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/wait.h> 30 #include <linux/blockgroup_lock.h> 31 #include <linux/percpu_counter.h> 32 #include <crypto/hash.h> 33 #ifdef __KERNEL__ 34 #include <linux/compat.h> 35 #endif 36 37 /* 38 * The fourth extended filesystem constants/structures 39 */ 40 41 /* 42 * Define EXT4FS_DEBUG to produce debug messages 43 */ 44 #undef EXT4FS_DEBUG 45 46 /* 47 * Debug code 48 */ 49 #ifdef EXT4FS_DEBUG 50 #define ext4_debug(f, a...) \ 51 do { \ 52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 53 __FILE__, __LINE__, __func__); \ 54 printk(KERN_DEBUG f, ## a); \ 55 } while (0) 56 #else 57 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 58 #endif 59 60 /* 61 * Turn on EXT_DEBUG to get lots of info about extents operations. 62 */ 63 #define EXT_DEBUG__ 64 #ifdef EXT_DEBUG 65 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 66 #else 67 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 68 #endif 69 70 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 71 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 72 73 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 75 76 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 78 79 /* data type for block offset of block group */ 80 typedef int ext4_grpblk_t; 81 82 /* data type for filesystem-wide blocks number */ 83 typedef unsigned long long ext4_fsblk_t; 84 85 /* data type for file logical block number */ 86 typedef __u32 ext4_lblk_t; 87 88 /* data type for block group number */ 89 typedef unsigned int ext4_group_t; 90 91 /* 92 * Flags used in mballoc's allocation_context flags field. 93 * 94 * Also used to show what's going on for debugging purposes when the 95 * flag field is exported via the traceport interface 96 */ 97 98 /* prefer goal again. length */ 99 #define EXT4_MB_HINT_MERGE 0x0001 100 /* blocks already reserved */ 101 #define EXT4_MB_HINT_RESERVED 0x0002 102 /* metadata is being allocated */ 103 #define EXT4_MB_HINT_METADATA 0x0004 104 /* first blocks in the file */ 105 #define EXT4_MB_HINT_FIRST 0x0008 106 /* search for the best chunk */ 107 #define EXT4_MB_HINT_BEST 0x0010 108 /* data is being allocated */ 109 #define EXT4_MB_HINT_DATA 0x0020 110 /* don't preallocate (for tails) */ 111 #define EXT4_MB_HINT_NOPREALLOC 0x0040 112 /* allocate for locality group */ 113 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 114 /* allocate goal blocks or none */ 115 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 116 /* goal is meaningful */ 117 #define EXT4_MB_HINT_TRY_GOAL 0x0200 118 /* blocks already pre-reserved by delayed allocation */ 119 #define EXT4_MB_DELALLOC_RESERVED 0x0400 120 /* We are doing stream allocation */ 121 #define EXT4_MB_STREAM_ALLOC 0x0800 122 /* Use reserved root blocks if needed */ 123 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 124 125 struct ext4_allocation_request { 126 /* target inode for block we're allocating */ 127 struct inode *inode; 128 /* how many blocks we want to allocate */ 129 unsigned int len; 130 /* logical block in target inode */ 131 ext4_lblk_t logical; 132 /* the closest logical allocated block to the left */ 133 ext4_lblk_t lleft; 134 /* the closest logical allocated block to the right */ 135 ext4_lblk_t lright; 136 /* phys. target (a hint) */ 137 ext4_fsblk_t goal; 138 /* phys. block for the closest logical allocated block to the left */ 139 ext4_fsblk_t pleft; 140 /* phys. block for the closest logical allocated block to the right */ 141 ext4_fsblk_t pright; 142 /* flags. see above EXT4_MB_HINT_* */ 143 unsigned int flags; 144 }; 145 146 /* 147 * Logical to physical block mapping, used by ext4_map_blocks() 148 * 149 * This structure is used to pass requests into ext4_map_blocks() as 150 * well as to store the information returned by ext4_map_blocks(). It 151 * takes less room on the stack than a struct buffer_head. 152 */ 153 #define EXT4_MAP_NEW (1 << BH_New) 154 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 155 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 156 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 157 #define EXT4_MAP_UNINIT (1 << BH_Uninit) 158 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of 159 * ext4_map_blocks wants to know whether or not the underlying cluster has 160 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that 161 * the requested mapping was from previously mapped (or delayed allocated) 162 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster 163 * should never appear on buffer_head's state flags. 164 */ 165 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster) 166 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 167 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 168 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER) 169 170 struct ext4_map_blocks { 171 ext4_fsblk_t m_pblk; 172 ext4_lblk_t m_lblk; 173 unsigned int m_len; 174 unsigned int m_flags; 175 }; 176 177 /* 178 * For delayed allocation tracking 179 */ 180 struct mpage_da_data { 181 struct inode *inode; 182 sector_t b_blocknr; /* start block number of extent */ 183 size_t b_size; /* size of extent */ 184 unsigned long b_state; /* state of the extent */ 185 unsigned long first_page, next_page; /* extent of pages */ 186 struct writeback_control *wbc; 187 int io_done; 188 int pages_written; 189 int retval; 190 }; 191 192 /* 193 * Flags for ext4_io_end->flags 194 */ 195 #define EXT4_IO_END_UNWRITTEN 0x0001 196 #define EXT4_IO_END_ERROR 0x0002 197 #define EXT4_IO_END_QUEUED 0x0004 198 #define EXT4_IO_END_DIRECT 0x0008 199 200 struct ext4_io_page { 201 struct page *p_page; 202 atomic_t p_count; 203 }; 204 205 #define MAX_IO_PAGES 128 206 207 /* 208 * For converting uninitialized extents on a work queue. 209 * 210 * 'page' is only used from the writepage() path; 'pages' is only used for 211 * buffered writes; they are used to keep page references until conversion 212 * takes place. For AIO/DIO, neither field is filled in. 213 */ 214 typedef struct ext4_io_end { 215 struct list_head list; /* per-file finished IO list */ 216 struct inode *inode; /* file being written to */ 217 unsigned int flag; /* unwritten or not */ 218 struct page *page; /* for writepage() path */ 219 loff_t offset; /* offset in the file */ 220 ssize_t size; /* size of the extent */ 221 struct work_struct work; /* data work queue */ 222 struct kiocb *iocb; /* iocb struct for AIO */ 223 int result; /* error value for AIO */ 224 int num_io_pages; /* for writepages() */ 225 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */ 226 } ext4_io_end_t; 227 228 struct ext4_io_submit { 229 int io_op; 230 struct bio *io_bio; 231 ext4_io_end_t *io_end; 232 struct ext4_io_page *io_page; 233 sector_t io_next_block; 234 }; 235 236 /* 237 * Special inodes numbers 238 */ 239 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 240 #define EXT4_ROOT_INO 2 /* Root inode */ 241 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 242 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 243 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 244 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 245 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 246 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 247 248 /* First non-reserved inode for old ext4 filesystems */ 249 #define EXT4_GOOD_OLD_FIRST_INO 11 250 251 /* 252 * Maximal count of links to a file 253 */ 254 #define EXT4_LINK_MAX 65000 255 256 /* 257 * Macro-instructions used to manage several block sizes 258 */ 259 #define EXT4_MIN_BLOCK_SIZE 1024 260 #define EXT4_MAX_BLOCK_SIZE 65536 261 #define EXT4_MIN_BLOCK_LOG_SIZE 10 262 #define EXT4_MAX_BLOCK_LOG_SIZE 16 263 #ifdef __KERNEL__ 264 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 265 #else 266 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 267 #endif 268 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 269 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 270 EXT4_SB(s)->s_cluster_bits) 271 #ifdef __KERNEL__ 272 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 273 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 274 #else 275 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 276 #endif 277 #ifdef __KERNEL__ 278 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 279 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 280 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 281 #else 282 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 283 EXT4_GOOD_OLD_INODE_SIZE : \ 284 (s)->s_inode_size) 285 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 286 EXT4_GOOD_OLD_FIRST_INO : \ 287 (s)->s_first_ino) 288 #endif 289 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 290 291 /* Translate a block number to a cluster number */ 292 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 293 /* Translate a cluster number to a block number */ 294 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 295 /* Translate # of blks to # of clusters */ 296 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 297 (sbi)->s_cluster_bits) 298 299 /* 300 * Structure of a blocks group descriptor 301 */ 302 struct ext4_group_desc 303 { 304 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 305 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 306 __le32 bg_inode_table_lo; /* Inodes table block */ 307 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 308 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 309 __le16 bg_used_dirs_count_lo; /* Directories count */ 310 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 311 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 312 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 313 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 314 __le16 bg_itable_unused_lo; /* Unused inodes count */ 315 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 316 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 317 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 318 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 319 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 320 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 321 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 322 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 323 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 324 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 325 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 326 __u32 bg_reserved; 327 }; 328 329 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 330 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 331 sizeof(__le16)) 332 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 333 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 334 sizeof(__le16)) 335 336 /* 337 * Structure of a flex block group info 338 */ 339 340 struct flex_groups { 341 atomic_t free_inodes; 342 atomic_t free_clusters; 343 atomic_t used_dirs; 344 }; 345 346 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 347 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 348 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 349 350 /* 351 * Macro-instructions used to manage group descriptors 352 */ 353 #define EXT4_MIN_DESC_SIZE 32 354 #define EXT4_MIN_DESC_SIZE_64BIT 64 355 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 356 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 357 #ifdef __KERNEL__ 358 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 359 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 361 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 362 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 363 #else 364 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 365 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 366 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 367 #endif 368 369 /* 370 * Constants relative to the data blocks 371 */ 372 #define EXT4_NDIR_BLOCKS 12 373 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 374 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 375 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 376 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 377 378 /* 379 * Inode flags 380 */ 381 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 382 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 383 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 384 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 385 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 386 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 387 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 388 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 389 /* Reserved for compression usage... */ 390 #define EXT4_DIRTY_FL 0x00000100 391 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 392 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 393 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 394 /* End compression flags --- maybe not all used */ 395 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 396 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 397 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 398 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 399 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 400 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 401 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 402 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 403 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 404 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 405 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 406 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 407 408 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 409 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */ 410 411 /* Flags that should be inherited by new inodes from their parent. */ 412 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 413 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 414 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 415 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 416 417 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 418 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 419 420 /* Flags that are appropriate for non-directories/regular files. */ 421 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 422 423 /* Mask out flags that are inappropriate for the given type of inode. */ 424 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 425 { 426 if (S_ISDIR(mode)) 427 return flags; 428 else if (S_ISREG(mode)) 429 return flags & EXT4_REG_FLMASK; 430 else 431 return flags & EXT4_OTHER_FLMASK; 432 } 433 434 /* 435 * Inode flags used for atomic set/get 436 */ 437 enum { 438 EXT4_INODE_SECRM = 0, /* Secure deletion */ 439 EXT4_INODE_UNRM = 1, /* Undelete */ 440 EXT4_INODE_COMPR = 2, /* Compress file */ 441 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 442 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 443 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 444 EXT4_INODE_NODUMP = 6, /* do not dump file */ 445 EXT4_INODE_NOATIME = 7, /* do not update atime */ 446 /* Reserved for compression usage... */ 447 EXT4_INODE_DIRTY = 8, 448 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 449 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 450 EXT4_INODE_ECOMPR = 11, /* Compression error */ 451 /* End compression flags --- maybe not all used */ 452 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 453 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 454 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 455 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 456 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 457 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 458 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 459 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 460 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 461 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 462 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 463 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 464 }; 465 466 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 467 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \ 468 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \ 469 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); } 470 471 /* 472 * Since it's pretty easy to mix up bit numbers and hex values, and we 473 * can't do a compile-time test for ENUM values, we use a run-time 474 * test to make sure that EXT4_XXX_FL is consistent with respect to 475 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop 476 * out so it won't cost any extra space in the compiled kernel image. 477 * But it's important that these values are the same, since we are 478 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL 479 * must be consistent with the values of FS_XXX_FL defined in 480 * include/linux/fs.h and the on-disk values found in ext2, ext3, and 481 * ext4 filesystems, and of course the values defined in e2fsprogs. 482 * 483 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 484 */ 485 static inline void ext4_check_flag_values(void) 486 { 487 CHECK_FLAG_VALUE(SECRM); 488 CHECK_FLAG_VALUE(UNRM); 489 CHECK_FLAG_VALUE(COMPR); 490 CHECK_FLAG_VALUE(SYNC); 491 CHECK_FLAG_VALUE(IMMUTABLE); 492 CHECK_FLAG_VALUE(APPEND); 493 CHECK_FLAG_VALUE(NODUMP); 494 CHECK_FLAG_VALUE(NOATIME); 495 CHECK_FLAG_VALUE(DIRTY); 496 CHECK_FLAG_VALUE(COMPRBLK); 497 CHECK_FLAG_VALUE(NOCOMPR); 498 CHECK_FLAG_VALUE(ECOMPR); 499 CHECK_FLAG_VALUE(INDEX); 500 CHECK_FLAG_VALUE(IMAGIC); 501 CHECK_FLAG_VALUE(JOURNAL_DATA); 502 CHECK_FLAG_VALUE(NOTAIL); 503 CHECK_FLAG_VALUE(DIRSYNC); 504 CHECK_FLAG_VALUE(TOPDIR); 505 CHECK_FLAG_VALUE(HUGE_FILE); 506 CHECK_FLAG_VALUE(EXTENTS); 507 CHECK_FLAG_VALUE(EA_INODE); 508 CHECK_FLAG_VALUE(EOFBLOCKS); 509 CHECK_FLAG_VALUE(INLINE_DATA); 510 CHECK_FLAG_VALUE(RESERVED); 511 } 512 513 /* Used to pass group descriptor data when online resize is done */ 514 struct ext4_new_group_input { 515 __u32 group; /* Group number for this data */ 516 __u64 block_bitmap; /* Absolute block number of block bitmap */ 517 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 518 __u64 inode_table; /* Absolute block number of inode table start */ 519 __u32 blocks_count; /* Total number of blocks in this group */ 520 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 521 __u16 unused; 522 }; 523 524 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 525 struct compat_ext4_new_group_input { 526 u32 group; 527 compat_u64 block_bitmap; 528 compat_u64 inode_bitmap; 529 compat_u64 inode_table; 530 u32 blocks_count; 531 u16 reserved_blocks; 532 u16 unused; 533 }; 534 #endif 535 536 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 537 struct ext4_new_group_data { 538 __u32 group; 539 __u64 block_bitmap; 540 __u64 inode_bitmap; 541 __u64 inode_table; 542 __u32 blocks_count; 543 __u16 reserved_blocks; 544 __u16 unused; 545 __u32 free_blocks_count; 546 }; 547 548 /* Indexes used to index group tables in ext4_new_group_data */ 549 enum { 550 BLOCK_BITMAP = 0, /* block bitmap */ 551 INODE_BITMAP, /* inode bitmap */ 552 INODE_TABLE, /* inode tables */ 553 GROUP_TABLE_COUNT, 554 }; 555 556 /* 557 * Flags used by ext4_map_blocks() 558 */ 559 /* Allocate any needed blocks and/or convert an unitialized 560 extent to be an initialized ext4 */ 561 #define EXT4_GET_BLOCKS_CREATE 0x0001 562 /* Request the creation of an unitialized extent */ 563 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 564 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 565 EXT4_GET_BLOCKS_CREATE) 566 /* Caller is from the delayed allocation writeout path, 567 so set the magic i_delalloc_reserve_flag after taking the 568 inode allocation semaphore for */ 569 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 570 /* caller is from the direct IO path, request to creation of an 571 unitialized extents if not allocated, split the uninitialized 572 extent if blocks has been preallocated already*/ 573 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 574 #define EXT4_GET_BLOCKS_CONVERT 0x0010 575 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 576 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 577 /* Convert extent to initialized after IO complete */ 578 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 579 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 580 /* Punch out blocks of an extent */ 581 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020 582 /* Don't normalize allocation size (used for fallocate) */ 583 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 584 /* Request will not result in inode size update (user for fallocate) */ 585 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 586 /* Do not take i_data_sem locking in ext4_map_blocks */ 587 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 588 589 /* 590 * Flags used by ext4_free_blocks 591 */ 592 #define EXT4_FREE_BLOCKS_METADATA 0x0001 593 #define EXT4_FREE_BLOCKS_FORGET 0x0002 594 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 595 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 596 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 597 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 598 599 /* 600 * Flags used by ext4_discard_partial_page_buffers 601 */ 602 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001 603 604 /* 605 * ioctl commands 606 */ 607 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 608 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 609 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 610 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 611 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 612 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 613 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 614 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 615 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 616 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 617 #define EXT4_IOC_MIGRATE _IO('f', 9) 618 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 619 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 620 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 621 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 622 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 623 624 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 625 /* 626 * ioctl commands in 32 bit emulation 627 */ 628 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 629 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 630 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 631 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 632 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 633 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 634 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 635 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 636 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 637 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 638 #endif 639 640 /* Max physical block we can address w/o extents */ 641 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 642 643 /* 644 * Structure of an inode on the disk 645 */ 646 struct ext4_inode { 647 __le16 i_mode; /* File mode */ 648 __le16 i_uid; /* Low 16 bits of Owner Uid */ 649 __le32 i_size_lo; /* Size in bytes */ 650 __le32 i_atime; /* Access time */ 651 __le32 i_ctime; /* Inode Change time */ 652 __le32 i_mtime; /* Modification time */ 653 __le32 i_dtime; /* Deletion Time */ 654 __le16 i_gid; /* Low 16 bits of Group Id */ 655 __le16 i_links_count; /* Links count */ 656 __le32 i_blocks_lo; /* Blocks count */ 657 __le32 i_flags; /* File flags */ 658 union { 659 struct { 660 __le32 l_i_version; 661 } linux1; 662 struct { 663 __u32 h_i_translator; 664 } hurd1; 665 struct { 666 __u32 m_i_reserved1; 667 } masix1; 668 } osd1; /* OS dependent 1 */ 669 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 670 __le32 i_generation; /* File version (for NFS) */ 671 __le32 i_file_acl_lo; /* File ACL */ 672 __le32 i_size_high; 673 __le32 i_obso_faddr; /* Obsoleted fragment address */ 674 union { 675 struct { 676 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 677 __le16 l_i_file_acl_high; 678 __le16 l_i_uid_high; /* these 2 fields */ 679 __le16 l_i_gid_high; /* were reserved2[0] */ 680 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 681 __le16 l_i_reserved; 682 } linux2; 683 struct { 684 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 685 __u16 h_i_mode_high; 686 __u16 h_i_uid_high; 687 __u16 h_i_gid_high; 688 __u32 h_i_author; 689 } hurd2; 690 struct { 691 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 692 __le16 m_i_file_acl_high; 693 __u32 m_i_reserved2[2]; 694 } masix2; 695 } osd2; /* OS dependent 2 */ 696 __le16 i_extra_isize; 697 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 698 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 699 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 700 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 701 __le32 i_crtime; /* File Creation time */ 702 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 703 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 704 }; 705 706 struct move_extent { 707 __u32 reserved; /* should be zero */ 708 __u32 donor_fd; /* donor file descriptor */ 709 __u64 orig_start; /* logical start offset in block for orig */ 710 __u64 donor_start; /* logical start offset in block for donor */ 711 __u64 len; /* block length to be moved */ 712 __u64 moved_len; /* moved block length */ 713 }; 714 715 #define EXT4_EPOCH_BITS 2 716 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 717 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 718 719 /* 720 * Extended fields will fit into an inode if the filesystem was formatted 721 * with large inodes (-I 256 or larger) and there are not currently any EAs 722 * consuming all of the available space. For new inodes we always reserve 723 * enough space for the kernel's known extended fields, but for inodes 724 * created with an old kernel this might not have been the case. None of 725 * the extended inode fields is critical for correct filesystem operation. 726 * This macro checks if a certain field fits in the inode. Note that 727 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 728 */ 729 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 730 ((offsetof(typeof(*ext4_inode), field) + \ 731 sizeof((ext4_inode)->field)) \ 732 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 733 (einode)->i_extra_isize)) \ 734 735 static inline __le32 ext4_encode_extra_time(struct timespec *time) 736 { 737 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 738 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 739 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 740 } 741 742 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 743 { 744 if (sizeof(time->tv_sec) > 4) 745 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 746 << 32; 747 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 748 } 749 750 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 751 do { \ 752 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 753 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 754 (raw_inode)->xtime ## _extra = \ 755 ext4_encode_extra_time(&(inode)->xtime); \ 756 } while (0) 757 758 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 759 do { \ 760 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 761 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 762 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 763 (raw_inode)->xtime ## _extra = \ 764 ext4_encode_extra_time(&(einode)->xtime); \ 765 } while (0) 766 767 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 768 do { \ 769 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 770 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 771 ext4_decode_extra_time(&(inode)->xtime, \ 772 raw_inode->xtime ## _extra); \ 773 else \ 774 (inode)->xtime.tv_nsec = 0; \ 775 } while (0) 776 777 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 778 do { \ 779 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 780 (einode)->xtime.tv_sec = \ 781 (signed)le32_to_cpu((raw_inode)->xtime); \ 782 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 783 ext4_decode_extra_time(&(einode)->xtime, \ 784 raw_inode->xtime ## _extra); \ 785 else \ 786 (einode)->xtime.tv_nsec = 0; \ 787 } while (0) 788 789 #define i_disk_version osd1.linux1.l_i_version 790 791 #if defined(__KERNEL__) || defined(__linux__) 792 #define i_reserved1 osd1.linux1.l_i_reserved1 793 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 794 #define i_blocks_high osd2.linux2.l_i_blocks_high 795 #define i_uid_low i_uid 796 #define i_gid_low i_gid 797 #define i_uid_high osd2.linux2.l_i_uid_high 798 #define i_gid_high osd2.linux2.l_i_gid_high 799 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 800 801 #elif defined(__GNU__) 802 803 #define i_translator osd1.hurd1.h_i_translator 804 #define i_uid_high osd2.hurd2.h_i_uid_high 805 #define i_gid_high osd2.hurd2.h_i_gid_high 806 #define i_author osd2.hurd2.h_i_author 807 808 #elif defined(__masix__) 809 810 #define i_reserved1 osd1.masix1.m_i_reserved1 811 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 812 #define i_reserved2 osd2.masix2.m_i_reserved2 813 814 #endif /* defined(__KERNEL__) || defined(__linux__) */ 815 816 /* 817 * storage for cached extent 818 * If ec_len == 0, then the cache is invalid. 819 * If ec_start == 0, then the cache represents a gap (null mapping) 820 */ 821 struct ext4_ext_cache { 822 ext4_fsblk_t ec_start; 823 ext4_lblk_t ec_block; 824 __u32 ec_len; /* must be 32bit to return holes */ 825 }; 826 827 #include "extents_status.h" 828 829 /* 830 * fourth extended file system inode data in memory 831 */ 832 struct ext4_inode_info { 833 __le32 i_data[15]; /* unconverted */ 834 __u32 i_dtime; 835 ext4_fsblk_t i_file_acl; 836 837 /* 838 * i_block_group is the number of the block group which contains 839 * this file's inode. Constant across the lifetime of the inode, 840 * it is ued for making block allocation decisions - we try to 841 * place a file's data blocks near its inode block, and new inodes 842 * near to their parent directory's inode. 843 */ 844 ext4_group_t i_block_group; 845 ext4_lblk_t i_dir_start_lookup; 846 #if (BITS_PER_LONG < 64) 847 unsigned long i_state_flags; /* Dynamic state flags */ 848 #endif 849 unsigned long i_flags; 850 851 /* 852 * Extended attributes can be read independently of the main file 853 * data. Taking i_mutex even when reading would cause contention 854 * between readers of EAs and writers of regular file data, so 855 * instead we synchronize on xattr_sem when reading or changing 856 * EAs. 857 */ 858 struct rw_semaphore xattr_sem; 859 860 struct list_head i_orphan; /* unlinked but open inodes */ 861 862 /* 863 * i_disksize keeps track of what the inode size is ON DISK, not 864 * in memory. During truncate, i_size is set to the new size by 865 * the VFS prior to calling ext4_truncate(), but the filesystem won't 866 * set i_disksize to 0 until the truncate is actually under way. 867 * 868 * The intent is that i_disksize always represents the blocks which 869 * are used by this file. This allows recovery to restart truncate 870 * on orphans if we crash during truncate. We actually write i_disksize 871 * into the on-disk inode when writing inodes out, instead of i_size. 872 * 873 * The only time when i_disksize and i_size may be different is when 874 * a truncate is in progress. The only things which change i_disksize 875 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 876 */ 877 loff_t i_disksize; 878 879 /* 880 * i_data_sem is for serialising ext4_truncate() against 881 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 882 * data tree are chopped off during truncate. We can't do that in 883 * ext4 because whenever we perform intermediate commits during 884 * truncate, the inode and all the metadata blocks *must* be in a 885 * consistent state which allows truncation of the orphans to restart 886 * during recovery. Hence we must fix the get_block-vs-truncate race 887 * by other means, so we have i_data_sem. 888 */ 889 struct rw_semaphore i_data_sem; 890 struct inode vfs_inode; 891 struct jbd2_inode *jinode; 892 893 struct ext4_ext_cache i_cached_extent; 894 /* 895 * File creation time. Its function is same as that of 896 * struct timespec i_{a,c,m}time in the generic inode. 897 */ 898 struct timespec i_crtime; 899 900 /* mballoc */ 901 struct list_head i_prealloc_list; 902 spinlock_t i_prealloc_lock; 903 904 /* extents status tree */ 905 struct ext4_es_tree i_es_tree; 906 rwlock_t i_es_lock; 907 908 /* ialloc */ 909 ext4_group_t i_last_alloc_group; 910 911 /* allocation reservation info for delalloc */ 912 /* In case of bigalloc, these refer to clusters rather than blocks */ 913 unsigned int i_reserved_data_blocks; 914 unsigned int i_reserved_meta_blocks; 915 unsigned int i_allocated_meta_blocks; 916 ext4_lblk_t i_da_metadata_calc_last_lblock; 917 int i_da_metadata_calc_len; 918 919 /* on-disk additional length */ 920 __u16 i_extra_isize; 921 922 /* Indicate the inline data space. */ 923 u16 i_inline_off; 924 u16 i_inline_size; 925 926 #ifdef CONFIG_QUOTA 927 /* quota space reservation, managed internally by quota code */ 928 qsize_t i_reserved_quota; 929 #endif 930 931 /* completed IOs that might need unwritten extents handling */ 932 struct list_head i_completed_io_list; 933 spinlock_t i_completed_io_lock; 934 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 935 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 936 937 spinlock_t i_block_reservation_lock; 938 939 /* 940 * Transactions that contain inode's metadata needed to complete 941 * fsync and fdatasync, respectively. 942 */ 943 tid_t i_sync_tid; 944 tid_t i_datasync_tid; 945 946 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 947 __u32 i_csum_seed; 948 }; 949 950 /* 951 * File system states 952 */ 953 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 954 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 955 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 956 957 /* 958 * Misc. filesystem flags 959 */ 960 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 961 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 962 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 963 964 /* 965 * Mount flags 966 */ 967 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 968 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 969 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 970 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 971 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 972 #define EXT4_MOUNT_ERRORS_MASK 0x00070 973 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 974 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 975 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 976 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 977 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 978 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 979 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 980 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 981 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 982 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 983 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 984 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 985 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 986 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 987 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 988 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 989 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 990 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 991 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */ 992 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 993 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 994 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 995 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 996 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 997 998 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 999 specified delalloc */ 1000 1001 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1002 ~EXT4_MOUNT_##opt 1003 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1004 EXT4_MOUNT_##opt 1005 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1006 EXT4_MOUNT_##opt) 1007 1008 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1009 ~EXT4_MOUNT2_##opt 1010 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1011 EXT4_MOUNT2_##opt 1012 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1013 EXT4_MOUNT2_##opt) 1014 1015 #define ext4_test_and_set_bit __test_and_set_bit_le 1016 #define ext4_set_bit __set_bit_le 1017 #define ext4_set_bit_atomic ext2_set_bit_atomic 1018 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1019 #define ext4_clear_bit __clear_bit_le 1020 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1021 #define ext4_test_bit test_bit_le 1022 #define ext4_find_next_zero_bit find_next_zero_bit_le 1023 #define ext4_find_next_bit find_next_bit_le 1024 1025 extern void ext4_set_bits(void *bm, int cur, int len); 1026 1027 /* 1028 * Maximal mount counts between two filesystem checks 1029 */ 1030 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1031 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1032 1033 /* 1034 * Behaviour when detecting errors 1035 */ 1036 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1037 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1038 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1039 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1040 1041 /* Metadata checksum algorithm codes */ 1042 #define EXT4_CRC32C_CHKSUM 1 1043 1044 /* 1045 * Structure of the super block 1046 */ 1047 struct ext4_super_block { 1048 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1049 __le32 s_blocks_count_lo; /* Blocks count */ 1050 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1051 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1052 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1053 __le32 s_first_data_block; /* First Data Block */ 1054 __le32 s_log_block_size; /* Block size */ 1055 __le32 s_log_cluster_size; /* Allocation cluster size */ 1056 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1057 __le32 s_clusters_per_group; /* # Clusters per group */ 1058 __le32 s_inodes_per_group; /* # Inodes per group */ 1059 __le32 s_mtime; /* Mount time */ 1060 /*30*/ __le32 s_wtime; /* Write time */ 1061 __le16 s_mnt_count; /* Mount count */ 1062 __le16 s_max_mnt_count; /* Maximal mount count */ 1063 __le16 s_magic; /* Magic signature */ 1064 __le16 s_state; /* File system state */ 1065 __le16 s_errors; /* Behaviour when detecting errors */ 1066 __le16 s_minor_rev_level; /* minor revision level */ 1067 /*40*/ __le32 s_lastcheck; /* time of last check */ 1068 __le32 s_checkinterval; /* max. time between checks */ 1069 __le32 s_creator_os; /* OS */ 1070 __le32 s_rev_level; /* Revision level */ 1071 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1072 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1073 /* 1074 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1075 * 1076 * Note: the difference between the compatible feature set and 1077 * the incompatible feature set is that if there is a bit set 1078 * in the incompatible feature set that the kernel doesn't 1079 * know about, it should refuse to mount the filesystem. 1080 * 1081 * e2fsck's requirements are more strict; if it doesn't know 1082 * about a feature in either the compatible or incompatible 1083 * feature set, it must abort and not try to meddle with 1084 * things it doesn't understand... 1085 */ 1086 __le32 s_first_ino; /* First non-reserved inode */ 1087 __le16 s_inode_size; /* size of inode structure */ 1088 __le16 s_block_group_nr; /* block group # of this superblock */ 1089 __le32 s_feature_compat; /* compatible feature set */ 1090 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1091 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1092 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1093 /*78*/ char s_volume_name[16]; /* volume name */ 1094 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1095 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1096 /* 1097 * Performance hints. Directory preallocation should only 1098 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1099 */ 1100 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1101 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1102 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1103 /* 1104 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1105 */ 1106 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1107 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1108 __le32 s_journal_dev; /* device number of journal file */ 1109 __le32 s_last_orphan; /* start of list of inodes to delete */ 1110 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1111 __u8 s_def_hash_version; /* Default hash version to use */ 1112 __u8 s_jnl_backup_type; 1113 __le16 s_desc_size; /* size of group descriptor */ 1114 /*100*/ __le32 s_default_mount_opts; 1115 __le32 s_first_meta_bg; /* First metablock block group */ 1116 __le32 s_mkfs_time; /* When the filesystem was created */ 1117 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1118 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1119 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1120 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1121 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1122 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1123 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1124 __le32 s_flags; /* Miscellaneous flags */ 1125 __le16 s_raid_stride; /* RAID stride */ 1126 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1127 __le64 s_mmp_block; /* Block for multi-mount protection */ 1128 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1129 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1130 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1131 __le16 s_reserved_pad; 1132 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1133 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1134 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1135 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1136 snapshot's future use */ 1137 __le32 s_snapshot_list; /* inode number of the head of the 1138 on-disk snapshot list */ 1139 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1140 __le32 s_error_count; /* number of fs errors */ 1141 __le32 s_first_error_time; /* first time an error happened */ 1142 __le32 s_first_error_ino; /* inode involved in first error */ 1143 __le64 s_first_error_block; /* block involved of first error */ 1144 __u8 s_first_error_func[32]; /* function where the error happened */ 1145 __le32 s_first_error_line; /* line number where error happened */ 1146 __le32 s_last_error_time; /* most recent time of an error */ 1147 __le32 s_last_error_ino; /* inode involved in last error */ 1148 __le32 s_last_error_line; /* line number where error happened */ 1149 __le64 s_last_error_block; /* block involved of last error */ 1150 __u8 s_last_error_func[32]; /* function where the error happened */ 1151 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1152 __u8 s_mount_opts[64]; 1153 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1154 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1155 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1156 __le32 s_reserved[108]; /* Padding to the end of the block */ 1157 __le32 s_checksum; /* crc32c(superblock) */ 1158 }; 1159 1160 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1161 1162 #ifdef __KERNEL__ 1163 1164 /* 1165 * run-time mount flags 1166 */ 1167 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1168 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1169 1170 /* 1171 * fourth extended-fs super-block data in memory 1172 */ 1173 struct ext4_sb_info { 1174 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1175 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1176 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1177 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1178 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1179 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1180 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1181 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1182 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1183 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1184 unsigned long s_overhead; /* # of fs overhead clusters */ 1185 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1186 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1187 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1188 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1189 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1190 struct buffer_head **s_group_desc; 1191 unsigned int s_mount_opt; 1192 unsigned int s_mount_opt2; 1193 unsigned int s_mount_flags; 1194 unsigned int s_def_mount_opt; 1195 ext4_fsblk_t s_sb_block; 1196 kuid_t s_resuid; 1197 kgid_t s_resgid; 1198 unsigned short s_mount_state; 1199 unsigned short s_pad; 1200 int s_addr_per_block_bits; 1201 int s_desc_per_block_bits; 1202 int s_inode_size; 1203 int s_first_ino; 1204 unsigned int s_inode_readahead_blks; 1205 unsigned int s_inode_goal; 1206 spinlock_t s_next_gen_lock; 1207 u32 s_next_generation; 1208 u32 s_hash_seed[4]; 1209 int s_def_hash_version; 1210 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1211 struct percpu_counter s_freeclusters_counter; 1212 struct percpu_counter s_freeinodes_counter; 1213 struct percpu_counter s_dirs_counter; 1214 struct percpu_counter s_dirtyclusters_counter; 1215 struct blockgroup_lock *s_blockgroup_lock; 1216 struct proc_dir_entry *s_proc; 1217 struct kobject s_kobj; 1218 struct completion s_kobj_unregister; 1219 struct super_block *s_sb; 1220 1221 /* Journaling */ 1222 struct journal_s *s_journal; 1223 struct list_head s_orphan; 1224 struct mutex s_orphan_lock; 1225 unsigned long s_resize_flags; /* Flags indicating if there 1226 is a resizer */ 1227 unsigned long s_commit_interval; 1228 u32 s_max_batch_time; 1229 u32 s_min_batch_time; 1230 struct block_device *journal_bdev; 1231 #ifdef CONFIG_QUOTA 1232 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 1233 int s_jquota_fmt; /* Format of quota to use */ 1234 #endif 1235 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1236 struct rb_root system_blks; 1237 1238 #ifdef EXTENTS_STATS 1239 /* ext4 extents stats */ 1240 unsigned long s_ext_min; 1241 unsigned long s_ext_max; 1242 unsigned long s_depth_max; 1243 spinlock_t s_ext_stats_lock; 1244 unsigned long s_ext_blocks; 1245 unsigned long s_ext_extents; 1246 #endif 1247 1248 /* for buddy allocator */ 1249 struct ext4_group_info ***s_group_info; 1250 struct inode *s_buddy_cache; 1251 spinlock_t s_md_lock; 1252 unsigned short *s_mb_offsets; 1253 unsigned int *s_mb_maxs; 1254 unsigned int s_group_info_size; 1255 1256 /* tunables */ 1257 unsigned long s_stripe; 1258 unsigned int s_mb_stream_request; 1259 unsigned int s_mb_max_to_scan; 1260 unsigned int s_mb_min_to_scan; 1261 unsigned int s_mb_stats; 1262 unsigned int s_mb_order2_reqs; 1263 unsigned int s_mb_group_prealloc; 1264 unsigned int s_max_writeback_mb_bump; 1265 unsigned int s_max_dir_size_kb; 1266 /* where last allocation was done - for stream allocation */ 1267 unsigned long s_mb_last_group; 1268 unsigned long s_mb_last_start; 1269 1270 /* stats for buddy allocator */ 1271 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1272 atomic_t s_bal_success; /* we found long enough chunks */ 1273 atomic_t s_bal_allocated; /* in blocks */ 1274 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1275 atomic_t s_bal_goals; /* goal hits */ 1276 atomic_t s_bal_breaks; /* too long searches */ 1277 atomic_t s_bal_2orders; /* 2^order hits */ 1278 spinlock_t s_bal_lock; 1279 unsigned long s_mb_buddies_generated; 1280 unsigned long long s_mb_generation_time; 1281 atomic_t s_mb_lost_chunks; 1282 atomic_t s_mb_preallocated; 1283 atomic_t s_mb_discarded; 1284 atomic_t s_lock_busy; 1285 1286 /* locality groups */ 1287 struct ext4_locality_group __percpu *s_locality_groups; 1288 1289 /* for write statistics */ 1290 unsigned long s_sectors_written_start; 1291 u64 s_kbytes_written; 1292 1293 /* the size of zero-out chunk */ 1294 unsigned int s_extent_max_zeroout_kb; 1295 1296 unsigned int s_log_groups_per_flex; 1297 struct flex_groups *s_flex_groups; 1298 ext4_group_t s_flex_groups_allocated; 1299 1300 /* workqueue for dio unwritten */ 1301 struct workqueue_struct *dio_unwritten_wq; 1302 1303 /* timer for periodic error stats printing */ 1304 struct timer_list s_err_report; 1305 1306 /* Lazy inode table initialization info */ 1307 struct ext4_li_request *s_li_request; 1308 /* Wait multiplier for lazy initialization thread */ 1309 unsigned int s_li_wait_mult; 1310 1311 /* Kernel thread for multiple mount protection */ 1312 struct task_struct *s_mmp_tsk; 1313 1314 /* record the last minlen when FITRIM is called. */ 1315 atomic_t s_last_trim_minblks; 1316 1317 /* Reference to checksum algorithm driver via cryptoapi */ 1318 struct crypto_shash *s_chksum_driver; 1319 1320 /* Precomputed FS UUID checksum for seeding other checksums */ 1321 __u32 s_csum_seed; 1322 }; 1323 1324 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1325 { 1326 return sb->s_fs_info; 1327 } 1328 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1329 { 1330 return container_of(inode, struct ext4_inode_info, vfs_inode); 1331 } 1332 1333 static inline struct timespec ext4_current_time(struct inode *inode) 1334 { 1335 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1336 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1337 } 1338 1339 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1340 { 1341 return ino == EXT4_ROOT_INO || 1342 ino == EXT4_USR_QUOTA_INO || 1343 ino == EXT4_GRP_QUOTA_INO || 1344 ino == EXT4_JOURNAL_INO || 1345 ino == EXT4_RESIZE_INO || 1346 (ino >= EXT4_FIRST_INO(sb) && 1347 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1348 } 1349 1350 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1351 struct ext4_io_end *io_end) 1352 { 1353 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1354 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1355 atomic_inc(&EXT4_I(inode)->i_unwritten); 1356 } 1357 } 1358 1359 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1360 { 1361 return inode->i_private; 1362 } 1363 1364 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1365 { 1366 inode->i_private = io; 1367 } 1368 1369 /* 1370 * Inode dynamic state flags 1371 */ 1372 enum { 1373 EXT4_STATE_JDATA, /* journaled data exists */ 1374 EXT4_STATE_NEW, /* inode is newly created */ 1375 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1376 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1377 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1378 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1379 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1380 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1381 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */ 1382 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1383 nolocking */ 1384 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1385 }; 1386 1387 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1388 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1389 { \ 1390 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1391 } \ 1392 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1393 { \ 1394 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1395 } \ 1396 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1397 { \ 1398 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1399 } 1400 1401 EXT4_INODE_BIT_FNS(flag, flags, 0) 1402 #if (BITS_PER_LONG < 64) 1403 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1404 1405 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1406 { 1407 (ei)->i_state_flags = 0; 1408 } 1409 #else 1410 EXT4_INODE_BIT_FNS(state, flags, 32) 1411 1412 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1413 { 1414 /* We depend on the fact that callers will set i_flags */ 1415 } 1416 #endif 1417 #else 1418 /* Assume that user mode programs are passing in an ext4fs superblock, not 1419 * a kernel struct super_block. This will allow us to call the feature-test 1420 * macros from user land. */ 1421 #define EXT4_SB(sb) (sb) 1422 #endif 1423 1424 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1425 1426 /* 1427 * Codes for operating systems 1428 */ 1429 #define EXT4_OS_LINUX 0 1430 #define EXT4_OS_HURD 1 1431 #define EXT4_OS_MASIX 2 1432 #define EXT4_OS_FREEBSD 3 1433 #define EXT4_OS_LITES 4 1434 1435 /* 1436 * Revision levels 1437 */ 1438 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1439 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1440 1441 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1442 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1443 1444 #define EXT4_GOOD_OLD_INODE_SIZE 128 1445 1446 /* 1447 * Feature set definitions 1448 */ 1449 1450 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1451 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1452 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1453 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1454 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1455 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1456 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1457 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1458 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1459 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1460 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1461 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1462 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1463 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1464 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1465 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1466 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1467 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1468 1469 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1470 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1471 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1472 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1473 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1474 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1475 1476 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1477 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1478 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1479 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1480 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1481 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1482 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1483 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1484 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1485 /* 1486 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1487 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1488 * all other data structures' checksums. However, the METADATA_CSUM and 1489 * GDT_CSUM bits are mutually exclusive. 1490 */ 1491 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1492 1493 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1494 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1495 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1496 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1497 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1498 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1499 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1500 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1501 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1502 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1503 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1504 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1505 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1506 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1507 1508 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1509 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1510 EXT4_FEATURE_INCOMPAT_META_BG) 1511 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1512 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1513 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1514 1515 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1516 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1517 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1518 EXT4_FEATURE_INCOMPAT_META_BG) 1519 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1520 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1521 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1522 1523 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1524 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1525 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1526 EXT4_FEATURE_INCOMPAT_META_BG| \ 1527 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1528 EXT4_FEATURE_INCOMPAT_64BIT| \ 1529 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1530 EXT4_FEATURE_INCOMPAT_MMP | \ 1531 EXT4_FEATURE_INCOMPAT_INLINE_DATA) 1532 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1533 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1534 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1535 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1536 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1537 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1538 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1539 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1540 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1541 EXT4_FEATURE_RO_COMPAT_QUOTA) 1542 1543 /* 1544 * Default values for user and/or group using reserved blocks 1545 */ 1546 #define EXT4_DEF_RESUID 0 1547 #define EXT4_DEF_RESGID 0 1548 1549 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1550 1551 /* 1552 * Default mount options 1553 */ 1554 #define EXT4_DEFM_DEBUG 0x0001 1555 #define EXT4_DEFM_BSDGROUPS 0x0002 1556 #define EXT4_DEFM_XATTR_USER 0x0004 1557 #define EXT4_DEFM_ACL 0x0008 1558 #define EXT4_DEFM_UID16 0x0010 1559 #define EXT4_DEFM_JMODE 0x0060 1560 #define EXT4_DEFM_JMODE_DATA 0x0020 1561 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1562 #define EXT4_DEFM_JMODE_WBACK 0x0060 1563 #define EXT4_DEFM_NOBARRIER 0x0100 1564 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1565 #define EXT4_DEFM_DISCARD 0x0400 1566 #define EXT4_DEFM_NODELALLOC 0x0800 1567 1568 /* 1569 * Default journal batch times 1570 */ 1571 #define EXT4_DEF_MIN_BATCH_TIME 0 1572 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1573 1574 /* 1575 * Minimum number of groups in a flexgroup before we separate out 1576 * directories into the first block group of a flexgroup 1577 */ 1578 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1579 1580 /* 1581 * Structure of a directory entry 1582 */ 1583 #define EXT4_NAME_LEN 255 1584 1585 struct ext4_dir_entry { 1586 __le32 inode; /* Inode number */ 1587 __le16 rec_len; /* Directory entry length */ 1588 __le16 name_len; /* Name length */ 1589 char name[EXT4_NAME_LEN]; /* File name */ 1590 }; 1591 1592 /* 1593 * The new version of the directory entry. Since EXT4 structures are 1594 * stored in intel byte order, and the name_len field could never be 1595 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1596 * file_type field. 1597 */ 1598 struct ext4_dir_entry_2 { 1599 __le32 inode; /* Inode number */ 1600 __le16 rec_len; /* Directory entry length */ 1601 __u8 name_len; /* Name length */ 1602 __u8 file_type; 1603 char name[EXT4_NAME_LEN]; /* File name */ 1604 }; 1605 1606 /* 1607 * This is a bogus directory entry at the end of each leaf block that 1608 * records checksums. 1609 */ 1610 struct ext4_dir_entry_tail { 1611 __le32 det_reserved_zero1; /* Pretend to be unused */ 1612 __le16 det_rec_len; /* 12 */ 1613 __u8 det_reserved_zero2; /* Zero name length */ 1614 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1615 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1616 }; 1617 1618 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1619 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1620 ((blocksize) - \ 1621 sizeof(struct ext4_dir_entry_tail)))) 1622 1623 /* 1624 * Ext4 directory file types. Only the low 3 bits are used. The 1625 * other bits are reserved for now. 1626 */ 1627 #define EXT4_FT_UNKNOWN 0 1628 #define EXT4_FT_REG_FILE 1 1629 #define EXT4_FT_DIR 2 1630 #define EXT4_FT_CHRDEV 3 1631 #define EXT4_FT_BLKDEV 4 1632 #define EXT4_FT_FIFO 5 1633 #define EXT4_FT_SOCK 6 1634 #define EXT4_FT_SYMLINK 7 1635 1636 #define EXT4_FT_MAX 8 1637 1638 #define EXT4_FT_DIR_CSUM 0xDE 1639 1640 /* 1641 * EXT4_DIR_PAD defines the directory entries boundaries 1642 * 1643 * NOTE: It must be a multiple of 4 1644 */ 1645 #define EXT4_DIR_PAD 4 1646 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1647 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1648 ~EXT4_DIR_ROUND) 1649 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1650 1651 /* 1652 * If we ever get support for fs block sizes > page_size, we'll need 1653 * to remove the #if statements in the next two functions... 1654 */ 1655 static inline unsigned int 1656 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1657 { 1658 unsigned len = le16_to_cpu(dlen); 1659 1660 #if (PAGE_CACHE_SIZE >= 65536) 1661 if (len == EXT4_MAX_REC_LEN || len == 0) 1662 return blocksize; 1663 return (len & 65532) | ((len & 3) << 16); 1664 #else 1665 return len; 1666 #endif 1667 } 1668 1669 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1670 { 1671 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1672 BUG(); 1673 #if (PAGE_CACHE_SIZE >= 65536) 1674 if (len < 65536) 1675 return cpu_to_le16(len); 1676 if (len == blocksize) { 1677 if (blocksize == 65536) 1678 return cpu_to_le16(EXT4_MAX_REC_LEN); 1679 else 1680 return cpu_to_le16(0); 1681 } 1682 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1683 #else 1684 return cpu_to_le16(len); 1685 #endif 1686 } 1687 1688 /* 1689 * Hash Tree Directory indexing 1690 * (c) Daniel Phillips, 2001 1691 */ 1692 1693 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1694 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1695 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1696 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1697 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1698 1699 /* Legal values for the dx_root hash_version field: */ 1700 1701 #define DX_HASH_LEGACY 0 1702 #define DX_HASH_HALF_MD4 1 1703 #define DX_HASH_TEA 2 1704 #define DX_HASH_LEGACY_UNSIGNED 3 1705 #define DX_HASH_HALF_MD4_UNSIGNED 4 1706 #define DX_HASH_TEA_UNSIGNED 5 1707 1708 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1709 const void *address, unsigned int length) 1710 { 1711 struct { 1712 struct shash_desc shash; 1713 char ctx[4]; 1714 } desc; 1715 int err; 1716 1717 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1718 1719 desc.shash.tfm = sbi->s_chksum_driver; 1720 desc.shash.flags = 0; 1721 *(u32 *)desc.ctx = crc; 1722 1723 err = crypto_shash_update(&desc.shash, address, length); 1724 BUG_ON(err); 1725 1726 return *(u32 *)desc.ctx; 1727 } 1728 1729 #ifdef __KERNEL__ 1730 1731 /* hash info structure used by the directory hash */ 1732 struct dx_hash_info 1733 { 1734 u32 hash; 1735 u32 minor_hash; 1736 int hash_version; 1737 u32 *seed; 1738 }; 1739 1740 1741 /* 32 and 64 bit signed EOF for dx directories */ 1742 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1743 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1744 1745 1746 /* 1747 * Control parameters used by ext4_htree_next_block 1748 */ 1749 #define HASH_NB_ALWAYS 1 1750 1751 1752 /* 1753 * Describe an inode's exact location on disk and in memory 1754 */ 1755 struct ext4_iloc 1756 { 1757 struct buffer_head *bh; 1758 unsigned long offset; 1759 ext4_group_t block_group; 1760 }; 1761 1762 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1763 { 1764 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1765 } 1766 1767 /* 1768 * This structure is stuffed into the struct file's private_data field 1769 * for directories. It is where we put information so that we can do 1770 * readdir operations in hash tree order. 1771 */ 1772 struct dir_private_info { 1773 struct rb_root root; 1774 struct rb_node *curr_node; 1775 struct fname *extra_fname; 1776 loff_t last_pos; 1777 __u32 curr_hash; 1778 __u32 curr_minor_hash; 1779 __u32 next_hash; 1780 }; 1781 1782 /* calculate the first block number of the group */ 1783 static inline ext4_fsblk_t 1784 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1785 { 1786 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1787 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1788 } 1789 1790 /* 1791 * Special error return code only used by dx_probe() and its callers. 1792 */ 1793 #define ERR_BAD_DX_DIR -75000 1794 1795 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr, 1796 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp); 1797 1798 /* 1799 * Timeout and state flag for lazy initialization inode thread. 1800 */ 1801 #define EXT4_DEF_LI_WAIT_MULT 10 1802 #define EXT4_DEF_LI_MAX_START_DELAY 5 1803 #define EXT4_LAZYINIT_QUIT 0x0001 1804 #define EXT4_LAZYINIT_RUNNING 0x0002 1805 1806 /* 1807 * Lazy inode table initialization info 1808 */ 1809 struct ext4_lazy_init { 1810 unsigned long li_state; 1811 struct list_head li_request_list; 1812 struct mutex li_list_mtx; 1813 }; 1814 1815 struct ext4_li_request { 1816 struct super_block *lr_super; 1817 struct ext4_sb_info *lr_sbi; 1818 ext4_group_t lr_next_group; 1819 struct list_head lr_request; 1820 unsigned long lr_next_sched; 1821 unsigned long lr_timeout; 1822 }; 1823 1824 struct ext4_features { 1825 struct kobject f_kobj; 1826 struct completion f_kobj_unregister; 1827 }; 1828 1829 /* 1830 * This structure will be used for multiple mount protection. It will be 1831 * written into the block number saved in the s_mmp_block field in the 1832 * superblock. Programs that check MMP should assume that if 1833 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1834 * to use the filesystem, regardless of how old the timestamp is. 1835 */ 1836 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1837 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1838 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1839 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1840 1841 struct mmp_struct { 1842 __le32 mmp_magic; /* Magic number for MMP */ 1843 __le32 mmp_seq; /* Sequence no. updated periodically */ 1844 1845 /* 1846 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1847 * purposes and do not affect the correctness of the algorithm 1848 */ 1849 __le64 mmp_time; /* Time last updated */ 1850 char mmp_nodename[64]; /* Node which last updated MMP block */ 1851 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1852 1853 /* 1854 * mmp_check_interval is used to verify if the MMP block has been 1855 * updated on the block device. The value is updated based on the 1856 * maximum time to write the MMP block during an update cycle. 1857 */ 1858 __le16 mmp_check_interval; 1859 1860 __le16 mmp_pad1; 1861 __le32 mmp_pad2[226]; 1862 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1863 }; 1864 1865 /* arguments passed to the mmp thread */ 1866 struct mmpd_data { 1867 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1868 struct super_block *sb; /* super block of the fs */ 1869 }; 1870 1871 /* 1872 * Check interval multiplier 1873 * The MMP block is written every update interval and initially checked every 1874 * update interval x the multiplier (the value is then adapted based on the 1875 * write latency). The reason is that writes can be delayed under load and we 1876 * don't want readers to incorrectly assume that the filesystem is no longer 1877 * in use. 1878 */ 1879 #define EXT4_MMP_CHECK_MULT 2UL 1880 1881 /* 1882 * Minimum interval for MMP checking in seconds. 1883 */ 1884 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1885 1886 /* 1887 * Maximum interval for MMP checking in seconds. 1888 */ 1889 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1890 1891 /* 1892 * Function prototypes 1893 */ 1894 1895 /* 1896 * Ok, these declarations are also in <linux/kernel.h> but none of the 1897 * ext4 source programs needs to include it so they are duplicated here. 1898 */ 1899 # define NORET_TYPE /**/ 1900 # define ATTRIB_NORET __attribute__((noreturn)) 1901 # define NORET_AND noreturn, 1902 1903 /* bitmap.c */ 1904 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1905 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1906 struct ext4_group_desc *gdp, 1907 struct buffer_head *bh, int sz); 1908 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1909 struct ext4_group_desc *gdp, 1910 struct buffer_head *bh, int sz); 1911 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1912 struct ext4_group_desc *gdp, 1913 struct buffer_head *bh); 1914 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1915 struct ext4_group_desc *gdp, 1916 struct buffer_head *bh); 1917 1918 /* balloc.c */ 1919 extern void ext4_validate_block_bitmap(struct super_block *sb, 1920 struct ext4_group_desc *desc, 1921 unsigned int block_group, 1922 struct buffer_head *bh); 1923 extern unsigned int ext4_block_group(struct super_block *sb, 1924 ext4_fsblk_t blocknr); 1925 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1926 ext4_fsblk_t blocknr); 1927 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1928 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1929 ext4_group_t group); 1930 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1931 ext4_fsblk_t goal, 1932 unsigned int flags, 1933 unsigned long *count, 1934 int *errp); 1935 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1936 s64 nclusters, unsigned int flags); 1937 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1938 extern void ext4_check_blocks_bitmap(struct super_block *); 1939 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1940 ext4_group_t block_group, 1941 struct buffer_head ** bh); 1942 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1943 1944 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1945 ext4_group_t block_group); 1946 extern int ext4_wait_block_bitmap(struct super_block *sb, 1947 ext4_group_t block_group, 1948 struct buffer_head *bh); 1949 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1950 ext4_group_t block_group); 1951 extern void ext4_init_block_bitmap(struct super_block *sb, 1952 struct buffer_head *bh, 1953 ext4_group_t group, 1954 struct ext4_group_desc *desc); 1955 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1956 ext4_group_t block_group, 1957 struct ext4_group_desc *gdp); 1958 extern unsigned ext4_num_overhead_clusters(struct super_block *sb, 1959 ext4_group_t block_group, 1960 struct ext4_group_desc *gdp); 1961 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1962 1963 /* dir.c */ 1964 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1965 struct file *, 1966 struct ext4_dir_entry_2 *, 1967 struct buffer_head *, char *, int, 1968 unsigned int); 1969 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 1970 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 1971 (de), (bh), (buf), (size), (offset))) 1972 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1973 __u32 minor_hash, 1974 struct ext4_dir_entry_2 *dirent); 1975 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1976 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 1977 struct buffer_head *bh, 1978 void *buf, int buf_size, 1979 const char *name, int namelen, 1980 struct ext4_dir_entry_2 **dest_de); 1981 void ext4_insert_dentry(struct inode *inode, 1982 struct ext4_dir_entry_2 *de, 1983 int buf_size, 1984 const char *name, int namelen); 1985 static inline void ext4_update_dx_flag(struct inode *inode) 1986 { 1987 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb, 1988 EXT4_FEATURE_COMPAT_DIR_INDEX)) 1989 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 1990 } 1991 static unsigned char ext4_filetype_table[] = { 1992 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 1993 }; 1994 1995 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 1996 { 1997 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) || 1998 (filetype >= EXT4_FT_MAX)) 1999 return DT_UNKNOWN; 2000 2001 return ext4_filetype_table[filetype]; 2002 } 2003 2004 /* fsync.c */ 2005 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2006 extern int ext4_flush_unwritten_io(struct inode *); 2007 2008 /* hash.c */ 2009 extern int ext4fs_dirhash(const char *name, int len, struct 2010 dx_hash_info *hinfo); 2011 2012 /* ialloc.c */ 2013 extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t, 2014 const struct qstr *qstr, __u32 goal, 2015 uid_t *owner); 2016 extern void ext4_free_inode(handle_t *, struct inode *); 2017 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2018 extern unsigned long ext4_count_free_inodes(struct super_block *); 2019 extern unsigned long ext4_count_dirs(struct super_block *); 2020 extern void ext4_check_inodes_bitmap(struct super_block *); 2021 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2022 extern int ext4_init_inode_table(struct super_block *sb, 2023 ext4_group_t group, int barrier); 2024 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2025 2026 /* mballoc.c */ 2027 extern long ext4_mb_stats; 2028 extern long ext4_mb_max_to_scan; 2029 extern int ext4_mb_init(struct super_block *); 2030 extern int ext4_mb_release(struct super_block *); 2031 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2032 struct ext4_allocation_request *, int *); 2033 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2034 extern void ext4_discard_preallocations(struct inode *); 2035 extern int __init ext4_init_mballoc(void); 2036 extern void ext4_exit_mballoc(void); 2037 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2038 struct buffer_head *bh, ext4_fsblk_t block, 2039 unsigned long count, int flags); 2040 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2041 ext4_group_t ngroups); 2042 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2043 ext4_group_t i, struct ext4_group_desc *desc); 2044 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2045 ext4_fsblk_t block, unsigned long count); 2046 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2047 2048 /* inode.c */ 2049 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 2050 ext4_lblk_t, int, int *); 2051 struct buffer_head *ext4_bread(handle_t *, struct inode *, 2052 ext4_lblk_t, int, int *); 2053 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2054 struct buffer_head *bh_result, int create); 2055 int ext4_get_block(struct inode *inode, sector_t iblock, 2056 struct buffer_head *bh_result, int create); 2057 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2058 struct buffer_head *bh, int create); 2059 int ext4_walk_page_buffers(handle_t *handle, 2060 struct buffer_head *head, 2061 unsigned from, 2062 unsigned to, 2063 int *partial, 2064 int (*fn)(handle_t *handle, 2065 struct buffer_head *bh)); 2066 int do_journal_get_write_access(handle_t *handle, 2067 struct buffer_head *bh); 2068 #define FALL_BACK_TO_NONDELALLOC 1 2069 #define CONVERT_INLINE_DATA 2 2070 2071 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2072 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2073 extern int ext4_setattr(struct dentry *, struct iattr *); 2074 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2075 struct kstat *stat); 2076 extern void ext4_evict_inode(struct inode *); 2077 extern void ext4_clear_inode(struct inode *); 2078 extern int ext4_sync_inode(handle_t *, struct inode *); 2079 extern void ext4_dirty_inode(struct inode *, int); 2080 extern int ext4_change_inode_journal_flag(struct inode *, int); 2081 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2082 extern int ext4_can_truncate(struct inode *inode); 2083 extern void ext4_truncate(struct inode *); 2084 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2085 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2086 extern void ext4_set_inode_flags(struct inode *); 2087 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2088 extern int ext4_alloc_da_blocks(struct inode *inode); 2089 extern void ext4_set_aops(struct inode *inode); 2090 extern int ext4_writepage_trans_blocks(struct inode *); 2091 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2092 extern int ext4_discard_partial_page_buffers(handle_t *handle, 2093 struct address_space *mapping, loff_t from, 2094 loff_t length, int flags); 2095 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2096 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2097 extern void ext4_da_update_reserve_space(struct inode *inode, 2098 int used, int quota_claim); 2099 2100 /* indirect.c */ 2101 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2102 struct ext4_map_blocks *map, int flags); 2103 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2104 const struct iovec *iov, loff_t offset, 2105 unsigned long nr_segs); 2106 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2107 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk); 2108 extern void ext4_ind_truncate(struct inode *inode); 2109 2110 /* ioctl.c */ 2111 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2112 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2113 2114 /* migrate.c */ 2115 extern int ext4_ext_migrate(struct inode *); 2116 2117 /* namei.c */ 2118 extern int ext4_dirent_csum_verify(struct inode *inode, 2119 struct ext4_dir_entry *dirent); 2120 extern int ext4_orphan_add(handle_t *, struct inode *); 2121 extern int ext4_orphan_del(handle_t *, struct inode *); 2122 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2123 __u32 start_minor_hash, __u32 *next_hash); 2124 extern int search_dir(struct buffer_head *bh, 2125 char *search_buf, 2126 int buf_size, 2127 struct inode *dir, 2128 const struct qstr *d_name, 2129 unsigned int offset, 2130 struct ext4_dir_entry_2 **res_dir); 2131 extern int ext4_generic_delete_entry(handle_t *handle, 2132 struct inode *dir, 2133 struct ext4_dir_entry_2 *de_del, 2134 struct buffer_head *bh, 2135 void *entry_buf, 2136 int buf_size, 2137 int csum_size); 2138 2139 /* resize.c */ 2140 extern int ext4_group_add(struct super_block *sb, 2141 struct ext4_new_group_data *input); 2142 extern int ext4_group_extend(struct super_block *sb, 2143 struct ext4_super_block *es, 2144 ext4_fsblk_t n_blocks_count); 2145 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2146 2147 /* super.c */ 2148 extern int ext4_calculate_overhead(struct super_block *sb); 2149 extern int ext4_superblock_csum_verify(struct super_block *sb, 2150 struct ext4_super_block *es); 2151 extern void ext4_superblock_csum_set(struct super_block *sb); 2152 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2153 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2154 extern void ext4_kvfree(void *ptr); 2155 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2156 ext4_group_t ngroup); 2157 extern __printf(4, 5) 2158 void __ext4_error(struct super_block *, const char *, unsigned int, 2159 const char *, ...); 2160 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \ 2161 __LINE__, ## message) 2162 extern __printf(5, 6) 2163 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2164 const char *, ...); 2165 extern __printf(5, 6) 2166 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2167 const char *, ...); 2168 extern void __ext4_std_error(struct super_block *, const char *, 2169 unsigned int, int); 2170 extern __printf(4, 5) 2171 void __ext4_abort(struct super_block *, const char *, unsigned int, 2172 const char *, ...); 2173 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \ 2174 __LINE__, ## message) 2175 extern __printf(4, 5) 2176 void __ext4_warning(struct super_block *, const char *, unsigned int, 2177 const char *, ...); 2178 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \ 2179 __LINE__, ## message) 2180 extern __printf(3, 4) 2181 void ext4_msg(struct super_block *, const char *, const char *, ...); 2182 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2183 const char *, unsigned int, const char *); 2184 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \ 2185 __LINE__, msg) 2186 extern __printf(7, 8) 2187 void __ext4_grp_locked_error(const char *, unsigned int, 2188 struct super_block *, ext4_group_t, 2189 unsigned long, ext4_fsblk_t, 2190 const char *, ...); 2191 #define ext4_grp_locked_error(sb, grp, message...) \ 2192 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message) 2193 extern void ext4_update_dynamic_rev(struct super_block *sb); 2194 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2195 __u32 compat); 2196 extern int ext4_update_rocompat_feature(handle_t *handle, 2197 struct super_block *sb, __u32 rocompat); 2198 extern int ext4_update_incompat_feature(handle_t *handle, 2199 struct super_block *sb, __u32 incompat); 2200 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2201 struct ext4_group_desc *bg); 2202 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2203 struct ext4_group_desc *bg); 2204 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2205 struct ext4_group_desc *bg); 2206 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2207 struct ext4_group_desc *bg); 2208 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2209 struct ext4_group_desc *bg); 2210 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2211 struct ext4_group_desc *bg); 2212 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2213 struct ext4_group_desc *bg); 2214 extern void ext4_block_bitmap_set(struct super_block *sb, 2215 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2216 extern void ext4_inode_bitmap_set(struct super_block *sb, 2217 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2218 extern void ext4_inode_table_set(struct super_block *sb, 2219 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2220 extern void ext4_free_group_clusters_set(struct super_block *sb, 2221 struct ext4_group_desc *bg, 2222 __u32 count); 2223 extern void ext4_free_inodes_set(struct super_block *sb, 2224 struct ext4_group_desc *bg, __u32 count); 2225 extern void ext4_used_dirs_set(struct super_block *sb, 2226 struct ext4_group_desc *bg, __u32 count); 2227 extern void ext4_itable_unused_set(struct super_block *sb, 2228 struct ext4_group_desc *bg, __u32 count); 2229 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2230 struct ext4_group_desc *gdp); 2231 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2232 struct ext4_group_desc *gdp); 2233 2234 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2235 { 2236 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2237 EXT4_FEATURE_RO_COMPAT_GDT_CSUM | 2238 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM); 2239 } 2240 2241 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2242 { 2243 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2244 le32_to_cpu(es->s_blocks_count_lo); 2245 } 2246 2247 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2248 { 2249 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2250 le32_to_cpu(es->s_r_blocks_count_lo); 2251 } 2252 2253 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2254 { 2255 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2256 le32_to_cpu(es->s_free_blocks_count_lo); 2257 } 2258 2259 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2260 ext4_fsblk_t blk) 2261 { 2262 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2263 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2264 } 2265 2266 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2267 ext4_fsblk_t blk) 2268 { 2269 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2270 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2271 } 2272 2273 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2274 ext4_fsblk_t blk) 2275 { 2276 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2277 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2278 } 2279 2280 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2281 { 2282 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2283 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2284 le32_to_cpu(raw_inode->i_size_lo); 2285 else 2286 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2287 } 2288 2289 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2290 { 2291 raw_inode->i_size_lo = cpu_to_le32(i_size); 2292 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2293 } 2294 2295 static inline 2296 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2297 ext4_group_t group) 2298 { 2299 struct ext4_group_info ***grp_info; 2300 long indexv, indexh; 2301 grp_info = EXT4_SB(sb)->s_group_info; 2302 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2303 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2304 return grp_info[indexv][indexh]; 2305 } 2306 2307 /* 2308 * Reading s_groups_count requires using smp_rmb() afterwards. See 2309 * the locking protocol documented in the comments of ext4_group_add() 2310 * in resize.c 2311 */ 2312 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2313 { 2314 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2315 2316 smp_rmb(); 2317 return ngroups; 2318 } 2319 2320 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2321 ext4_group_t block_group) 2322 { 2323 return block_group >> sbi->s_log_groups_per_flex; 2324 } 2325 2326 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2327 { 2328 return 1 << sbi->s_log_groups_per_flex; 2329 } 2330 2331 #define ext4_std_error(sb, errno) \ 2332 do { \ 2333 if ((errno)) \ 2334 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2335 } while (0) 2336 2337 #ifdef CONFIG_SMP 2338 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2339 * counters. So we need to make sure we have free clusters more 2340 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2341 */ 2342 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2343 #else 2344 #define EXT4_FREECLUSTERS_WATERMARK 0 2345 #endif 2346 2347 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2348 { 2349 /* 2350 * XXX: replace with spinlock if seen contended -bzzz 2351 */ 2352 down_write(&EXT4_I(inode)->i_data_sem); 2353 if (newsize > EXT4_I(inode)->i_disksize) 2354 EXT4_I(inode)->i_disksize = newsize; 2355 up_write(&EXT4_I(inode)->i_data_sem); 2356 return ; 2357 } 2358 2359 struct ext4_group_info { 2360 unsigned long bb_state; 2361 struct rb_root bb_free_root; 2362 ext4_grpblk_t bb_first_free; /* first free block */ 2363 ext4_grpblk_t bb_free; /* total free blocks */ 2364 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2365 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2366 struct list_head bb_prealloc_list; 2367 #ifdef DOUBLE_CHECK 2368 void *bb_bitmap; 2369 #endif 2370 struct rw_semaphore alloc_sem; 2371 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2372 * regions, index is order. 2373 * bb_counters[3] = 5 means 2374 * 5 free 8-block regions. */ 2375 }; 2376 2377 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2378 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2379 2380 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2381 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2382 2383 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2384 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2385 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2386 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2387 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2388 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2389 2390 #define EXT4_MAX_CONTENTION 8 2391 #define EXT4_CONTENTION_THRESHOLD 2 2392 2393 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2394 ext4_group_t group) 2395 { 2396 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2397 } 2398 2399 /* 2400 * Returns true if the filesystem is busy enough that attempts to 2401 * access the block group locks has run into contention. 2402 */ 2403 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2404 { 2405 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2406 } 2407 2408 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2409 { 2410 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2411 if (spin_trylock(lock)) 2412 /* 2413 * We're able to grab the lock right away, so drop the 2414 * lock contention counter. 2415 */ 2416 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2417 else { 2418 /* 2419 * The lock is busy, so bump the contention counter, 2420 * and then wait on the spin lock. 2421 */ 2422 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2423 EXT4_MAX_CONTENTION); 2424 spin_lock(lock); 2425 } 2426 } 2427 2428 static inline void ext4_unlock_group(struct super_block *sb, 2429 ext4_group_t group) 2430 { 2431 spin_unlock(ext4_group_lock_ptr(sb, group)); 2432 } 2433 2434 /* 2435 * Block validity checking 2436 */ 2437 #define ext4_check_indirect_blockref(inode, bh) \ 2438 ext4_check_blockref(__func__, __LINE__, inode, \ 2439 (__le32 *)(bh)->b_data, \ 2440 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2441 2442 #define ext4_ind_check_inode(inode) \ 2443 ext4_check_blockref(__func__, __LINE__, inode, \ 2444 EXT4_I(inode)->i_data, \ 2445 EXT4_NDIR_BLOCKS) 2446 2447 /* 2448 * Inodes and files operations 2449 */ 2450 2451 /* dir.c */ 2452 extern const struct file_operations ext4_dir_operations; 2453 2454 /* file.c */ 2455 extern const struct inode_operations ext4_file_inode_operations; 2456 extern const struct file_operations ext4_file_operations; 2457 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2458 extern void ext4_unwritten_wait(struct inode *inode); 2459 2460 /* namei.c */ 2461 extern const struct inode_operations ext4_dir_inode_operations; 2462 extern const struct inode_operations ext4_special_inode_operations; 2463 extern struct dentry *ext4_get_parent(struct dentry *child); 2464 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 2465 struct ext4_dir_entry_2 *de, 2466 int blocksize, int csum_size, 2467 unsigned int parent_ino, int dotdot_real_len); 2468 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 2469 unsigned int blocksize); 2470 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 2471 struct inode *inode, 2472 struct buffer_head *bh); 2473 2474 /* symlink.c */ 2475 extern const struct inode_operations ext4_symlink_inode_operations; 2476 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2477 2478 /* block_validity */ 2479 extern void ext4_release_system_zone(struct super_block *sb); 2480 extern int ext4_setup_system_zone(struct super_block *sb); 2481 extern int __init ext4_init_system_zone(void); 2482 extern void ext4_exit_system_zone(void); 2483 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2484 ext4_fsblk_t start_blk, 2485 unsigned int count); 2486 extern int ext4_check_blockref(const char *, unsigned int, 2487 struct inode *, __le32 *, unsigned int); 2488 2489 /* extents.c */ 2490 struct ext4_ext_path; 2491 struct ext4_extent; 2492 2493 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2494 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2495 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 2496 int chunk); 2497 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2498 struct ext4_map_blocks *map, int flags); 2499 extern void ext4_ext_truncate(struct inode *); 2500 extern int ext4_ext_punch_hole(struct file *file, loff_t offset, 2501 loff_t length); 2502 extern void ext4_ext_init(struct super_block *); 2503 extern void ext4_ext_release(struct super_block *); 2504 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2505 loff_t len); 2506 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 2507 ssize_t len); 2508 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2509 struct ext4_map_blocks *map, int flags); 2510 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 2511 ext4_lblk_t lblocks); 2512 extern int ext4_extent_tree_init(handle_t *, struct inode *); 2513 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 2514 int num, 2515 struct ext4_ext_path *path); 2516 extern int ext4_can_extents_be_merged(struct inode *inode, 2517 struct ext4_extent *ex1, 2518 struct ext4_extent *ex2); 2519 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 2520 struct ext4_ext_path *, 2521 struct ext4_extent *, int); 2522 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t, 2523 struct ext4_ext_path *); 2524 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 2525 extern int ext4_ext_check_inode(struct inode *inode); 2526 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 2527 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2528 __u64 start, __u64 len); 2529 2530 2531 /* move_extent.c */ 2532 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2533 __u64 start_orig, __u64 start_donor, 2534 __u64 len, __u64 *moved_len); 2535 2536 /* page-io.c */ 2537 extern int __init ext4_init_pageio(void); 2538 extern void ext4_add_complete_io(ext4_io_end_t *io_end); 2539 extern void ext4_exit_pageio(void); 2540 extern void ext4_ioend_wait(struct inode *); 2541 extern void ext4_free_io_end(ext4_io_end_t *io); 2542 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2543 extern void ext4_io_submit(struct ext4_io_submit *io); 2544 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2545 struct page *page, 2546 int len, 2547 struct writeback_control *wbc); 2548 2549 /* mmp.c */ 2550 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2551 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp); 2552 extern int ext4_mmp_csum_verify(struct super_block *sb, 2553 struct mmp_struct *mmp); 2554 2555 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */ 2556 enum ext4_state_bits { 2557 BH_Uninit /* blocks are allocated but uninitialized on disk */ 2558 = BH_JBDPrivateStart, 2559 BH_AllocFromCluster, /* allocated blocks were part of already 2560 * allocated cluster. Note that this flag will 2561 * never, ever appear in a buffer_head's state 2562 * flag. See EXT4_MAP_FROM_CLUSTER to see where 2563 * this is used. */ 2564 }; 2565 2566 BUFFER_FNS(Uninit, uninit) 2567 TAS_BUFFER_FNS(Uninit, uninit) 2568 2569 /* 2570 * Add new method to test wether block and inode bitmaps are properly 2571 * initialized. With uninit_bg reading the block from disk is not enough 2572 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2573 */ 2574 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2575 2576 static inline int bitmap_uptodate(struct buffer_head *bh) 2577 { 2578 return (buffer_uptodate(bh) && 2579 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2580 } 2581 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2582 { 2583 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2584 } 2585 2586 /* 2587 * Disable DIO read nolock optimization, so new dioreaders will be forced 2588 * to grab i_mutex 2589 */ 2590 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2591 { 2592 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2593 smp_mb(); 2594 } 2595 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2596 { 2597 smp_mb(); 2598 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2599 } 2600 2601 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2602 2603 /* For ioend & aio unwritten conversion wait queues */ 2604 #define EXT4_WQ_HASH_SZ 37 2605 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2606 EXT4_WQ_HASH_SZ]) 2607 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2608 EXT4_WQ_HASH_SZ]) 2609 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2610 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2611 2612 #define EXT4_RESIZING 0 2613 extern int ext4_resize_begin(struct super_block *sb); 2614 extern void ext4_resize_end(struct super_block *sb); 2615 2616 #endif /* __KERNEL__ */ 2617 2618 #endif /* _EXT4_H */ 2619