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 45 #include <linux/compiler.h> 46 47 /* 48 * The fourth extended filesystem constants/structures 49 */ 50 51 /* 52 * with AGGRESSIVE_CHECK allocator runs consistency checks over 53 * structures. these checks slow things down a lot 54 */ 55 #define AGGRESSIVE_CHECK__ 56 57 /* 58 * with DOUBLE_CHECK defined mballoc creates persistent in-core 59 * bitmaps, maintains and uses them to check for double allocations 60 */ 61 #define DOUBLE_CHECK__ 62 63 /* 64 * Define EXT4FS_DEBUG to produce debug messages 65 */ 66 #undef EXT4FS_DEBUG 67 68 /* 69 * Debug code 70 */ 71 #ifdef EXT4FS_DEBUG 72 #define ext4_debug(f, a...) \ 73 do { \ 74 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 75 __FILE__, __LINE__, __func__); \ 76 printk(KERN_DEBUG f, ## a); \ 77 } while (0) 78 #else 79 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 80 #endif 81 82 /* 83 * Turn on EXT_DEBUG to get lots of info about extents operations. 84 */ 85 #define EXT_DEBUG__ 86 #ifdef EXT_DEBUG 87 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 88 #else 89 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 90 #endif 91 92 /* data type for block offset of block group */ 93 typedef int ext4_grpblk_t; 94 95 /* data type for filesystem-wide blocks number */ 96 typedef unsigned long long ext4_fsblk_t; 97 98 /* data type for file logical block number */ 99 typedef __u32 ext4_lblk_t; 100 101 /* data type for block group number */ 102 typedef unsigned int ext4_group_t; 103 104 enum SHIFT_DIRECTION { 105 SHIFT_LEFT = 0, 106 SHIFT_RIGHT, 107 }; 108 109 /* 110 * Flags used in mballoc's allocation_context flags field. 111 * 112 * Also used to show what's going on for debugging purposes when the 113 * flag field is exported via the traceport interface 114 */ 115 116 /* prefer goal again. length */ 117 #define EXT4_MB_HINT_MERGE 0x0001 118 /* blocks already reserved */ 119 #define EXT4_MB_HINT_RESERVED 0x0002 120 /* metadata is being allocated */ 121 #define EXT4_MB_HINT_METADATA 0x0004 122 /* first blocks in the file */ 123 #define EXT4_MB_HINT_FIRST 0x0008 124 /* search for the best chunk */ 125 #define EXT4_MB_HINT_BEST 0x0010 126 /* data is being allocated */ 127 #define EXT4_MB_HINT_DATA 0x0020 128 /* don't preallocate (for tails) */ 129 #define EXT4_MB_HINT_NOPREALLOC 0x0040 130 /* allocate for locality group */ 131 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 132 /* allocate goal blocks or none */ 133 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 134 /* goal is meaningful */ 135 #define EXT4_MB_HINT_TRY_GOAL 0x0200 136 /* blocks already pre-reserved by delayed allocation */ 137 #define EXT4_MB_DELALLOC_RESERVED 0x0400 138 /* We are doing stream allocation */ 139 #define EXT4_MB_STREAM_ALLOC 0x0800 140 /* Use reserved root blocks if needed */ 141 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 142 /* Use blocks from reserved pool */ 143 #define EXT4_MB_USE_RESERVED 0x2000 144 145 struct ext4_allocation_request { 146 /* target inode for block we're allocating */ 147 struct inode *inode; 148 /* how many blocks we want to allocate */ 149 unsigned int len; 150 /* logical block in target inode */ 151 ext4_lblk_t logical; 152 /* the closest logical allocated block to the left */ 153 ext4_lblk_t lleft; 154 /* the closest logical allocated block to the right */ 155 ext4_lblk_t lright; 156 /* phys. target (a hint) */ 157 ext4_fsblk_t goal; 158 /* phys. block for the closest logical allocated block to the left */ 159 ext4_fsblk_t pleft; 160 /* phys. block for the closest logical allocated block to the right */ 161 ext4_fsblk_t pright; 162 /* flags. see above EXT4_MB_HINT_* */ 163 unsigned int flags; 164 }; 165 166 /* 167 * Logical to physical block mapping, used by ext4_map_blocks() 168 * 169 * This structure is used to pass requests into ext4_map_blocks() as 170 * well as to store the information returned by ext4_map_blocks(). It 171 * takes less room on the stack than a struct buffer_head. 172 */ 173 #define EXT4_MAP_NEW (1 << BH_New) 174 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 175 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 176 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 177 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 178 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 179 180 struct ext4_map_blocks { 181 ext4_fsblk_t m_pblk; 182 ext4_lblk_t m_lblk; 183 unsigned int m_len; 184 unsigned int m_flags; 185 }; 186 187 /* 188 * Flags for ext4_io_end->flags 189 */ 190 #define EXT4_IO_END_UNWRITTEN 0x0001 191 192 /* 193 * For converting unwritten extents on a work queue. 'handle' is used for 194 * buffered writeback. 195 */ 196 typedef struct ext4_io_end { 197 struct list_head list; /* per-file finished IO list */ 198 handle_t *handle; /* handle reserved for extent 199 * conversion */ 200 struct inode *inode; /* file being written to */ 201 struct bio *bio; /* Linked list of completed 202 * bios covering the extent */ 203 unsigned int flag; /* unwritten or not */ 204 atomic_t count; /* reference counter */ 205 loff_t offset; /* offset in the file */ 206 ssize_t size; /* size of the extent */ 207 } ext4_io_end_t; 208 209 struct ext4_io_submit { 210 struct writeback_control *io_wbc; 211 struct bio *io_bio; 212 ext4_io_end_t *io_end; 213 sector_t io_next_block; 214 }; 215 216 /* 217 * Special inodes numbers 218 */ 219 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 220 #define EXT4_ROOT_INO 2 /* Root inode */ 221 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 222 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 223 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 224 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 225 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 226 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 227 228 /* First non-reserved inode for old ext4 filesystems */ 229 #define EXT4_GOOD_OLD_FIRST_INO 11 230 231 /* 232 * Maximal count of links to a file 233 */ 234 #define EXT4_LINK_MAX 65000 235 236 /* 237 * Macro-instructions used to manage several block sizes 238 */ 239 #define EXT4_MIN_BLOCK_SIZE 1024 240 #define EXT4_MAX_BLOCK_SIZE 65536 241 #define EXT4_MIN_BLOCK_LOG_SIZE 10 242 #define EXT4_MAX_BLOCK_LOG_SIZE 16 243 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 244 #ifdef __KERNEL__ 245 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 246 #else 247 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 248 #endif 249 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 250 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 251 EXT4_SB(s)->s_cluster_bits) 252 #ifdef __KERNEL__ 253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 254 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 255 #else 256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 257 #endif 258 #ifdef __KERNEL__ 259 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 260 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 261 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 262 #else 263 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 264 EXT4_GOOD_OLD_INODE_SIZE : \ 265 (s)->s_inode_size) 266 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 267 EXT4_GOOD_OLD_FIRST_INO : \ 268 (s)->s_first_ino) 269 #endif 270 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 271 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 272 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 273 blkbits)) 274 275 /* Translate a block number to a cluster number */ 276 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 277 /* Translate a cluster number to a block number */ 278 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 279 /* Translate # of blks to # of clusters */ 280 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 281 (sbi)->s_cluster_bits) 282 /* Mask out the low bits to get the starting block of the cluster */ 283 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 285 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 287 /* Get the cluster offset */ 288 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 290 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 292 293 /* 294 * Structure of a blocks group descriptor 295 */ 296 struct ext4_group_desc 297 { 298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 300 __le32 bg_inode_table_lo; /* Inodes table block */ 301 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 302 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 303 __le16 bg_used_dirs_count_lo; /* Directories count */ 304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 305 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 306 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 307 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 308 __le16 bg_itable_unused_lo; /* Unused inodes count */ 309 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 310 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 311 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 312 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 313 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 314 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 315 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 316 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 317 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 318 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 319 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 320 __u32 bg_reserved; 321 }; 322 323 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 324 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 325 sizeof(__le16)) 326 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 327 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 328 sizeof(__le16)) 329 330 /* 331 * Structure of a flex block group info 332 */ 333 334 struct flex_groups { 335 atomic64_t free_clusters; 336 atomic_t free_inodes; 337 atomic_t used_dirs; 338 }; 339 340 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 341 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 342 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 343 344 /* 345 * Macro-instructions used to manage group descriptors 346 */ 347 #define EXT4_MIN_DESC_SIZE 32 348 #define EXT4_MIN_DESC_SIZE_64BIT 64 349 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 350 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 351 #ifdef __KERNEL__ 352 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 353 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 354 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 355 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 356 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 357 #else 358 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 359 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 360 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 361 #endif 362 363 /* 364 * Constants relative to the data blocks 365 */ 366 #define EXT4_NDIR_BLOCKS 12 367 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 368 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 369 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 370 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 371 372 /* 373 * Inode flags 374 */ 375 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 376 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 377 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 378 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 379 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 380 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 381 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 382 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 383 /* Reserved for compression usage... */ 384 #define EXT4_DIRTY_FL 0x00000100 385 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 386 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 387 /* nb: was previously EXT2_ECOMPR_FL */ 388 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 389 /* End compression flags --- maybe not all used */ 390 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 391 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 392 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 393 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 394 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 395 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 396 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 397 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 398 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 399 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 400 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 401 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 402 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */ 403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 404 405 #define EXT4_FL_USER_VISIBLE 0x704BDFFF /* User visible flags */ 406 #define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */ 407 408 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 409 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 410 EXT4_IMMUTABLE_FL | \ 411 EXT4_APPEND_FL | \ 412 EXT4_NODUMP_FL | \ 413 EXT4_NOATIME_FL | \ 414 EXT4_PROJINHERIT_FL) 415 416 /* Flags that should be inherited by new inodes from their parent. */ 417 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 418 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 419 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 420 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 421 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL) 422 423 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 424 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 425 EXT4_PROJINHERIT_FL)) 426 427 /* Flags that are appropriate for non-directories/regular files. */ 428 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 429 430 /* The only flags that should be swapped */ 431 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 432 433 /* Mask out flags that are inappropriate for the given type of inode. */ 434 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 435 { 436 if (S_ISDIR(mode)) 437 return flags; 438 else if (S_ISREG(mode)) 439 return flags & EXT4_REG_FLMASK; 440 else 441 return flags & EXT4_OTHER_FLMASK; 442 } 443 444 /* 445 * Inode flags used for atomic set/get 446 */ 447 enum { 448 EXT4_INODE_SECRM = 0, /* Secure deletion */ 449 EXT4_INODE_UNRM = 1, /* Undelete */ 450 EXT4_INODE_COMPR = 2, /* Compress file */ 451 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 452 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 453 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 454 EXT4_INODE_NODUMP = 6, /* do not dump file */ 455 EXT4_INODE_NOATIME = 7, /* do not update atime */ 456 /* Reserved for compression usage... */ 457 EXT4_INODE_DIRTY = 8, 458 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 459 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 460 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 461 /* End compression flags --- maybe not all used */ 462 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 463 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 464 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 465 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 466 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 467 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 468 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 469 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 470 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 471 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 472 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 473 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 474 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 475 }; 476 477 /* 478 * Since it's pretty easy to mix up bit numbers and hex values, we use a 479 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 480 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 481 * any extra space in the compiled kernel image, otherwise, the build will fail. 482 * It's important that these values are the same, since we are using 483 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 484 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 485 * values found in ext2, ext3 and ext4 filesystems, and of course the values 486 * defined in e2fsprogs. 487 * 488 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 489 */ 490 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 491 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 492 493 static inline void ext4_check_flag_values(void) 494 { 495 CHECK_FLAG_VALUE(SECRM); 496 CHECK_FLAG_VALUE(UNRM); 497 CHECK_FLAG_VALUE(COMPR); 498 CHECK_FLAG_VALUE(SYNC); 499 CHECK_FLAG_VALUE(IMMUTABLE); 500 CHECK_FLAG_VALUE(APPEND); 501 CHECK_FLAG_VALUE(NODUMP); 502 CHECK_FLAG_VALUE(NOATIME); 503 CHECK_FLAG_VALUE(DIRTY); 504 CHECK_FLAG_VALUE(COMPRBLK); 505 CHECK_FLAG_VALUE(NOCOMPR); 506 CHECK_FLAG_VALUE(ENCRYPT); 507 CHECK_FLAG_VALUE(INDEX); 508 CHECK_FLAG_VALUE(IMAGIC); 509 CHECK_FLAG_VALUE(JOURNAL_DATA); 510 CHECK_FLAG_VALUE(NOTAIL); 511 CHECK_FLAG_VALUE(DIRSYNC); 512 CHECK_FLAG_VALUE(TOPDIR); 513 CHECK_FLAG_VALUE(HUGE_FILE); 514 CHECK_FLAG_VALUE(EXTENTS); 515 CHECK_FLAG_VALUE(EA_INODE); 516 CHECK_FLAG_VALUE(EOFBLOCKS); 517 CHECK_FLAG_VALUE(INLINE_DATA); 518 CHECK_FLAG_VALUE(PROJINHERIT); 519 CHECK_FLAG_VALUE(RESERVED); 520 } 521 522 /* Used to pass group descriptor data when online resize is done */ 523 struct ext4_new_group_input { 524 __u32 group; /* Group number for this data */ 525 __u64 block_bitmap; /* Absolute block number of block bitmap */ 526 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 527 __u64 inode_table; /* Absolute block number of inode table start */ 528 __u32 blocks_count; /* Total number of blocks in this group */ 529 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 530 __u16 unused; 531 }; 532 533 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 534 struct compat_ext4_new_group_input { 535 u32 group; 536 compat_u64 block_bitmap; 537 compat_u64 inode_bitmap; 538 compat_u64 inode_table; 539 u32 blocks_count; 540 u16 reserved_blocks; 541 u16 unused; 542 }; 543 #endif 544 545 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 546 struct ext4_new_group_data { 547 __u32 group; 548 __u64 block_bitmap; 549 __u64 inode_bitmap; 550 __u64 inode_table; 551 __u32 blocks_count; 552 __u16 reserved_blocks; 553 __u16 mdata_blocks; 554 __u32 free_clusters_count; 555 }; 556 557 /* Indexes used to index group tables in ext4_new_group_data */ 558 enum { 559 BLOCK_BITMAP = 0, /* block bitmap */ 560 INODE_BITMAP, /* inode bitmap */ 561 INODE_TABLE, /* inode tables */ 562 GROUP_TABLE_COUNT, 563 }; 564 565 /* 566 * Flags used by ext4_map_blocks() 567 */ 568 /* Allocate any needed blocks and/or convert an unwritten 569 extent to be an initialized ext4 */ 570 #define EXT4_GET_BLOCKS_CREATE 0x0001 571 /* Request the creation of an unwritten extent */ 572 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 573 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 574 EXT4_GET_BLOCKS_CREATE) 575 /* Caller is from the delayed allocation writeout path 576 * finally doing the actual allocation of delayed blocks */ 577 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 578 /* caller is from the direct IO path, request to creation of an 579 unwritten extents if not allocated, split the unwritten 580 extent if blocks has been preallocated already*/ 581 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 582 #define EXT4_GET_BLOCKS_CONVERT 0x0010 583 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 584 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 585 /* Convert extent to initialized after IO complete */ 586 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 587 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 588 /* Eventual metadata allocation (due to growing extent tree) 589 * should not fail, so try to use reserved blocks for that.*/ 590 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 591 /* Don't normalize allocation size (used for fallocate) */ 592 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 593 /* Request will not result in inode size update (user for fallocate) */ 594 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 595 /* Convert written extents to unwritten */ 596 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 597 /* Write zeros to newly created written extents */ 598 #define EXT4_GET_BLOCKS_ZERO 0x0200 599 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 600 EXT4_GET_BLOCKS_ZERO) 601 /* Caller will submit data before dropping transaction handle. This 602 * allows jbd2 to avoid submitting data before commit. */ 603 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 604 605 /* 606 * The bit position of these flags must not overlap with any of the 607 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 608 * read_extent_tree_block(), ext4_split_extent_at(), 609 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 610 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 611 * caching the extents when reading from the extent tree while a 612 * truncate or punch hole operation is in progress. 613 */ 614 #define EXT4_EX_NOCACHE 0x40000000 615 #define EXT4_EX_FORCE_CACHE 0x20000000 616 617 /* 618 * Flags used by ext4_free_blocks 619 */ 620 #define EXT4_FREE_BLOCKS_METADATA 0x0001 621 #define EXT4_FREE_BLOCKS_FORGET 0x0002 622 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 623 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 624 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 625 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 626 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 627 628 /* 629 * ioctl commands 630 */ 631 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 632 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 633 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 634 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 635 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 636 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 637 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 638 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 639 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 640 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 641 #define EXT4_IOC_MIGRATE _IO('f', 9) 642 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 643 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 644 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 645 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 646 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 647 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 648 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 649 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 650 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 651 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 652 /* ioctl codes 19--39 are reserved for fscrypt */ 653 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 654 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 655 656 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 657 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 658 659 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 660 661 /* 662 * Flags for going down operation 663 */ 664 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 665 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 666 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 667 668 /* 669 * Flags returned by EXT4_IOC_GETSTATE 670 * 671 * We only expose to userspace a subset of the state flags in 672 * i_state_flags 673 */ 674 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 675 #define EXT4_STATE_FLAG_NEW 0x00000002 676 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 677 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 678 679 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 680 /* 681 * ioctl commands in 32 bit emulation 682 */ 683 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 684 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 685 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 686 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 687 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 688 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 689 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 690 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 691 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 692 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 693 #endif 694 695 /* Max physical block we can address w/o extents */ 696 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 697 698 /* Max logical block we can support */ 699 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF 700 701 /* 702 * Structure of an inode on the disk 703 */ 704 struct ext4_inode { 705 __le16 i_mode; /* File mode */ 706 __le16 i_uid; /* Low 16 bits of Owner Uid */ 707 __le32 i_size_lo; /* Size in bytes */ 708 __le32 i_atime; /* Access time */ 709 __le32 i_ctime; /* Inode Change time */ 710 __le32 i_mtime; /* Modification time */ 711 __le32 i_dtime; /* Deletion Time */ 712 __le16 i_gid; /* Low 16 bits of Group Id */ 713 __le16 i_links_count; /* Links count */ 714 __le32 i_blocks_lo; /* Blocks count */ 715 __le32 i_flags; /* File flags */ 716 union { 717 struct { 718 __le32 l_i_version; 719 } linux1; 720 struct { 721 __u32 h_i_translator; 722 } hurd1; 723 struct { 724 __u32 m_i_reserved1; 725 } masix1; 726 } osd1; /* OS dependent 1 */ 727 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 728 __le32 i_generation; /* File version (for NFS) */ 729 __le32 i_file_acl_lo; /* File ACL */ 730 __le32 i_size_high; 731 __le32 i_obso_faddr; /* Obsoleted fragment address */ 732 union { 733 struct { 734 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 735 __le16 l_i_file_acl_high; 736 __le16 l_i_uid_high; /* these 2 fields */ 737 __le16 l_i_gid_high; /* were reserved2[0] */ 738 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 739 __le16 l_i_reserved; 740 } linux2; 741 struct { 742 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 743 __u16 h_i_mode_high; 744 __u16 h_i_uid_high; 745 __u16 h_i_gid_high; 746 __u32 h_i_author; 747 } hurd2; 748 struct { 749 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 750 __le16 m_i_file_acl_high; 751 __u32 m_i_reserved2[2]; 752 } masix2; 753 } osd2; /* OS dependent 2 */ 754 __le16 i_extra_isize; 755 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 756 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 757 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 758 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 759 __le32 i_crtime; /* File Creation time */ 760 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 761 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 762 __le32 i_projid; /* Project ID */ 763 }; 764 765 struct move_extent { 766 __u32 reserved; /* should be zero */ 767 __u32 donor_fd; /* donor file descriptor */ 768 __u64 orig_start; /* logical start offset in block for orig */ 769 __u64 donor_start; /* logical start offset in block for donor */ 770 __u64 len; /* block length to be moved */ 771 __u64 moved_len; /* moved block length */ 772 }; 773 774 #define EXT4_EPOCH_BITS 2 775 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 776 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 777 778 /* 779 * Extended fields will fit into an inode if the filesystem was formatted 780 * with large inodes (-I 256 or larger) and there are not currently any EAs 781 * consuming all of the available space. For new inodes we always reserve 782 * enough space for the kernel's known extended fields, but for inodes 783 * created with an old kernel this might not have been the case. None of 784 * the extended inode fields is critical for correct filesystem operation. 785 * This macro checks if a certain field fits in the inode. Note that 786 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 787 */ 788 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 789 ((offsetof(typeof(*ext4_inode), field) + \ 790 sizeof((ext4_inode)->field)) \ 791 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 792 (einode)->i_extra_isize)) \ 793 794 /* 795 * We use an encoding that preserves the times for extra epoch "00": 796 * 797 * extra msb of adjust for signed 798 * epoch 32-bit 32-bit tv_sec to 799 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 800 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 801 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 802 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 803 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 804 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 805 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 806 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 807 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 808 * 809 * Note that previous versions of the kernel on 64-bit systems would 810 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 811 * 1970. e2fsck will correct this, assuming that it is run on the 812 * affected filesystem before 2242. 813 */ 814 815 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 816 { 817 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 818 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 819 } 820 821 static inline void ext4_decode_extra_time(struct timespec64 *time, 822 __le32 extra) 823 { 824 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) { 825 826 #if 1 827 /* Handle legacy encoding of pre-1970 dates with epoch 828 * bits 1,1. (This backwards compatibility may be removed 829 * at the discretion of the ext4 developers.) 830 */ 831 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 832 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 833 extra_bits = 0; 834 time->tv_sec += extra_bits << 32; 835 #else 836 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 837 #endif 838 } 839 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 840 } 841 842 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 843 do { \ 844 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 845 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 846 (raw_inode)->xtime ## _extra = \ 847 ext4_encode_extra_time(&(inode)->xtime); \ 848 } \ 849 } while (0) 850 851 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 852 do { \ 853 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 854 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 855 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 856 (raw_inode)->xtime ## _extra = \ 857 ext4_encode_extra_time(&(einode)->xtime); \ 858 } while (0) 859 860 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 861 do { \ 862 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 863 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 864 ext4_decode_extra_time(&(inode)->xtime, \ 865 raw_inode->xtime ## _extra); \ 866 } \ 867 else \ 868 (inode)->xtime.tv_nsec = 0; \ 869 } while (0) 870 871 872 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 873 do { \ 874 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 875 (einode)->xtime.tv_sec = \ 876 (signed)le32_to_cpu((raw_inode)->xtime); \ 877 else \ 878 (einode)->xtime.tv_sec = 0; \ 879 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 880 ext4_decode_extra_time(&(einode)->xtime, \ 881 raw_inode->xtime ## _extra); \ 882 else \ 883 (einode)->xtime.tv_nsec = 0; \ 884 } while (0) 885 886 #define i_disk_version osd1.linux1.l_i_version 887 888 #if defined(__KERNEL__) || defined(__linux__) 889 #define i_reserved1 osd1.linux1.l_i_reserved1 890 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 891 #define i_blocks_high osd2.linux2.l_i_blocks_high 892 #define i_uid_low i_uid 893 #define i_gid_low i_gid 894 #define i_uid_high osd2.linux2.l_i_uid_high 895 #define i_gid_high osd2.linux2.l_i_gid_high 896 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 897 898 #elif defined(__GNU__) 899 900 #define i_translator osd1.hurd1.h_i_translator 901 #define i_uid_high osd2.hurd2.h_i_uid_high 902 #define i_gid_high osd2.hurd2.h_i_gid_high 903 #define i_author osd2.hurd2.h_i_author 904 905 #elif defined(__masix__) 906 907 #define i_reserved1 osd1.masix1.m_i_reserved1 908 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 909 #define i_reserved2 osd2.masix2.m_i_reserved2 910 911 #endif /* defined(__KERNEL__) || defined(__linux__) */ 912 913 #include "extents_status.h" 914 915 /* 916 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 917 * 918 * These are needed to avoid lockdep false positives when we need to 919 * allocate blocks to the quota inode during ext4_map_blocks(), while 920 * holding i_data_sem for a normal (non-quota) inode. Since we don't 921 * do quota tracking for the quota inode, this avoids deadlock (as 922 * well as infinite recursion, since it isn't turtles all the way 923 * down...) 924 * 925 * I_DATA_SEM_NORMAL - Used for most inodes 926 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 927 * where the second inode has larger inode number 928 * than the first 929 * I_DATA_SEM_QUOTA - Used for quota inodes only 930 */ 931 enum { 932 I_DATA_SEM_NORMAL = 0, 933 I_DATA_SEM_OTHER, 934 I_DATA_SEM_QUOTA, 935 }; 936 937 938 /* 939 * fourth extended file system inode data in memory 940 */ 941 struct ext4_inode_info { 942 __le32 i_data[15]; /* unconverted */ 943 __u32 i_dtime; 944 ext4_fsblk_t i_file_acl; 945 946 /* 947 * i_block_group is the number of the block group which contains 948 * this file's inode. Constant across the lifetime of the inode, 949 * it is used for making block allocation decisions - we try to 950 * place a file's data blocks near its inode block, and new inodes 951 * near to their parent directory's inode. 952 */ 953 ext4_group_t i_block_group; 954 ext4_lblk_t i_dir_start_lookup; 955 #if (BITS_PER_LONG < 64) 956 unsigned long i_state_flags; /* Dynamic state flags */ 957 #endif 958 unsigned long i_flags; 959 960 /* 961 * Extended attributes can be read independently of the main file 962 * data. Taking i_mutex even when reading would cause contention 963 * between readers of EAs and writers of regular file data, so 964 * instead we synchronize on xattr_sem when reading or changing 965 * EAs. 966 */ 967 struct rw_semaphore xattr_sem; 968 969 struct list_head i_orphan; /* unlinked but open inodes */ 970 971 /* 972 * i_disksize keeps track of what the inode size is ON DISK, not 973 * in memory. During truncate, i_size is set to the new size by 974 * the VFS prior to calling ext4_truncate(), but the filesystem won't 975 * set i_disksize to 0 until the truncate is actually under way. 976 * 977 * The intent is that i_disksize always represents the blocks which 978 * are used by this file. This allows recovery to restart truncate 979 * on orphans if we crash during truncate. We actually write i_disksize 980 * into the on-disk inode when writing inodes out, instead of i_size. 981 * 982 * The only time when i_disksize and i_size may be different is when 983 * a truncate is in progress. The only things which change i_disksize 984 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 985 */ 986 loff_t i_disksize; 987 988 /* 989 * i_data_sem is for serialising ext4_truncate() against 990 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 991 * data tree are chopped off during truncate. We can't do that in 992 * ext4 because whenever we perform intermediate commits during 993 * truncate, the inode and all the metadata blocks *must* be in a 994 * consistent state which allows truncation of the orphans to restart 995 * during recovery. Hence we must fix the get_block-vs-truncate race 996 * by other means, so we have i_data_sem. 997 */ 998 struct rw_semaphore i_data_sem; 999 /* 1000 * i_mmap_sem is for serializing page faults with truncate / punch hole 1001 * operations. We have to make sure that new page cannot be faulted in 1002 * a section of the inode that is being punched. We cannot easily use 1003 * i_data_sem for this since we need protection for the whole punch 1004 * operation and i_data_sem ranks below transaction start so we have 1005 * to occasionally drop it. 1006 */ 1007 struct rw_semaphore i_mmap_sem; 1008 struct inode vfs_inode; 1009 struct jbd2_inode *jinode; 1010 1011 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1012 1013 /* 1014 * File creation time. Its function is same as that of 1015 * struct timespec64 i_{a,c,m}time in the generic inode. 1016 */ 1017 struct timespec64 i_crtime; 1018 1019 /* mballoc */ 1020 struct list_head i_prealloc_list; 1021 spinlock_t i_prealloc_lock; 1022 1023 /* extents status tree */ 1024 struct ext4_es_tree i_es_tree; 1025 rwlock_t i_es_lock; 1026 struct list_head i_es_list; 1027 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1028 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1029 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1030 extents to shrink. Protected by 1031 i_es_lock */ 1032 1033 /* ialloc */ 1034 ext4_group_t i_last_alloc_group; 1035 1036 /* allocation reservation info for delalloc */ 1037 /* In case of bigalloc, this refer to clusters rather than blocks */ 1038 unsigned int i_reserved_data_blocks; 1039 ext4_lblk_t i_da_metadata_calc_last_lblock; 1040 int i_da_metadata_calc_len; 1041 1042 /* pending cluster reservations for bigalloc file systems */ 1043 struct ext4_pending_tree i_pending_tree; 1044 1045 /* on-disk additional length */ 1046 __u16 i_extra_isize; 1047 1048 /* Indicate the inline data space. */ 1049 u16 i_inline_off; 1050 u16 i_inline_size; 1051 1052 #ifdef CONFIG_QUOTA 1053 /* quota space reservation, managed internally by quota code */ 1054 qsize_t i_reserved_quota; 1055 #endif 1056 1057 /* Lock protecting lists below */ 1058 spinlock_t i_completed_io_lock; 1059 /* 1060 * Completed IOs that need unwritten extents handling and have 1061 * transaction reserved 1062 */ 1063 struct list_head i_rsv_conversion_list; 1064 struct work_struct i_rsv_conversion_work; 1065 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1066 1067 spinlock_t i_block_reservation_lock; 1068 1069 /* 1070 * Transactions that contain inode's metadata needed to complete 1071 * fsync and fdatasync, respectively. 1072 */ 1073 tid_t i_sync_tid; 1074 tid_t i_datasync_tid; 1075 1076 #ifdef CONFIG_QUOTA 1077 struct dquot *i_dquot[MAXQUOTAS]; 1078 #endif 1079 1080 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1081 __u32 i_csum_seed; 1082 1083 kprojid_t i_projid; 1084 }; 1085 1086 /* 1087 * File system states 1088 */ 1089 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1090 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1091 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1092 1093 /* 1094 * Misc. filesystem flags 1095 */ 1096 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1097 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1098 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1099 1100 /* 1101 * Mount flags set via mount options or defaults 1102 */ 1103 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1104 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1105 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1106 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1107 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1108 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1109 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1110 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1111 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1112 #ifdef CONFIG_FS_DAX 1113 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1114 #else 1115 #define EXT4_MOUNT_DAX 0 1116 #endif 1117 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1118 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1119 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1120 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1121 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1122 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1123 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1124 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1125 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1126 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1127 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1128 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1129 * enable enforcement for hidden 1130 * quota files */ 1131 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1132 * enforcement for hidden quota 1133 * files */ 1134 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1135 * enforcement */ 1136 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1137 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1138 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1139 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1140 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1141 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1142 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1143 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1144 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1145 1146 /* 1147 * Mount flags set either automatically (could not be set by mount option) 1148 * based on per file system feature or property or in special cases such as 1149 * distinguishing between explicit mount option definition and default. 1150 */ 1151 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1152 specified delalloc */ 1153 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1154 size of blocksize * 8 1155 blocks */ 1156 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1157 file systems */ 1158 1159 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1160 specified journal checksum */ 1161 1162 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1163 ~EXT4_MOUNT_##opt 1164 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1165 EXT4_MOUNT_##opt 1166 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1167 EXT4_MOUNT_##opt) 1168 1169 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1170 ~EXT4_MOUNT2_##opt 1171 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1172 EXT4_MOUNT2_##opt 1173 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1174 EXT4_MOUNT2_##opt) 1175 1176 #define ext4_test_and_set_bit __test_and_set_bit_le 1177 #define ext4_set_bit __set_bit_le 1178 #define ext4_set_bit_atomic ext2_set_bit_atomic 1179 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1180 #define ext4_clear_bit __clear_bit_le 1181 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1182 #define ext4_test_bit test_bit_le 1183 #define ext4_find_next_zero_bit find_next_zero_bit_le 1184 #define ext4_find_next_bit find_next_bit_le 1185 1186 extern void ext4_set_bits(void *bm, int cur, int len); 1187 1188 /* 1189 * Maximal mount counts between two filesystem checks 1190 */ 1191 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1192 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1193 1194 /* 1195 * Behaviour when detecting errors 1196 */ 1197 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1198 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1199 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1200 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1201 1202 /* Metadata checksum algorithm codes */ 1203 #define EXT4_CRC32C_CHKSUM 1 1204 1205 /* 1206 * Structure of the super block 1207 */ 1208 struct ext4_super_block { 1209 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1210 __le32 s_blocks_count_lo; /* Blocks count */ 1211 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1212 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1213 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1214 __le32 s_first_data_block; /* First Data Block */ 1215 __le32 s_log_block_size; /* Block size */ 1216 __le32 s_log_cluster_size; /* Allocation cluster size */ 1217 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1218 __le32 s_clusters_per_group; /* # Clusters per group */ 1219 __le32 s_inodes_per_group; /* # Inodes per group */ 1220 __le32 s_mtime; /* Mount time */ 1221 /*30*/ __le32 s_wtime; /* Write time */ 1222 __le16 s_mnt_count; /* Mount count */ 1223 __le16 s_max_mnt_count; /* Maximal mount count */ 1224 __le16 s_magic; /* Magic signature */ 1225 __le16 s_state; /* File system state */ 1226 __le16 s_errors; /* Behaviour when detecting errors */ 1227 __le16 s_minor_rev_level; /* minor revision level */ 1228 /*40*/ __le32 s_lastcheck; /* time of last check */ 1229 __le32 s_checkinterval; /* max. time between checks */ 1230 __le32 s_creator_os; /* OS */ 1231 __le32 s_rev_level; /* Revision level */ 1232 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1233 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1234 /* 1235 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1236 * 1237 * Note: the difference between the compatible feature set and 1238 * the incompatible feature set is that if there is a bit set 1239 * in the incompatible feature set that the kernel doesn't 1240 * know about, it should refuse to mount the filesystem. 1241 * 1242 * e2fsck's requirements are more strict; if it doesn't know 1243 * about a feature in either the compatible or incompatible 1244 * feature set, it must abort and not try to meddle with 1245 * things it doesn't understand... 1246 */ 1247 __le32 s_first_ino; /* First non-reserved inode */ 1248 __le16 s_inode_size; /* size of inode structure */ 1249 __le16 s_block_group_nr; /* block group # of this superblock */ 1250 __le32 s_feature_compat; /* compatible feature set */ 1251 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1252 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1253 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1254 /*78*/ char s_volume_name[16]; /* volume name */ 1255 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1256 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1257 /* 1258 * Performance hints. Directory preallocation should only 1259 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1260 */ 1261 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1262 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1263 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1264 /* 1265 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1266 */ 1267 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1268 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1269 __le32 s_journal_dev; /* device number of journal file */ 1270 __le32 s_last_orphan; /* start of list of inodes to delete */ 1271 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1272 __u8 s_def_hash_version; /* Default hash version to use */ 1273 __u8 s_jnl_backup_type; 1274 __le16 s_desc_size; /* size of group descriptor */ 1275 /*100*/ __le32 s_default_mount_opts; 1276 __le32 s_first_meta_bg; /* First metablock block group */ 1277 __le32 s_mkfs_time; /* When the filesystem was created */ 1278 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1279 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1280 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1281 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1282 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1283 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1284 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1285 __le32 s_flags; /* Miscellaneous flags */ 1286 __le16 s_raid_stride; /* RAID stride */ 1287 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1288 __le64 s_mmp_block; /* Block for multi-mount protection */ 1289 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1290 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1291 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1292 __u8 s_encryption_level; /* versioning level for encryption */ 1293 __u8 s_reserved_pad; /* Padding to next 32bits */ 1294 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1295 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1296 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1297 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1298 snapshot's future use */ 1299 __le32 s_snapshot_list; /* inode number of the head of the 1300 on-disk snapshot list */ 1301 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1302 __le32 s_error_count; /* number of fs errors */ 1303 __le32 s_first_error_time; /* first time an error happened */ 1304 __le32 s_first_error_ino; /* inode involved in first error */ 1305 __le64 s_first_error_block; /* block involved of first error */ 1306 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1307 __le32 s_first_error_line; /* line number where error happened */ 1308 __le32 s_last_error_time; /* most recent time of an error */ 1309 __le32 s_last_error_ino; /* inode involved in last error */ 1310 __le32 s_last_error_line; /* line number where error happened */ 1311 __le64 s_last_error_block; /* block involved of last error */ 1312 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1313 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1314 __u8 s_mount_opts[64]; 1315 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1316 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1317 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1318 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1319 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1320 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1321 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1322 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1323 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1324 __u8 s_wtime_hi; 1325 __u8 s_mtime_hi; 1326 __u8 s_mkfs_time_hi; 1327 __u8 s_lastcheck_hi; 1328 __u8 s_first_error_time_hi; 1329 __u8 s_last_error_time_hi; 1330 __u8 s_pad[2]; 1331 __le16 s_encoding; /* Filename charset encoding */ 1332 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1333 __le32 s_reserved[95]; /* Padding to the end of the block */ 1334 __le32 s_checksum; /* crc32c(superblock) */ 1335 }; 1336 1337 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1338 1339 #ifdef __KERNEL__ 1340 1341 /* 1342 * run-time mount flags 1343 */ 1344 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1345 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1346 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1347 1348 #ifdef CONFIG_FS_ENCRYPTION 1349 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1350 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1351 #else 1352 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1353 #endif 1354 1355 /* Number of quota types we support */ 1356 #define EXT4_MAXQUOTAS 3 1357 1358 #define EXT4_ENC_UTF8_12_1 1 1359 1360 /* 1361 * Flags for ext4_sb_info.s_encoding_flags. 1362 */ 1363 #define EXT4_ENC_STRICT_MODE_FL (1 << 0) 1364 1365 #define ext4_has_strict_mode(sbi) \ 1366 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL) 1367 1368 /* 1369 * fourth extended-fs super-block data in memory 1370 */ 1371 struct ext4_sb_info { 1372 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1373 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1374 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1375 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1376 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1377 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1378 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1379 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1380 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1381 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1382 unsigned long s_overhead; /* # of fs overhead clusters */ 1383 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1384 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1385 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1386 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1387 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1388 struct buffer_head **s_group_desc; 1389 unsigned int s_mount_opt; 1390 unsigned int s_mount_opt2; 1391 unsigned int s_mount_flags; 1392 unsigned int s_def_mount_opt; 1393 ext4_fsblk_t s_sb_block; 1394 atomic64_t s_resv_clusters; 1395 kuid_t s_resuid; 1396 kgid_t s_resgid; 1397 unsigned short s_mount_state; 1398 unsigned short s_pad; 1399 int s_addr_per_block_bits; 1400 int s_desc_per_block_bits; 1401 int s_inode_size; 1402 int s_first_ino; 1403 unsigned int s_inode_readahead_blks; 1404 unsigned int s_inode_goal; 1405 u32 s_hash_seed[4]; 1406 int s_def_hash_version; 1407 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1408 struct percpu_counter s_freeclusters_counter; 1409 struct percpu_counter s_freeinodes_counter; 1410 struct percpu_counter s_dirs_counter; 1411 struct percpu_counter s_dirtyclusters_counter; 1412 struct blockgroup_lock *s_blockgroup_lock; 1413 struct proc_dir_entry *s_proc; 1414 struct kobject s_kobj; 1415 struct completion s_kobj_unregister; 1416 struct super_block *s_sb; 1417 #ifdef CONFIG_UNICODE 1418 struct unicode_map *s_encoding; 1419 __u16 s_encoding_flags; 1420 #endif 1421 1422 /* Journaling */ 1423 struct journal_s *s_journal; 1424 struct list_head s_orphan; 1425 struct mutex s_orphan_lock; 1426 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1427 unsigned long s_commit_interval; 1428 u32 s_max_batch_time; 1429 u32 s_min_batch_time; 1430 struct block_device *journal_bdev; 1431 #ifdef CONFIG_QUOTA 1432 /* Names of quota files with journalled quota */ 1433 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1434 int s_jquota_fmt; /* Format of quota to use */ 1435 #endif 1436 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1437 struct rb_root system_blks; 1438 1439 #ifdef EXTENTS_STATS 1440 /* ext4 extents stats */ 1441 unsigned long s_ext_min; 1442 unsigned long s_ext_max; 1443 unsigned long s_depth_max; 1444 spinlock_t s_ext_stats_lock; 1445 unsigned long s_ext_blocks; 1446 unsigned long s_ext_extents; 1447 #endif 1448 1449 /* for buddy allocator */ 1450 struct ext4_group_info ***s_group_info; 1451 struct inode *s_buddy_cache; 1452 spinlock_t s_md_lock; 1453 unsigned short *s_mb_offsets; 1454 unsigned int *s_mb_maxs; 1455 unsigned int s_group_info_size; 1456 unsigned int s_mb_free_pending; 1457 struct list_head s_freed_data_list; /* List of blocks to be freed 1458 after commit completed */ 1459 1460 /* tunables */ 1461 unsigned long s_stripe; 1462 unsigned int s_mb_stream_request; 1463 unsigned int s_mb_max_to_scan; 1464 unsigned int s_mb_min_to_scan; 1465 unsigned int s_mb_stats; 1466 unsigned int s_mb_order2_reqs; 1467 unsigned int s_mb_group_prealloc; 1468 unsigned int s_max_dir_size_kb; 1469 /* where last allocation was done - for stream allocation */ 1470 unsigned long s_mb_last_group; 1471 unsigned long s_mb_last_start; 1472 1473 /* stats for buddy allocator */ 1474 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1475 atomic_t s_bal_success; /* we found long enough chunks */ 1476 atomic_t s_bal_allocated; /* in blocks */ 1477 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1478 atomic_t s_bal_goals; /* goal hits */ 1479 atomic_t s_bal_breaks; /* too long searches */ 1480 atomic_t s_bal_2orders; /* 2^order hits */ 1481 spinlock_t s_bal_lock; 1482 unsigned long s_mb_buddies_generated; 1483 unsigned long long s_mb_generation_time; 1484 atomic_t s_mb_lost_chunks; 1485 atomic_t s_mb_preallocated; 1486 atomic_t s_mb_discarded; 1487 atomic_t s_lock_busy; 1488 1489 /* locality groups */ 1490 struct ext4_locality_group __percpu *s_locality_groups; 1491 1492 /* for write statistics */ 1493 unsigned long s_sectors_written_start; 1494 u64 s_kbytes_written; 1495 1496 /* the size of zero-out chunk */ 1497 unsigned int s_extent_max_zeroout_kb; 1498 1499 unsigned int s_log_groups_per_flex; 1500 struct flex_groups *s_flex_groups; 1501 ext4_group_t s_flex_groups_allocated; 1502 1503 /* workqueue for reserved extent conversions (buffered io) */ 1504 struct workqueue_struct *rsv_conversion_wq; 1505 1506 /* timer for periodic error stats printing */ 1507 struct timer_list s_err_report; 1508 1509 /* Lazy inode table initialization info */ 1510 struct ext4_li_request *s_li_request; 1511 /* Wait multiplier for lazy initialization thread */ 1512 unsigned int s_li_wait_mult; 1513 1514 /* Kernel thread for multiple mount protection */ 1515 struct task_struct *s_mmp_tsk; 1516 1517 /* record the last minlen when FITRIM is called. */ 1518 atomic_t s_last_trim_minblks; 1519 1520 /* Reference to checksum algorithm driver via cryptoapi */ 1521 struct crypto_shash *s_chksum_driver; 1522 1523 /* Precomputed FS UUID checksum for seeding other checksums */ 1524 __u32 s_csum_seed; 1525 1526 /* Reclaim extents from extent status tree */ 1527 struct shrinker s_es_shrinker; 1528 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1529 long s_es_nr_inode; 1530 struct ext4_es_stats s_es_stats; 1531 struct mb_cache *s_ea_block_cache; 1532 struct mb_cache *s_ea_inode_cache; 1533 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1534 1535 /* Ratelimit ext4 messages. */ 1536 struct ratelimit_state s_err_ratelimit_state; 1537 struct ratelimit_state s_warning_ratelimit_state; 1538 struct ratelimit_state s_msg_ratelimit_state; 1539 1540 /* Barrier between changing inodes' journal flags and writepages ops. */ 1541 struct percpu_rw_semaphore s_journal_flag_rwsem; 1542 struct dax_device *s_daxdev; 1543 }; 1544 1545 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1546 { 1547 return sb->s_fs_info; 1548 } 1549 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1550 { 1551 return container_of(inode, struct ext4_inode_info, vfs_inode); 1552 } 1553 1554 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1555 { 1556 return ino == EXT4_ROOT_INO || 1557 (ino >= EXT4_FIRST_INO(sb) && 1558 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1559 } 1560 1561 /* 1562 * Inode dynamic state flags 1563 */ 1564 enum { 1565 EXT4_STATE_JDATA, /* journaled data exists */ 1566 EXT4_STATE_NEW, /* inode is newly created */ 1567 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1568 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1569 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1570 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1571 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1572 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1573 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1574 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1575 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1576 }; 1577 1578 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1579 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1580 { \ 1581 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1582 } \ 1583 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1584 { \ 1585 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1586 } \ 1587 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1588 { \ 1589 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1590 } 1591 1592 /* Add these declarations here only so that these functions can be 1593 * found by name. Otherwise, they are very hard to locate. */ 1594 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1595 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1596 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1597 EXT4_INODE_BIT_FNS(flag, flags, 0) 1598 1599 /* Add these declarations here only so that these functions can be 1600 * found by name. Otherwise, they are very hard to locate. */ 1601 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1602 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1603 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1604 #if (BITS_PER_LONG < 64) 1605 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1606 1607 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1608 { 1609 (ei)->i_state_flags = 0; 1610 } 1611 #else 1612 EXT4_INODE_BIT_FNS(state, flags, 32) 1613 1614 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1615 { 1616 /* We depend on the fact that callers will set i_flags */ 1617 } 1618 #endif 1619 #else 1620 /* Assume that user mode programs are passing in an ext4fs superblock, not 1621 * a kernel struct super_block. This will allow us to call the feature-test 1622 * macros from user land. */ 1623 #define EXT4_SB(sb) (sb) 1624 #endif 1625 1626 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1627 1628 /* 1629 * Codes for operating systems 1630 */ 1631 #define EXT4_OS_LINUX 0 1632 #define EXT4_OS_HURD 1 1633 #define EXT4_OS_MASIX 2 1634 #define EXT4_OS_FREEBSD 3 1635 #define EXT4_OS_LITES 4 1636 1637 /* 1638 * Revision levels 1639 */ 1640 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1641 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1642 1643 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1644 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1645 1646 #define EXT4_GOOD_OLD_INODE_SIZE 128 1647 1648 /* 1649 * Feature set definitions 1650 */ 1651 1652 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1653 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1654 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1655 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1656 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1657 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1658 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1659 1660 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1661 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1662 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1663 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1664 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1665 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1666 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1667 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1668 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1669 /* 1670 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1671 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1672 * all other data structures' checksums. However, the METADATA_CSUM and 1673 * GDT_CSUM bits are mutually exclusive. 1674 */ 1675 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1676 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1677 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1678 1679 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1680 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1681 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1682 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1683 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1684 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1685 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1686 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1687 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1688 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1689 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1690 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1691 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1692 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1693 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1694 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1695 1696 extern void ext4_update_dynamic_rev(struct super_block *sb); 1697 1698 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1699 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1700 { \ 1701 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1702 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1703 } \ 1704 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1705 { \ 1706 ext4_update_dynamic_rev(sb); \ 1707 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1708 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1709 } \ 1710 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1711 { \ 1712 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1713 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1714 } 1715 1716 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1717 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1718 { \ 1719 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1720 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1721 } \ 1722 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1723 { \ 1724 ext4_update_dynamic_rev(sb); \ 1725 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1726 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1727 } \ 1728 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1729 { \ 1730 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1731 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1732 } 1733 1734 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1735 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1736 { \ 1737 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1738 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1739 } \ 1740 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1741 { \ 1742 ext4_update_dynamic_rev(sb); \ 1743 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1744 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1745 } \ 1746 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1747 { \ 1748 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1749 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1750 } 1751 1752 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1753 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1754 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1755 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1756 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1757 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1758 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1759 1760 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1761 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1762 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1763 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1764 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1765 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1766 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1767 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1768 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1769 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1770 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1771 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 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 1831 #define EXTN_FEATURE_FUNCS(ver) \ 1832 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1833 { \ 1834 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1835 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1836 } \ 1837 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1838 { \ 1839 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1840 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1841 } \ 1842 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1843 { \ 1844 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1845 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1846 } 1847 1848 EXTN_FEATURE_FUNCS(2) 1849 EXTN_FEATURE_FUNCS(3) 1850 EXTN_FEATURE_FUNCS(4) 1851 1852 static inline bool ext4_has_compat_features(struct super_block *sb) 1853 { 1854 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1855 } 1856 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1857 { 1858 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1859 } 1860 static inline bool ext4_has_incompat_features(struct super_block *sb) 1861 { 1862 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1863 } 1864 1865 /* 1866 * Superblock flags 1867 */ 1868 #define EXT4_FLAGS_RESIZING 0 1869 #define EXT4_FLAGS_SHUTDOWN 1 1870 1871 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1872 { 1873 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1874 } 1875 1876 1877 /* 1878 * Default values for user and/or group using reserved blocks 1879 */ 1880 #define EXT4_DEF_RESUID 0 1881 #define EXT4_DEF_RESGID 0 1882 1883 /* 1884 * Default project ID 1885 */ 1886 #define EXT4_DEF_PROJID 0 1887 1888 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1889 1890 /* 1891 * Default mount options 1892 */ 1893 #define EXT4_DEFM_DEBUG 0x0001 1894 #define EXT4_DEFM_BSDGROUPS 0x0002 1895 #define EXT4_DEFM_XATTR_USER 0x0004 1896 #define EXT4_DEFM_ACL 0x0008 1897 #define EXT4_DEFM_UID16 0x0010 1898 #define EXT4_DEFM_JMODE 0x0060 1899 #define EXT4_DEFM_JMODE_DATA 0x0020 1900 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1901 #define EXT4_DEFM_JMODE_WBACK 0x0060 1902 #define EXT4_DEFM_NOBARRIER 0x0100 1903 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1904 #define EXT4_DEFM_DISCARD 0x0400 1905 #define EXT4_DEFM_NODELALLOC 0x0800 1906 1907 /* 1908 * Default journal batch times 1909 */ 1910 #define EXT4_DEF_MIN_BATCH_TIME 0 1911 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1912 1913 /* 1914 * Minimum number of groups in a flexgroup before we separate out 1915 * directories into the first block group of a flexgroup 1916 */ 1917 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1918 1919 /* 1920 * Structure of a directory entry 1921 */ 1922 #define EXT4_NAME_LEN 255 1923 1924 struct ext4_dir_entry { 1925 __le32 inode; /* Inode number */ 1926 __le16 rec_len; /* Directory entry length */ 1927 __le16 name_len; /* Name length */ 1928 char name[EXT4_NAME_LEN]; /* File name */ 1929 }; 1930 1931 /* 1932 * The new version of the directory entry. Since EXT4 structures are 1933 * stored in intel byte order, and the name_len field could never be 1934 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1935 * file_type field. 1936 */ 1937 struct ext4_dir_entry_2 { 1938 __le32 inode; /* Inode number */ 1939 __le16 rec_len; /* Directory entry length */ 1940 __u8 name_len; /* Name length */ 1941 __u8 file_type; 1942 char name[EXT4_NAME_LEN]; /* File name */ 1943 }; 1944 1945 /* 1946 * This is a bogus directory entry at the end of each leaf block that 1947 * records checksums. 1948 */ 1949 struct ext4_dir_entry_tail { 1950 __le32 det_reserved_zero1; /* Pretend to be unused */ 1951 __le16 det_rec_len; /* 12 */ 1952 __u8 det_reserved_zero2; /* Zero name length */ 1953 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1954 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1955 }; 1956 1957 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1958 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1959 ((blocksize) - \ 1960 sizeof(struct ext4_dir_entry_tail)))) 1961 1962 /* 1963 * Ext4 directory file types. Only the low 3 bits are used. The 1964 * other bits are reserved for now. 1965 */ 1966 #define EXT4_FT_UNKNOWN 0 1967 #define EXT4_FT_REG_FILE 1 1968 #define EXT4_FT_DIR 2 1969 #define EXT4_FT_CHRDEV 3 1970 #define EXT4_FT_BLKDEV 4 1971 #define EXT4_FT_FIFO 5 1972 #define EXT4_FT_SOCK 6 1973 #define EXT4_FT_SYMLINK 7 1974 1975 #define EXT4_FT_MAX 8 1976 1977 #define EXT4_FT_DIR_CSUM 0xDE 1978 1979 /* 1980 * EXT4_DIR_PAD defines the directory entries boundaries 1981 * 1982 * NOTE: It must be a multiple of 4 1983 */ 1984 #define EXT4_DIR_PAD 4 1985 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1986 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1987 ~EXT4_DIR_ROUND) 1988 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1989 1990 /* 1991 * If we ever get support for fs block sizes > page_size, we'll need 1992 * to remove the #if statements in the next two functions... 1993 */ 1994 static inline unsigned int 1995 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1996 { 1997 unsigned len = le16_to_cpu(dlen); 1998 1999 #if (PAGE_SIZE >= 65536) 2000 if (len == EXT4_MAX_REC_LEN || len == 0) 2001 return blocksize; 2002 return (len & 65532) | ((len & 3) << 16); 2003 #else 2004 return len; 2005 #endif 2006 } 2007 2008 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2009 { 2010 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2011 BUG(); 2012 #if (PAGE_SIZE >= 65536) 2013 if (len < 65536) 2014 return cpu_to_le16(len); 2015 if (len == blocksize) { 2016 if (blocksize == 65536) 2017 return cpu_to_le16(EXT4_MAX_REC_LEN); 2018 else 2019 return cpu_to_le16(0); 2020 } 2021 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2022 #else 2023 return cpu_to_le16(len); 2024 #endif 2025 } 2026 2027 /* 2028 * Hash Tree Directory indexing 2029 * (c) Daniel Phillips, 2001 2030 */ 2031 2032 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2033 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2034 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2035 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2036 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2037 2038 /* Legal values for the dx_root hash_version field: */ 2039 2040 #define DX_HASH_LEGACY 0 2041 #define DX_HASH_HALF_MD4 1 2042 #define DX_HASH_TEA 2 2043 #define DX_HASH_LEGACY_UNSIGNED 3 2044 #define DX_HASH_HALF_MD4_UNSIGNED 4 2045 #define DX_HASH_TEA_UNSIGNED 5 2046 2047 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2048 const void *address, unsigned int length) 2049 { 2050 struct { 2051 struct shash_desc shash; 2052 char ctx[4]; 2053 } desc; 2054 2055 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2056 2057 desc.shash.tfm = sbi->s_chksum_driver; 2058 *(u32 *)desc.ctx = crc; 2059 2060 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2061 2062 return *(u32 *)desc.ctx; 2063 } 2064 2065 #ifdef __KERNEL__ 2066 2067 /* hash info structure used by the directory hash */ 2068 struct dx_hash_info 2069 { 2070 u32 hash; 2071 u32 minor_hash; 2072 int hash_version; 2073 u32 *seed; 2074 }; 2075 2076 2077 /* 32 and 64 bit signed EOF for dx directories */ 2078 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2079 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2080 2081 2082 /* 2083 * Control parameters used by ext4_htree_next_block 2084 */ 2085 #define HASH_NB_ALWAYS 1 2086 2087 struct ext4_filename { 2088 const struct qstr *usr_fname; 2089 struct fscrypt_str disk_name; 2090 struct dx_hash_info hinfo; 2091 #ifdef CONFIG_FS_ENCRYPTION 2092 struct fscrypt_str crypto_buf; 2093 #endif 2094 #ifdef CONFIG_UNICODE 2095 struct fscrypt_str cf_name; 2096 #endif 2097 }; 2098 2099 #define fname_name(p) ((p)->disk_name.name) 2100 #define fname_len(p) ((p)->disk_name.len) 2101 2102 /* 2103 * Describe an inode's exact location on disk and in memory 2104 */ 2105 struct ext4_iloc 2106 { 2107 struct buffer_head *bh; 2108 unsigned long offset; 2109 ext4_group_t block_group; 2110 }; 2111 2112 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2113 { 2114 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2115 } 2116 2117 static inline bool ext4_is_quota_file(struct inode *inode) 2118 { 2119 return IS_NOQUOTA(inode) && 2120 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2121 } 2122 2123 /* 2124 * This structure is stuffed into the struct file's private_data field 2125 * for directories. It is where we put information so that we can do 2126 * readdir operations in hash tree order. 2127 */ 2128 struct dir_private_info { 2129 struct rb_root root; 2130 struct rb_node *curr_node; 2131 struct fname *extra_fname; 2132 loff_t last_pos; 2133 __u32 curr_hash; 2134 __u32 curr_minor_hash; 2135 __u32 next_hash; 2136 }; 2137 2138 /* calculate the first block number of the group */ 2139 static inline ext4_fsblk_t 2140 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2141 { 2142 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2143 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2144 } 2145 2146 /* 2147 * Special error return code only used by dx_probe() and its callers. 2148 */ 2149 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2150 2151 /* htree levels for ext4 */ 2152 #define EXT4_HTREE_LEVEL_COMPAT 2 2153 #define EXT4_HTREE_LEVEL 3 2154 2155 static inline int ext4_dir_htree_level(struct super_block *sb) 2156 { 2157 return ext4_has_feature_largedir(sb) ? 2158 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2159 } 2160 2161 /* 2162 * Timeout and state flag for lazy initialization inode thread. 2163 */ 2164 #define EXT4_DEF_LI_WAIT_MULT 10 2165 #define EXT4_DEF_LI_MAX_START_DELAY 5 2166 #define EXT4_LAZYINIT_QUIT 0x0001 2167 #define EXT4_LAZYINIT_RUNNING 0x0002 2168 2169 /* 2170 * Lazy inode table initialization info 2171 */ 2172 struct ext4_lazy_init { 2173 unsigned long li_state; 2174 struct list_head li_request_list; 2175 struct mutex li_list_mtx; 2176 }; 2177 2178 struct ext4_li_request { 2179 struct super_block *lr_super; 2180 struct ext4_sb_info *lr_sbi; 2181 ext4_group_t lr_next_group; 2182 struct list_head lr_request; 2183 unsigned long lr_next_sched; 2184 unsigned long lr_timeout; 2185 }; 2186 2187 struct ext4_features { 2188 struct kobject f_kobj; 2189 struct completion f_kobj_unregister; 2190 }; 2191 2192 /* 2193 * This structure will be used for multiple mount protection. It will be 2194 * written into the block number saved in the s_mmp_block field in the 2195 * superblock. Programs that check MMP should assume that if 2196 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2197 * to use the filesystem, regardless of how old the timestamp is. 2198 */ 2199 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2200 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2201 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2202 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2203 2204 struct mmp_struct { 2205 __le32 mmp_magic; /* Magic number for MMP */ 2206 __le32 mmp_seq; /* Sequence no. updated periodically */ 2207 2208 /* 2209 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2210 * purposes and do not affect the correctness of the algorithm 2211 */ 2212 __le64 mmp_time; /* Time last updated */ 2213 char mmp_nodename[64]; /* Node which last updated MMP block */ 2214 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2215 2216 /* 2217 * mmp_check_interval is used to verify if the MMP block has been 2218 * updated on the block device. The value is updated based on the 2219 * maximum time to write the MMP block during an update cycle. 2220 */ 2221 __le16 mmp_check_interval; 2222 2223 __le16 mmp_pad1; 2224 __le32 mmp_pad2[226]; 2225 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2226 }; 2227 2228 /* arguments passed to the mmp thread */ 2229 struct mmpd_data { 2230 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2231 struct super_block *sb; /* super block of the fs */ 2232 }; 2233 2234 /* 2235 * Check interval multiplier 2236 * The MMP block is written every update interval and initially checked every 2237 * update interval x the multiplier (the value is then adapted based on the 2238 * write latency). The reason is that writes can be delayed under load and we 2239 * don't want readers to incorrectly assume that the filesystem is no longer 2240 * in use. 2241 */ 2242 #define EXT4_MMP_CHECK_MULT 2UL 2243 2244 /* 2245 * Minimum interval for MMP checking in seconds. 2246 */ 2247 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2248 2249 /* 2250 * Maximum interval for MMP checking in seconds. 2251 */ 2252 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2253 2254 /* 2255 * Function prototypes 2256 */ 2257 2258 /* 2259 * Ok, these declarations are also in <linux/kernel.h> but none of the 2260 * ext4 source programs needs to include it so they are duplicated here. 2261 */ 2262 # define NORET_TYPE /**/ 2263 # define ATTRIB_NORET __attribute__((noreturn)) 2264 # define NORET_AND noreturn, 2265 2266 /* bitmap.c */ 2267 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2268 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2269 struct ext4_group_desc *gdp, 2270 struct buffer_head *bh, int sz); 2271 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2272 struct ext4_group_desc *gdp, 2273 struct buffer_head *bh, int sz); 2274 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2275 struct ext4_group_desc *gdp, 2276 struct buffer_head *bh); 2277 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2278 struct ext4_group_desc *gdp, 2279 struct buffer_head *bh); 2280 2281 /* balloc.c */ 2282 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2283 ext4_fsblk_t blocknr, 2284 ext4_group_t *blockgrpp, 2285 ext4_grpblk_t *offsetp); 2286 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2287 ext4_fsblk_t block); 2288 2289 extern unsigned int ext4_block_group(struct super_block *sb, 2290 ext4_fsblk_t blocknr); 2291 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2292 ext4_fsblk_t blocknr); 2293 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2294 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2295 ext4_group_t group); 2296 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2297 ext4_fsblk_t goal, 2298 unsigned int flags, 2299 unsigned long *count, 2300 int *errp); 2301 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2302 s64 nclusters, unsigned int flags); 2303 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2304 extern void ext4_check_blocks_bitmap(struct super_block *); 2305 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2306 ext4_group_t block_group, 2307 struct buffer_head ** bh); 2308 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2309 2310 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2311 ext4_group_t block_group); 2312 extern int ext4_wait_block_bitmap(struct super_block *sb, 2313 ext4_group_t block_group, 2314 struct buffer_head *bh); 2315 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2316 ext4_group_t block_group); 2317 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2318 ext4_group_t block_group, 2319 struct ext4_group_desc *gdp); 2320 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2321 2322 #ifdef CONFIG_UNICODE 2323 extern void ext4_fname_setup_ci_filename(struct inode *dir, 2324 const struct qstr *iname, 2325 struct fscrypt_str *fname); 2326 #endif 2327 2328 #ifdef CONFIG_FS_ENCRYPTION 2329 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst, 2330 const struct fscrypt_name *src) 2331 { 2332 memset(dst, 0, sizeof(*dst)); 2333 2334 dst->usr_fname = src->usr_fname; 2335 dst->disk_name = src->disk_name; 2336 dst->hinfo.hash = src->hash; 2337 dst->hinfo.minor_hash = src->minor_hash; 2338 dst->crypto_buf = src->crypto_buf; 2339 } 2340 2341 static inline int ext4_fname_setup_filename(struct inode *dir, 2342 const struct qstr *iname, 2343 int lookup, 2344 struct ext4_filename *fname) 2345 { 2346 struct fscrypt_name name; 2347 int err; 2348 2349 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2350 if (err) 2351 return err; 2352 2353 ext4_fname_from_fscrypt_name(fname, &name); 2354 2355 #ifdef CONFIG_UNICODE 2356 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2357 #endif 2358 return 0; 2359 } 2360 2361 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2362 struct dentry *dentry, 2363 struct ext4_filename *fname) 2364 { 2365 struct fscrypt_name name; 2366 int err; 2367 2368 err = fscrypt_prepare_lookup(dir, dentry, &name); 2369 if (err) 2370 return err; 2371 2372 ext4_fname_from_fscrypt_name(fname, &name); 2373 2374 #ifdef CONFIG_UNICODE 2375 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name); 2376 #endif 2377 return 0; 2378 } 2379 2380 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2381 { 2382 struct fscrypt_name name; 2383 2384 name.crypto_buf = fname->crypto_buf; 2385 fscrypt_free_filename(&name); 2386 2387 fname->crypto_buf.name = NULL; 2388 fname->usr_fname = NULL; 2389 fname->disk_name.name = NULL; 2390 2391 #ifdef CONFIG_UNICODE 2392 kfree(fname->cf_name.name); 2393 fname->cf_name.name = NULL; 2394 #endif 2395 } 2396 #else /* !CONFIG_FS_ENCRYPTION */ 2397 static inline int ext4_fname_setup_filename(struct inode *dir, 2398 const struct qstr *iname, 2399 int lookup, 2400 struct ext4_filename *fname) 2401 { 2402 fname->usr_fname = iname; 2403 fname->disk_name.name = (unsigned char *) iname->name; 2404 fname->disk_name.len = iname->len; 2405 2406 #ifdef CONFIG_UNICODE 2407 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2408 #endif 2409 2410 return 0; 2411 } 2412 2413 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2414 struct dentry *dentry, 2415 struct ext4_filename *fname) 2416 { 2417 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2418 } 2419 2420 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2421 { 2422 #ifdef CONFIG_UNICODE 2423 kfree(fname->cf_name.name); 2424 fname->cf_name.name = NULL; 2425 #endif 2426 } 2427 #endif /* !CONFIG_FS_ENCRYPTION */ 2428 2429 /* dir.c */ 2430 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2431 struct file *, 2432 struct ext4_dir_entry_2 *, 2433 struct buffer_head *, char *, int, 2434 unsigned int); 2435 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2436 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2437 (de), (bh), (buf), (size), (offset))) 2438 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2439 __u32 minor_hash, 2440 struct ext4_dir_entry_2 *dirent, 2441 struct fscrypt_str *ent_name); 2442 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2443 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2444 struct buffer_head *bh, 2445 void *buf, int buf_size, 2446 struct ext4_filename *fname, 2447 struct ext4_dir_entry_2 **dest_de); 2448 void ext4_insert_dentry(struct inode *inode, 2449 struct ext4_dir_entry_2 *de, 2450 int buf_size, 2451 struct ext4_filename *fname); 2452 static inline void ext4_update_dx_flag(struct inode *inode) 2453 { 2454 if (!ext4_has_feature_dir_index(inode->i_sb)) 2455 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2456 } 2457 static const unsigned char ext4_filetype_table[] = { 2458 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2459 }; 2460 2461 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2462 { 2463 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2464 return DT_UNKNOWN; 2465 2466 return ext4_filetype_table[filetype]; 2467 } 2468 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2469 void *buf, int buf_size); 2470 2471 /* fsync.c */ 2472 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2473 2474 /* hash.c */ 2475 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2476 struct dx_hash_info *hinfo); 2477 2478 /* ialloc.c */ 2479 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2480 const struct qstr *qstr, __u32 goal, 2481 uid_t *owner, __u32 i_flags, 2482 int handle_type, unsigned int line_no, 2483 int nblocks); 2484 2485 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2486 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2487 i_flags, 0, 0, 0) 2488 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2489 type, nblocks) \ 2490 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2491 0, (type), __LINE__, (nblocks)) 2492 2493 2494 extern void ext4_free_inode(handle_t *, struct inode *); 2495 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2496 extern unsigned long ext4_count_free_inodes(struct super_block *); 2497 extern unsigned long ext4_count_dirs(struct super_block *); 2498 extern void ext4_check_inodes_bitmap(struct super_block *); 2499 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2500 extern int ext4_init_inode_table(struct super_block *sb, 2501 ext4_group_t group, int barrier); 2502 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2503 2504 /* mballoc.c */ 2505 extern const struct seq_operations ext4_mb_seq_groups_ops; 2506 extern long ext4_mb_stats; 2507 extern long ext4_mb_max_to_scan; 2508 extern int ext4_mb_init(struct super_block *); 2509 extern int ext4_mb_release(struct super_block *); 2510 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2511 struct ext4_allocation_request *, int *); 2512 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2513 extern void ext4_discard_preallocations(struct inode *); 2514 extern int __init ext4_init_mballoc(void); 2515 extern void ext4_exit_mballoc(void); 2516 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2517 struct buffer_head *bh, ext4_fsblk_t block, 2518 unsigned long count, int flags); 2519 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2520 ext4_group_t ngroups); 2521 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2522 ext4_group_t i, struct ext4_group_desc *desc); 2523 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2524 ext4_fsblk_t block, unsigned long count); 2525 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2526 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2527 2528 /* inode.c */ 2529 int ext4_inode_is_fast_symlink(struct inode *inode); 2530 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2531 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2532 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2533 bool wait, struct buffer_head **bhs); 2534 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2535 struct buffer_head *bh_result, int create); 2536 int ext4_get_block(struct inode *inode, sector_t iblock, 2537 struct buffer_head *bh_result, int create); 2538 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2539 struct buffer_head *bh_result, int create); 2540 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2541 struct buffer_head *bh, int create); 2542 int ext4_walk_page_buffers(handle_t *handle, 2543 struct buffer_head *head, 2544 unsigned from, 2545 unsigned to, 2546 int *partial, 2547 int (*fn)(handle_t *handle, 2548 struct buffer_head *bh)); 2549 int do_journal_get_write_access(handle_t *handle, 2550 struct buffer_head *bh); 2551 #define FALL_BACK_TO_NONDELALLOC 1 2552 #define CONVERT_INLINE_DATA 2 2553 2554 typedef enum { 2555 EXT4_IGET_NORMAL = 0, 2556 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2557 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2558 } ext4_iget_flags; 2559 2560 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2561 ext4_iget_flags flags, const char *function, 2562 unsigned int line); 2563 2564 #define ext4_iget(sb, ino, flags) \ 2565 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2566 2567 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2568 extern int ext4_setattr(struct dentry *, struct iattr *); 2569 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2570 extern void ext4_evict_inode(struct inode *); 2571 extern void ext4_clear_inode(struct inode *); 2572 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2573 extern int ext4_sync_inode(handle_t *, struct inode *); 2574 extern void ext4_dirty_inode(struct inode *, int); 2575 extern int ext4_change_inode_journal_flag(struct inode *, int); 2576 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2577 extern int ext4_inode_attach_jinode(struct inode *inode); 2578 extern int ext4_can_truncate(struct inode *inode); 2579 extern int ext4_truncate(struct inode *); 2580 extern int ext4_break_layouts(struct inode *); 2581 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2582 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2583 extern void ext4_set_inode_flags(struct inode *); 2584 extern int ext4_alloc_da_blocks(struct inode *inode); 2585 extern void ext4_set_aops(struct inode *inode); 2586 extern int ext4_writepage_trans_blocks(struct inode *); 2587 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2588 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2589 loff_t lstart, loff_t lend); 2590 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2591 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2592 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2593 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2594 extern void ext4_da_release_space(struct inode *inode, int to_free); 2595 extern void ext4_da_update_reserve_space(struct inode *inode, 2596 int used, int quota_claim); 2597 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2598 ext4_fsblk_t pblk, ext4_lblk_t len); 2599 2600 /* indirect.c */ 2601 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2602 struct ext4_map_blocks *map, int flags); 2603 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2604 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2605 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2606 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2607 ext4_lblk_t start, ext4_lblk_t end); 2608 2609 /* ioctl.c */ 2610 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2611 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2612 2613 /* migrate.c */ 2614 extern int ext4_ext_migrate(struct inode *); 2615 extern int ext4_ind_migrate(struct inode *inode); 2616 2617 /* namei.c */ 2618 extern int ext4_dirblock_csum_verify(struct inode *inode, 2619 struct buffer_head *bh); 2620 extern int ext4_orphan_add(handle_t *, struct inode *); 2621 extern int ext4_orphan_del(handle_t *, struct inode *); 2622 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2623 __u32 start_minor_hash, __u32 *next_hash); 2624 extern int ext4_search_dir(struct buffer_head *bh, 2625 char *search_buf, 2626 int buf_size, 2627 struct inode *dir, 2628 struct ext4_filename *fname, 2629 unsigned int offset, 2630 struct ext4_dir_entry_2 **res_dir); 2631 extern int ext4_generic_delete_entry(handle_t *handle, 2632 struct inode *dir, 2633 struct ext4_dir_entry_2 *de_del, 2634 struct buffer_head *bh, 2635 void *entry_buf, 2636 int buf_size, 2637 int csum_size); 2638 extern bool ext4_empty_dir(struct inode *inode); 2639 2640 /* resize.c */ 2641 extern int ext4_group_add(struct super_block *sb, 2642 struct ext4_new_group_data *input); 2643 extern int ext4_group_extend(struct super_block *sb, 2644 struct ext4_super_block *es, 2645 ext4_fsblk_t n_blocks_count); 2646 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2647 2648 /* super.c */ 2649 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2650 sector_t block, int op_flags); 2651 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2652 extern int ext4_calculate_overhead(struct super_block *sb); 2653 extern void ext4_superblock_csum_set(struct super_block *sb); 2654 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2655 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2656 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2657 ext4_group_t ngroup); 2658 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2659 char nbuf[16]); 2660 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2661 ext4_group_t block_group, 2662 unsigned int flags); 2663 2664 extern __printf(4, 5) 2665 void __ext4_error(struct super_block *, const char *, unsigned int, 2666 const char *, ...); 2667 extern __printf(5, 6) 2668 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2669 const char *, ...); 2670 extern __printf(5, 6) 2671 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2672 const char *, ...); 2673 extern void __ext4_std_error(struct super_block *, const char *, 2674 unsigned int, int); 2675 extern __printf(4, 5) 2676 void __ext4_abort(struct super_block *, const char *, unsigned int, 2677 const char *, ...); 2678 extern __printf(4, 5) 2679 void __ext4_warning(struct super_block *, const char *, unsigned int, 2680 const char *, ...); 2681 extern __printf(4, 5) 2682 void __ext4_warning_inode(const struct inode *inode, const char *function, 2683 unsigned int line, const char *fmt, ...); 2684 extern __printf(3, 4) 2685 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2686 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2687 const char *, unsigned int, const char *); 2688 extern __printf(7, 8) 2689 void __ext4_grp_locked_error(const char *, unsigned int, 2690 struct super_block *, ext4_group_t, 2691 unsigned long, ext4_fsblk_t, 2692 const char *, ...); 2693 2694 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2695 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2696 2697 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2698 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2699 2700 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2701 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2702 2703 #ifdef CONFIG_PRINTK 2704 2705 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2706 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2707 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2708 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2709 #define ext4_error(sb, fmt, ...) \ 2710 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2711 #define ext4_abort(sb, fmt, ...) \ 2712 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2713 #define ext4_warning(sb, fmt, ...) \ 2714 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2715 #define ext4_warning_inode(inode, fmt, ...) \ 2716 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2717 #define ext4_msg(sb, level, fmt, ...) \ 2718 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2719 #define dump_mmp_msg(sb, mmp, msg) \ 2720 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2721 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2722 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2723 fmt, ##__VA_ARGS__) 2724 2725 #else 2726 2727 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2728 do { \ 2729 no_printk(fmt, ##__VA_ARGS__); \ 2730 __ext4_error_inode(inode, "", 0, block, " "); \ 2731 } while (0) 2732 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2733 do { \ 2734 no_printk(fmt, ##__VA_ARGS__); \ 2735 __ext4_error_file(file, "", 0, block, " "); \ 2736 } while (0) 2737 #define ext4_error(sb, fmt, ...) \ 2738 do { \ 2739 no_printk(fmt, ##__VA_ARGS__); \ 2740 __ext4_error(sb, "", 0, " "); \ 2741 } while (0) 2742 #define ext4_abort(sb, fmt, ...) \ 2743 do { \ 2744 no_printk(fmt, ##__VA_ARGS__); \ 2745 __ext4_abort(sb, "", 0, " "); \ 2746 } while (0) 2747 #define ext4_warning(sb, fmt, ...) \ 2748 do { \ 2749 no_printk(fmt, ##__VA_ARGS__); \ 2750 __ext4_warning(sb, "", 0, " "); \ 2751 } while (0) 2752 #define ext4_warning_inode(inode, fmt, ...) \ 2753 do { \ 2754 no_printk(fmt, ##__VA_ARGS__); \ 2755 __ext4_warning_inode(inode, "", 0, " "); \ 2756 } while (0) 2757 #define ext4_msg(sb, level, fmt, ...) \ 2758 do { \ 2759 no_printk(fmt, ##__VA_ARGS__); \ 2760 __ext4_msg(sb, "", " "); \ 2761 } while (0) 2762 #define dump_mmp_msg(sb, mmp, msg) \ 2763 __dump_mmp_msg(sb, mmp, "", 0, "") 2764 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2765 do { \ 2766 no_printk(fmt, ##__VA_ARGS__); \ 2767 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2768 } while (0) 2769 2770 #endif 2771 2772 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2773 __u32 compat); 2774 extern int ext4_update_rocompat_feature(handle_t *handle, 2775 struct super_block *sb, __u32 rocompat); 2776 extern int ext4_update_incompat_feature(handle_t *handle, 2777 struct super_block *sb, __u32 incompat); 2778 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2779 struct ext4_group_desc *bg); 2780 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2781 struct ext4_group_desc *bg); 2782 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2783 struct ext4_group_desc *bg); 2784 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2785 struct ext4_group_desc *bg); 2786 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2787 struct ext4_group_desc *bg); 2788 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2789 struct ext4_group_desc *bg); 2790 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2791 struct ext4_group_desc *bg); 2792 extern void ext4_block_bitmap_set(struct super_block *sb, 2793 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2794 extern void ext4_inode_bitmap_set(struct super_block *sb, 2795 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2796 extern void ext4_inode_table_set(struct super_block *sb, 2797 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2798 extern void ext4_free_group_clusters_set(struct super_block *sb, 2799 struct ext4_group_desc *bg, 2800 __u32 count); 2801 extern void ext4_free_inodes_set(struct super_block *sb, 2802 struct ext4_group_desc *bg, __u32 count); 2803 extern void ext4_used_dirs_set(struct super_block *sb, 2804 struct ext4_group_desc *bg, __u32 count); 2805 extern void ext4_itable_unused_set(struct super_block *sb, 2806 struct ext4_group_desc *bg, __u32 count); 2807 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2808 struct ext4_group_desc *gdp); 2809 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2810 struct ext4_group_desc *gdp); 2811 extern int ext4_register_li_request(struct super_block *sb, 2812 ext4_group_t first_not_zeroed); 2813 2814 static inline int ext4_has_metadata_csum(struct super_block *sb) 2815 { 2816 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2817 !EXT4_SB(sb)->s_chksum_driver); 2818 2819 return ext4_has_feature_metadata_csum(sb) && 2820 (EXT4_SB(sb)->s_chksum_driver != NULL); 2821 } 2822 2823 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2824 { 2825 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2826 } 2827 2828 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2829 { 2830 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2831 le32_to_cpu(es->s_blocks_count_lo); 2832 } 2833 2834 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2835 { 2836 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2837 le32_to_cpu(es->s_r_blocks_count_lo); 2838 } 2839 2840 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2841 { 2842 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2843 le32_to_cpu(es->s_free_blocks_count_lo); 2844 } 2845 2846 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2847 ext4_fsblk_t blk) 2848 { 2849 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2850 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2851 } 2852 2853 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2854 ext4_fsblk_t blk) 2855 { 2856 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2857 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2858 } 2859 2860 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2861 ext4_fsblk_t blk) 2862 { 2863 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2864 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2865 } 2866 2867 static inline loff_t ext4_isize(struct super_block *sb, 2868 struct ext4_inode *raw_inode) 2869 { 2870 if (ext4_has_feature_largedir(sb) || 2871 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2872 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2873 le32_to_cpu(raw_inode->i_size_lo); 2874 2875 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2876 } 2877 2878 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2879 { 2880 raw_inode->i_size_lo = cpu_to_le32(i_size); 2881 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2882 } 2883 2884 static inline 2885 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2886 ext4_group_t group) 2887 { 2888 struct ext4_group_info ***grp_info; 2889 long indexv, indexh; 2890 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2891 grp_info = EXT4_SB(sb)->s_group_info; 2892 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2893 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2894 return grp_info[indexv][indexh]; 2895 } 2896 2897 /* 2898 * Reading s_groups_count requires using smp_rmb() afterwards. See 2899 * the locking protocol documented in the comments of ext4_group_add() 2900 * in resize.c 2901 */ 2902 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2903 { 2904 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2905 2906 smp_rmb(); 2907 return ngroups; 2908 } 2909 2910 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2911 ext4_group_t block_group) 2912 { 2913 return block_group >> sbi->s_log_groups_per_flex; 2914 } 2915 2916 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2917 { 2918 return 1 << sbi->s_log_groups_per_flex; 2919 } 2920 2921 #define ext4_std_error(sb, errno) \ 2922 do { \ 2923 if ((errno)) \ 2924 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2925 } while (0) 2926 2927 #ifdef CONFIG_SMP 2928 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2929 * counters. So we need to make sure we have free clusters more 2930 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2931 */ 2932 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2933 #else 2934 #define EXT4_FREECLUSTERS_WATERMARK 0 2935 #endif 2936 2937 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2938 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2939 { 2940 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2941 !inode_is_locked(inode)); 2942 down_write(&EXT4_I(inode)->i_data_sem); 2943 if (newsize > EXT4_I(inode)->i_disksize) 2944 EXT4_I(inode)->i_disksize = newsize; 2945 up_write(&EXT4_I(inode)->i_data_sem); 2946 } 2947 2948 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2949 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2950 { 2951 int changed = 0; 2952 2953 if (newsize > inode->i_size) { 2954 i_size_write(inode, newsize); 2955 changed = 1; 2956 } 2957 if (newsize > EXT4_I(inode)->i_disksize) { 2958 ext4_update_i_disksize(inode, newsize); 2959 changed |= 2; 2960 } 2961 return changed; 2962 } 2963 2964 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2965 loff_t len); 2966 2967 struct ext4_group_info { 2968 unsigned long bb_state; 2969 struct rb_root bb_free_root; 2970 ext4_grpblk_t bb_first_free; /* first free block */ 2971 ext4_grpblk_t bb_free; /* total free blocks */ 2972 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2973 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2974 struct list_head bb_prealloc_list; 2975 #ifdef DOUBLE_CHECK 2976 void *bb_bitmap; 2977 #endif 2978 struct rw_semaphore alloc_sem; 2979 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2980 * regions, index is order. 2981 * bb_counters[3] = 5 means 2982 * 5 free 8-block regions. */ 2983 }; 2984 2985 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2986 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2987 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2988 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2989 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2990 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2991 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2992 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2993 2994 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2995 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2996 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2997 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2998 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2999 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3000 3001 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3002 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3003 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3004 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3005 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3006 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3007 3008 #define EXT4_MAX_CONTENTION 8 3009 #define EXT4_CONTENTION_THRESHOLD 2 3010 3011 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3012 ext4_group_t group) 3013 { 3014 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3015 } 3016 3017 /* 3018 * Returns true if the filesystem is busy enough that attempts to 3019 * access the block group locks has run into contention. 3020 */ 3021 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3022 { 3023 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3024 } 3025 3026 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3027 { 3028 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3029 if (spin_trylock(lock)) 3030 /* 3031 * We're able to grab the lock right away, so drop the 3032 * lock contention counter. 3033 */ 3034 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3035 else { 3036 /* 3037 * The lock is busy, so bump the contention counter, 3038 * and then wait on the spin lock. 3039 */ 3040 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3041 EXT4_MAX_CONTENTION); 3042 spin_lock(lock); 3043 } 3044 } 3045 3046 static inline void ext4_unlock_group(struct super_block *sb, 3047 ext4_group_t group) 3048 { 3049 spin_unlock(ext4_group_lock_ptr(sb, group)); 3050 } 3051 3052 /* 3053 * Block validity checking 3054 */ 3055 #define ext4_check_indirect_blockref(inode, bh) \ 3056 ext4_check_blockref(__func__, __LINE__, inode, \ 3057 (__le32 *)(bh)->b_data, \ 3058 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3059 3060 #define ext4_ind_check_inode(inode) \ 3061 ext4_check_blockref(__func__, __LINE__, inode, \ 3062 EXT4_I(inode)->i_data, \ 3063 EXT4_NDIR_BLOCKS) 3064 3065 /* 3066 * Inodes and files operations 3067 */ 3068 3069 /* dir.c */ 3070 extern const struct file_operations ext4_dir_operations; 3071 3072 #ifdef CONFIG_UNICODE 3073 extern const struct dentry_operations ext4_dentry_ops; 3074 #endif 3075 3076 /* file.c */ 3077 extern const struct inode_operations ext4_file_inode_operations; 3078 extern const struct file_operations ext4_file_operations; 3079 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3080 3081 /* inline.c */ 3082 extern int ext4_get_max_inline_size(struct inode *inode); 3083 extern int ext4_find_inline_data_nolock(struct inode *inode); 3084 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3085 unsigned int len); 3086 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3087 3088 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3089 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3090 struct inode *inode, 3091 loff_t pos, unsigned len, 3092 unsigned flags, 3093 struct page **pagep); 3094 extern int ext4_write_inline_data_end(struct inode *inode, 3095 loff_t pos, unsigned len, 3096 unsigned copied, 3097 struct page *page); 3098 extern struct buffer_head * 3099 ext4_journalled_write_inline_data(struct inode *inode, 3100 unsigned len, 3101 struct page *page); 3102 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3103 struct inode *inode, 3104 loff_t pos, unsigned len, 3105 unsigned flags, 3106 struct page **pagep, 3107 void **fsdata); 3108 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3109 unsigned len, unsigned copied, 3110 struct page *page); 3111 extern int ext4_try_add_inline_entry(handle_t *handle, 3112 struct ext4_filename *fname, 3113 struct inode *dir, struct inode *inode); 3114 extern int ext4_try_create_inline_dir(handle_t *handle, 3115 struct inode *parent, 3116 struct inode *inode); 3117 extern int ext4_read_inline_dir(struct file *filp, 3118 struct dir_context *ctx, 3119 int *has_inline_data); 3120 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3121 struct inode *dir, ext4_lblk_t block, 3122 struct dx_hash_info *hinfo, 3123 __u32 start_hash, __u32 start_minor_hash, 3124 int *has_inline_data); 3125 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3126 struct ext4_filename *fname, 3127 struct ext4_dir_entry_2 **res_dir, 3128 int *has_inline_data); 3129 extern int ext4_delete_inline_entry(handle_t *handle, 3130 struct inode *dir, 3131 struct ext4_dir_entry_2 *de_del, 3132 struct buffer_head *bh, 3133 int *has_inline_data); 3134 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3135 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3136 struct ext4_dir_entry_2 **parent_de, 3137 int *retval); 3138 extern int ext4_inline_data_fiemap(struct inode *inode, 3139 struct fiemap_extent_info *fieinfo, 3140 int *has_inline, __u64 start, __u64 len); 3141 3142 struct iomap; 3143 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3144 3145 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3146 3147 extern int ext4_convert_inline_data(struct inode *inode); 3148 3149 static inline int ext4_has_inline_data(struct inode *inode) 3150 { 3151 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3152 EXT4_I(inode)->i_inline_off; 3153 } 3154 3155 /* namei.c */ 3156 extern const struct inode_operations ext4_dir_inode_operations; 3157 extern const struct inode_operations ext4_special_inode_operations; 3158 extern struct dentry *ext4_get_parent(struct dentry *child); 3159 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3160 struct ext4_dir_entry_2 *de, 3161 int blocksize, int csum_size, 3162 unsigned int parent_ino, int dotdot_real_len); 3163 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3164 unsigned int blocksize); 3165 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3166 struct buffer_head *bh); 3167 extern int ext4_ci_compare(const struct inode *parent, 3168 const struct qstr *fname, 3169 const struct qstr *entry, bool quick); 3170 3171 #define S_SHIFT 12 3172 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3173 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3174 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3175 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3176 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3177 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3178 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3179 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3180 }; 3181 3182 static inline void ext4_set_de_type(struct super_block *sb, 3183 struct ext4_dir_entry_2 *de, 3184 umode_t mode) { 3185 if (ext4_has_feature_filetype(sb)) 3186 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3187 } 3188 3189 /* readpages.c */ 3190 extern int ext4_mpage_readpages(struct address_space *mapping, 3191 struct list_head *pages, struct page *page, 3192 unsigned nr_pages, bool is_readahead); 3193 3194 /* symlink.c */ 3195 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3196 extern const struct inode_operations ext4_symlink_inode_operations; 3197 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3198 3199 /* sysfs.c */ 3200 extern int ext4_register_sysfs(struct super_block *sb); 3201 extern void ext4_unregister_sysfs(struct super_block *sb); 3202 extern int __init ext4_init_sysfs(void); 3203 extern void ext4_exit_sysfs(void); 3204 3205 /* block_validity */ 3206 extern void ext4_release_system_zone(struct super_block *sb); 3207 extern int ext4_setup_system_zone(struct super_block *sb); 3208 extern int __init ext4_init_system_zone(void); 3209 extern void ext4_exit_system_zone(void); 3210 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3211 ext4_fsblk_t start_blk, 3212 unsigned int count); 3213 extern int ext4_check_blockref(const char *, unsigned int, 3214 struct inode *, __le32 *, unsigned int); 3215 3216 /* extents.c */ 3217 struct ext4_ext_path; 3218 struct ext4_extent; 3219 3220 /* 3221 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3222 * __le32. 3223 */ 3224 #define EXT_MAX_BLOCKS 0xffffffff 3225 3226 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3227 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3228 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3229 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3230 struct ext4_map_blocks *map, int flags); 3231 extern int ext4_ext_truncate(handle_t *, struct inode *); 3232 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3233 ext4_lblk_t end); 3234 extern void ext4_ext_init(struct super_block *); 3235 extern void ext4_ext_release(struct super_block *); 3236 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3237 loff_t len); 3238 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3239 loff_t offset, ssize_t len); 3240 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3241 struct ext4_map_blocks *map, int flags); 3242 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3243 ext4_lblk_t lblocks); 3244 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3245 int num, 3246 struct ext4_ext_path *path); 3247 extern int ext4_can_extents_be_merged(struct inode *inode, 3248 struct ext4_extent *ex1, 3249 struct ext4_extent *ex2); 3250 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3251 struct ext4_ext_path **, 3252 struct ext4_extent *, int); 3253 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3254 struct ext4_ext_path **, 3255 int flags); 3256 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3257 extern int ext4_ext_check_inode(struct inode *inode); 3258 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3259 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3260 __u64 start, __u64 len); 3261 extern int ext4_ext_precache(struct inode *inode); 3262 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3263 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3264 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3265 struct inode *inode2, ext4_lblk_t lblk1, 3266 ext4_lblk_t lblk2, ext4_lblk_t count, 3267 int mark_unwritten,int *err); 3268 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3269 3270 /* move_extent.c */ 3271 extern void ext4_double_down_write_data_sem(struct inode *first, 3272 struct inode *second); 3273 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3274 struct inode *donor_inode); 3275 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3276 __u64 start_orig, __u64 start_donor, 3277 __u64 len, __u64 *moved_len); 3278 3279 /* page-io.c */ 3280 extern int __init ext4_init_pageio(void); 3281 extern void ext4_exit_pageio(void); 3282 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3283 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3284 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3285 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3286 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3287 struct writeback_control *wbc); 3288 extern void ext4_end_io_rsv_work(struct work_struct *work); 3289 extern void ext4_io_submit(struct ext4_io_submit *io); 3290 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3291 struct page *page, 3292 int len, 3293 struct writeback_control *wbc, 3294 bool keep_towrite); 3295 3296 /* mmp.c */ 3297 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3298 3299 /* 3300 * Add new method to test whether block and inode bitmaps are properly 3301 * initialized. With uninit_bg reading the block from disk is not enough 3302 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3303 */ 3304 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3305 3306 static inline int bitmap_uptodate(struct buffer_head *bh) 3307 { 3308 return (buffer_uptodate(bh) && 3309 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3310 } 3311 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3312 { 3313 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3314 } 3315 3316 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3317 3318 /* For ioend & aio unwritten conversion wait queues */ 3319 #define EXT4_WQ_HASH_SZ 37 3320 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3321 EXT4_WQ_HASH_SZ]) 3322 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3323 3324 extern int ext4_resize_begin(struct super_block *sb); 3325 extern void ext4_resize_end(struct super_block *sb); 3326 3327 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3328 struct ext4_io_end *io_end) 3329 { 3330 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3331 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3332 atomic_inc(&EXT4_I(inode)->i_unwritten); 3333 } 3334 } 3335 3336 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3337 { 3338 struct inode *inode = io_end->inode; 3339 3340 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3341 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3342 /* Wake up anyone waiting on unwritten extent conversion */ 3343 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3344 wake_up_all(ext4_ioend_wq(inode)); 3345 } 3346 } 3347 3348 extern const struct iomap_ops ext4_iomap_ops; 3349 3350 #endif /* __KERNEL__ */ 3351 3352 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3353 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3354 3355 #endif /* _EXT4_H */ 3356