1 /* 2 * ext4.h 3 * 4 * Copyright (C) 1992, 1993, 1994, 1995 5 * Remy Card (card@masi.ibp.fr) 6 * Laboratoire MASI - Institut Blaise Pascal 7 * Universite Pierre et Marie Curie (Paris VI) 8 * 9 * from 10 * 11 * linux/include/linux/minix_fs.h 12 * 13 * Copyright (C) 1991, 1992 Linus Torvalds 14 */ 15 16 #ifndef _EXT4_H 17 #define _EXT4_H 18 19 #include <linux/types.h> 20 #include <linux/blkdev.h> 21 #include <linux/magic.h> 22 #include <linux/jbd2.h> 23 #include <linux/quota.h> 24 #include <linux/rwsem.h> 25 #include <linux/rbtree.h> 26 #include <linux/seqlock.h> 27 #include <linux/mutex.h> 28 #include <linux/timer.h> 29 #include <linux/version.h> 30 #include <linux/wait.h> 31 #include <linux/sched/signal.h> 32 #include <linux/blockgroup_lock.h> 33 #include <linux/percpu_counter.h> 34 #include <linux/ratelimit.h> 35 #include <crypto/hash.h> 36 #include <linux/falloc.h> 37 #include <linux/percpu-rwsem.h> 38 #ifdef __KERNEL__ 39 #include <linux/compat.h> 40 #endif 41 42 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION) 43 #include <linux/fscrypt.h> 44 45 /* 46 * The fourth extended filesystem constants/structures 47 */ 48 49 /* 50 * with AGGRESSIVE_CHECK allocator runs consistency checks over 51 * structures. these checks slow things down a lot 52 */ 53 #define AGGRESSIVE_CHECK__ 54 55 /* 56 * with DOUBLE_CHECK defined mballoc creates persistent in-core 57 * bitmaps, maintains and uses them to check for double allocations 58 */ 59 #define DOUBLE_CHECK__ 60 61 /* 62 * Define EXT4FS_DEBUG to produce debug messages 63 */ 64 #undef EXT4FS_DEBUG 65 66 /* 67 * Debug code 68 */ 69 #ifdef EXT4FS_DEBUG 70 #define ext4_debug(f, a...) \ 71 do { \ 72 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 73 __FILE__, __LINE__, __func__); \ 74 printk(KERN_DEBUG f, ## a); \ 75 } while (0) 76 #else 77 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 78 #endif 79 80 /* 81 * Turn on EXT_DEBUG to get lots of info about extents operations. 82 */ 83 #define EXT_DEBUG__ 84 #ifdef EXT_DEBUG 85 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 86 #else 87 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 88 #endif 89 90 /* data type for block offset of block group */ 91 typedef int ext4_grpblk_t; 92 93 /* data type for filesystem-wide blocks number */ 94 typedef unsigned long long ext4_fsblk_t; 95 96 /* data type for file logical block number */ 97 typedef __u32 ext4_lblk_t; 98 99 /* data type for block group number */ 100 typedef unsigned int ext4_group_t; 101 102 enum SHIFT_DIRECTION { 103 SHIFT_LEFT = 0, 104 SHIFT_RIGHT, 105 }; 106 107 /* 108 * Flags used in mballoc's allocation_context flags field. 109 * 110 * Also used to show what's going on for debugging purposes when the 111 * flag field is exported via the traceport interface 112 */ 113 114 /* prefer goal again. length */ 115 #define EXT4_MB_HINT_MERGE 0x0001 116 /* blocks already reserved */ 117 #define EXT4_MB_HINT_RESERVED 0x0002 118 /* metadata is being allocated */ 119 #define EXT4_MB_HINT_METADATA 0x0004 120 /* first blocks in the file */ 121 #define EXT4_MB_HINT_FIRST 0x0008 122 /* search for the best chunk */ 123 #define EXT4_MB_HINT_BEST 0x0010 124 /* data is being allocated */ 125 #define EXT4_MB_HINT_DATA 0x0020 126 /* don't preallocate (for tails) */ 127 #define EXT4_MB_HINT_NOPREALLOC 0x0040 128 /* allocate for locality group */ 129 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 130 /* allocate goal blocks or none */ 131 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 132 /* goal is meaningful */ 133 #define EXT4_MB_HINT_TRY_GOAL 0x0200 134 /* blocks already pre-reserved by delayed allocation */ 135 #define EXT4_MB_DELALLOC_RESERVED 0x0400 136 /* We are doing stream allocation */ 137 #define EXT4_MB_STREAM_ALLOC 0x0800 138 /* Use reserved root blocks if needed */ 139 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 140 /* Use blocks from reserved pool */ 141 #define EXT4_MB_USE_RESERVED 0x2000 142 143 struct ext4_allocation_request { 144 /* target inode for block we're allocating */ 145 struct inode *inode; 146 /* how many blocks we want to allocate */ 147 unsigned int len; 148 /* logical block in target inode */ 149 ext4_lblk_t logical; 150 /* the closest logical allocated block to the left */ 151 ext4_lblk_t lleft; 152 /* the closest logical allocated block to the right */ 153 ext4_lblk_t lright; 154 /* phys. target (a hint) */ 155 ext4_fsblk_t goal; 156 /* phys. block for the closest logical allocated block to the left */ 157 ext4_fsblk_t pleft; 158 /* phys. block for the closest logical allocated block to the right */ 159 ext4_fsblk_t pright; 160 /* flags. see above EXT4_MB_HINT_* */ 161 unsigned int flags; 162 }; 163 164 /* 165 * Logical to physical block mapping, used by ext4_map_blocks() 166 * 167 * This structure is used to pass requests into ext4_map_blocks() as 168 * well as to store the information returned by ext4_map_blocks(). It 169 * takes less room on the stack than a struct buffer_head. 170 */ 171 #define EXT4_MAP_NEW (1 << BH_New) 172 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 173 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 174 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 175 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 176 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 177 178 struct ext4_map_blocks { 179 ext4_fsblk_t m_pblk; 180 ext4_lblk_t m_lblk; 181 unsigned int m_len; 182 unsigned int m_flags; 183 }; 184 185 /* 186 * Flags for ext4_io_end->flags 187 */ 188 #define EXT4_IO_END_UNWRITTEN 0x0001 189 190 /* 191 * For converting unwritten extents on a work queue. 'handle' is used for 192 * buffered writeback. 193 */ 194 typedef struct ext4_io_end { 195 struct list_head list; /* per-file finished IO list */ 196 handle_t *handle; /* handle reserved for extent 197 * conversion */ 198 struct inode *inode; /* file being written to */ 199 struct bio *bio; /* Linked list of completed 200 * bios covering the extent */ 201 unsigned int flag; /* unwritten or not */ 202 atomic_t count; /* reference counter */ 203 loff_t offset; /* offset in the file */ 204 ssize_t size; /* size of the extent */ 205 } ext4_io_end_t; 206 207 struct ext4_io_submit { 208 struct writeback_control *io_wbc; 209 struct bio *io_bio; 210 ext4_io_end_t *io_end; 211 sector_t io_next_block; 212 }; 213 214 /* 215 * Special inodes numbers 216 */ 217 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 218 #define EXT4_ROOT_INO 2 /* Root inode */ 219 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 220 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 221 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 222 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 223 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 224 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 225 226 /* First non-reserved inode for old ext4 filesystems */ 227 #define EXT4_GOOD_OLD_FIRST_INO 11 228 229 /* 230 * Maximal count of links to a file 231 */ 232 #define EXT4_LINK_MAX 65000 233 234 /* 235 * Macro-instructions used to manage several block sizes 236 */ 237 #define EXT4_MIN_BLOCK_SIZE 1024 238 #define EXT4_MAX_BLOCK_SIZE 65536 239 #define EXT4_MIN_BLOCK_LOG_SIZE 10 240 #define EXT4_MAX_BLOCK_LOG_SIZE 16 241 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 242 #ifdef __KERNEL__ 243 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 244 #else 245 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 246 #endif 247 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 248 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 249 EXT4_SB(s)->s_cluster_bits) 250 #ifdef __KERNEL__ 251 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 252 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 253 #else 254 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 255 #endif 256 #ifdef __KERNEL__ 257 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 258 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 259 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 260 #else 261 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 262 EXT4_GOOD_OLD_INODE_SIZE : \ 263 (s)->s_inode_size) 264 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 265 EXT4_GOOD_OLD_FIRST_INO : \ 266 (s)->s_first_ino) 267 #endif 268 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 269 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 270 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 271 blkbits)) 272 273 /* Translate a block number to a cluster number */ 274 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 275 /* Translate a cluster number to a block number */ 276 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 277 /* Translate # of blks to # of clusters */ 278 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 279 (sbi)->s_cluster_bits) 280 /* Mask out the low bits to get the starting block of the cluster */ 281 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 282 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 283 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 284 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 285 /* Get the cluster offset */ 286 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 287 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 288 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 289 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 290 291 /* 292 * Structure of a blocks group descriptor 293 */ 294 struct ext4_group_desc 295 { 296 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 297 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 298 __le32 bg_inode_table_lo; /* Inodes table block */ 299 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 300 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 301 __le16 bg_used_dirs_count_lo; /* Directories count */ 302 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 303 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 304 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 305 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 306 __le16 bg_itable_unused_lo; /* Unused inodes count */ 307 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 308 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 309 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 310 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 311 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 312 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 313 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 314 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 315 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 316 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 317 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 318 __u32 bg_reserved; 319 }; 320 321 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 322 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 323 sizeof(__le16)) 324 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 325 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 326 sizeof(__le16)) 327 328 /* 329 * Structure of a flex block group info 330 */ 331 332 struct flex_groups { 333 atomic64_t free_clusters; 334 atomic_t free_inodes; 335 atomic_t used_dirs; 336 }; 337 338 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 339 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 340 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 341 342 /* 343 * Macro-instructions used to manage group descriptors 344 */ 345 #define EXT4_MIN_DESC_SIZE 32 346 #define EXT4_MIN_DESC_SIZE_64BIT 64 347 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 348 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 349 #ifdef __KERNEL__ 350 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 351 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 352 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 353 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 354 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 355 #else 356 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 357 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 358 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 359 #endif 360 361 /* 362 * Constants relative to the data blocks 363 */ 364 #define EXT4_NDIR_BLOCKS 12 365 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 366 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 367 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 368 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 369 370 /* 371 * Inode flags 372 */ 373 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 374 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 375 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 376 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 377 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 378 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 379 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 380 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 381 /* Reserved for compression usage... */ 382 #define EXT4_DIRTY_FL 0x00000100 383 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 384 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 385 /* nb: was previously EXT2_ECOMPR_FL */ 386 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 387 /* End compression flags --- maybe not all used */ 388 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 389 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 390 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 391 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 392 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 393 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 394 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 395 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 396 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 397 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 398 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 399 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 400 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 401 402 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 403 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 404 405 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 406 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 407 EXT4_IMMUTABLE_FL | \ 408 EXT4_APPEND_FL | \ 409 EXT4_NODUMP_FL | \ 410 EXT4_NOATIME_FL | \ 411 EXT4_PROJINHERIT_FL) 412 413 /* Flags that should be inherited by new inodes from their parent. */ 414 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 415 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 416 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 417 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 418 EXT4_PROJINHERIT_FL) 419 420 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 421 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 422 423 /* Flags that are appropriate for non-directories/regular files. */ 424 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 425 426 /* Mask out flags that are inappropriate for the given type of inode. */ 427 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 428 { 429 if (S_ISDIR(mode)) 430 return flags; 431 else if (S_ISREG(mode)) 432 return flags & EXT4_REG_FLMASK; 433 else 434 return flags & EXT4_OTHER_FLMASK; 435 } 436 437 /* 438 * Inode flags used for atomic set/get 439 */ 440 enum { 441 EXT4_INODE_SECRM = 0, /* Secure deletion */ 442 EXT4_INODE_UNRM = 1, /* Undelete */ 443 EXT4_INODE_COMPR = 2, /* Compress file */ 444 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 445 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 446 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 447 EXT4_INODE_NODUMP = 6, /* do not dump file */ 448 EXT4_INODE_NOATIME = 7, /* do not update atime */ 449 /* Reserved for compression usage... */ 450 EXT4_INODE_DIRTY = 8, 451 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 452 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 453 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 454 /* End compression flags --- maybe not all used */ 455 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 456 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 457 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 458 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 459 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 460 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 461 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 462 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 463 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 464 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 465 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 466 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 467 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 468 }; 469 470 /* 471 * Since it's pretty easy to mix up bit numbers and hex values, we use a 472 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 473 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 474 * any extra space in the compiled kernel image, otherwise, the build will fail. 475 * It's important that these values are the same, since we are using 476 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 477 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 478 * values found in ext2, ext3 and ext4 filesystems, and of course the values 479 * defined in e2fsprogs. 480 * 481 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 482 */ 483 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 484 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 485 486 static inline void ext4_check_flag_values(void) 487 { 488 CHECK_FLAG_VALUE(SECRM); 489 CHECK_FLAG_VALUE(UNRM); 490 CHECK_FLAG_VALUE(COMPR); 491 CHECK_FLAG_VALUE(SYNC); 492 CHECK_FLAG_VALUE(IMMUTABLE); 493 CHECK_FLAG_VALUE(APPEND); 494 CHECK_FLAG_VALUE(NODUMP); 495 CHECK_FLAG_VALUE(NOATIME); 496 CHECK_FLAG_VALUE(DIRTY); 497 CHECK_FLAG_VALUE(COMPRBLK); 498 CHECK_FLAG_VALUE(NOCOMPR); 499 CHECK_FLAG_VALUE(ENCRYPT); 500 CHECK_FLAG_VALUE(INDEX); 501 CHECK_FLAG_VALUE(IMAGIC); 502 CHECK_FLAG_VALUE(JOURNAL_DATA); 503 CHECK_FLAG_VALUE(NOTAIL); 504 CHECK_FLAG_VALUE(DIRSYNC); 505 CHECK_FLAG_VALUE(TOPDIR); 506 CHECK_FLAG_VALUE(HUGE_FILE); 507 CHECK_FLAG_VALUE(EXTENTS); 508 CHECK_FLAG_VALUE(EA_INODE); 509 CHECK_FLAG_VALUE(EOFBLOCKS); 510 CHECK_FLAG_VALUE(INLINE_DATA); 511 CHECK_FLAG_VALUE(PROJINHERIT); 512 CHECK_FLAG_VALUE(RESERVED); 513 } 514 515 /* Used to pass group descriptor data when online resize is done */ 516 struct ext4_new_group_input { 517 __u32 group; /* Group number for this data */ 518 __u64 block_bitmap; /* Absolute block number of block bitmap */ 519 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 520 __u64 inode_table; /* Absolute block number of inode table start */ 521 __u32 blocks_count; /* Total number of blocks in this group */ 522 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 523 __u16 unused; 524 }; 525 526 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 527 struct compat_ext4_new_group_input { 528 u32 group; 529 compat_u64 block_bitmap; 530 compat_u64 inode_bitmap; 531 compat_u64 inode_table; 532 u32 blocks_count; 533 u16 reserved_blocks; 534 u16 unused; 535 }; 536 #endif 537 538 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 539 struct ext4_new_group_data { 540 __u32 group; 541 __u64 block_bitmap; 542 __u64 inode_bitmap; 543 __u64 inode_table; 544 __u32 blocks_count; 545 __u16 reserved_blocks; 546 __u16 unused; 547 __u32 free_blocks_count; 548 }; 549 550 /* Indexes used to index group tables in ext4_new_group_data */ 551 enum { 552 BLOCK_BITMAP = 0, /* block bitmap */ 553 INODE_BITMAP, /* inode bitmap */ 554 INODE_TABLE, /* inode tables */ 555 GROUP_TABLE_COUNT, 556 }; 557 558 /* 559 * Flags used by ext4_map_blocks() 560 */ 561 /* Allocate any needed blocks and/or convert an unwritten 562 extent to be an initialized ext4 */ 563 #define EXT4_GET_BLOCKS_CREATE 0x0001 564 /* Request the creation of an unwritten extent */ 565 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 566 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 567 EXT4_GET_BLOCKS_CREATE) 568 /* Caller is from the delayed allocation writeout path 569 * finally doing the actual allocation of delayed blocks */ 570 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 571 /* caller is from the direct IO path, request to creation of an 572 unwritten extents if not allocated, split the unwritten 573 extent if blocks has been preallocated already*/ 574 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 575 #define EXT4_GET_BLOCKS_CONVERT 0x0010 576 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 577 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 578 /* Convert extent to initialized after IO complete */ 579 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 580 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 581 /* Eventual metadata allocation (due to growing extent tree) 582 * should not fail, so try to use reserved blocks for that.*/ 583 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 584 /* Don't normalize allocation size (used for fallocate) */ 585 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 586 /* Request will not result in inode size update (user for fallocate) */ 587 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 588 /* Convert written extents to unwritten */ 589 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 590 /* Write zeros to newly created written extents */ 591 #define EXT4_GET_BLOCKS_ZERO 0x0200 592 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 593 EXT4_GET_BLOCKS_ZERO) 594 /* Caller will submit data before dropping transaction handle. This 595 * allows jbd2 to avoid submitting data before commit. */ 596 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 597 598 /* 599 * The bit position of these flags must not overlap with any of the 600 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 601 * read_extent_tree_block(), ext4_split_extent_at(), 602 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 603 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 604 * caching the extents when reading from the extent tree while a 605 * truncate or punch hole operation is in progress. 606 */ 607 #define EXT4_EX_NOCACHE 0x40000000 608 #define EXT4_EX_FORCE_CACHE 0x20000000 609 610 /* 611 * Flags used by ext4_free_blocks 612 */ 613 #define EXT4_FREE_BLOCKS_METADATA 0x0001 614 #define EXT4_FREE_BLOCKS_FORGET 0x0002 615 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 616 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 617 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 618 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 619 620 /* 621 * ioctl commands 622 */ 623 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 624 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 625 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 626 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 627 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 628 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 629 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 630 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 631 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 632 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 633 #define EXT4_IOC_MIGRATE _IO('f', 9) 634 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 635 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 636 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 637 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 638 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 639 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 640 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 641 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 642 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 643 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 644 645 #ifndef FS_IOC_FSGETXATTR 646 /* Until the uapi changes get merged for project quota... */ 647 648 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr) 649 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr) 650 651 /* 652 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR. 653 */ 654 struct fsxattr { 655 __u32 fsx_xflags; /* xflags field value (get/set) */ 656 __u32 fsx_extsize; /* extsize field value (get/set)*/ 657 __u32 fsx_nextents; /* nextents field value (get) */ 658 __u32 fsx_projid; /* project identifier (get/set) */ 659 unsigned char fsx_pad[12]; 660 }; 661 662 /* 663 * Flags for the fsx_xflags field 664 */ 665 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */ 666 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */ 667 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */ 668 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */ 669 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */ 670 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */ 671 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */ 672 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */ 673 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */ 674 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */ 675 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */ 676 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */ 677 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */ 678 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */ 679 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */ 680 #endif /* !defined(FS_IOC_FSGETXATTR) */ 681 682 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 683 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 684 685 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 686 687 /* 688 * Flags for going down operation 689 */ 690 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 691 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 692 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 693 694 695 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 696 /* 697 * ioctl commands in 32 bit emulation 698 */ 699 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 700 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 701 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 702 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 703 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 704 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 705 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 706 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 707 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 708 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 709 #endif 710 711 /* Max physical block we can address w/o extents */ 712 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 713 714 /* 715 * Structure of an inode on the disk 716 */ 717 struct ext4_inode { 718 __le16 i_mode; /* File mode */ 719 __le16 i_uid; /* Low 16 bits of Owner Uid */ 720 __le32 i_size_lo; /* Size in bytes */ 721 __le32 i_atime; /* Access time */ 722 __le32 i_ctime; /* Inode Change time */ 723 __le32 i_mtime; /* Modification time */ 724 __le32 i_dtime; /* Deletion Time */ 725 __le16 i_gid; /* Low 16 bits of Group Id */ 726 __le16 i_links_count; /* Links count */ 727 __le32 i_blocks_lo; /* Blocks count */ 728 __le32 i_flags; /* File flags */ 729 union { 730 struct { 731 __le32 l_i_version; 732 } linux1; 733 struct { 734 __u32 h_i_translator; 735 } hurd1; 736 struct { 737 __u32 m_i_reserved1; 738 } masix1; 739 } osd1; /* OS dependent 1 */ 740 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 741 __le32 i_generation; /* File version (for NFS) */ 742 __le32 i_file_acl_lo; /* File ACL */ 743 __le32 i_size_high; 744 __le32 i_obso_faddr; /* Obsoleted fragment address */ 745 union { 746 struct { 747 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 748 __le16 l_i_file_acl_high; 749 __le16 l_i_uid_high; /* these 2 fields */ 750 __le16 l_i_gid_high; /* were reserved2[0] */ 751 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 752 __le16 l_i_reserved; 753 } linux2; 754 struct { 755 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 756 __u16 h_i_mode_high; 757 __u16 h_i_uid_high; 758 __u16 h_i_gid_high; 759 __u32 h_i_author; 760 } hurd2; 761 struct { 762 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 763 __le16 m_i_file_acl_high; 764 __u32 m_i_reserved2[2]; 765 } masix2; 766 } osd2; /* OS dependent 2 */ 767 __le16 i_extra_isize; 768 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 769 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 770 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 771 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 772 __le32 i_crtime; /* File Creation time */ 773 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 774 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 775 __le32 i_projid; /* Project ID */ 776 }; 777 778 struct move_extent { 779 __u32 reserved; /* should be zero */ 780 __u32 donor_fd; /* donor file descriptor */ 781 __u64 orig_start; /* logical start offset in block for orig */ 782 __u64 donor_start; /* logical start offset in block for donor */ 783 __u64 len; /* block length to be moved */ 784 __u64 moved_len; /* moved block length */ 785 }; 786 787 #define EXT4_EPOCH_BITS 2 788 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 789 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 790 791 /* 792 * Extended fields will fit into an inode if the filesystem was formatted 793 * with large inodes (-I 256 or larger) and there are not currently any EAs 794 * consuming all of the available space. For new inodes we always reserve 795 * enough space for the kernel's known extended fields, but for inodes 796 * created with an old kernel this might not have been the case. None of 797 * the extended inode fields is critical for correct filesystem operation. 798 * This macro checks if a certain field fits in the inode. Note that 799 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 800 */ 801 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 802 ((offsetof(typeof(*ext4_inode), field) + \ 803 sizeof((ext4_inode)->field)) \ 804 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 805 (einode)->i_extra_isize)) \ 806 807 /* 808 * We use an encoding that preserves the times for extra epoch "00": 809 * 810 * extra msb of adjust for signed 811 * epoch 32-bit 32-bit tv_sec to 812 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 813 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 814 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 815 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 816 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 817 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 818 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 819 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 820 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 821 * 822 * Note that previous versions of the kernel on 64-bit systems would 823 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 824 * 1970. e2fsck will correct this, assuming that it is run on the 825 * affected filesystem before 2242. 826 */ 827 828 static inline __le32 ext4_encode_extra_time(struct timespec *time) 829 { 830 u32 extra = sizeof(time->tv_sec) > 4 ? 831 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0; 832 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 833 } 834 835 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 836 { 837 if (unlikely(sizeof(time->tv_sec) > 4 && 838 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) { 839 840 #if 1 841 /* Handle legacy encoding of pre-1970 dates with epoch 842 * bits 1,1. (This backwards compatibility may be removed 843 * at the discretion of the ext4 developers.) 844 */ 845 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 846 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 847 extra_bits = 0; 848 time->tv_sec += extra_bits << 32; 849 #else 850 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 851 #endif 852 } 853 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 854 } 855 856 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 857 do { \ 858 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 859 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 860 (raw_inode)->xtime ## _extra = \ 861 ext4_encode_extra_time(&(inode)->xtime); \ 862 } while (0) 863 864 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 865 do { \ 866 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 867 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 868 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 869 (raw_inode)->xtime ## _extra = \ 870 ext4_encode_extra_time(&(einode)->xtime); \ 871 } while (0) 872 873 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 874 do { \ 875 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 876 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 877 ext4_decode_extra_time(&(inode)->xtime, \ 878 raw_inode->xtime ## _extra); \ 879 else \ 880 (inode)->xtime.tv_nsec = 0; \ 881 } while (0) 882 883 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 884 do { \ 885 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 886 (einode)->xtime.tv_sec = \ 887 (signed)le32_to_cpu((raw_inode)->xtime); \ 888 else \ 889 (einode)->xtime.tv_sec = 0; \ 890 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 891 ext4_decode_extra_time(&(einode)->xtime, \ 892 raw_inode->xtime ## _extra); \ 893 else \ 894 (einode)->xtime.tv_nsec = 0; \ 895 } while (0) 896 897 #define i_disk_version osd1.linux1.l_i_version 898 899 #if defined(__KERNEL__) || defined(__linux__) 900 #define i_reserved1 osd1.linux1.l_i_reserved1 901 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 902 #define i_blocks_high osd2.linux2.l_i_blocks_high 903 #define i_uid_low i_uid 904 #define i_gid_low i_gid 905 #define i_uid_high osd2.linux2.l_i_uid_high 906 #define i_gid_high osd2.linux2.l_i_gid_high 907 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 908 909 #elif defined(__GNU__) 910 911 #define i_translator osd1.hurd1.h_i_translator 912 #define i_uid_high osd2.hurd2.h_i_uid_high 913 #define i_gid_high osd2.hurd2.h_i_gid_high 914 #define i_author osd2.hurd2.h_i_author 915 916 #elif defined(__masix__) 917 918 #define i_reserved1 osd1.masix1.m_i_reserved1 919 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 920 #define i_reserved2 osd2.masix2.m_i_reserved2 921 922 #endif /* defined(__KERNEL__) || defined(__linux__) */ 923 924 #include "extents_status.h" 925 926 /* 927 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 928 * 929 * These are needed to avoid lockdep false positives when we need to 930 * allocate blocks to the quota inode during ext4_map_blocks(), while 931 * holding i_data_sem for a normal (non-quota) inode. Since we don't 932 * do quota tracking for the quota inode, this avoids deadlock (as 933 * well as infinite recursion, since it isn't turtles all the way 934 * down...) 935 * 936 * I_DATA_SEM_NORMAL - Used for most inodes 937 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 938 * where the second inode has larger inode number 939 * than the first 940 * I_DATA_SEM_QUOTA - Used for quota inodes only 941 */ 942 enum { 943 I_DATA_SEM_NORMAL = 0, 944 I_DATA_SEM_OTHER, 945 I_DATA_SEM_QUOTA, 946 }; 947 948 949 /* 950 * fourth extended file system inode data in memory 951 */ 952 struct ext4_inode_info { 953 __le32 i_data[15]; /* unconverted */ 954 __u32 i_dtime; 955 ext4_fsblk_t i_file_acl; 956 957 /* 958 * i_block_group is the number of the block group which contains 959 * this file's inode. Constant across the lifetime of the inode, 960 * it is used for making block allocation decisions - we try to 961 * place a file's data blocks near its inode block, and new inodes 962 * near to their parent directory's inode. 963 */ 964 ext4_group_t i_block_group; 965 ext4_lblk_t i_dir_start_lookup; 966 #if (BITS_PER_LONG < 64) 967 unsigned long i_state_flags; /* Dynamic state flags */ 968 #endif 969 unsigned long i_flags; 970 971 /* 972 * Extended attributes can be read independently of the main file 973 * data. Taking i_mutex even when reading would cause contention 974 * between readers of EAs and writers of regular file data, so 975 * instead we synchronize on xattr_sem when reading or changing 976 * EAs. 977 */ 978 struct rw_semaphore xattr_sem; 979 980 struct list_head i_orphan; /* unlinked but open inodes */ 981 982 /* 983 * i_disksize keeps track of what the inode size is ON DISK, not 984 * in memory. During truncate, i_size is set to the new size by 985 * the VFS prior to calling ext4_truncate(), but the filesystem won't 986 * set i_disksize to 0 until the truncate is actually under way. 987 * 988 * The intent is that i_disksize always represents the blocks which 989 * are used by this file. This allows recovery to restart truncate 990 * on orphans if we crash during truncate. We actually write i_disksize 991 * into the on-disk inode when writing inodes out, instead of i_size. 992 * 993 * The only time when i_disksize and i_size may be different is when 994 * a truncate is in progress. The only things which change i_disksize 995 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 996 */ 997 loff_t i_disksize; 998 999 /* 1000 * i_data_sem is for serialising ext4_truncate() against 1001 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1002 * data tree are chopped off during truncate. We can't do that in 1003 * ext4 because whenever we perform intermediate commits during 1004 * truncate, the inode and all the metadata blocks *must* be in a 1005 * consistent state which allows truncation of the orphans to restart 1006 * during recovery. Hence we must fix the get_block-vs-truncate race 1007 * by other means, so we have i_data_sem. 1008 */ 1009 struct rw_semaphore i_data_sem; 1010 /* 1011 * i_mmap_sem is for serializing page faults with truncate / punch hole 1012 * operations. We have to make sure that new page cannot be faulted in 1013 * a section of the inode that is being punched. We cannot easily use 1014 * i_data_sem for this since we need protection for the whole punch 1015 * operation and i_data_sem ranks below transaction start so we have 1016 * to occasionally drop it. 1017 */ 1018 struct rw_semaphore i_mmap_sem; 1019 struct inode vfs_inode; 1020 struct jbd2_inode *jinode; 1021 1022 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1023 1024 /* 1025 * File creation time. Its function is same as that of 1026 * struct timespec i_{a,c,m}time in the generic inode. 1027 */ 1028 struct timespec i_crtime; 1029 1030 /* mballoc */ 1031 struct list_head i_prealloc_list; 1032 spinlock_t i_prealloc_lock; 1033 1034 /* extents status tree */ 1035 struct ext4_es_tree i_es_tree; 1036 rwlock_t i_es_lock; 1037 struct list_head i_es_list; 1038 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1039 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1040 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1041 extents to shrink. Protected by 1042 i_es_lock */ 1043 1044 /* ialloc */ 1045 ext4_group_t i_last_alloc_group; 1046 1047 /* allocation reservation info for delalloc */ 1048 /* In case of bigalloc, this refer to clusters rather than blocks */ 1049 unsigned int i_reserved_data_blocks; 1050 ext4_lblk_t i_da_metadata_calc_last_lblock; 1051 int i_da_metadata_calc_len; 1052 1053 /* on-disk additional length */ 1054 __u16 i_extra_isize; 1055 1056 /* Indicate the inline data space. */ 1057 u16 i_inline_off; 1058 u16 i_inline_size; 1059 1060 #ifdef CONFIG_QUOTA 1061 /* quota space reservation, managed internally by quota code */ 1062 qsize_t i_reserved_quota; 1063 #endif 1064 1065 /* Lock protecting lists below */ 1066 spinlock_t i_completed_io_lock; 1067 /* 1068 * Completed IOs that need unwritten extents handling and have 1069 * transaction reserved 1070 */ 1071 struct list_head i_rsv_conversion_list; 1072 struct work_struct i_rsv_conversion_work; 1073 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1074 1075 spinlock_t i_block_reservation_lock; 1076 1077 /* 1078 * Transactions that contain inode's metadata needed to complete 1079 * fsync and fdatasync, respectively. 1080 */ 1081 tid_t i_sync_tid; 1082 tid_t i_datasync_tid; 1083 1084 #ifdef CONFIG_QUOTA 1085 struct dquot *i_dquot[MAXQUOTAS]; 1086 #endif 1087 1088 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1089 __u32 i_csum_seed; 1090 1091 kprojid_t i_projid; 1092 }; 1093 1094 /* 1095 * File system states 1096 */ 1097 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1098 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1099 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1100 1101 /* 1102 * Misc. filesystem flags 1103 */ 1104 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1105 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1106 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1107 1108 /* 1109 * Mount flags set via mount options or defaults 1110 */ 1111 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1112 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1113 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1114 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1115 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1116 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1117 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1118 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1119 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1120 #ifdef CONFIG_FS_DAX 1121 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1122 #else 1123 #define EXT4_MOUNT_DAX 0 1124 #endif 1125 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1126 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1127 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1128 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1129 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1130 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1131 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1132 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1133 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1134 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1135 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1136 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1137 * enable enforcement for hidden 1138 * quota files */ 1139 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1140 * enforcement for hidden quota 1141 * files */ 1142 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1143 * enforcement */ 1144 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1145 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1146 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1147 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1148 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1149 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1150 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1151 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1152 1153 /* 1154 * Mount flags set either automatically (could not be set by mount option) 1155 * based on per file system feature or property or in special cases such as 1156 * distinguishing between explicit mount option definition and default. 1157 */ 1158 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1159 specified delalloc */ 1160 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1161 size of blocksize * 8 1162 blocks */ 1163 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1164 file systems */ 1165 1166 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1167 specified journal checksum */ 1168 1169 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1170 ~EXT4_MOUNT_##opt 1171 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1172 EXT4_MOUNT_##opt 1173 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1174 EXT4_MOUNT_##opt) 1175 1176 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1177 ~EXT4_MOUNT2_##opt 1178 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1179 EXT4_MOUNT2_##opt 1180 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1181 EXT4_MOUNT2_##opt) 1182 1183 #define ext4_test_and_set_bit __test_and_set_bit_le 1184 #define ext4_set_bit __set_bit_le 1185 #define ext4_set_bit_atomic ext2_set_bit_atomic 1186 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1187 #define ext4_clear_bit __clear_bit_le 1188 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1189 #define ext4_test_bit test_bit_le 1190 #define ext4_find_next_zero_bit find_next_zero_bit_le 1191 #define ext4_find_next_bit find_next_bit_le 1192 1193 extern void ext4_set_bits(void *bm, int cur, int len); 1194 1195 /* 1196 * Maximal mount counts between two filesystem checks 1197 */ 1198 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1199 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1200 1201 /* 1202 * Behaviour when detecting errors 1203 */ 1204 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1205 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1206 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1207 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1208 1209 /* Metadata checksum algorithm codes */ 1210 #define EXT4_CRC32C_CHKSUM 1 1211 1212 /* 1213 * Structure of the super block 1214 */ 1215 struct ext4_super_block { 1216 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1217 __le32 s_blocks_count_lo; /* Blocks count */ 1218 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1219 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1220 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1221 __le32 s_first_data_block; /* First Data Block */ 1222 __le32 s_log_block_size; /* Block size */ 1223 __le32 s_log_cluster_size; /* Allocation cluster size */ 1224 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1225 __le32 s_clusters_per_group; /* # Clusters per group */ 1226 __le32 s_inodes_per_group; /* # Inodes per group */ 1227 __le32 s_mtime; /* Mount time */ 1228 /*30*/ __le32 s_wtime; /* Write time */ 1229 __le16 s_mnt_count; /* Mount count */ 1230 __le16 s_max_mnt_count; /* Maximal mount count */ 1231 __le16 s_magic; /* Magic signature */ 1232 __le16 s_state; /* File system state */ 1233 __le16 s_errors; /* Behaviour when detecting errors */ 1234 __le16 s_minor_rev_level; /* minor revision level */ 1235 /*40*/ __le32 s_lastcheck; /* time of last check */ 1236 __le32 s_checkinterval; /* max. time between checks */ 1237 __le32 s_creator_os; /* OS */ 1238 __le32 s_rev_level; /* Revision level */ 1239 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1240 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1241 /* 1242 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1243 * 1244 * Note: the difference between the compatible feature set and 1245 * the incompatible feature set is that if there is a bit set 1246 * in the incompatible feature set that the kernel doesn't 1247 * know about, it should refuse to mount the filesystem. 1248 * 1249 * e2fsck's requirements are more strict; if it doesn't know 1250 * about a feature in either the compatible or incompatible 1251 * feature set, it must abort and not try to meddle with 1252 * things it doesn't understand... 1253 */ 1254 __le32 s_first_ino; /* First non-reserved inode */ 1255 __le16 s_inode_size; /* size of inode structure */ 1256 __le16 s_block_group_nr; /* block group # of this superblock */ 1257 __le32 s_feature_compat; /* compatible feature set */ 1258 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1259 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1260 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1261 /*78*/ char s_volume_name[16]; /* volume name */ 1262 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1263 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1264 /* 1265 * Performance hints. Directory preallocation should only 1266 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1267 */ 1268 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1269 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1270 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1271 /* 1272 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1273 */ 1274 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1275 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1276 __le32 s_journal_dev; /* device number of journal file */ 1277 __le32 s_last_orphan; /* start of list of inodes to delete */ 1278 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1279 __u8 s_def_hash_version; /* Default hash version to use */ 1280 __u8 s_jnl_backup_type; 1281 __le16 s_desc_size; /* size of group descriptor */ 1282 /*100*/ __le32 s_default_mount_opts; 1283 __le32 s_first_meta_bg; /* First metablock block group */ 1284 __le32 s_mkfs_time; /* When the filesystem was created */ 1285 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1286 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1287 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1288 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1289 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1290 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1291 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1292 __le32 s_flags; /* Miscellaneous flags */ 1293 __le16 s_raid_stride; /* RAID stride */ 1294 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1295 __le64 s_mmp_block; /* Block for multi-mount protection */ 1296 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1297 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1298 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1299 __u8 s_encryption_level; /* versioning level for encryption */ 1300 __u8 s_reserved_pad; /* Padding to next 32bits */ 1301 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1302 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1303 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1304 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1305 snapshot's future use */ 1306 __le32 s_snapshot_list; /* inode number of the head of the 1307 on-disk snapshot list */ 1308 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1309 __le32 s_error_count; /* number of fs errors */ 1310 __le32 s_first_error_time; /* first time an error happened */ 1311 __le32 s_first_error_ino; /* inode involved in first error */ 1312 __le64 s_first_error_block; /* block involved of first error */ 1313 __u8 s_first_error_func[32]; /* function where the error happened */ 1314 __le32 s_first_error_line; /* line number where error happened */ 1315 __le32 s_last_error_time; /* most recent time of an error */ 1316 __le32 s_last_error_ino; /* inode involved in last error */ 1317 __le32 s_last_error_line; /* line number where error happened */ 1318 __le64 s_last_error_block; /* block involved of last error */ 1319 __u8 s_last_error_func[32]; /* function where the error happened */ 1320 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1321 __u8 s_mount_opts[64]; 1322 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1323 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1324 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1325 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1326 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1327 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1328 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1329 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1330 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1331 __le32 s_reserved[98]; /* Padding to the end of the block */ 1332 __le32 s_checksum; /* crc32c(superblock) */ 1333 }; 1334 1335 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1336 1337 #ifdef __KERNEL__ 1338 1339 /* 1340 * run-time mount flags 1341 */ 1342 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1343 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1344 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1345 1346 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1347 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1348 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1349 #else 1350 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1351 #endif 1352 1353 /* Number of quota types we support */ 1354 #define EXT4_MAXQUOTAS 3 1355 1356 /* 1357 * fourth extended-fs super-block data in memory 1358 */ 1359 struct ext4_sb_info { 1360 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1361 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1362 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1363 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1364 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1365 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1366 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1367 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1368 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1369 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1370 unsigned long s_overhead; /* # of fs overhead clusters */ 1371 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1372 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1373 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1374 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1375 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1376 struct buffer_head **s_group_desc; 1377 unsigned int s_mount_opt; 1378 unsigned int s_mount_opt2; 1379 unsigned int s_mount_flags; 1380 unsigned int s_def_mount_opt; 1381 ext4_fsblk_t s_sb_block; 1382 atomic64_t s_resv_clusters; 1383 kuid_t s_resuid; 1384 kgid_t s_resgid; 1385 unsigned short s_mount_state; 1386 unsigned short s_pad; 1387 int s_addr_per_block_bits; 1388 int s_desc_per_block_bits; 1389 int s_inode_size; 1390 int s_first_ino; 1391 unsigned int s_inode_readahead_blks; 1392 unsigned int s_inode_goal; 1393 spinlock_t s_next_gen_lock; 1394 u32 s_next_generation; 1395 u32 s_hash_seed[4]; 1396 int s_def_hash_version; 1397 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1398 struct percpu_counter s_freeclusters_counter; 1399 struct percpu_counter s_freeinodes_counter; 1400 struct percpu_counter s_dirs_counter; 1401 struct percpu_counter s_dirtyclusters_counter; 1402 struct blockgroup_lock *s_blockgroup_lock; 1403 struct proc_dir_entry *s_proc; 1404 struct kobject s_kobj; 1405 struct completion s_kobj_unregister; 1406 struct super_block *s_sb; 1407 1408 /* Journaling */ 1409 struct journal_s *s_journal; 1410 struct list_head s_orphan; 1411 struct mutex s_orphan_lock; 1412 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1413 unsigned long s_commit_interval; 1414 u32 s_max_batch_time; 1415 u32 s_min_batch_time; 1416 struct block_device *journal_bdev; 1417 #ifdef CONFIG_QUOTA 1418 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1419 int s_jquota_fmt; /* Format of quota to use */ 1420 #endif 1421 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1422 struct rb_root system_blks; 1423 1424 #ifdef EXTENTS_STATS 1425 /* ext4 extents stats */ 1426 unsigned long s_ext_min; 1427 unsigned long s_ext_max; 1428 unsigned long s_depth_max; 1429 spinlock_t s_ext_stats_lock; 1430 unsigned long s_ext_blocks; 1431 unsigned long s_ext_extents; 1432 #endif 1433 1434 /* for buddy allocator */ 1435 struct ext4_group_info ***s_group_info; 1436 struct inode *s_buddy_cache; 1437 spinlock_t s_md_lock; 1438 unsigned short *s_mb_offsets; 1439 unsigned int *s_mb_maxs; 1440 unsigned int s_group_info_size; 1441 unsigned int s_mb_free_pending; 1442 struct list_head s_freed_data_list; /* List of blocks to be freed 1443 after commit completed */ 1444 1445 /* tunables */ 1446 unsigned long s_stripe; 1447 unsigned int s_mb_stream_request; 1448 unsigned int s_mb_max_to_scan; 1449 unsigned int s_mb_min_to_scan; 1450 unsigned int s_mb_stats; 1451 unsigned int s_mb_order2_reqs; 1452 unsigned int s_mb_group_prealloc; 1453 unsigned int s_max_dir_size_kb; 1454 /* where last allocation was done - for stream allocation */ 1455 unsigned long s_mb_last_group; 1456 unsigned long s_mb_last_start; 1457 1458 /* stats for buddy allocator */ 1459 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1460 atomic_t s_bal_success; /* we found long enough chunks */ 1461 atomic_t s_bal_allocated; /* in blocks */ 1462 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1463 atomic_t s_bal_goals; /* goal hits */ 1464 atomic_t s_bal_breaks; /* too long searches */ 1465 atomic_t s_bal_2orders; /* 2^order hits */ 1466 spinlock_t s_bal_lock; 1467 unsigned long s_mb_buddies_generated; 1468 unsigned long long s_mb_generation_time; 1469 atomic_t s_mb_lost_chunks; 1470 atomic_t s_mb_preallocated; 1471 atomic_t s_mb_discarded; 1472 atomic_t s_lock_busy; 1473 1474 /* locality groups */ 1475 struct ext4_locality_group __percpu *s_locality_groups; 1476 1477 /* for write statistics */ 1478 unsigned long s_sectors_written_start; 1479 u64 s_kbytes_written; 1480 1481 /* the size of zero-out chunk */ 1482 unsigned int s_extent_max_zeroout_kb; 1483 1484 unsigned int s_log_groups_per_flex; 1485 struct flex_groups *s_flex_groups; 1486 ext4_group_t s_flex_groups_allocated; 1487 1488 /* workqueue for reserved extent conversions (buffered io) */ 1489 struct workqueue_struct *rsv_conversion_wq; 1490 1491 /* timer for periodic error stats printing */ 1492 struct timer_list s_err_report; 1493 1494 /* Lazy inode table initialization info */ 1495 struct ext4_li_request *s_li_request; 1496 /* Wait multiplier for lazy initialization thread */ 1497 unsigned int s_li_wait_mult; 1498 1499 /* Kernel thread for multiple mount protection */ 1500 struct task_struct *s_mmp_tsk; 1501 1502 /* record the last minlen when FITRIM is called. */ 1503 atomic_t s_last_trim_minblks; 1504 1505 /* Reference to checksum algorithm driver via cryptoapi */ 1506 struct crypto_shash *s_chksum_driver; 1507 1508 /* Precomputed FS UUID checksum for seeding other checksums */ 1509 __u32 s_csum_seed; 1510 1511 /* Reclaim extents from extent status tree */ 1512 struct shrinker s_es_shrinker; 1513 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1514 long s_es_nr_inode; 1515 struct ext4_es_stats s_es_stats; 1516 struct mb_cache *s_ea_block_cache; 1517 struct mb_cache *s_ea_inode_cache; 1518 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1519 1520 /* Ratelimit ext4 messages. */ 1521 struct ratelimit_state s_err_ratelimit_state; 1522 struct ratelimit_state s_warning_ratelimit_state; 1523 struct ratelimit_state s_msg_ratelimit_state; 1524 1525 /* Barrier between changing inodes' journal flags and writepages ops. */ 1526 struct percpu_rw_semaphore s_journal_flag_rwsem; 1527 struct dax_device *s_daxdev; 1528 }; 1529 1530 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1531 { 1532 return sb->s_fs_info; 1533 } 1534 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1535 { 1536 return container_of(inode, struct ext4_inode_info, vfs_inode); 1537 } 1538 1539 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1540 { 1541 return ino == EXT4_ROOT_INO || 1542 ino == EXT4_USR_QUOTA_INO || 1543 ino == EXT4_GRP_QUOTA_INO || 1544 ino == EXT4_BOOT_LOADER_INO || 1545 ino == EXT4_JOURNAL_INO || 1546 ino == EXT4_RESIZE_INO || 1547 (ino >= EXT4_FIRST_INO(sb) && 1548 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1549 } 1550 1551 /* 1552 * Inode dynamic state flags 1553 */ 1554 enum { 1555 EXT4_STATE_JDATA, /* journaled data exists */ 1556 EXT4_STATE_NEW, /* inode is newly created */ 1557 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1558 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1559 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1560 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1561 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1562 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1563 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1564 nolocking */ 1565 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1566 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1567 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1568 }; 1569 1570 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1571 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1572 { \ 1573 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1574 } \ 1575 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1576 { \ 1577 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1578 } \ 1579 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1580 { \ 1581 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1582 } 1583 1584 /* Add these declarations here only so that these functions can be 1585 * found by name. Otherwise, they are very hard to locate. */ 1586 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1587 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1588 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1589 EXT4_INODE_BIT_FNS(flag, flags, 0) 1590 1591 /* Add these declarations here only so that these functions can be 1592 * found by name. Otherwise, they are very hard to locate. */ 1593 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1594 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1595 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1596 #if (BITS_PER_LONG < 64) 1597 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1598 1599 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1600 { 1601 (ei)->i_state_flags = 0; 1602 } 1603 #else 1604 EXT4_INODE_BIT_FNS(state, flags, 32) 1605 1606 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1607 { 1608 /* We depend on the fact that callers will set i_flags */ 1609 } 1610 #endif 1611 #else 1612 /* Assume that user mode programs are passing in an ext4fs superblock, not 1613 * a kernel struct super_block. This will allow us to call the feature-test 1614 * macros from user land. */ 1615 #define EXT4_SB(sb) (sb) 1616 #endif 1617 1618 /* 1619 * Returns true if the inode is inode is encrypted 1620 */ 1621 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1622 1623 /* 1624 * Codes for operating systems 1625 */ 1626 #define EXT4_OS_LINUX 0 1627 #define EXT4_OS_HURD 1 1628 #define EXT4_OS_MASIX 2 1629 #define EXT4_OS_FREEBSD 3 1630 #define EXT4_OS_LITES 4 1631 1632 /* 1633 * Revision levels 1634 */ 1635 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1636 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1637 1638 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1639 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1640 1641 #define EXT4_GOOD_OLD_INODE_SIZE 128 1642 1643 /* 1644 * Feature set definitions 1645 */ 1646 1647 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1648 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1649 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1650 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1651 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1652 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1653 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1654 1655 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1656 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1657 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1658 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1659 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1660 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1661 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1662 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1663 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1664 /* 1665 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1666 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1667 * all other data structures' checksums. However, the METADATA_CSUM and 1668 * GDT_CSUM bits are mutually exclusive. 1669 */ 1670 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1671 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1672 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1673 1674 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1675 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1676 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1677 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1678 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1679 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1680 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1681 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1682 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1683 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1684 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1685 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1686 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1687 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1688 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1689 1690 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1691 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1692 { \ 1693 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1694 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1695 } \ 1696 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1697 { \ 1698 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1699 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1700 } \ 1701 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1702 { \ 1703 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1704 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1705 } 1706 1707 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1708 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1709 { \ 1710 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1711 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1712 } \ 1713 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1714 { \ 1715 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1716 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1717 } \ 1718 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1719 { \ 1720 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1721 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1722 } 1723 1724 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1725 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1726 { \ 1727 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1728 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1729 } \ 1730 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1731 { \ 1732 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1733 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1734 } \ 1735 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1736 { \ 1737 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1738 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1739 } 1740 1741 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1742 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1743 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1744 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1745 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1746 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1747 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1748 1749 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1750 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1751 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1752 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1753 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1754 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1755 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1756 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1757 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1758 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1759 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1760 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1761 1762 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1763 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1764 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1765 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1766 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1767 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1768 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1769 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1770 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1771 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1772 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1773 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1774 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1775 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1776 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1777 1778 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1779 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1780 EXT4_FEATURE_INCOMPAT_META_BG) 1781 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1782 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1783 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1784 1785 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1786 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1787 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1788 EXT4_FEATURE_INCOMPAT_META_BG) 1789 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1790 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1791 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1792 1793 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1794 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1795 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1796 EXT4_FEATURE_INCOMPAT_META_BG| \ 1797 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1798 EXT4_FEATURE_INCOMPAT_64BIT| \ 1799 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1800 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1801 EXT4_FEATURE_INCOMPAT_MMP | \ 1802 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1803 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 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 int err; 2042 2043 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2044 2045 desc.shash.tfm = sbi->s_chksum_driver; 2046 desc.shash.flags = 0; 2047 *(u32 *)desc.ctx = crc; 2048 2049 err = crypto_shash_update(&desc.shash, address, length); 2050 BUG_ON(err); 2051 2052 return *(u32 *)desc.ctx; 2053 } 2054 2055 #ifdef __KERNEL__ 2056 2057 /* hash info structure used by the directory hash */ 2058 struct dx_hash_info 2059 { 2060 u32 hash; 2061 u32 minor_hash; 2062 int hash_version; 2063 u32 *seed; 2064 }; 2065 2066 2067 /* 32 and 64 bit signed EOF for dx directories */ 2068 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2069 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2070 2071 2072 /* 2073 * Control parameters used by ext4_htree_next_block 2074 */ 2075 #define HASH_NB_ALWAYS 1 2076 2077 struct ext4_filename { 2078 const struct qstr *usr_fname; 2079 struct fscrypt_str disk_name; 2080 struct dx_hash_info hinfo; 2081 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2082 struct fscrypt_str crypto_buf; 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 static inline bool ext4_encrypted_inode(struct inode *inode) 2310 { 2311 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2312 } 2313 2314 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2315 static inline int ext4_fname_setup_filename(struct inode *dir, 2316 const struct qstr *iname, 2317 int lookup, struct ext4_filename *fname) 2318 { 2319 struct fscrypt_name name; 2320 int err; 2321 2322 memset(fname, 0, sizeof(struct ext4_filename)); 2323 2324 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2325 2326 fname->usr_fname = name.usr_fname; 2327 fname->disk_name = name.disk_name; 2328 fname->hinfo.hash = name.hash; 2329 fname->hinfo.minor_hash = name.minor_hash; 2330 fname->crypto_buf = name.crypto_buf; 2331 return err; 2332 } 2333 2334 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2335 { 2336 struct fscrypt_name name; 2337 2338 name.crypto_buf = fname->crypto_buf; 2339 fscrypt_free_filename(&name); 2340 2341 fname->crypto_buf.name = NULL; 2342 fname->usr_fname = NULL; 2343 fname->disk_name.name = NULL; 2344 } 2345 #else 2346 static inline int ext4_fname_setup_filename(struct inode *dir, 2347 const struct qstr *iname, 2348 int lookup, struct ext4_filename *fname) 2349 { 2350 fname->usr_fname = iname; 2351 fname->disk_name.name = (unsigned char *) iname->name; 2352 fname->disk_name.len = iname->len; 2353 return 0; 2354 } 2355 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2356 2357 #endif 2358 2359 /* dir.c */ 2360 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2361 struct file *, 2362 struct ext4_dir_entry_2 *, 2363 struct buffer_head *, char *, int, 2364 unsigned int); 2365 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2366 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2367 (de), (bh), (buf), (size), (offset))) 2368 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2369 __u32 minor_hash, 2370 struct ext4_dir_entry_2 *dirent, 2371 struct fscrypt_str *ent_name); 2372 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2373 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2374 struct buffer_head *bh, 2375 void *buf, int buf_size, 2376 struct ext4_filename *fname, 2377 struct ext4_dir_entry_2 **dest_de); 2378 void ext4_insert_dentry(struct inode *inode, 2379 struct ext4_dir_entry_2 *de, 2380 int buf_size, 2381 struct ext4_filename *fname); 2382 static inline void ext4_update_dx_flag(struct inode *inode) 2383 { 2384 if (!ext4_has_feature_dir_index(inode->i_sb)) 2385 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2386 } 2387 static const unsigned char ext4_filetype_table[] = { 2388 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2389 }; 2390 2391 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2392 { 2393 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2394 return DT_UNKNOWN; 2395 2396 return ext4_filetype_table[filetype]; 2397 } 2398 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2399 void *buf, int buf_size); 2400 2401 /* fsync.c */ 2402 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2403 2404 /* hash.c */ 2405 extern int ext4fs_dirhash(const char *name, int len, struct 2406 dx_hash_info *hinfo); 2407 2408 /* ialloc.c */ 2409 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2410 const struct qstr *qstr, __u32 goal, 2411 uid_t *owner, __u32 i_flags, 2412 int handle_type, unsigned int line_no, 2413 int nblocks); 2414 2415 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2416 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2417 i_flags, 0, 0, 0) 2418 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2419 type, nblocks) \ 2420 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2421 0, (type), __LINE__, (nblocks)) 2422 2423 2424 extern void ext4_free_inode(handle_t *, struct inode *); 2425 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2426 extern unsigned long ext4_count_free_inodes(struct super_block *); 2427 extern unsigned long ext4_count_dirs(struct super_block *); 2428 extern void ext4_check_inodes_bitmap(struct super_block *); 2429 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2430 extern int ext4_init_inode_table(struct super_block *sb, 2431 ext4_group_t group, int barrier); 2432 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2433 2434 /* mballoc.c */ 2435 extern const struct file_operations ext4_seq_mb_groups_fops; 2436 extern long ext4_mb_stats; 2437 extern long ext4_mb_max_to_scan; 2438 extern int ext4_mb_init(struct super_block *); 2439 extern int ext4_mb_release(struct super_block *); 2440 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2441 struct ext4_allocation_request *, int *); 2442 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2443 extern void ext4_discard_preallocations(struct inode *); 2444 extern int __init ext4_init_mballoc(void); 2445 extern void ext4_exit_mballoc(void); 2446 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2447 struct buffer_head *bh, ext4_fsblk_t block, 2448 unsigned long count, int flags); 2449 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2450 ext4_group_t ngroups); 2451 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2452 ext4_group_t i, struct ext4_group_desc *desc); 2453 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2454 ext4_fsblk_t block, unsigned long count); 2455 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2456 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2457 2458 /* inode.c */ 2459 int ext4_inode_is_fast_symlink(struct inode *inode); 2460 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2461 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2462 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2463 bool wait, struct buffer_head **bhs); 2464 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2465 struct buffer_head *bh_result, int create); 2466 int ext4_get_block(struct inode *inode, sector_t iblock, 2467 struct buffer_head *bh_result, int create); 2468 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2469 struct buffer_head *bh_result, int create); 2470 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2471 struct buffer_head *bh, int create); 2472 int ext4_walk_page_buffers(handle_t *handle, 2473 struct buffer_head *head, 2474 unsigned from, 2475 unsigned to, 2476 int *partial, 2477 int (*fn)(handle_t *handle, 2478 struct buffer_head *bh)); 2479 int do_journal_get_write_access(handle_t *handle, 2480 struct buffer_head *bh); 2481 #define FALL_BACK_TO_NONDELALLOC 1 2482 #define CONVERT_INLINE_DATA 2 2483 2484 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2485 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2486 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2487 extern int ext4_setattr(struct dentry *, struct iattr *); 2488 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2489 extern void ext4_evict_inode(struct inode *); 2490 extern void ext4_clear_inode(struct inode *); 2491 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2492 extern int ext4_sync_inode(handle_t *, struct inode *); 2493 extern void ext4_dirty_inode(struct inode *, int); 2494 extern int ext4_change_inode_journal_flag(struct inode *, int); 2495 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2496 extern int ext4_inode_attach_jinode(struct inode *inode); 2497 extern int ext4_can_truncate(struct inode *inode); 2498 extern int ext4_truncate(struct inode *); 2499 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2500 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2501 extern void ext4_set_inode_flags(struct inode *); 2502 extern int ext4_alloc_da_blocks(struct inode *inode); 2503 extern void ext4_set_aops(struct inode *inode); 2504 extern int ext4_writepage_trans_blocks(struct inode *); 2505 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2506 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2507 loff_t lstart, loff_t lend); 2508 extern int ext4_page_mkwrite(struct vm_fault *vmf); 2509 extern int ext4_filemap_fault(struct vm_fault *vmf); 2510 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2511 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2512 extern void ext4_da_update_reserve_space(struct inode *inode, 2513 int used, int quota_claim); 2514 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2515 ext4_fsblk_t pblk, ext4_lblk_t len); 2516 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk, 2517 unsigned int map_len, 2518 struct extent_status *result); 2519 2520 /* indirect.c */ 2521 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2522 struct ext4_map_blocks *map, int flags); 2523 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2524 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2525 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2526 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2527 ext4_lblk_t start, ext4_lblk_t end); 2528 2529 /* ioctl.c */ 2530 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2531 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2532 2533 /* migrate.c */ 2534 extern int ext4_ext_migrate(struct inode *); 2535 extern int ext4_ind_migrate(struct inode *inode); 2536 2537 /* namei.c */ 2538 extern int ext4_dirent_csum_verify(struct inode *inode, 2539 struct ext4_dir_entry *dirent); 2540 extern int ext4_orphan_add(handle_t *, struct inode *); 2541 extern int ext4_orphan_del(handle_t *, struct inode *); 2542 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2543 __u32 start_minor_hash, __u32 *next_hash); 2544 extern int ext4_search_dir(struct buffer_head *bh, 2545 char *search_buf, 2546 int buf_size, 2547 struct inode *dir, 2548 struct ext4_filename *fname, 2549 unsigned int offset, 2550 struct ext4_dir_entry_2 **res_dir); 2551 extern int ext4_generic_delete_entry(handle_t *handle, 2552 struct inode *dir, 2553 struct ext4_dir_entry_2 *de_del, 2554 struct buffer_head *bh, 2555 void *entry_buf, 2556 int buf_size, 2557 int csum_size); 2558 extern bool ext4_empty_dir(struct inode *inode); 2559 2560 /* resize.c */ 2561 extern int ext4_group_add(struct super_block *sb, 2562 struct ext4_new_group_data *input); 2563 extern int ext4_group_extend(struct super_block *sb, 2564 struct ext4_super_block *es, 2565 ext4_fsblk_t n_blocks_count); 2566 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2567 2568 /* super.c */ 2569 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2570 extern int ext4_calculate_overhead(struct super_block *sb); 2571 extern void ext4_superblock_csum_set(struct super_block *sb); 2572 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2573 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2574 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2575 ext4_group_t ngroup); 2576 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2577 char nbuf[16]); 2578 2579 extern __printf(4, 5) 2580 void __ext4_error(struct super_block *, const char *, unsigned int, 2581 const char *, ...); 2582 extern __printf(5, 6) 2583 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2584 const char *, ...); 2585 extern __printf(5, 6) 2586 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2587 const char *, ...); 2588 extern void __ext4_std_error(struct super_block *, const char *, 2589 unsigned int, int); 2590 extern __printf(4, 5) 2591 void __ext4_abort(struct super_block *, const char *, unsigned int, 2592 const char *, ...); 2593 extern __printf(4, 5) 2594 void __ext4_warning(struct super_block *, const char *, unsigned int, 2595 const char *, ...); 2596 extern __printf(4, 5) 2597 void __ext4_warning_inode(const struct inode *inode, const char *function, 2598 unsigned int line, const char *fmt, ...); 2599 extern __printf(3, 4) 2600 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2601 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2602 const char *, unsigned int, const char *); 2603 extern __printf(7, 8) 2604 void __ext4_grp_locked_error(const char *, unsigned int, 2605 struct super_block *, ext4_group_t, 2606 unsigned long, ext4_fsblk_t, 2607 const char *, ...); 2608 2609 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2610 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2611 2612 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2613 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2614 2615 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2616 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2617 2618 #ifdef CONFIG_PRINTK 2619 2620 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2621 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2622 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2623 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2624 #define ext4_error(sb, fmt, ...) \ 2625 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2626 #define ext4_abort(sb, fmt, ...) \ 2627 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2628 #define ext4_warning(sb, fmt, ...) \ 2629 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2630 #define ext4_warning_inode(inode, fmt, ...) \ 2631 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2632 #define ext4_msg(sb, level, fmt, ...) \ 2633 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2634 #define dump_mmp_msg(sb, mmp, msg) \ 2635 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2636 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2637 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2638 fmt, ##__VA_ARGS__) 2639 2640 #else 2641 2642 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2643 do { \ 2644 no_printk(fmt, ##__VA_ARGS__); \ 2645 __ext4_error_inode(inode, "", 0, block, " "); \ 2646 } while (0) 2647 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2648 do { \ 2649 no_printk(fmt, ##__VA_ARGS__); \ 2650 __ext4_error_file(file, "", 0, block, " "); \ 2651 } while (0) 2652 #define ext4_error(sb, fmt, ...) \ 2653 do { \ 2654 no_printk(fmt, ##__VA_ARGS__); \ 2655 __ext4_error(sb, "", 0, " "); \ 2656 } while (0) 2657 #define ext4_abort(sb, fmt, ...) \ 2658 do { \ 2659 no_printk(fmt, ##__VA_ARGS__); \ 2660 __ext4_abort(sb, "", 0, " "); \ 2661 } while (0) 2662 #define ext4_warning(sb, fmt, ...) \ 2663 do { \ 2664 no_printk(fmt, ##__VA_ARGS__); \ 2665 __ext4_warning(sb, "", 0, " "); \ 2666 } while (0) 2667 #define ext4_warning_inode(inode, fmt, ...) \ 2668 do { \ 2669 no_printk(fmt, ##__VA_ARGS__); \ 2670 __ext4_warning_inode(inode, "", 0, " "); \ 2671 } while (0) 2672 #define ext4_msg(sb, level, fmt, ...) \ 2673 do { \ 2674 no_printk(fmt, ##__VA_ARGS__); \ 2675 __ext4_msg(sb, "", " "); \ 2676 } while (0) 2677 #define dump_mmp_msg(sb, mmp, msg) \ 2678 __dump_mmp_msg(sb, mmp, "", 0, "") 2679 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2680 do { \ 2681 no_printk(fmt, ##__VA_ARGS__); \ 2682 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2683 } while (0) 2684 2685 #endif 2686 2687 extern void ext4_update_dynamic_rev(struct super_block *sb); 2688 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2689 __u32 compat); 2690 extern int ext4_update_rocompat_feature(handle_t *handle, 2691 struct super_block *sb, __u32 rocompat); 2692 extern int ext4_update_incompat_feature(handle_t *handle, 2693 struct super_block *sb, __u32 incompat); 2694 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2695 struct ext4_group_desc *bg); 2696 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2697 struct ext4_group_desc *bg); 2698 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2699 struct ext4_group_desc *bg); 2700 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2701 struct ext4_group_desc *bg); 2702 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2703 struct ext4_group_desc *bg); 2704 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2705 struct ext4_group_desc *bg); 2706 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2707 struct ext4_group_desc *bg); 2708 extern void ext4_block_bitmap_set(struct super_block *sb, 2709 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2710 extern void ext4_inode_bitmap_set(struct super_block *sb, 2711 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2712 extern void ext4_inode_table_set(struct super_block *sb, 2713 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2714 extern void ext4_free_group_clusters_set(struct super_block *sb, 2715 struct ext4_group_desc *bg, 2716 __u32 count); 2717 extern void ext4_free_inodes_set(struct super_block *sb, 2718 struct ext4_group_desc *bg, __u32 count); 2719 extern void ext4_used_dirs_set(struct super_block *sb, 2720 struct ext4_group_desc *bg, __u32 count); 2721 extern void ext4_itable_unused_set(struct super_block *sb, 2722 struct ext4_group_desc *bg, __u32 count); 2723 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2724 struct ext4_group_desc *gdp); 2725 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2726 struct ext4_group_desc *gdp); 2727 extern int ext4_register_li_request(struct super_block *sb, 2728 ext4_group_t first_not_zeroed); 2729 2730 static inline int ext4_has_metadata_csum(struct super_block *sb) 2731 { 2732 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2733 !EXT4_SB(sb)->s_chksum_driver); 2734 2735 return ext4_has_feature_metadata_csum(sb) && 2736 (EXT4_SB(sb)->s_chksum_driver != NULL); 2737 } 2738 2739 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2740 { 2741 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2742 } 2743 2744 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2745 { 2746 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2747 le32_to_cpu(es->s_blocks_count_lo); 2748 } 2749 2750 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2751 { 2752 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2753 le32_to_cpu(es->s_r_blocks_count_lo); 2754 } 2755 2756 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2757 { 2758 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2759 le32_to_cpu(es->s_free_blocks_count_lo); 2760 } 2761 2762 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2763 ext4_fsblk_t blk) 2764 { 2765 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2766 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2767 } 2768 2769 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2770 ext4_fsblk_t blk) 2771 { 2772 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2773 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2774 } 2775 2776 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2777 ext4_fsblk_t blk) 2778 { 2779 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2780 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2781 } 2782 2783 static inline loff_t ext4_isize(struct super_block *sb, 2784 struct ext4_inode *raw_inode) 2785 { 2786 if (ext4_has_feature_largedir(sb) || 2787 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2788 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2789 le32_to_cpu(raw_inode->i_size_lo); 2790 2791 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2792 } 2793 2794 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2795 { 2796 raw_inode->i_size_lo = cpu_to_le32(i_size); 2797 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2798 } 2799 2800 static inline 2801 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2802 ext4_group_t group) 2803 { 2804 struct ext4_group_info ***grp_info; 2805 long indexv, indexh; 2806 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2807 grp_info = EXT4_SB(sb)->s_group_info; 2808 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2809 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2810 return grp_info[indexv][indexh]; 2811 } 2812 2813 /* 2814 * Reading s_groups_count requires using smp_rmb() afterwards. See 2815 * the locking protocol documented in the comments of ext4_group_add() 2816 * in resize.c 2817 */ 2818 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2819 { 2820 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2821 2822 smp_rmb(); 2823 return ngroups; 2824 } 2825 2826 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2827 ext4_group_t block_group) 2828 { 2829 return block_group >> sbi->s_log_groups_per_flex; 2830 } 2831 2832 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2833 { 2834 return 1 << sbi->s_log_groups_per_flex; 2835 } 2836 2837 #define ext4_std_error(sb, errno) \ 2838 do { \ 2839 if ((errno)) \ 2840 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2841 } while (0) 2842 2843 #ifdef CONFIG_SMP 2844 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2845 * counters. So we need to make sure we have free clusters more 2846 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2847 */ 2848 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2849 #else 2850 #define EXT4_FREECLUSTERS_WATERMARK 0 2851 #endif 2852 2853 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2854 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2855 { 2856 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2857 !inode_is_locked(inode)); 2858 down_write(&EXT4_I(inode)->i_data_sem); 2859 if (newsize > EXT4_I(inode)->i_disksize) 2860 EXT4_I(inode)->i_disksize = newsize; 2861 up_write(&EXT4_I(inode)->i_data_sem); 2862 } 2863 2864 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2865 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2866 { 2867 int changed = 0; 2868 2869 if (newsize > inode->i_size) { 2870 i_size_write(inode, newsize); 2871 changed = 1; 2872 } 2873 if (newsize > EXT4_I(inode)->i_disksize) { 2874 ext4_update_i_disksize(inode, newsize); 2875 changed |= 2; 2876 } 2877 return changed; 2878 } 2879 2880 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2881 loff_t len); 2882 2883 struct ext4_group_info { 2884 unsigned long bb_state; 2885 struct rb_root bb_free_root; 2886 ext4_grpblk_t bb_first_free; /* first free block */ 2887 ext4_grpblk_t bb_free; /* total free blocks */ 2888 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2889 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2890 struct list_head bb_prealloc_list; 2891 #ifdef DOUBLE_CHECK 2892 void *bb_bitmap; 2893 #endif 2894 struct rw_semaphore alloc_sem; 2895 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2896 * regions, index is order. 2897 * bb_counters[3] = 5 means 2898 * 5 free 8-block regions. */ 2899 }; 2900 2901 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2902 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2903 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2904 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2905 2906 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2907 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2908 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2909 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2910 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2911 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2912 2913 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2914 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2915 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2916 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2917 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2918 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2919 2920 #define EXT4_MAX_CONTENTION 8 2921 #define EXT4_CONTENTION_THRESHOLD 2 2922 2923 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2924 ext4_group_t group) 2925 { 2926 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2927 } 2928 2929 /* 2930 * Returns true if the filesystem is busy enough that attempts to 2931 * access the block group locks has run into contention. 2932 */ 2933 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2934 { 2935 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2936 } 2937 2938 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2939 { 2940 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2941 if (spin_trylock(lock)) 2942 /* 2943 * We're able to grab the lock right away, so drop the 2944 * lock contention counter. 2945 */ 2946 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2947 else { 2948 /* 2949 * The lock is busy, so bump the contention counter, 2950 * and then wait on the spin lock. 2951 */ 2952 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2953 EXT4_MAX_CONTENTION); 2954 spin_lock(lock); 2955 } 2956 } 2957 2958 static inline void ext4_unlock_group(struct super_block *sb, 2959 ext4_group_t group) 2960 { 2961 spin_unlock(ext4_group_lock_ptr(sb, group)); 2962 } 2963 2964 /* 2965 * Block validity checking 2966 */ 2967 #define ext4_check_indirect_blockref(inode, bh) \ 2968 ext4_check_blockref(__func__, __LINE__, inode, \ 2969 (__le32 *)(bh)->b_data, \ 2970 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2971 2972 #define ext4_ind_check_inode(inode) \ 2973 ext4_check_blockref(__func__, __LINE__, inode, \ 2974 EXT4_I(inode)->i_data, \ 2975 EXT4_NDIR_BLOCKS) 2976 2977 /* 2978 * Inodes and files operations 2979 */ 2980 2981 /* dir.c */ 2982 extern const struct file_operations ext4_dir_operations; 2983 2984 /* file.c */ 2985 extern const struct inode_operations ext4_file_inode_operations; 2986 extern const struct file_operations ext4_file_operations; 2987 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2988 2989 /* inline.c */ 2990 extern int ext4_get_max_inline_size(struct inode *inode); 2991 extern int ext4_find_inline_data_nolock(struct inode *inode); 2992 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2993 unsigned int len); 2994 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2995 2996 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2997 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2998 struct inode *inode, 2999 loff_t pos, unsigned len, 3000 unsigned flags, 3001 struct page **pagep); 3002 extern int ext4_write_inline_data_end(struct inode *inode, 3003 loff_t pos, unsigned len, 3004 unsigned copied, 3005 struct page *page); 3006 extern struct buffer_head * 3007 ext4_journalled_write_inline_data(struct inode *inode, 3008 unsigned len, 3009 struct page *page); 3010 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3011 struct inode *inode, 3012 loff_t pos, unsigned len, 3013 unsigned flags, 3014 struct page **pagep, 3015 void **fsdata); 3016 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3017 unsigned len, unsigned copied, 3018 struct page *page); 3019 extern int ext4_try_add_inline_entry(handle_t *handle, 3020 struct ext4_filename *fname, 3021 struct inode *dir, struct inode *inode); 3022 extern int ext4_try_create_inline_dir(handle_t *handle, 3023 struct inode *parent, 3024 struct inode *inode); 3025 extern int ext4_read_inline_dir(struct file *filp, 3026 struct dir_context *ctx, 3027 int *has_inline_data); 3028 extern int htree_inlinedir_to_tree(struct file *dir_file, 3029 struct inode *dir, ext4_lblk_t block, 3030 struct dx_hash_info *hinfo, 3031 __u32 start_hash, __u32 start_minor_hash, 3032 int *has_inline_data); 3033 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3034 struct ext4_filename *fname, 3035 struct ext4_dir_entry_2 **res_dir, 3036 int *has_inline_data); 3037 extern int ext4_delete_inline_entry(handle_t *handle, 3038 struct inode *dir, 3039 struct ext4_dir_entry_2 *de_del, 3040 struct buffer_head *bh, 3041 int *has_inline_data); 3042 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3043 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3044 struct ext4_dir_entry_2 **parent_de, 3045 int *retval); 3046 extern int ext4_inline_data_fiemap(struct inode *inode, 3047 struct fiemap_extent_info *fieinfo, 3048 int *has_inline, __u64 start, __u64 len); 3049 extern int ext4_try_to_evict_inline_data(handle_t *handle, 3050 struct inode *inode, 3051 int needed); 3052 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3053 3054 extern int ext4_convert_inline_data(struct inode *inode); 3055 3056 static inline int ext4_has_inline_data(struct inode *inode) 3057 { 3058 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3059 EXT4_I(inode)->i_inline_off; 3060 } 3061 3062 /* namei.c */ 3063 extern const struct inode_operations ext4_dir_inode_operations; 3064 extern const struct inode_operations ext4_special_inode_operations; 3065 extern struct dentry *ext4_get_parent(struct dentry *child); 3066 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3067 struct ext4_dir_entry_2 *de, 3068 int blocksize, int csum_size, 3069 unsigned int parent_ino, int dotdot_real_len); 3070 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3071 unsigned int blocksize); 3072 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3073 struct inode *inode, 3074 struct buffer_head *bh); 3075 #define S_SHIFT 12 3076 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3077 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3078 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3079 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3080 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3081 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3082 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3083 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3084 }; 3085 3086 static inline void ext4_set_de_type(struct super_block *sb, 3087 struct ext4_dir_entry_2 *de, 3088 umode_t mode) { 3089 if (ext4_has_feature_filetype(sb)) 3090 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3091 } 3092 3093 /* readpages.c */ 3094 extern int ext4_mpage_readpages(struct address_space *mapping, 3095 struct list_head *pages, struct page *page, 3096 unsigned nr_pages); 3097 3098 /* symlink.c */ 3099 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3100 extern const struct inode_operations ext4_symlink_inode_operations; 3101 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3102 3103 /* sysfs.c */ 3104 extern int ext4_register_sysfs(struct super_block *sb); 3105 extern void ext4_unregister_sysfs(struct super_block *sb); 3106 extern int __init ext4_init_sysfs(void); 3107 extern void ext4_exit_sysfs(void); 3108 3109 /* block_validity */ 3110 extern void ext4_release_system_zone(struct super_block *sb); 3111 extern int ext4_setup_system_zone(struct super_block *sb); 3112 extern int __init ext4_init_system_zone(void); 3113 extern void ext4_exit_system_zone(void); 3114 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3115 ext4_fsblk_t start_blk, 3116 unsigned int count); 3117 extern int ext4_check_blockref(const char *, unsigned int, 3118 struct inode *, __le32 *, unsigned int); 3119 3120 /* extents.c */ 3121 struct ext4_ext_path; 3122 struct ext4_extent; 3123 3124 /* 3125 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3126 * __le32. 3127 */ 3128 #define EXT_MAX_BLOCKS 0xffffffff 3129 3130 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3131 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3132 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3133 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3134 struct ext4_map_blocks *map, int flags); 3135 extern int ext4_ext_truncate(handle_t *, struct inode *); 3136 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3137 ext4_lblk_t end); 3138 extern void ext4_ext_init(struct super_block *); 3139 extern void ext4_ext_release(struct super_block *); 3140 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3141 loff_t len); 3142 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3143 loff_t offset, ssize_t len); 3144 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3145 struct ext4_map_blocks *map, int flags); 3146 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3147 ext4_lblk_t lblocks); 3148 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3149 int num, 3150 struct ext4_ext_path *path); 3151 extern int ext4_can_extents_be_merged(struct inode *inode, 3152 struct ext4_extent *ex1, 3153 struct ext4_extent *ex2); 3154 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3155 struct ext4_ext_path **, 3156 struct ext4_extent *, int); 3157 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3158 struct ext4_ext_path **, 3159 int flags); 3160 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3161 extern int ext4_ext_check_inode(struct inode *inode); 3162 extern int ext4_find_delalloc_range(struct inode *inode, 3163 ext4_lblk_t lblk_start, 3164 ext4_lblk_t lblk_end); 3165 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3166 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3167 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3168 __u64 start, __u64 len); 3169 extern int ext4_ext_precache(struct inode *inode); 3170 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3171 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3172 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3173 struct inode *inode2, ext4_lblk_t lblk1, 3174 ext4_lblk_t lblk2, ext4_lblk_t count, 3175 int mark_unwritten,int *err); 3176 3177 /* move_extent.c */ 3178 extern void ext4_double_down_write_data_sem(struct inode *first, 3179 struct inode *second); 3180 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3181 struct inode *donor_inode); 3182 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3183 __u64 start_orig, __u64 start_donor, 3184 __u64 len, __u64 *moved_len); 3185 3186 /* page-io.c */ 3187 extern int __init ext4_init_pageio(void); 3188 extern void ext4_exit_pageio(void); 3189 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3190 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3191 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3192 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3193 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3194 struct writeback_control *wbc); 3195 extern void ext4_end_io_rsv_work(struct work_struct *work); 3196 extern void ext4_io_submit(struct ext4_io_submit *io); 3197 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3198 struct page *page, 3199 int len, 3200 struct writeback_control *wbc, 3201 bool keep_towrite); 3202 3203 /* mmp.c */ 3204 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3205 3206 /* 3207 * Add new method to test whether block and inode bitmaps are properly 3208 * initialized. With uninit_bg reading the block from disk is not enough 3209 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3210 */ 3211 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3212 3213 static inline int bitmap_uptodate(struct buffer_head *bh) 3214 { 3215 return (buffer_uptodate(bh) && 3216 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3217 } 3218 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3219 { 3220 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3221 } 3222 3223 /* 3224 * Disable DIO read nolock optimization, so new dioreaders will be forced 3225 * to grab i_mutex 3226 */ 3227 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 3228 { 3229 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3230 smp_mb(); 3231 } 3232 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 3233 { 3234 smp_mb(); 3235 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3236 } 3237 3238 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3239 3240 /* For ioend & aio unwritten conversion wait queues */ 3241 #define EXT4_WQ_HASH_SZ 37 3242 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3243 EXT4_WQ_HASH_SZ]) 3244 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3245 3246 extern int ext4_resize_begin(struct super_block *sb); 3247 extern void ext4_resize_end(struct super_block *sb); 3248 3249 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3250 struct ext4_io_end *io_end) 3251 { 3252 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3253 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3254 atomic_inc(&EXT4_I(inode)->i_unwritten); 3255 } 3256 } 3257 3258 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3259 { 3260 struct inode *inode = io_end->inode; 3261 3262 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3263 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3264 /* Wake up anyone waiting on unwritten extent conversion */ 3265 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3266 wake_up_all(ext4_ioend_wq(inode)); 3267 } 3268 } 3269 3270 extern const struct iomap_ops ext4_iomap_ops; 3271 3272 #endif /* __KERNEL__ */ 3273 3274 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3275 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3276 3277 #endif /* _EXT4_H */ 3278