 __le16 i_mode; /* File mode */ 697 __le16 i_uid; /* Low 16 bits of Owner Uid */ 698 __le32 i_size_lo; /* Size in bytes */ 699 __le32 i_atime; /* Access time */ 700 __le32 i_ctime; /* Inode Change time */ 701 __le32 i_mtime; /* Modification time */ 702 __le32 i_dtime; /* Deletion Time */ 703 __le16 i_gid; /* Low 16 bits of Group Id */ 704 __le16 i_links_count; /* Links count */ 705 __le32 i_blocks_lo; /* Blocks count */ 706 __le32 i_flags; /* File flags */ 707 union { 708 struct { 709 __le32 l_i_version; 710 } linux1; 711 struct { 712 __u32 h_i_translator; 713 } hurd1; 714 struct { 715 __u32 m_i_reserved1; 716 } masix1; 717 } osd1; /* OS dependent 1 */ 718 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 719 __le32 i_generation; /* File version (for NFS) */ 720 __le32 i_file_acl_lo; /* File ACL */ 721 __le32 i_size_high; 722 __le32 i_obso_faddr; /* Obsoleted fragment address */ 723 union { 724 struct { 725 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 726 __le16 l_i_file_acl_high; 727 __le16 l_i_uid_high; /* these 2 fields */ 728 __le16 l_i_gid_high; /* were reserved2[0] */ 729 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 730 __le16 l_i_reserved; 731 } linux2; 732 struct { 733 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 734 __u16 h_i_mode_high; 735 __u16 h_i_uid_high; 736 __u16 h_i_gid_high; 737 __u32 h_i_author; 738 } hurd2; 739 struct { 740 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 741 __le16 m_i_file_acl_high; 742 __u32 m_i_reserved2[2]; 743 } masix2; 744 } osd2; /* OS dependent 2 */ 745 __le16 i_extra_isize; 746 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 747 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 748 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 749 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 750 __le32 i_crtime; /* File Creation time */ 751 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 752 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 753 __le32 i_projid; /* Project ID */ 754 }; 755 756 struct move_extent { 757 __u32 reserved; /* should be zero */ 758 __u32 donor_fd; /* donor file descriptor */ 759 __u64 orig_start; /* logical start offset in block for orig */ 760 __u64 donor_start; /* logical start offset in block for donor */ 761 __u64 len; /* block length to be moved */ 762 __u64 moved_len; /* moved block length */ 763 }; 764 765 #define EXT4_EPOCH_BITS 2 766 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 767 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 768 769 /* 770 * Extended fields will fit into an inode if the filesystem was formatted 771 * with large inodes (-I 256 or larger) and there are not currently any EAs 772 * consuming all of the available space. For new inodes we always reserve 773 * enough space for the kernel's known extended fields, but for inodes 774 * created with an old kernel this might not have been the case. None of 775 * the extended inode fields is critical for correct filesystem operation. 776 * This macro checks if a certain field fits in the inode. Note that 777 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 778 */ 779 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 780 ((offsetof(typeof(*ext4_inode), field) + \ 781 sizeof((ext4_inode)->field)) \ 782 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 783 (einode)->i_extra_isize)) \ 784 785 /* 786 * We use an encoding that preserves the times for extra epoch "00": 787 * 788 * extra msb of adjust for signed 789 * epoch 32-bit 32-bit tv_sec to 790 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 791 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 792 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 793 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 794 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 795 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 796 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 797 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 798 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 799 * 800 * Note that previous versions of the kernel on 64-bit systems would 801 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 802 * 1970. e2fsck will correct this, assuming that it is run on the 803 * affected filesystem before 2242. 804 */ 805 806 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 807 { 808 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 809 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 810 } 811 812 static inline void ext4_decode_extra_time(struct timespec64 *time, 813 __le32 extra) 814 { 815 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) { 816 817 #if 1 818 /* Handle legacy encoding of pre-1970 dates with epoch 819 * bits 1,1. (This backwards compatibility may be removed 820 * at the discretion of the ext4 developers.) 821 */ 822 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 823 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 824 extra_bits = 0; 825 time->tv_sec += extra_bits << 32; 826 #else 827 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 828 #endif 829 } 830 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 831 } 832 833 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 834 do { \ 835 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 836 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 837 (raw_inode)->xtime ## _extra = \ 838 ext4_encode_extra_time(&(inode)->xtime); \ 839 } \ 840 } while (0) 841 842 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 843 do { \ 844 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 845 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 846 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 847 (raw_inode)->xtime ## _extra = \ 848 ext4_encode_extra_time(&(einode)->xtime); \ 849 } while (0) 850 851 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 852 do { \ 853 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 854 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 855 ext4_decode_extra_time(&(inode)->xtime, \ 856 raw_inode->xtime ## _extra); \ 857 } \ 858 else \ 859 (inode)->xtime.tv_nsec = 0; \ 860 } while (0) 861 862 863 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 864 do { \ 865 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 866 (einode)->xtime.tv_sec = \ 867 (signed)le32_to_cpu((raw_inode)->xtime); \ 868 else \ 869 (einode)->xtime.tv_sec = 0; \ 870 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 871 ext4_decode_extra_time(&(einode)->xtime, \ 872 raw_inode->xtime ## _extra); \ 873 else \ 874 (einode)->xtime.tv_nsec = 0; \ 875 } while (0) 876 877 #define i_disk_version osd1.linux1.l_i_version 878 879 #if defined(__KERNEL__) || defined(__linux__) 880 #define i_reserved1 osd1.linux1.l_i_reserved1 881 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 882 #define i_blocks_high osd2.linux2.l_i_blocks_high 883 #define i_uid_low i_uid 884 #define i_gid_low i_gid 885 #define i_uid_high osd2.linux2.l_i_uid_high 886 #define i_gid_high osd2.linux2.l_i_gid_high 887 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 888 889 #elif defined(__GNU__) 890 891 #define i_translator osd1.hurd1.h_i_translator 892 #define i_uid_high osd2.hurd2.h_i_uid_high 893 #define i_gid_high osd2.hurd2.h_i_gid_high 894 #define i_author osd2.hurd2.h_i_author 895 896 #elif defined(__masix__) 897 898 #define i_reserved1 osd1.masix1.m_i_reserved1 899 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 900 #define i_reserved2 osd2.masix2.m_i_reserved2 901 902 #endif /* defined(__KERNEL__) || defined(__linux__) */ 903 904 #include "extents_status.h" 905 906 /* 907 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 908 * 909 * These are needed to avoid lockdep false positives when we need to 910 * allocate blocks to the quota inode during ext4_map_blocks(), while 911 * holding i_data_sem for a normal (non-quota) inode. Since we don't 912 * do quota tracking for the quota inode, this avoids deadlock (as 913 * well as infinite recursion, since it isn't turtles all the way 914 * down...) 915 * 916 * I_DATA_SEM_NORMAL - Used for most inodes 917 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 918 * where the second inode has larger inode number 919 * than the first 920 * I_DATA_SEM_QUOTA - Used for quota inodes only 921 */ 922 enum { 923 I_DATA_SEM_NORMAL = 0, 924 I_DATA_SEM_OTHER, 925 I_DATA_SEM_QUOTA, 926 }; 927 928 929 /* 930 * fourth extended file system inode data in memory 931 */ 932 struct ext4_inode_info { 933 __le32 i_data[15]; /* unconverted */ 934 __u32 i_dtime; 935 ext4_fsblk_t i_file_acl; 936 937 /* 938 * i_block_group is the number of the block group which contains 939 * this file's inode. Constant across the lifetime of the inode, 940 * it is used for making block allocation decisions - we try to 941 * place a file's data blocks near its inode block, and new inodes 942 * near to their parent directory's inode. 943 */ 944 ext4_group_t i_block_group; 945 ext4_lblk_t i_dir_start_lookup; 946 #if (BITS_PER_LONG < 64) 947 unsigned long i_state_flags; /* Dynamic state flags */ 948 #endif 949 unsigned long i_flags; 950 951 /* 952 * Extended attributes can be read independently of the main file 953 * data. Taking i_mutex even when reading would cause contention 954 * between readers of EAs and writers of regular file data, so 955 * instead we synchronize on xattr_sem when reading or changing 956 * EAs. 957 */ 958 struct rw_semaphore xattr_sem; 959 960 struct list_head i_orphan; /* unlinked but open inodes */ 961 962 /* 963 * i_disksize keeps track of what the inode size is ON DISK, not 964 * in memory. During truncate, i_size is set to the new size by 965 * the VFS prior to calling ext4_truncate(), but the filesystem won't 966 * set i_disksize to 0 until the truncate is actually under way. 967 * 968 * The intent is that i_disksize always represents the blocks which 969 * are used by this file. This allows recovery to restart truncate 970 * on orphans if we crash during truncate. We actually write i_disksize 971 * into the on-disk inode when writing inodes out, instead of i_size. 972 * 973 * The only time when i_disksize and i_size may be different is when 974 * a truncate is in progress. The only things which change i_disksize 975 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 976 */ 977 loff_t i_disksize; 978 979 /* 980 * i_data_sem is for serialising ext4_truncate() against 981 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 982 * data tree are chopped off during truncate. We can't do that in 983 * ext4 because whenever we perform intermediate commits during 984 * truncate, the inode and all the metadata blocks *must* be in a 985 * consistent state which allows truncation of the orphans to restart 986 * during recovery. Hence we must fix the get_block-vs-truncate race 987 * by other means, so we have i_data_sem. 988 */ 989 struct rw_semaphore i_data_sem; 990 /* 991 * i_mmap_sem is for serializing page faults with truncate / punch hole 992 * operations. We have to make sure that new page cannot be faulted in 993 * a section of the inode that is being punched. We cannot easily use 994 * i_data_sem for this since we need protection for the whole punch 995 * operation and i_data_sem ranks below transaction start so we have 996 * to occasionally drop it. 997 */ 998 struct rw_semaphore i_mmap_sem; 999 struct inode vfs_inode; 1000 struct jbd2_inode *jinode; 1001 1002 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1003 1004 /* 1005 * File creation time. Its function is same as that of 1006 * struct timespec64 i_{a,c,m}time in the generic inode. 1007 */ 1008 struct timespec64 i_crtime; 1009 1010 /* mballoc */ 1011 struct list_head i_prealloc_list; 1012 spinlock_t i_prealloc_lock; 1013 1014 /* extents status tree */ 1015 struct ext4_es_tree i_es_tree; 1016 rwlock_t i_es_lock; 1017 struct list_head i_es_list; 1018 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1019 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1020 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1021 extents to shrink. Protected by 1022 i_es_lock */ 1023 1024 /* ialloc */ 1025 ext4_group_t i_last_alloc_group; 1026 1027 /* allocation reservation info for delalloc */ 1028 /* In case of bigalloc, this refer to clusters rather than blocks */ 1029 unsigned int i_reserved_data_blocks; 1030 ext4_lblk_t i_da_metadata_calc_last_lblock; 1031 int i_da_metadata_calc_len; 1032 1033 /* pending cluster reservations for bigalloc file systems */ 1034 struct ext4_pending_tree i_pending_tree; 1035 1036 /* on-disk additional length */ 1037 __u16 i_extra_isize; 1038 1039 /* Indicate the inline data space. */ 1040 u16 i_inline_off; 1041 u16 i_inline_size; 1042 1043 #ifdef CONFIG_QUOTA 1044 /* quota space reservation, managed internally by quota code */ 1045 qsize_t i_reserved_quota; 1046 #endif 1047 1048 /* Lock protecting lists below */ 1049 spinlock_t i_completed_io_lock; 1050 /* 1051 * Completed IOs that need unwritten extents handling and have 1052 * transaction reserved 1053 */ 1054 struct list_head i_rsv_conversion_list; 1055 struct work_struct i_rsv_conversion_work; 1056 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1057 1058 spinlock_t i_block_reservation_lock; 1059 1060 /* 1061 * Transactions that contain inode's metadata needed to complete 1062 * fsync and fdatasync, respectively. 1063 */ 1064 tid_t i_sync_tid; 1065 tid_t i_datasync_tid; 1066 1067 #ifdef CONFIG_QUOTA 1068 struct dquot *i_dquot[MAXQUOTAS]; 1069 #endif 1070 1071 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1072 __u32 i_csum_seed; 1073 1074 kprojid_t i_projid; 1075 }; 1076 1077 /* 1078 * File system states 1079 */ 1080 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1081 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1082 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1083 1084 /* 1085 * Misc. filesystem flags 1086 */ 1087 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1088 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1089 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1090 1091 /* 1092 * Mount flags set via mount options or defaults 1093 */ 1094 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1095 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1096 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1097 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1098 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1099 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1100 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1101 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1102 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1103 #ifdef CONFIG_FS_DAX 1104 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1105 #else 1106 #define EXT4_MOUNT_DAX 0 1107 #endif 1108 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1109 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1110 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1111 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1112 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1113 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1114 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1115 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1116 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1117 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1118 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1119 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1120 * enable enforcement for hidden 1121 * quota files */ 1122 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1123 * enforcement for hidden quota 1124 * files */ 1125 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1126 * enforcement */ 1127 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1128 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1129 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1130 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1131 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1132 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1133 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1134 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1135 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1136 1137 /* 1138 * Mount flags set either automatically (could not be set by mount option) 1139 * based on per file system feature or property or in special cases such as 1140 * distinguishing between explicit mount option definition and default. 1141 */ 1142 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1143 specified delalloc */ 1144 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1145 size of blocksize * 8 1146 blocks */ 1147 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1148 file systems */ 1149 1150 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1151 specified journal checksum */ 1152 1153 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1154 ~EXT4_MOUNT_##opt 1155 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1156 EXT4_MOUNT_##opt 1157 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1158 EXT4_MOUNT_##opt) 1159 1160 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1161 ~EXT4_MOUNT2_##opt 1162 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1163 EXT4_MOUNT2_##opt 1164 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1165 EXT4_MOUNT2_##opt) 1166 1167 #define ext4_test_and_set_bit __test_and_set_bit_le 1168 #define ext4_set_bit __set_bit_le 1169 #define ext4_set_bit_atomic ext2_set_bit_atomic 1170 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1171 #define ext4_clear_bit __clear_bit_le 1172 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1173 #define ext4_test_bit test_bit_le 1174 #define ext4_find_next_zero_bit find_next_zero_bit_le 1175 #define ext4_find_next_bit find_next_bit_le 1176 1177 extern void ext4_set_bits(void *bm, int cur, int len); 1178 1179 /* 1180 * Maximal mount counts between two filesystem checks 1181 */ 1182 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1183 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1184 1185 /* 1186 * Behaviour when detecting errors 1187 */ 1188 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1189 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1190 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1191 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1192 1193 /* Metadata checksum algorithm codes */ 1194 #define EXT4_CRC32C_CHKSUM 1 1195 1196 /* 1197 * Structure of the super block 1198 */ 1199 struct ext4_super_block { 1200 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1201 __le32 s_blocks_count_lo; /* Blocks count */ 1202 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1203 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1204 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1205 __le32 s_first_data_block; /* First Data Block */ 1206 __le32 s_log_block_size; /* Block size */ 1207 __le32 s_log_cluster_size; /* Allocation cluster size */ 1208 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1209 __le32 s_clusters_per_group; /* # Clusters per group */ 1210 __le32 s_inodes_per_group; /* # Inodes per group */ 1211 __le32 s_mtime; /* Mount time */ 1212 /*30*/ __le32 s_wtime; /* Write time */ 1213 __le16 s_mnt_count; /* Mount count */ 1214 __le16 s_max_mnt_count; /* Maximal mount count */ 1215 __le16 s_magic; /* Magic signature */ 1216 __le16 s_state; /* File system state */ 1217 __le16 s_errors; /* Behaviour when detecting errors */ 1218 __le16 s_minor_rev_level; /* minor revision level */ 1219 /*40*/ __le32 s_lastcheck; /* time of last check */ 1220 __le32 s_checkinterval; /* max. time between checks */ 1221 __le32 s_creator_os; /* OS */ 1222 __le32 s_rev_level; /* Revision level */ 1223 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1224 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1225 /* 1226 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1227 * 1228 * Note: the difference between the compatible feature set and 1229 * the incompatible feature set is that if there is a bit set 1230 * in the incompatible feature set that the kernel doesn't 1231 * know about, it should refuse to mount the filesystem. 1232 * 1233 * e2fsck's requirements are more strict; if it doesn't know 1234 * about a feature in either the compatible or incompatible 1235 * feature set, it must abort and not try to meddle with 1236 * things it doesn't understand... 1237 */ 1238 __le32 s_first_ino; /* First non-reserved inode */ 1239 __le16 s_inode_size; /* size of inode structure */ 1240 __le16 s_block_group_nr; /* block group # of this superblock */ 1241 __le32 s_feature_compat; /* compatible feature set */ 1242 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1243 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1244 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1245 /*78*/ char s_volume_name[16]; /* volume name */ 1246 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1247 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1248 /* 1249 * Performance hints. Directory preallocation should only 1250 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1251 */ 1252 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1253 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1254 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1255 /* 1256 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1257 */ 1258 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1259 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1260 __le32 s_journal_dev; /* device number of journal file */ 1261 __le32 s_last_orphan; /* start of list of inodes to delete */ 1262 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1263 __u8 s_def_hash_version; /* Default hash version to use */ 1264 __u8 s_jnl_backup_type; 1265 __le16 s_desc_size; /* size of group descriptor */ 1266 /*100*/ __le32 s_default_mount_opts; 1267 __le32 s_first_meta_bg; /* First metablock block group */ 1268 __le32 s_mkfs_time; /* When the filesystem was created */ 1269 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1270 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1271 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1272 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1273 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1274 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1275 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1276 __le32 s_flags; /* Miscellaneous flags */ 1277 __le16 s_raid_stride; /* RAID stride */ 1278 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1279 __le64 s_mmp_block; /* Block for multi-mount protection */ 1280 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1281 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1282 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1283 __u8 s_encryption_level; /* versioning level for encryption */ 1284 __u8 s_reserved_pad; /* Padding to next 32bits */ 1285 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1286 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1287 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1288 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1289 snapshot's future use */ 1290 __le32 s_snapshot_list; /* inode number of the head of the 1291 on-disk snapshot list */ 1292 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1293 __le32 s_error_count; /* number of fs errors */ 1294 __le32 s_first_error_time; /* first time an error happened */ 1295 __le32 s_first_error_ino; /* inode involved in first error */ 1296 __le64 s_first_error_block; /* block involved of first error */ 1297 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1298 __le32 s_first_error_line; /* line number where error happened */ 1299 __le32 s_last_error_time; /* most recent time of an error */ 1300 __le32 s_last_error_ino; /* inode involved in last error */ 1301 __le32 s_last_error_line; /* line number where error happened */ 1302 __le64 s_last_error_block; /* block involved of last error */ 1303 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1304 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1305 __u8 s_mount_opts[64]; 1306 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1307 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1308 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1309 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1310 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1311 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1312 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1313 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1314 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1315 __u8 s_wtime_hi; 1316 __u8 s_mtime_hi; 1317 __u8 s_mkfs_time_hi; 1318 __u8 s_lastcheck_hi; 1319 __u8 s_first_error_time_hi; 1320 __u8 s_last_error_time_hi; 1321 __u8 s_pad[2]; 1322 __le16 s_encoding; /* Filename charset encoding */ 1323 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1324 __le32 s_reserved[95]; /* Padding to the end of the block */ 1325 __le32 s_checksum; /* crc32c(superblock) */ 1326 }; 1327 1328 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1329 1330 #ifdef __KERNEL__ 1331 1332 /* 1333 * run-time mount flags 1334 */ 1335 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1336 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1337 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1338 1339 #ifdef CONFIG_FS_ENCRYPTION 1340 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1341 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1342 #else 1343 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1344 #endif 1345 1346 /* Number of quota types we support */ 1347 #define EXT4_MAXQUOTAS 3 1348 1349 #define EXT4_ENC_UTF8_12_1 1 1350 1351 /* 1352 * Flags for ext4_sb_info.s_encoding_flags. 1353 */ 1354 #define EXT4_ENC_STRICT_MODE_FL (1 << 0) 1355 1356 #define ext4_has_strict_mode(sbi) \ 1357 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL) 1358 1359 /* 1360 * fourth extended-fs super-block data in memory 1361 */ 1362 struct ext4_sb_info { 1363 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1364 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1365 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1366 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1367 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1368 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1369 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1370 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1371 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1372 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1373 unsigned long s_overhead; /* # of fs overhead clusters */ 1374 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1375 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1376 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1377 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1378 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1379 struct buffer_head **s_group_desc; 1380 unsigned int s_mount_opt; 1381 unsigned int s_mount_opt2; 1382 unsigned int s_mount_flags; 1383 unsigned int s_def_mount_opt; 1384 ext4_fsblk_t s_sb_block; 1385 atomic64_t s_resv_clusters; 1386 kuid_t s_resuid; 1387 kgid_t s_resgid; 1388 unsigned short s_mount_state; 1389 unsigned short s_pad; 1390 int s_addr_per_block_bits; 1391 int s_desc_per_block_bits; 1392 int s_inode_size; 1393 int s_first_ino; 1394 unsigned int s_inode_readahead_blks; 1395 unsigned int s_inode_goal; 1396 u32 s_hash_seed[4]; 1397 int s_def_hash_version; 1398 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1399 struct percpu_counter s_freeclusters_counter; 1400 struct percpu_counter s_freeinodes_counter; 1401 struct percpu_counter s_dirs_counter; 1402 struct percpu_counter s_dirtyclusters_counter; 1403 struct blockgroup_lock *s_blockgroup_lock; 1404 struct proc_dir_entry *s_proc; 1405 struct kobject s_kobj; 1406 struct completion s_kobj_unregister; 1407 struct super_block *s_sb; 1408 #ifdef CONFIG_UNICODE 1409 struct unicode_map *s_encoding; 1410 __u16 s_encoding_flags; 1411 #endif 1412 1413 /* Journaling */ 1414 struct journal_s *s_journal; 1415 struct list_head s_orphan; 1416 struct mutex s_orphan_lock; 1417 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1418 unsigned long s_commit_interval; 1419 u32 s_max_batch_time; 1420 u32 s_min_batch_time; 1421 struct block_device *journal_bdev; 1422 #ifdef CONFIG_QUOTA 1423 /* Names of quota files with journalled quota */ 1424 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1425 int s_jquota_fmt; /* Format of quota to use */ 1426 #endif 1427 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1428 struct rb_root system_blks; 1429 1430 #ifdef EXTENTS_STATS 1431 /* ext4 extents stats */ 1432 unsigned long s_ext_min; 1433 unsigned long s_ext_max; 1434 unsigned long s_depth_max; 1435 spinlock_t s_ext_stats_lock; 1436 unsigned long s_ext_blocks; 1437 unsigned long s_ext_extents; 1438 #endif 1439 1440 /* for buddy allocator */ 1441 struct ext4_group_info ***s_group_info; 1442 struct inode *s_buddy_cache; 1443 spinlock_t s_md_lock; 1444 unsigned short *s_mb_offsets; 1445 unsigned int *s_mb_maxs; 1446 unsigned int s_group_info_size; 1447 unsigned int s_mb_free_pending; 1448 struct list_head s_freed_data_list; /* List of blocks to be freed 1449 after commit completed */ 1450 1451 /* tunables */ 1452 unsigned long s_stripe; 1453 unsigned int s_mb_stream_request; 1454 unsigned int s_mb_max_to_scan; 1455 unsigned int s_mb_min_to_scan; 1456 unsigned int s_mb_stats; 1457 unsigned int s_mb_order2_reqs; 1458 unsigned int s_mb_group_prealloc; 1459 unsigned int s_max_dir_size_kb; 1460 /* where last allocation was done - for stream allocation */ 1461 unsigned long s_mb_last_group; 1462 unsigned long s_mb_last_start; 1463 1464 /* stats for buddy allocator */ 1465 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1466 atomic_t s_bal_success; /* we found long enough chunks */ 1467 atomic_t s_bal_allocated; /* in blocks */ 1468 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1469 atomic_t s_bal_goals; /* goal hits */ 1470 atomic_t s_bal_breaks; /* too long searches */ 1471 atomic_t s_bal_2orders; /* 2^order hits */ 1472 spinlock_t s_bal_lock; 1473 unsigned long s_mb_buddies_generated; 1474 unsigned long long s_mb_generation_time; 1475 atomic_t s_mb_lost_chunks; 1476 atomic_t s_mb_preallocated; 1477 atomic_t s_mb_discarded; 1478 atomic_t s_lock_busy; 1479 1480 /* locality groups */ 1481 struct ext4_locality_group __percpu *s_locality_groups; 1482 1483 /* for write statistics */ 1484 unsigned long s_sectors_written_start; 1485 u64 s_kbytes_written; 1486 1487 /* the size of zero-out chunk */ 1488 unsigned int s_extent_max_zeroout_kb; 1489 1490 unsigned int s_log_groups_per_flex; 1491 struct flex_groups *s_flex_groups; 1492 ext4_group_t s_flex_groups_allocated; 1493 1494 /* workqueue for reserved extent conversions (buffered io) */ 1495 struct workqueue_struct *rsv_conversion_wq; 1496 1497 /* timer for periodic error stats printing */ 1498 struct timer_list s_err_report; 1499 1500 /* Lazy inode table initialization info */ 1501 struct ext4_li_request *s_li_request; 1502 /* Wait multiplier for lazy initialization thread */ 1503 unsigned int s_li_wait_mult; 1504 1505 /* Kernel thread for multiple mount protection */ 1506 struct task_struct *s_mmp_tsk; 1507 1508 /* record the last minlen when FITRIM is called. */ 1509 atomic_t s_last_trim_minblks; 1510 1511 /* Reference to checksum algorithm driver via cryptoapi */ 1512 struct crypto_shash *s_chksum_driver; 1513 1514 /* Precomputed FS UUID checksum for seeding other checksums */ 1515 __u32 s_csum_seed; 1516 1517 /* Reclaim extents from extent status tree */ 1518 struct shrinker s_es_shrinker; 1519 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1520 long s_es_nr_inode; 1521 struct ext4_es_stats s_es_stats; 1522 struct mb_cache *s_ea_block_cache; 1523 struct mb_cache *s_ea_inode_cache; 1524 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1525 1526 /* Ratelimit ext4 messages. */ 1527 struct ratelimit_state s_err_ratelimit_state; 1528 struct ratelimit_state s_warning_ratelimit_state; 1529 struct ratelimit_state s_msg_ratelimit_state; 1530 1531 /* Barrier between changing inodes' journal flags and writepages ops. */ 1532 struct percpu_rw_semaphore s_journal_flag_rwsem; 1533 struct dax_device *s_daxdev; 1534 }; 1535 1536 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1537 { 1538 return sb->s_fs_info; 1539 } 1540 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1541 { 1542 return container_of(inode, struct ext4_inode_info, vfs_inode); 1543 } 1544 1545 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1546 { 1547 return ino == EXT4_ROOT_INO || 1548 (ino >= EXT4_FIRST_INO(sb) && 1549 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1550 } 1551 1552 /* 1553 * Inode dynamic state flags 1554 */ 1555 enum { 1556 EXT4_STATE_JDATA, /* journaled data exists */ 1557 EXT4_STATE_NEW, /* inode is newly created */ 1558 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1559 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1560 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1561 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1562 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1563 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1564 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1565 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1566 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1567 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1568 }; 1569 1570 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1571 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1572 { \ 1573 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1574 } \ 1575 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1576 { \ 1577 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1578 } \ 1579 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1580 { \ 1581 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1582 } 1583 1584 /* Add these declarations here only so that these functions can be 1585 * found by name. Otherwise, they are very hard to locate. */ 1586 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1587 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1588 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1589 EXT4_INODE_BIT_FNS(flag, flags, 0) 1590 1591 /* Add these declarations here only so that these functions can be 1592 * found by name. Otherwise, they are very hard to locate. */ 1593 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1594 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1595 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1596 #if (BITS_PER_LONG < 64) 1597 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1598 1599 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1600 { 1601 (ei)->i_state_flags = 0; 1602 } 1603 #else 1604 EXT4_INODE_BIT_FNS(state, flags, 32) 1605 1606 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1607 { 1608 /* We depend on the fact that callers will set i_flags */ 1609 } 1610 #endif 1611 #else 1612 /* Assume that user mode programs are passing in an ext4fs superblock, not 1613 * a kernel struct super_block. This will allow us to call the feature-test 1614 * macros from user land. */ 1615 #define EXT4_SB(sb) (sb) 1616 #endif 1617 1618 static inline bool ext4_verity_in_progress(struct inode *inode) 1619 { 1620 return IS_ENABLED(CONFIG_FS_VERITY) && 1621 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1622 } 1623 1624 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1625 1626 /* 1627 * Codes for operating systems 1628 */ 1629 #define EXT4_OS_LINUX 0 1630 #define EXT4_OS_HURD 1 1631 #define EXT4_OS_MASIX 2 1632 #define EXT4_OS_FREEBSD 3 1633 #define EXT4_OS_LITES 4 1634 1635 /* 1636 * Revision levels 1637 */ 1638 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1639 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1640 1641 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1642 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1643 1644 #define EXT4_GOOD_OLD_INODE_SIZE 128 1645 1646 /* 1647 * Feature set definitions 1648 */ 1649 1650 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1651 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1652 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1653 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1654 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1655 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1656 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1657 1658 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1659 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1660 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1661 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1662 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1663 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1664 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1665 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1666 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1667 /* 1668 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1669 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1670 * all other data structures' checksums. However, the METADATA_CSUM and 1671 * GDT_CSUM bits are mutually exclusive. 1672 */ 1673 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1674 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1675 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1676 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1677 1678 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1679 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1680 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1681 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1682 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1683 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1684 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1685 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1686 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1687 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1688 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1689 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1690 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1691 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1692 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1693 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1694 1695 extern void ext4_update_dynamic_rev(struct super_block *sb); 1696 1697 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1698 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1699 { \ 1700 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1701 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1702 } \ 1703 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1704 { \ 1705 ext4_update_dynamic_rev(sb); \ 1706 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1707 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1708 } \ 1709 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1710 { \ 1711 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1712 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1713 } 1714 1715 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1716 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1717 { \ 1718 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1719 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1720 } \ 1721 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1722 { \ 1723 ext4_update_dynamic_rev(sb); \ 1724 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1725 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1726 } \ 1727 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1728 { \ 1729 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1730 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1731 } 1732 1733 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1734 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1735 { \ 1736 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1737 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1738 } \ 1739 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1740 { \ 1741 ext4_update_dynamic_rev(sb); \ 1742 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1743 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1744 } \ 1745 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1746 { \ 1747 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1748 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1749 } 1750 1751 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1752 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1753 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1754 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1755 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1756 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1757 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1758 1759 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1760 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1761 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1762 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1763 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1764 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1765 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1766 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1767 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1768 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1769 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1770 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1771 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 1772 1773 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1774 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1775 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1776 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1777 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1778 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1779 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1780 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1781 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1782 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1783 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1784 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1785 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1786 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1787 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1788 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 1789 1790 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1791 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1792 EXT4_FEATURE_INCOMPAT_META_BG) 1793 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1794 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1795 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1796 1797 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1798 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1799 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1800 EXT4_FEATURE_INCOMPAT_META_BG) 1801 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1802 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1803 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1804 1805 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1806 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1807 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1808 EXT4_FEATURE_INCOMPAT_META_BG| \ 1809 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1810 EXT4_FEATURE_INCOMPAT_64BIT| \ 1811 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1812 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1813 EXT4_FEATURE_INCOMPAT_MMP | \ 1814 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1815 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1816 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 1817 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1818 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1819 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1820 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1821 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1822 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1823 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1824 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1825 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1826 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1827 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1828 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1829 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 1830 EXT4_FEATURE_RO_COMPAT_VERITY) 1831 1832 #define EXTN_FEATURE_FUNCS(ver) \ 1833 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1834 { \ 1835 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1836 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1837 } \ 1838 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1839 { \ 1840 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1841 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1842 } \ 1843 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1844 { \ 1845 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1846 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1847 } 1848 1849 EXTN_FEATURE_FUNCS(2) 1850 EXTN_FEATURE_FUNCS(3) 1851 EXTN_FEATURE_FUNCS(4) 1852 1853 static inline bool ext4_has_compat_features(struct super_block *sb) 1854 { 1855 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1856 } 1857 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1858 { 1859 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1860 } 1861 static inline bool ext4_has_incompat_features(struct super_block *sb) 1862 { 1863 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1864 } 1865 1866 /* 1867 * Superblock flags 1868 */ 1869 #define EXT4_FLAGS_RESIZING 0 1870 #define EXT4_FLAGS_SHUTDOWN 1 1871 1872 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1873 { 1874 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1875 } 1876 1877 1878 /* 1879 * Default values for user and/or group using reserved blocks 1880 */ 1881 #define EXT4_DEF_RESUID 0 1882 #define EXT4_DEF_RESGID 0 1883 1884 /* 1885 * Default project ID 1886 */ 1887 #define EXT4_DEF_PROJID 0 1888 1889 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1890 1891 /* 1892 * Default mount options 1893 */ 1894 #define EXT4_DEFM_DEBUG 0x0001 1895 #define EXT4_DEFM_BSDGROUPS 0x0002 1896 #define EXT4_DEFM_XATTR_USER 0x0004 1897 #define EXT4_DEFM_ACL 0x0008 1898 #define EXT4_DEFM_UID16 0x0010 1899 #define EXT4_DEFM_JMODE 0x0060 1900 #define EXT4_DEFM_JMODE_DATA 0x0020 1901 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1902 #define EXT4_DEFM_JMODE_WBACK 0x0060 1903 #define EXT4_DEFM_NOBARRIER 0x0100 1904 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1905 #define EXT4_DEFM_DISCARD 0x0400 1906 #define EXT4_DEFM_NODELALLOC 0x0800 1907 1908 /* 1909 * Default journal batch times 1910 */ 1911 #define EXT4_DEF_MIN_BATCH_TIME 0 1912 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1913 1914 /* 1915 * Minimum number of groups in a flexgroup before we separate out 1916 * directories into the first block group of a flexgroup 1917 */ 1918 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1919 1920 /* 1921 * Structure of a directory entry 1922 */ 1923 #define EXT4_NAME_LEN 255 1924 1925 struct ext4_dir_entry { 1926 __le32 inode; /* Inode number */ 1927 __le16 rec_len; /* Directory entry length */ 1928 __le16 name_len; /* Name length */ 1929 char name[EXT4_NAME_LEN]; /* File name */ 1930 }; 1931 1932 /* 1933 * The new version of the directory entry. Since EXT4 structures are 1934 * stored in intel byte order, and the name_len field could never be 1935 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1936 * file_type field. 1937 */ 1938 struct ext4_dir_entry_2 { 1939 __le32 inode; /* Inode number */ 1940 __le16 rec_len; /* Directory entry length */ 1941 __u8 name_len; /* Name length */ 1942 __u8 file_type; 1943 char name[EXT4_NAME_LEN]; /* File name */ 1944 }; 1945 1946 /* 1947 * This is a bogus directory entry at the end of each leaf block that 1948 * records checksums. 1949 */ 1950 struct ext4_dir_entry_tail { 1951 __le32 det_reserved_zero1; /* Pretend to be unused */ 1952 __le16 det_rec_len; /* 12 */ 1953 __u8 det_reserved_zero2; /* Zero name length */ 1954 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1955 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1956 }; 1957 1958 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1959 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1960 ((blocksize) - \ 1961 sizeof(struct ext4_dir_entry_tail)))) 1962 1963 /* 1964 * Ext4 directory file types. Only the low 3 bits are used. The 1965 * other bits are reserved for now. 1966 */ 1967 #define EXT4_FT_UNKNOWN 0 1968 #define EXT4_FT_REG_FILE 1 1969 #define EXT4_FT_DIR 2 1970 #define EXT4_FT_CHRDEV 3 1971 #define EXT4_FT_BLKDEV 4 1972 #define EXT4_FT_FIFO 5 1973 #define EXT4_FT_SOCK 6 1974 #define EXT4_FT_SYMLINK 7 1975 1976 #define EXT4_FT_MAX 8 1977 1978 #define EXT4_FT_DIR_CSUM 0xDE 1979 1980 /* 1981 * EXT4_DIR_PAD defines the directory entries boundaries 1982 * 1983 * NOTE: It must be a multiple of 4 1984 */ 1985 #define EXT4_DIR_PAD 4 1986 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1987 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1988 ~EXT4_DIR_ROUND) 1989 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1990 1991 /* 1992 * If we ever get support for fs block sizes > page_size, we'll need 1993 * to remove the #if statements in the next two functions... 1994 */ 1995 static inline unsigned int 1996 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1997 { 1998 unsigned len = le16_to_cpu(dlen); 1999 2000 #if (PAGE_SIZE >= 65536) 2001 if (len == EXT4_MAX_REC_LEN || len == 0) 2002 return blocksize; 2003 return (len & 65532) | ((len & 3) << 16); 2004 #else 2005 return len; 2006 #endif 2007 } 2008 2009 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2010 { 2011 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2012 BUG(); 2013 #if (PAGE_SIZE >= 65536) 2014 if (len < 65536) 2015 return cpu_to_le16(len); 2016 if (len == blocksize) { 2017 if (blocksize == 65536) 2018 return cpu_to_le16(EXT4_MAX_REC_LEN); 2019 else 2020 return cpu_to_le16(0); 2021 } 2022 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2023 #else 2024 return cpu_to_le16(len); 2025 #endif 2026 } 2027 2028 /* 2029 * Hash Tree Directory indexing 2030 * (c) Daniel Phillips, 2001 2031 */ 2032 2033 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2034 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2035 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2036 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2037 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2038 2039 /* Legal values for the dx_root hash_version field: */ 2040 2041 #define DX_HASH_LEGACY 0 2042 #define DX_HASH_HALF_MD4 1 2043 #define DX_HASH_TEA 2 2044 #define DX_HASH_LEGACY_UNSIGNED 3 2045 #define DX_HASH_HALF_MD4_UNSIGNED 4 2046 #define DX_HASH_TEA_UNSIGNED 5 2047 2048 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2049 const void *address, unsigned int length) 2050 { 2051 struct { 2052 struct shash_desc shash; 2053 char ctx[4]; 2054 } desc; 2055 2056 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2057 2058 desc.shash.tfm = sbi->s_chksum_driver; 2059 *(u32 *)desc.ctx = crc; 2060 2061 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2062 2063 return *(u32 *)desc.ctx; 2064 } 2065 2066 #ifdef __KERNEL__ 2067 2068 /* hash info structure used by the directory hash */ 2069 struct dx_hash_info 2070 { 2071 u32 hash; 2072 u32 minor_hash; 2073 int hash_version; 2074 u32 *seed; 2075 }; 2076 2077 2078 /* 32 and 64 bit signed EOF for dx directories */ 2079 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2080 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2081 2082 2083 /* 2084 * Control parameters used by ext4_htree_next_block 2085 */ 2086 #define HASH_NB_ALWAYS 1 2087 2088 struct ext4_filename { 2089 const struct qstr *usr_fname; 2090 struct fscrypt_str disk_name; 2091 struct dx_hash_info hinfo; 2092 #ifdef CONFIG_FS_ENCRYPTION 2093 struct fscrypt_str crypto_buf; 2094 #endif 2095 #ifdef CONFIG_UNICODE 2096 struct fscrypt_str cf_name; 2097 #endif 2098 }; 2099 2100 #define fname_name(p) ((p)->disk_name.name) 2101 #define fname_len(p) ((p)->disk_name.len) 2102 2103 /* 2104 * Describe an inode's exact location on disk and in memory 2105 */ 2106 struct ext4_iloc 2107 { 2108 struct buffer_head *bh; 2109 unsigned long offset; 2110 ext4_group_t block_group; 2111 }; 2112 2113 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2114 { 2115 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2116 } 2117 2118 static inline bool ext4_is_quota_file(struct inode *inode) 2119 { 2120 return IS_NOQUOTA(inode) && 2121 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2122 } 2123 2124 /* 2125 * This structure is stuffed into the struct file's private_data field 2126 * for directories. It is where we put information so that we can do 2127 * readdir operations in hash tree order. 2128 */ 2129 struct dir_private_info { 2130 struct rb_root root; 2131 struct rb_node *curr_node; 2132 struct fname *extra_fname; 2133 loff_t last_pos; 2134 __u32 curr_hash; 2135 __u32 curr_minor_hash; 2136 __u32 next_hash; 2137 }; 2138 2139 /* calculate the first block number of the group */ 2140 static inline ext4_fsblk_t 2141 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2142 { 2143 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2144 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2145 } 2146 2147 /* 2148 * Special error return code only used by dx_probe() and its callers. 2149 */ 2150 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2151 2152 /* htree levels for ext4 */ 2153 #define EXT4_HTREE_LEVEL_COMPAT 2 2154 #define EXT4_HTREE_LEVEL 3 2155 2156 static inline int ext4_dir_htree_level(struct super_block *sb) 2157 { 2158 return ext4_has_feature_largedir(sb) ? 2159 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2160 } 2161 2162 /* 2163 * Timeout and state flag for lazy initialization inode thread. 2164 */ 2165 #define EXT4_DEF_LI_WAIT_MULT 10 2166 #define EXT4_DEF_LI_MAX_START_DELAY 5 2167 #define EXT4_LAZYINIT_QUIT 0x0001 2168 #define EXT4_LAZYINIT_RUNNING 0x0002 2169 2170 /* 2171 * Lazy inode table initialization info 2172 */ 2173 struct ext4_lazy_init { 2174 unsigned long li_state; 2175 struct list_head li_request_list; 2176 struct mutex li_list_mtx; 2177 }; 2178 2179 struct ext4_li_request { 2180 struct super_block *lr_super; 2181 struct ext4_sb_info *lr_sbi; 2182 ext4_group_t lr_next_group; 2183 struct list_head lr_request; 2184 unsigned long lr_next_sched; 2185 unsigned long lr_timeout; 2186 }; 2187 2188 struct ext4_features { 2189 struct kobject f_kobj; 2190 struct completion f_kobj_unregister; 2191 }; 2192 2193 /* 2194 * This structure will be used for multiple mount protection. It will be 2195 * written into the block number saved in the s_mmp_block field in the 2196 * superblock. Programs that check MMP should assume that if 2197 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2198 * to use the filesystem, regardless of how old the timestamp is. 2199 */ 2200 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2201 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2202 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2203 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2204 2205 struct mmp_struct { 2206 __le32 mmp_magic; /* Magic number for MMP */ 2207 __le32 mmp_seq; /* Sequence no. updated periodically */ 2208 2209 /* 2210 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2211 * purposes and do not affect the correctness of the algorithm 2212 */ 2213 __le64 mmp_time; /* Time last updated */ 2214 char mmp_nodename[64]; /* Node which last updated MMP block */ 2215 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2216 2217 /* 2218 * mmp_check_interval is used to verify if the MMP block has been 2219 * updated on the block device. The value is updated based on the 2220 * maximum time to write the MMP block during an update cycle. 2221 */ 2222 __le16 mmp_check_interval; 2223 2224 __le16 mmp_pad1; 2225 __le32 mmp_pad2[226]; 2226 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2227 }; 2228 2229 /* arguments passed to the mmp thread */ 2230 struct mmpd_data { 2231 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2232 struct super_block *sb; /* super block of the fs */ 2233 }; 2234 2235 /* 2236 * Check interval multiplier 2237 * The MMP block is written every update interval and initially checked every 2238 * update interval x the multiplier (the value is then adapted based on the 2239 * write latency). The reason is that writes can be delayed under load and we 2240 * don't want readers to incorrectly assume that the filesystem is no longer 2241 * in use. 2242 */ 2243 #define EXT4_MMP_CHECK_MULT 2UL 2244 2245 /* 2246 * Minimum interval for MMP checking in seconds. 2247 */ 2248 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2249 2250 /* 2251 * Maximum interval for MMP checking in seconds. 2252 */ 2253 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2254 2255 /* 2256 * Function prototypes 2257 */ 2258 2259 /* 2260 * Ok, these declarations are also in <linux/kernel.h> but none of the 2261 * ext4 source programs needs to include it so they are duplicated here. 2262 */ 2263 # define NORET_TYPE /**/ 2264 # define ATTRIB_NORET __attribute__((noreturn)) 2265 # define NORET_AND noreturn, 2266 2267 /* bitmap.c */ 2268 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2269 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2270 struct ext4_group_desc *gdp, 2271 struct buffer_head *bh, int sz); 2272 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2273 struct ext4_group_desc *gdp, 2274 struct buffer_head *bh, int sz); 2275 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2276 struct ext4_group_desc *gdp, 2277 struct buffer_head *bh); 2278 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2279 struct ext4_group_desc *gdp, 2280 struct buffer_head *bh); 2281 2282 /* balloc.c */ 2283 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2284 ext4_fsblk_t blocknr, 2285 ext4_group_t *blockgrpp, 2286 ext4_grpblk_t *offsetp); 2287 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2288 ext4_fsblk_t block); 2289 2290 extern unsigned int ext4_block_group(struct super_block *sb, 2291 ext4_fsblk_t blocknr); 2292 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2293 ext4_fsblk_t blocknr); 2294 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2295 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2296 ext4_group_t group); 2297 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2298 ext4_fsblk_t goal, 2299 unsigned int flags, 2300 unsigned long *count, 2301 int *errp); 2302 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2303 s64 nclusters, unsigned int flags); 2304 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2305 extern void ext4_check_blocks_bitmap(struct super_block *); 2306 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2307 ext4_group_t block_group, 2308 struct buffer_head ** bh); 2309 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2310 2311 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2312 ext4_group_t block_group); 2313 extern int ext4_wait_block_bitmap(struct super_block *sb, 2314 ext4_group_t block_group, 2315 struct buffer_head *bh); 2316 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2317 ext4_group_t block_group); 2318 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2319 ext4_group_t block_group, 2320 struct ext4_group_desc *gdp); 2321 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2322 2323 #ifdef CONFIG_UNICODE 2324 extern void ext4_fname_setup_ci_filename(struct inode *dir, 2325 const struct qstr *iname, 2326 struct fscrypt_str *fname); 2327 #endif 2328 2329 #ifdef CONFIG_FS_ENCRYPTION 2330 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst, 2331 const struct fscrypt_name *src) 2332 { 2333 memset(dst, 0, sizeof(*dst)); 2334 2335 dst->usr_fname = src->usr_fname; 2336 dst->disk_name = src->disk_name; 2337 dst->hinfo.hash = src->hash; 2338 dst->hinfo.minor_hash = src->minor_hash; 2339 dst->crypto_buf = src->crypto_buf; 2340 } 2341 2342 static inline int ext4_fname_setup_filename(struct inode *dir, 2343 const struct qstr *iname, 2344 int lookup, 2345 struct ext4_filename *fname) 2346 { 2347 struct fscrypt_name name; 2348 int err; 2349 2350 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2351 if (err) 2352 return err; 2353 2354 ext4_fname_from_fscrypt_name(fname, &name); 2355 2356 #ifdef CONFIG_UNICODE 2357 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2358 #endif 2359 return 0; 2360 } 2361 2362 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2363 struct dentry *dentry, 2364 struct ext4_filename *fname) 2365 { 2366 struct fscrypt_name name; 2367 int err; 2368 2369 err = fscrypt_prepare_lookup(dir, dentry, &name); 2370 if (err) 2371 return err; 2372 2373 ext4_fname_from_fscrypt_name(fname, &name); 2374 2375 #ifdef CONFIG_UNICODE 2376 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name); 2377 #endif 2378 return 0; 2379 } 2380 2381 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2382 { 2383 struct fscrypt_name name; 2384 2385 name.crypto_buf = fname->crypto_buf; 2386 fscrypt_free_filename(&name); 2387 2388 fname->crypto_buf.name = NULL; 2389 fname->usr_fname = NULL; 2390 fname->disk_name.name = NULL; 2391 2392 #ifdef CONFIG_UNICODE 2393 kfree(fname->cf_name.name); 2394 fname->cf_name.name = NULL; 2395 #endif 2396 } 2397 #else /* !CONFIG_FS_ENCRYPTION */ 2398 static inline int ext4_fname_setup_filename(struct inode *dir, 2399 const struct qstr *iname, 2400 int lookup, 2401 struct ext4_filename *fname) 2402 { 2403 fname->usr_fname = iname; 2404 fname->disk_name.name = (unsigned char *) iname->name; 2405 fname->disk_name.len = iname->len; 2406 2407 #ifdef CONFIG_UNICODE 2408 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2409 #endif 2410 2411 return 0; 2412 } 2413 2414 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2415 struct dentry *dentry, 2416 struct ext4_filename *fname) 2417 { 2418 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2419 } 2420 2421 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2422 { 2423 #ifdef CONFIG_UNICODE 2424 kfree(fname->cf_name.name); 2425 fname->cf_name.name = NULL; 2426 #endif 2427 } 2428 #endif /* !CONFIG_FS_ENCRYPTION */ 2429 2430 /* dir.c */ 2431 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2432 struct file *, 2433 struct ext4_dir_entry_2 *, 2434 struct buffer_head *, char *, int, 2435 unsigned int); 2436 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2437 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2438 (de), (bh), (buf), (size), (offset))) 2439 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2440 __u32 minor_hash, 2441 struct ext4_dir_entry_2 *dirent, 2442 struct fscrypt_str *ent_name); 2443 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2444 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2445 struct buffer_head *bh, 2446 void *buf, int buf_size, 2447 struct ext4_filename *fname, 2448 struct ext4_dir_entry_2 **dest_de); 2449 void ext4_insert_dentry(struct inode *inode, 2450 struct ext4_dir_entry_2 *de, 2451 int buf_size, 2452 struct ext4_filename *fname); 2453 static inline void ext4_update_dx_flag(struct inode *inode) 2454 { 2455 if (!ext4_has_feature_dir_index(inode->i_sb)) 2456 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2457 } 2458 static const unsigned char ext4_filetype_table[] = { 2459 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2460 }; 2461 2462 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2463 { 2464 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2465 return DT_UNKNOWN; 2466 2467 return ext4_filetype_table[filetype]; 2468 } 2469 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2470 void *buf, int buf_size); 2471 2472 /* fsync.c */ 2473 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2474 2475 /* hash.c */ 2476 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2477 struct dx_hash_info *hinfo); 2478 2479 /* ialloc.c */ 2480 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2481 const struct qstr *qstr, __u32 goal, 2482 uid_t *owner, __u32 i_flags, 2483 int handle_type, unsigned int line_no, 2484 int nblocks); 2485 2486 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2487 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2488 i_flags, 0, 0, 0) 2489 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2490 type, nblocks) \ 2491 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2492 0, (type), __LINE__, (nblocks)) 2493 2494 2495 extern void ext4_free_inode(handle_t *, struct inode *); 2496 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2497 extern unsigned long ext4_count_free_inodes(struct super_block *); 2498 extern unsigned long ext4_count_dirs(struct super_block *); 2499 extern void ext4_check_inodes_bitmap(struct super_block *); 2500 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2501 extern int ext4_init_inode_table(struct super_block *sb, 2502 ext4_group_t group, int barrier); 2503 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2504 2505 /* mballoc.c */ 2506 extern const struct seq_operations ext4_mb_seq_groups_ops; 2507 extern long ext4_mb_stats; 2508 extern long ext4_mb_max_to_scan; 2509 extern int ext4_mb_init(struct super_block *); 2510 extern int ext4_mb_release(struct super_block *); 2511 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2512 struct ext4_allocation_request *, int *); 2513 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2514 extern void ext4_discard_preallocations(struct inode *); 2515 extern int __init ext4_init_mballoc(void); 2516 extern void ext4_exit_mballoc(void); 2517 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2518 struct buffer_head *bh, ext4_fsblk_t block, 2519 unsigned long count, int flags); 2520 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2521 ext4_group_t ngroups); 2522 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2523 ext4_group_t i, struct ext4_group_desc *desc); 2524 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2525 ext4_fsblk_t block, unsigned long count); 2526 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2527 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2528 2529 /* inode.c */ 2530 int ext4_inode_is_fast_symlink(struct inode *inode); 2531 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2532 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2533 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2534 bool wait, struct buffer_head **bhs); 2535 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2536 struct buffer_head *bh_result, int create); 2537 int ext4_get_block(struct inode *inode, sector_t iblock, 2538 struct buffer_head *bh_result, int create); 2539 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2540 struct buffer_head *bh_result, int create); 2541 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2542 struct buffer_head *bh, int create); 2543 int ext4_walk_page_buffers(handle_t *handle, 2544 struct buffer_head *head, 2545 unsigned from, 2546 unsigned to, 2547 int *partial, 2548 int (*fn)(handle_t *handle, 2549 struct buffer_head *bh)); 2550 int do_journal_get_write_access(handle_t *handle, 2551 struct buffer_head *bh); 2552 #define FALL_BACK_TO_NONDELALLOC 1 2553 #define CONVERT_INLINE_DATA 2 2554 2555 typedef enum { 2556 EXT4_IGET_NORMAL = 0, 2557 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2558 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2559 } ext4_iget_flags; 2560 2561 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2562 ext4_iget_flags flags, const char *function, 2563 unsigned int line); 2564 2565 #define ext4_iget(sb, ino, flags) \ 2566 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2567 2568 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2569 extern int ext4_setattr(struct dentry *, struct iattr *); 2570 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2571 extern void ext4_evict_inode(struct inode *); 2572 extern void ext4_clear_inode(struct inode *); 2573 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2574 extern int ext4_sync_inode(handle_t *, struct inode *); 2575 extern void ext4_dirty_inode(struct inode *, int); 2576 extern int ext4_change_inode_journal_flag(struct inode *, int); 2577 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2578 extern int ext4_inode_attach_jinode(struct inode *inode); 2579 extern int ext4_can_truncate(struct inode *inode); 2580 extern int ext4_truncate(struct inode *); 2581 extern int ext4_break_layouts(struct inode *); 2582 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2583 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2584 extern void ext4_set_inode_flags(struct inode *); 2585 extern int ext4_alloc_da_blocks(struct inode *inode); 2586 extern void ext4_set_aops(struct inode *inode); 2587 extern int ext4_writepage_trans_blocks(struct inode *); 2588 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2589 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2590 loff_t lstart, loff_t lend); 2591 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2592 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2593 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2594 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2595 extern void ext4_da_release_space(struct inode *inode, int to_free); 2596 extern void ext4_da_update_reserve_space(struct inode *inode, 2597 int used, int quota_claim); 2598 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2599 ext4_fsblk_t pblk, ext4_lblk_t len); 2600 2601 /* indirect.c */ 2602 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2603 struct ext4_map_blocks *map, int flags); 2604 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2605 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2606 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2607 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2608 ext4_lblk_t start, ext4_lblk_t end); 2609 2610 /* ioctl.c */ 2611 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2612 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2613 2614 /* migrate.c */ 2615 extern int ext4_ext_migrate(struct inode *); 2616 extern int ext4_ind_migrate(struct inode *inode); 2617 2618 /* namei.c */ 2619 extern int ext4_dirblock_csum_verify(struct inode *inode, 2620 struct buffer_head *bh); 2621 extern int ext4_orphan_add(handle_t *, struct inode *); 2622 extern int ext4_orphan_del(handle_t *, struct inode *); 2623 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2624 __u32 start_minor_hash, __u32 *next_hash); 2625 extern int ext4_search_dir(struct buffer_head *bh, 2626 char *search_buf, 2627 int buf_size, 2628 struct inode *dir, 2629 struct ext4_filename *fname, 2630 unsigned int offset, 2631 struct ext4_dir_entry_2 **res_dir); 2632 extern int ext4_generic_delete_entry(handle_t *handle, 2633 struct inode *dir, 2634 struct ext4_dir_entry_2 *de_del, 2635 struct buffer_head *bh, 2636 void *entry_buf, 2637 int buf_size, 2638 int csum_size); 2639 extern bool ext4_empty_dir(struct inode *inode); 2640 2641 /* resize.c */ 2642 extern int ext4_group_add(struct super_block *sb, 2643 struct ext4_new_group_data *input); 2644 extern int ext4_group_extend(struct super_block *sb, 2645 struct ext4_super_block *es, 2646 ext4_fsblk_t n_blocks_count); 2647 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2648 2649 /* super.c */ 2650 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2651 sector_t block, int op_flags); 2652 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2653 extern int ext4_calculate_overhead(struct super_block *sb); 2654 extern void ext4_superblock_csum_set(struct super_block *sb); 2655 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2656 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2657 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2658 ext4_group_t ngroup); 2659 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2660 char nbuf[16]); 2661 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2662 ext4_group_t block_group, 2663 unsigned int flags); 2664 2665 extern __printf(4, 5) 2666 void __ext4_error(struct super_block *, const char *, unsigned int, 2667 const char *, ...); 2668 extern __printf(5, 6) 2669 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2670 const char *, ...); 2671 extern __printf(5, 6) 2672 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2673 const char *, ...); 2674 extern void __ext4_std_error(struct super_block *, const char *, 2675 unsigned int, int); 2676 extern __printf(4, 5) 2677 void __ext4_abort(struct super_block *, const char *, unsigned int, 2678 const char *, ...); 2679 extern __printf(4, 5) 2680 void __ext4_warning(struct super_block *, const char *, unsigned int, 2681 const char *, ...); 2682 extern __printf(4, 5) 2683 void __ext4_warning_inode(const struct inode *inode, const char *function, 2684 unsigned int line, const char *fmt, ...); 2685 extern __printf(3, 4) 2686 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2687 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2688 const char *, unsigned int, const char *); 2689 extern __printf(7, 8) 2690 void __ext4_grp_locked_error(const char *, unsigned int, 2691 struct super_block *, ext4_group_t, 2692 unsigned long, ext4_fsblk_t, 2693 const char *, ...); 2694 2695 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2696 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2697 2698 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2699 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2700 2701 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2702 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2703 2704 #ifdef CONFIG_PRINTK 2705 2706 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2707 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2708 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2709 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2710 #define ext4_error(sb, fmt, ...) \ 2711 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2712 #define ext4_abort(sb, fmt, ...) \ 2713 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2714 #define ext4_warning(sb, fmt, ...) \ 2715 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2716 #define ext4_warning_inode(inode, fmt, ...) \ 2717 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2718 #define ext4_msg(sb, level, fmt, ...) \ 2719 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2720 #define dump_mmp_msg(sb, mmp, msg) \ 2721 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2722 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2723 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2724 fmt, ##__VA_ARGS__) 2725 2726 #else 2727 2728 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2729 do { \ 2730 no_printk(fmt, ##__VA_ARGS__); \ 2731 __ext4_error_inode(inode, "", 0, block, " "); \ 2732 } while (0) 2733 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2734 do { \ 2735 no_printk(fmt, ##__VA_ARGS__); \ 2736 __ext4_error_file(file, "", 0, block, " "); \ 2737 } while (0) 2738 #define ext4_error(sb, fmt, ...) \ 2739 do { \ 2740 no_printk(fmt, ##__VA_ARGS__); \ 2741 __ext4_error(sb, "", 0, " "); \ 2742 } while (0) 2743 #define ext4_abort(sb, fmt, ...) \ 2744 do { \ 2745 no_printk(fmt, ##__VA_ARGS__); \ 2746 __ext4_abort(sb, "", 0, " "); \ 2747 } while (0) 2748 #define ext4_warning(sb, fmt, ...) \ 2749 do { \ 2750 no_printk(fmt, ##__VA_ARGS__); \ 2751 __ext4_warning(sb, "", 0, " "); \ 2752 } while (0) 2753 #define ext4_warning_inode(inode, fmt, ...) \ 2754 do { \ 2755 no_printk(fmt, ##__VA_ARGS__); \ 2756 __ext4_warning_inode(inode, "", 0, " "); \ 2757 } while (0) 2758 #define ext4_msg(sb, level, fmt, ...) \ 2759 do { \ 2760 no_printk(fmt, ##__VA_ARGS__); \ 2761 __ext4_msg(sb, "", " "); \ 2762 } while (0) 2763 #define dump_mmp_msg(sb, mmp, msg) \ 2764 __dump_mmp_msg(sb, mmp, "", 0, "") 2765 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2766 do { \ 2767 no_printk(fmt, ##__VA_ARGS__); \ 2768 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2769 } while (0) 2770 2771 #endif 2772 2773 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2774 __u32 compat); 2775 extern int ext4_update_rocompat_feature(handle_t *handle, 2776 struct super_block *sb, __u32 rocompat); 2777 extern int ext4_update_incompat_feature(handle_t *handle, 2778 struct super_block *sb, __u32 incompat); 2779 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2780 struct ext4_group_desc *bg); 2781 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2782 struct ext4_group_desc *bg); 2783 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2784 struct ext4_group_desc *bg); 2785 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2786 struct ext4_group_desc *bg); 2787 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2788 struct ext4_group_desc *bg); 2789 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2790 struct ext4_group_desc *bg); 2791 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2792 struct ext4_group_desc *bg); 2793 extern void ext4_block_bitmap_set(struct super_block *sb, 2794 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2795 extern void ext4_inode_bitmap_set(struct super_block *sb, 2796 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2797 extern void ext4_inode_table_set(struct super_block *sb, 2798 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2799 extern void ext4_free_group_clusters_set(struct super_block *sb, 2800 struct ext4_group_desc *bg, 2801 __u32 count); 2802 extern void ext4_free_inodes_set(struct super_block *sb, 2803 struct ext4_group_desc *bg, __u32 count); 2804 extern void ext4_used_dirs_set(struct super_block *sb, 2805 struct ext4_group_desc *bg, __u32 count); 2806 extern void ext4_itable_unused_set(struct super_block *sb, 2807 struct ext4_group_desc *bg, __u32 count); 2808 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2809 struct ext4_group_desc *gdp); 2810 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2811 struct ext4_group_desc *gdp); 2812 extern int ext4_register_li_request(struct super_block *sb, 2813 ext4_group_t first_not_zeroed); 2814 2815 static inline int ext4_has_metadata_csum(struct super_block *sb) 2816 { 2817 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2818 !EXT4_SB(sb)->s_chksum_driver); 2819 2820 return ext4_has_feature_metadata_csum(sb) && 2821 (EXT4_SB(sb)->s_chksum_driver != NULL); 2822 } 2823 2824 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2825 { 2826 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2827 } 2828 2829 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2830 { 2831 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2832 le32_to_cpu(es->s_blocks_count_lo); 2833 } 2834 2835 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2836 { 2837 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2838 le32_to_cpu(es->s_r_blocks_count_lo); 2839 } 2840 2841 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2842 { 2843 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2844 le32_to_cpu(es->s_free_blocks_count_lo); 2845 } 2846 2847 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2848 ext4_fsblk_t blk) 2849 { 2850 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2851 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2852 } 2853 2854 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2855 ext4_fsblk_t blk) 2856 { 2857 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2858 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2859 } 2860 2861 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2862 ext4_fsblk_t blk) 2863 { 2864 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2865 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2866 } 2867 2868 static inline loff_t ext4_isize(struct super_block *sb, 2869 struct ext4_inode *raw_inode) 2870 { 2871 if (ext4_has_feature_largedir(sb) || 2872 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2873 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2874 le32_to_cpu(raw_inode->i_size_lo); 2875 2876 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2877 } 2878 2879 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2880 { 2881 raw_inode->i_size_lo = cpu_to_le32(i_size); 2882 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2883 } 2884 2885 static inline 2886 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2887 ext4_group_t group) 2888 { 2889 struct ext4_group_info ***grp_info; 2890 long indexv, indexh; 2891 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2892 grp_info = EXT4_SB(sb)->s_group_info; 2893 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2894 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2895 return grp_info[indexv][indexh]; 2896 } 2897 2898 /* 2899 * Reading s_groups_count requires using smp_rmb() afterwards. See 2900 * the locking protocol documented in the comments of ext4_group_add() 2901 * in resize.c 2902 */ 2903 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2904 { 2905 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2906 2907 smp_rmb(); 2908 return ngroups; 2909 } 2910 2911 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2912 ext4_group_t block_group) 2913 { 2914 return block_group >> sbi->s_log_groups_per_flex; 2915 } 2916 2917 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2918 { 2919 return 1 << sbi->s_log_groups_per_flex; 2920 } 2921 2922 #define ext4_std_error(sb, errno) \ 2923 do { \ 2924 if ((errno)) \ 2925 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2926 } while (0) 2927 2928 #ifdef CONFIG_SMP 2929 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2930 * counters. So we need to make sure we have free clusters more 2931 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2932 */ 2933 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2934 #else 2935 #define EXT4_FREECLUSTERS_WATERMARK 0 2936 #endif 2937 2938 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2939 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2940 { 2941 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2942 !inode_is_locked(inode)); 2943 down_write(&EXT4_I(inode)->i_data_sem); 2944 if (newsize > EXT4_I(inode)->i_disksize) 2945 EXT4_I(inode)->i_disksize = newsize; 2946 up_write(&EXT4_I(inode)->i_data_sem); 2947 } 2948 2949 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2950 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2951 { 2952 int changed = 0; 2953 2954 if (newsize > inode->i_size) { 2955 i_size_write(inode, newsize); 2956 changed = 1; 2957 } 2958 if (newsize > EXT4_I(inode)->i_disksize) { 2959 ext4_update_i_disksize(inode, newsize); 2960 changed |= 2; 2961 } 2962 return changed; 2963 } 2964 2965 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2966 loff_t len); 2967 2968 struct ext4_group_info { 2969 unsigned long bb_state; 2970 struct rb_root bb_free_root; 2971 ext4_grpblk_t bb_first_free; /* first free block */ 2972 ext4_grpblk_t bb_free; /* total free blocks */ 2973 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2974 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2975 struct list_head bb_prealloc_list; 2976 #ifdef DOUBLE_CHECK 2977 void *bb_bitmap; 2978 #endif 2979 struct rw_semaphore alloc_sem; 2980 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2981 * regions, index is order. 2982 * bb_counters[3] = 5 means 2983 * 5 free 8-block regions. */ 2984 }; 2985 2986 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2987 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2988 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2989 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2990 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2991 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2992 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2993 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2994 2995 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2996 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2997 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2998 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2999 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3000 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3001 3002 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3003 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3004 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3005 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3006 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3007 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3008 3009 #define EXT4_MAX_CONTENTION 8 3010 #define EXT4_CONTENTION_THRESHOLD 2 3011 3012 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3013 ext4_group_t group) 3014 { 3015 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3016 } 3017 3018 /* 3019 * Returns true if the filesystem is busy enough that attempts to 3020 * access the block group locks has run into contention. 3021 */ 3022 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3023 { 3024 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3025 } 3026 3027 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3028 { 3029 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3030 if (spin_trylock(lock)) 3031 /* 3032 * We're able to grab the lock right away, so drop the 3033 * lock contention counter. 3034 */ 3035 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3036 else { 3037 /* 3038 * The lock is busy, so bump the contention counter, 3039 * and then wait on the spin lock. 3040 */ 3041 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3042 EXT4_MAX_CONTENTION); 3043 spin_lock(lock); 3044 } 3045 } 3046 3047 static inline void ext4_unlock_group(struct super_block *sb, 3048 ext4_group_t group) 3049 { 3050 spin_unlock(ext4_group_lock_ptr(sb, group)); 3051 } 3052 3053 /* 3054 * Block validity checking 3055 */ 3056 #define ext4_check_indirect_blockref(inode, bh) \ 3057 ext4_check_blockref(__func__, __LINE__, inode, \ 3058 (__le32 *)(bh)->b_data, \ 3059 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3060 3061 #define ext4_ind_check_inode(inode) \ 3062 ext4_check_blockref(__func__, __LINE__, inode, \ 3063 EXT4_I(inode)->i_data, \ 3064 EXT4_NDIR_BLOCKS) 3065 3066 /* 3067 * Inodes and files operations 3068 */ 3069 3070 /* dir.c */ 3071 extern const struct file_operations ext4_dir_operations; 3072 3073 #ifdef CONFIG_UNICODE 3074 extern const struct dentry_operations ext4_dentry_ops; 3075 #endif 3076 3077 /* file.c */ 3078 extern const struct inode_operations ext4_file_inode_operations; 3079 extern const struct file_operations ext4_file_operations; 3080 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3081 3082 /* inline.c */ 3083 extern int ext4_get_max_inline_size(struct inode *inode); 3084 extern int ext4_find_inline_data_nolock(struct inode *inode); 3085 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3086 unsigned int len); 3087 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3088 3089 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3090 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3091 struct inode *inode, 3092 loff_t pos, unsigned len, 3093 unsigned flags, 3094 struct page **pagep); 3095 extern int ext4_write_inline_data_end(struct inode *inode, 3096 loff_t pos, unsigned len, 3097 unsigned copied, 3098 struct page *page); 3099 extern struct buffer_head * 3100 ext4_journalled_write_inline_data(struct inode *inode, 3101 unsigned len, 3102 struct page *page); 3103 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3104 struct inode *inode, 3105 loff_t pos, unsigned len, 3106 unsigned flags, 3107 struct page **pagep, 3108 void **fsdata); 3109 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3110 unsigned len, unsigned copied, 3111 struct page *page); 3112 extern int ext4_try_add_inline_entry(handle_t *handle, 3113 struct ext4_filename *fname, 3114 struct inode *dir, struct inode *inode); 3115 extern int ext4_try_create_inline_dir(handle_t *handle, 3116 struct inode *parent, 3117 struct inode *inode); 3118 extern int ext4_read_inline_dir(struct file *filp, 3119 struct dir_context *ctx, 3120 int *has_inline_data); 3121 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3122 struct inode *dir, ext4_lblk_t block, 3123 struct dx_hash_info *hinfo, 3124 __u32 start_hash, __u32 start_minor_hash, 3125 int *has_inline_data); 3126 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3127 struct ext4_filename *fname, 3128 struct ext4_dir_entry_2 **res_dir, 3129 int *has_inline_data); 3130 extern int ext4_delete_inline_entry(handle_t *handle, 3131 struct inode *dir, 3132 struct ext4_dir_entry_2 *de_del, 3133 struct buffer_head *bh, 3134 int *has_inline_data); 3135 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3136 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3137 struct ext4_dir_entry_2 **parent_de, 3138 int *retval); 3139 extern int ext4_inline_data_fiemap(struct inode *inode, 3140 struct fiemap_extent_info *fieinfo, 3141 int *has_inline, __u64 start, __u64 len); 3142 3143 struct iomap; 3144 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3145 3146 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3147 3148 extern int ext4_convert_inline_data(struct inode *inode); 3149 3150 static inline int ext4_has_inline_data(struct inode *inode) 3151 { 3152 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3153 EXT4_I(inode)->i_inline_off; 3154 } 3155 3156 /* namei.c */ 3157 extern const struct inode_operations ext4_dir_inode_operations; 3158 extern const struct inode_operations ext4_special_inode_operations; 3159 extern struct dentry *ext4_get_parent(struct dentry *child); 3160 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3161 struct ext4_dir_entry_2 *de, 3162 int blocksize, int csum_size, 3163 unsigned int parent_ino, int dotdot_real_len); 3164 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3165 unsigned int blocksize); 3166 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3167 struct buffer_head *bh); 3168 extern int ext4_ci_compare(const struct inode *parent, 3169 const struct qstr *fname, 3170 const struct qstr *entry, bool quick); 3171 3172 #define S_SHIFT 12 3173 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3174 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3175 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3176 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3177 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3178 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3179 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3180 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3181 }; 3182 3183 static inline void ext4_set_de_type(struct super_block *sb, 3184 struct ext4_dir_entry_2 *de, 3185 umode_t mode) { 3186 if (ext4_has_feature_filetype(sb)) 3187 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3188 } 3189 3190 /* readpages.c */ 3191 extern int ext4_mpage_readpages(struct address_space *mapping, 3192 struct list_head *pages, struct page *page, 3193 unsigned nr_pages, bool is_readahead); 3194 extern int __init ext4_init_post_read_processing(void); 3195 extern void ext4_exit_post_read_processing(void); 3196 3197 /* symlink.c */ 3198 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3199 extern const struct inode_operations ext4_symlink_inode_operations; 3200 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3201 3202 /* sysfs.c */ 3203 extern int ext4_register_sysfs(struct super_block *sb); 3204 extern void ext4_unregister_sysfs(struct super_block *sb); 3205 extern int __init ext4_init_sysfs(void); 3206 extern void ext4_exit_sysfs(void); 3207 3208 /* block_validity */ 3209 extern void ext4_release_system_zone(struct super_block *sb); 3210 extern int ext4_setup_system_zone(struct super_block *sb); 3211 extern int __init ext4_init_system_zone(void); 3212 extern void ext4_exit_system_zone(void); 3213 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3214 ext4_fsblk_t start_blk, 3215 unsigned int count); 3216 extern int ext4_check_blockref(const char *, unsigned int, 3217 struct inode *, __le32 *, unsigned int); 3218 3219 /* extents.c */ 3220 struct ext4_ext_path; 3221 struct ext4_extent; 3222 3223 /* 3224 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3225 * __le32. 3226 */ 3227 #define EXT_MAX_BLOCKS 0xffffffff 3228 3229 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3230 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3231 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3232 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3233 struct ext4_map_blocks *map, int flags); 3234 extern int ext4_ext_truncate(handle_t *, struct inode *); 3235 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3236 ext4_lblk_t end); 3237 extern void ext4_ext_init(struct super_block *); 3238 extern void ext4_ext_release(struct super_block *); 3239 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3240 loff_t len); 3241 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3242 loff_t offset, ssize_t len); 3243 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3244 struct ext4_map_blocks *map, int flags); 3245 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3246 ext4_lblk_t lblocks); 3247 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3248 int num, 3249 struct ext4_ext_path *path); 3250 extern int ext4_can_extents_be_merged(struct inode *inode, 3251 struct ext4_extent *ex1, 3252 struct ext4_extent *ex2); 3253 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3254 struct ext4_ext_path **, 3255 struct ext4_extent *, int); 3256 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3257 struct ext4_ext_path **, 3258 int flags); 3259 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3260 extern int ext4_ext_check_inode(struct inode *inode); 3261 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3262 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3263 __u64 start, __u64 len); 3264 extern int ext4_ext_precache(struct inode *inode); 3265 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3266 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3267 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3268 struct inode *inode2, ext4_lblk_t lblk1, 3269 ext4_lblk_t lblk2, ext4_lblk_t count, 3270 int mark_unwritten,int *err); 3271 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3272 3273 /* move_extent.c */ 3274 extern void ext4_double_down_write_data_sem(struct inode *first, 3275 struct inode *second); 3276 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3277 struct inode *donor_inode); 3278 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3279 __u64 start_orig, __u64 start_donor, 3280 __u64 len, __u64 *moved_len); 3281 3282 /* page-io.c */ 3283 extern int __init ext4_init_pageio(void); 3284 extern void ext4_exit_pageio(void); 3285 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3286 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3287 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3288 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3289 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3290 struct writeback_control *wbc); 3291 extern void ext4_end_io_rsv_work(struct work_struct *work); 3292 extern void ext4_io_submit(struct ext4_io_submit *io); 3293 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3294 struct page *page, 3295 int len, 3296 struct writeback_control *wbc, 3297 bool keep_towrite); 3298 3299 /* mmp.c */ 3300 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3301 3302 /* verity.c */ 3303 extern const struct fsverity_operations ext4_verityops; 3304 3305 /* 3306 * Add new method to test whether block and inode bitmaps are properly 3307 * initialized. With uninit_bg reading the block from disk is not enough 3308 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3309 */ 3310 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3311 3312 static inline int bitmap_uptodate(struct buffer_head *bh) 3313 { 3314 return (buffer_uptodate(bh) && 3315 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3316 } 3317 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3318 { 3319 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3320 } 3321 3322 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3323 3324 /* For ioend & aio unwritten conversion wait queues */ 3325 #define EXT4_WQ_HASH_SZ 37 3326 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3327 EXT4_WQ_HASH_SZ]) 3328 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3329 3330 extern int ext4_resize_begin(struct super_block *sb); 3331 extern void ext4_resize_end(struct super_block *sb); 3332 3333 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3334 struct ext4_io_end *io_end) 3335 { 3336 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3337 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3338 atomic_inc(&EXT4_I(inode)->i_unwritten); 3339 } 3340 } 3341 3342 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3343 { 3344 struct inode *inode = io_end->inode; 3345 3346 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3347 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3348 /* Wake up anyone waiting on unwritten extent conversion */ 3349 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3350 wake_up_all(ext4_ioend_wq(inode)); 3351 } 3352 } 3353 3354 extern const struct iomap_ops ext4_iomap_ops; 3355 3356 #endif /* __KERNEL__ */ 3357 3358 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3359 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3360 3361 #endif /* _EXT4_H */ 3362