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