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 #ifdef __KERNEL__ 33 #include <linux/compat.h> 34 #endif 35 36 /* 37 * The fourth extended filesystem constants/structures 38 */ 39 40 /* 41 * Define EXT4FS_DEBUG to produce debug messages 42 */ 43 #undef EXT4FS_DEBUG 44 45 /* 46 * Debug code 47 */ 48 #ifdef EXT4FS_DEBUG 49 #define ext4_debug(f, a...) \ 50 do { \ 51 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 52 __FILE__, __LINE__, __func__); \ 53 printk(KERN_DEBUG f, ## a); \ 54 } while (0) 55 #else 56 #define ext4_debug(f, a...) do {} while (0) 57 #endif 58 59 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 60 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 61 62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 63 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 64 65 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 66 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 67 68 /* data type for block offset of block group */ 69 typedef int ext4_grpblk_t; 70 71 /* data type for filesystem-wide blocks number */ 72 typedef unsigned long long ext4_fsblk_t; 73 74 /* data type for file logical block number */ 75 typedef __u32 ext4_lblk_t; 76 77 /* data type for block group number */ 78 typedef unsigned int ext4_group_t; 79 80 /* 81 * Flags used in mballoc's allocation_context flags field. 82 * 83 * Also used to show what's going on for debugging purposes when the 84 * flag field is exported via the traceport interface 85 */ 86 87 /* prefer goal again. length */ 88 #define EXT4_MB_HINT_MERGE 0x0001 89 /* blocks already reserved */ 90 #define EXT4_MB_HINT_RESERVED 0x0002 91 /* metadata is being allocated */ 92 #define EXT4_MB_HINT_METADATA 0x0004 93 /* first blocks in the file */ 94 #define EXT4_MB_HINT_FIRST 0x0008 95 /* search for the best chunk */ 96 #define EXT4_MB_HINT_BEST 0x0010 97 /* data is being allocated */ 98 #define EXT4_MB_HINT_DATA 0x0020 99 /* don't preallocate (for tails) */ 100 #define EXT4_MB_HINT_NOPREALLOC 0x0040 101 /* allocate for locality group */ 102 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 103 /* allocate goal blocks or none */ 104 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 105 /* goal is meaningful */ 106 #define EXT4_MB_HINT_TRY_GOAL 0x0200 107 /* blocks already pre-reserved by delayed allocation */ 108 #define EXT4_MB_DELALLOC_RESERVED 0x0400 109 /* We are doing stream allocation */ 110 #define EXT4_MB_STREAM_ALLOC 0x0800 111 /* Use reserved root blocks if needed */ 112 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 113 114 struct ext4_allocation_request { 115 /* target inode for block we're allocating */ 116 struct inode *inode; 117 /* how many blocks we want to allocate */ 118 unsigned int len; 119 /* logical block in target inode */ 120 ext4_lblk_t logical; 121 /* the closest logical allocated block to the left */ 122 ext4_lblk_t lleft; 123 /* the closest logical allocated block to the right */ 124 ext4_lblk_t lright; 125 /* phys. target (a hint) */ 126 ext4_fsblk_t goal; 127 /* phys. block for the closest logical allocated block to the left */ 128 ext4_fsblk_t pleft; 129 /* phys. block for the closest logical allocated block to the right */ 130 ext4_fsblk_t pright; 131 /* flags. see above EXT4_MB_HINT_* */ 132 unsigned int flags; 133 }; 134 135 /* 136 * Logical to physical block mapping, used by ext4_map_blocks() 137 * 138 * This structure is used to pass requests into ext4_map_blocks() as 139 * well as to store the information returned by ext4_map_blocks(). It 140 * takes less room on the stack than a struct buffer_head. 141 */ 142 #define EXT4_MAP_NEW (1 << BH_New) 143 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 144 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 145 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 146 #define EXT4_MAP_UNINIT (1 << BH_Uninit) 147 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of 148 * ext4_map_blocks wants to know whether or not the underlying cluster has 149 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that 150 * the requested mapping was from previously mapped (or delayed allocated) 151 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster 152 * should never appear on buffer_head's state flags. 153 */ 154 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster) 155 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 156 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 157 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER) 158 159 struct ext4_map_blocks { 160 ext4_fsblk_t m_pblk; 161 ext4_lblk_t m_lblk; 162 unsigned int m_len; 163 unsigned int m_flags; 164 }; 165 166 /* 167 * For delayed allocation tracking 168 */ 169 struct mpage_da_data { 170 struct inode *inode; 171 sector_t b_blocknr; /* start block number of extent */ 172 size_t b_size; /* size of extent */ 173 unsigned long b_state; /* state of the extent */ 174 unsigned long first_page, next_page; /* extent of pages */ 175 struct writeback_control *wbc; 176 int io_done; 177 int pages_written; 178 int retval; 179 }; 180 181 /* 182 * Flags for ext4_io_end->flags 183 */ 184 #define EXT4_IO_END_UNWRITTEN 0x0001 185 #define EXT4_IO_END_ERROR 0x0002 186 #define EXT4_IO_END_QUEUED 0x0004 187 #define EXT4_IO_END_DIRECT 0x0008 188 #define EXT4_IO_END_IN_FSYNC 0x0010 189 190 struct ext4_io_page { 191 struct page *p_page; 192 atomic_t p_count; 193 }; 194 195 #define MAX_IO_PAGES 128 196 197 typedef struct ext4_io_end { 198 struct list_head list; /* per-file finished IO list */ 199 struct inode *inode; /* file being written to */ 200 unsigned int flag; /* unwritten or not */ 201 struct page *page; /* page struct for buffer write */ 202 loff_t offset; /* offset in the file */ 203 ssize_t size; /* size of the extent */ 204 struct work_struct work; /* data work queue */ 205 struct kiocb *iocb; /* iocb struct for AIO */ 206 int result; /* error value for AIO */ 207 int num_io_pages; 208 struct ext4_io_page *pages[MAX_IO_PAGES]; 209 } ext4_io_end_t; 210 211 struct ext4_io_submit { 212 int io_op; 213 struct bio *io_bio; 214 ext4_io_end_t *io_end; 215 struct ext4_io_page *io_page; 216 sector_t io_next_block; 217 }; 218 219 /* 220 * Special inodes numbers 221 */ 222 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 223 #define EXT4_ROOT_INO 2 /* Root inode */ 224 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 225 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 226 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 227 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 228 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 229 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 230 231 /* First non-reserved inode for old ext4 filesystems */ 232 #define EXT4_GOOD_OLD_FIRST_INO 11 233 234 /* 235 * Maximal count of links to a file 236 */ 237 #define EXT4_LINK_MAX 65000 238 239 /* 240 * Macro-instructions used to manage several block sizes 241 */ 242 #define EXT4_MIN_BLOCK_SIZE 1024 243 #define EXT4_MAX_BLOCK_SIZE 65536 244 #define EXT4_MIN_BLOCK_LOG_SIZE 10 245 #define EXT4_MAX_BLOCK_LOG_SIZE 16 246 #ifdef __KERNEL__ 247 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 248 #else 249 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 250 #endif 251 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 252 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 253 EXT4_SB(s)->s_cluster_bits) 254 #ifdef __KERNEL__ 255 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 256 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 257 #else 258 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 259 #endif 260 #ifdef __KERNEL__ 261 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 262 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 263 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 264 #else 265 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 266 EXT4_GOOD_OLD_INODE_SIZE : \ 267 (s)->s_inode_size) 268 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 269 EXT4_GOOD_OLD_FIRST_INO : \ 270 (s)->s_first_ino) 271 #endif 272 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 273 274 /* Translate a block number to a cluster number */ 275 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 276 /* Translate a cluster number to a block number */ 277 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 278 /* Translate # of blks to # of clusters */ 279 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 280 (sbi)->s_cluster_bits) 281 282 /* 283 * Structure of a blocks group descriptor 284 */ 285 struct ext4_group_desc 286 { 287 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 288 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 289 __le32 bg_inode_table_lo; /* Inodes table block */ 290 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 291 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 292 __le16 bg_used_dirs_count_lo; /* Directories count */ 293 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 294 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */ 295 __le16 bg_itable_unused_lo; /* Unused inodes count */ 296 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 297 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 298 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 299 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 300 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 301 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 302 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 303 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 304 __u32 bg_reserved2[3]; 305 }; 306 307 /* 308 * Structure of a flex block group info 309 */ 310 311 struct flex_groups { 312 atomic_t free_inodes; 313 atomic_t free_clusters; 314 atomic_t used_dirs; 315 }; 316 317 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 318 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 319 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 320 321 /* 322 * Macro-instructions used to manage group descriptors 323 */ 324 #define EXT4_MIN_DESC_SIZE 32 325 #define EXT4_MIN_DESC_SIZE_64BIT 64 326 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 327 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 328 #ifdef __KERNEL__ 329 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 330 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 331 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 332 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 333 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 334 #else 335 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 336 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 337 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 338 #endif 339 340 /* 341 * Constants relative to the data blocks 342 */ 343 #define EXT4_NDIR_BLOCKS 12 344 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 345 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 346 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 347 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 348 349 /* 350 * Inode flags 351 */ 352 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 353 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 354 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 355 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 356 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 357 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 358 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 359 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 360 /* Reserved for compression usage... */ 361 #define EXT4_DIRTY_FL 0x00000100 362 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 363 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 364 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 365 /* End compression flags --- maybe not all used */ 366 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 367 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 368 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 369 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 370 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 371 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 372 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 373 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 374 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 375 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 376 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 377 378 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 379 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */ 380 381 /* Flags that should be inherited by new inodes from their parent. */ 382 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 383 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 384 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 385 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 386 387 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 388 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 389 390 /* Flags that are appropriate for non-directories/regular files. */ 391 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 392 393 /* Mask out flags that are inappropriate for the given type of inode. */ 394 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 395 { 396 if (S_ISDIR(mode)) 397 return flags; 398 else if (S_ISREG(mode)) 399 return flags & EXT4_REG_FLMASK; 400 else 401 return flags & EXT4_OTHER_FLMASK; 402 } 403 404 /* 405 * Inode flags used for atomic set/get 406 */ 407 enum { 408 EXT4_INODE_SECRM = 0, /* Secure deletion */ 409 EXT4_INODE_UNRM = 1, /* Undelete */ 410 EXT4_INODE_COMPR = 2, /* Compress file */ 411 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 412 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 413 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 414 EXT4_INODE_NODUMP = 6, /* do not dump file */ 415 EXT4_INODE_NOATIME = 7, /* do not update atime */ 416 /* Reserved for compression usage... */ 417 EXT4_INODE_DIRTY = 8, 418 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 419 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 420 EXT4_INODE_ECOMPR = 11, /* Compression error */ 421 /* End compression flags --- maybe not all used */ 422 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 423 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 424 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 425 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 426 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 427 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 428 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 429 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 430 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 431 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 432 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 433 }; 434 435 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 436 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \ 437 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \ 438 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); } 439 440 /* 441 * Since it's pretty easy to mix up bit numbers and hex values, and we 442 * can't do a compile-time test for ENUM values, we use a run-time 443 * test to make sure that EXT4_XXX_FL is consistent with respect to 444 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop 445 * out so it won't cost any extra space in the compiled kernel image. 446 * But it's important that these values are the same, since we are 447 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL 448 * must be consistent with the values of FS_XXX_FL defined in 449 * include/linux/fs.h and the on-disk values found in ext2, ext3, and 450 * ext4 filesystems, and of course the values defined in e2fsprogs. 451 * 452 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 453 */ 454 static inline void ext4_check_flag_values(void) 455 { 456 CHECK_FLAG_VALUE(SECRM); 457 CHECK_FLAG_VALUE(UNRM); 458 CHECK_FLAG_VALUE(COMPR); 459 CHECK_FLAG_VALUE(SYNC); 460 CHECK_FLAG_VALUE(IMMUTABLE); 461 CHECK_FLAG_VALUE(APPEND); 462 CHECK_FLAG_VALUE(NODUMP); 463 CHECK_FLAG_VALUE(NOATIME); 464 CHECK_FLAG_VALUE(DIRTY); 465 CHECK_FLAG_VALUE(COMPRBLK); 466 CHECK_FLAG_VALUE(NOCOMPR); 467 CHECK_FLAG_VALUE(ECOMPR); 468 CHECK_FLAG_VALUE(INDEX); 469 CHECK_FLAG_VALUE(IMAGIC); 470 CHECK_FLAG_VALUE(JOURNAL_DATA); 471 CHECK_FLAG_VALUE(NOTAIL); 472 CHECK_FLAG_VALUE(DIRSYNC); 473 CHECK_FLAG_VALUE(TOPDIR); 474 CHECK_FLAG_VALUE(HUGE_FILE); 475 CHECK_FLAG_VALUE(EXTENTS); 476 CHECK_FLAG_VALUE(EA_INODE); 477 CHECK_FLAG_VALUE(EOFBLOCKS); 478 CHECK_FLAG_VALUE(RESERVED); 479 } 480 481 /* Used to pass group descriptor data when online resize is done */ 482 struct ext4_new_group_input { 483 __u32 group; /* Group number for this data */ 484 __u64 block_bitmap; /* Absolute block number of block bitmap */ 485 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 486 __u64 inode_table; /* Absolute block number of inode table start */ 487 __u32 blocks_count; /* Total number of blocks in this group */ 488 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 489 __u16 unused; 490 }; 491 492 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 493 struct compat_ext4_new_group_input { 494 u32 group; 495 compat_u64 block_bitmap; 496 compat_u64 inode_bitmap; 497 compat_u64 inode_table; 498 u32 blocks_count; 499 u16 reserved_blocks; 500 u16 unused; 501 }; 502 #endif 503 504 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 505 struct ext4_new_group_data { 506 __u32 group; 507 __u64 block_bitmap; 508 __u64 inode_bitmap; 509 __u64 inode_table; 510 __u32 blocks_count; 511 __u16 reserved_blocks; 512 __u16 unused; 513 __u32 free_blocks_count; 514 }; 515 516 /* Indexes used to index group tables in ext4_new_group_data */ 517 enum { 518 BLOCK_BITMAP = 0, /* block bitmap */ 519 INODE_BITMAP, /* inode bitmap */ 520 INODE_TABLE, /* inode tables */ 521 GROUP_TABLE_COUNT, 522 }; 523 524 /* 525 * Flags used by ext4_map_blocks() 526 */ 527 /* Allocate any needed blocks and/or convert an unitialized 528 extent to be an initialized ext4 */ 529 #define EXT4_GET_BLOCKS_CREATE 0x0001 530 /* Request the creation of an unitialized extent */ 531 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 532 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 533 EXT4_GET_BLOCKS_CREATE) 534 /* Caller is from the delayed allocation writeout path, 535 so set the magic i_delalloc_reserve_flag after taking the 536 inode allocation semaphore for */ 537 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 538 /* caller is from the direct IO path, request to creation of an 539 unitialized extents if not allocated, split the uninitialized 540 extent if blocks has been preallocated already*/ 541 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 542 #define EXT4_GET_BLOCKS_CONVERT 0x0010 543 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 544 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 545 /* Convert extent to initialized after IO complete */ 546 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 547 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 548 /* Punch out blocks of an extent */ 549 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020 550 /* Don't normalize allocation size (used for fallocate) */ 551 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 552 /* Request will not result in inode size update (user for fallocate) */ 553 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 554 555 /* 556 * Flags used by ext4_free_blocks 557 */ 558 #define EXT4_FREE_BLOCKS_METADATA 0x0001 559 #define EXT4_FREE_BLOCKS_FORGET 0x0002 560 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 561 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 562 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 563 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 564 565 /* 566 * Flags used by ext4_discard_partial_page_buffers 567 */ 568 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001 569 570 /* 571 * ioctl commands 572 */ 573 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 574 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 575 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 576 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 577 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 578 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 579 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 580 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 581 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 582 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 583 #define EXT4_IOC_MIGRATE _IO('f', 9) 584 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 585 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 586 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 587 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 588 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 589 590 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 591 /* 592 * ioctl commands in 32 bit emulation 593 */ 594 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 595 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 596 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 597 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 598 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 599 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 600 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 601 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 602 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 603 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 604 #endif 605 606 /* Max physical block we can address w/o extents */ 607 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 608 609 /* 610 * Structure of an inode on the disk 611 */ 612 struct ext4_inode { 613 __le16 i_mode; /* File mode */ 614 __le16 i_uid; /* Low 16 bits of Owner Uid */ 615 __le32 i_size_lo; /* Size in bytes */ 616 __le32 i_atime; /* Access time */ 617 __le32 i_ctime; /* Inode Change time */ 618 __le32 i_mtime; /* Modification time */ 619 __le32 i_dtime; /* Deletion Time */ 620 __le16 i_gid; /* Low 16 bits of Group Id */ 621 __le16 i_links_count; /* Links count */ 622 __le32 i_blocks_lo; /* Blocks count */ 623 __le32 i_flags; /* File flags */ 624 union { 625 struct { 626 __le32 l_i_version; 627 } linux1; 628 struct { 629 __u32 h_i_translator; 630 } hurd1; 631 struct { 632 __u32 m_i_reserved1; 633 } masix1; 634 } osd1; /* OS dependent 1 */ 635 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 636 __le32 i_generation; /* File version (for NFS) */ 637 __le32 i_file_acl_lo; /* File ACL */ 638 __le32 i_size_high; 639 __le32 i_obso_faddr; /* Obsoleted fragment address */ 640 union { 641 struct { 642 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 643 __le16 l_i_file_acl_high; 644 __le16 l_i_uid_high; /* these 2 fields */ 645 __le16 l_i_gid_high; /* were reserved2[0] */ 646 __u32 l_i_reserved2; 647 } linux2; 648 struct { 649 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 650 __u16 h_i_mode_high; 651 __u16 h_i_uid_high; 652 __u16 h_i_gid_high; 653 __u32 h_i_author; 654 } hurd2; 655 struct { 656 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 657 __le16 m_i_file_acl_high; 658 __u32 m_i_reserved2[2]; 659 } masix2; 660 } osd2; /* OS dependent 2 */ 661 __le16 i_extra_isize; 662 __le16 i_pad1; 663 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 664 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 665 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 666 __le32 i_crtime; /* File Creation time */ 667 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 668 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 669 }; 670 671 struct move_extent { 672 __u32 reserved; /* should be zero */ 673 __u32 donor_fd; /* donor file descriptor */ 674 __u64 orig_start; /* logical start offset in block for orig */ 675 __u64 donor_start; /* logical start offset in block for donor */ 676 __u64 len; /* block length to be moved */ 677 __u64 moved_len; /* moved block length */ 678 }; 679 680 #define EXT4_EPOCH_BITS 2 681 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 682 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 683 684 /* 685 * Extended fields will fit into an inode if the filesystem was formatted 686 * with large inodes (-I 256 or larger) and there are not currently any EAs 687 * consuming all of the available space. For new inodes we always reserve 688 * enough space for the kernel's known extended fields, but for inodes 689 * created with an old kernel this might not have been the case. None of 690 * the extended inode fields is critical for correct filesystem operation. 691 * This macro checks if a certain field fits in the inode. Note that 692 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 693 */ 694 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 695 ((offsetof(typeof(*ext4_inode), field) + \ 696 sizeof((ext4_inode)->field)) \ 697 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 698 (einode)->i_extra_isize)) \ 699 700 static inline __le32 ext4_encode_extra_time(struct timespec *time) 701 { 702 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 703 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 704 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 705 } 706 707 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 708 { 709 if (sizeof(time->tv_sec) > 4) 710 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 711 << 32; 712 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 713 } 714 715 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 716 do { \ 717 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 718 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 719 (raw_inode)->xtime ## _extra = \ 720 ext4_encode_extra_time(&(inode)->xtime); \ 721 } while (0) 722 723 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 724 do { \ 725 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 726 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 727 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 728 (raw_inode)->xtime ## _extra = \ 729 ext4_encode_extra_time(&(einode)->xtime); \ 730 } while (0) 731 732 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 733 do { \ 734 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 735 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 736 ext4_decode_extra_time(&(inode)->xtime, \ 737 raw_inode->xtime ## _extra); \ 738 else \ 739 (inode)->xtime.tv_nsec = 0; \ 740 } while (0) 741 742 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 743 do { \ 744 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 745 (einode)->xtime.tv_sec = \ 746 (signed)le32_to_cpu((raw_inode)->xtime); \ 747 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 748 ext4_decode_extra_time(&(einode)->xtime, \ 749 raw_inode->xtime ## _extra); \ 750 else \ 751 (einode)->xtime.tv_nsec = 0; \ 752 } while (0) 753 754 #define i_disk_version osd1.linux1.l_i_version 755 756 #if defined(__KERNEL__) || defined(__linux__) 757 #define i_reserved1 osd1.linux1.l_i_reserved1 758 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 759 #define i_blocks_high osd2.linux2.l_i_blocks_high 760 #define i_uid_low i_uid 761 #define i_gid_low i_gid 762 #define i_uid_high osd2.linux2.l_i_uid_high 763 #define i_gid_high osd2.linux2.l_i_gid_high 764 #define i_reserved2 osd2.linux2.l_i_reserved2 765 766 #elif defined(__GNU__) 767 768 #define i_translator osd1.hurd1.h_i_translator 769 #define i_uid_high osd2.hurd2.h_i_uid_high 770 #define i_gid_high osd2.hurd2.h_i_gid_high 771 #define i_author osd2.hurd2.h_i_author 772 773 #elif defined(__masix__) 774 775 #define i_reserved1 osd1.masix1.m_i_reserved1 776 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 777 #define i_reserved2 osd2.masix2.m_i_reserved2 778 779 #endif /* defined(__KERNEL__) || defined(__linux__) */ 780 781 /* 782 * storage for cached extent 783 * If ec_len == 0, then the cache is invalid. 784 * If ec_start == 0, then the cache represents a gap (null mapping) 785 */ 786 struct ext4_ext_cache { 787 ext4_fsblk_t ec_start; 788 ext4_lblk_t ec_block; 789 __u32 ec_len; /* must be 32bit to return holes */ 790 }; 791 792 /* 793 * fourth extended file system inode data in memory 794 */ 795 struct ext4_inode_info { 796 __le32 i_data[15]; /* unconverted */ 797 __u32 i_dtime; 798 ext4_fsblk_t i_file_acl; 799 800 /* 801 * i_block_group is the number of the block group which contains 802 * this file's inode. Constant across the lifetime of the inode, 803 * it is ued for making block allocation decisions - we try to 804 * place a file's data blocks near its inode block, and new inodes 805 * near to their parent directory's inode. 806 */ 807 ext4_group_t i_block_group; 808 ext4_lblk_t i_dir_start_lookup; 809 #if (BITS_PER_LONG < 64) 810 unsigned long i_state_flags; /* Dynamic state flags */ 811 #endif 812 unsigned long i_flags; 813 814 #ifdef CONFIG_EXT4_FS_XATTR 815 /* 816 * Extended attributes can be read independently of the main file 817 * data. Taking i_mutex even when reading would cause contention 818 * between readers of EAs and writers of regular file data, so 819 * instead we synchronize on xattr_sem when reading or changing 820 * EAs. 821 */ 822 struct rw_semaphore xattr_sem; 823 #endif 824 825 struct list_head i_orphan; /* unlinked but open inodes */ 826 827 /* 828 * i_disksize keeps track of what the inode size is ON DISK, not 829 * in memory. During truncate, i_size is set to the new size by 830 * the VFS prior to calling ext4_truncate(), but the filesystem won't 831 * set i_disksize to 0 until the truncate is actually under way. 832 * 833 * The intent is that i_disksize always represents the blocks which 834 * are used by this file. This allows recovery to restart truncate 835 * on orphans if we crash during truncate. We actually write i_disksize 836 * into the on-disk inode when writing inodes out, instead of i_size. 837 * 838 * The only time when i_disksize and i_size may be different is when 839 * a truncate is in progress. The only things which change i_disksize 840 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 841 */ 842 loff_t i_disksize; 843 844 /* 845 * i_data_sem is for serialising ext4_truncate() against 846 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 847 * data tree are chopped off during truncate. We can't do that in 848 * ext4 because whenever we perform intermediate commits during 849 * truncate, the inode and all the metadata blocks *must* be in a 850 * consistent state which allows truncation of the orphans to restart 851 * during recovery. Hence we must fix the get_block-vs-truncate race 852 * by other means, so we have i_data_sem. 853 */ 854 struct rw_semaphore i_data_sem; 855 struct inode vfs_inode; 856 struct jbd2_inode *jinode; 857 858 struct ext4_ext_cache i_cached_extent; 859 /* 860 * File creation time. Its function is same as that of 861 * struct timespec i_{a,c,m}time in the generic inode. 862 */ 863 struct timespec i_crtime; 864 865 /* mballoc */ 866 struct list_head i_prealloc_list; 867 spinlock_t i_prealloc_lock; 868 869 /* ialloc */ 870 ext4_group_t i_last_alloc_group; 871 872 /* allocation reservation info for delalloc */ 873 /* In case of bigalloc, these refer to clusters rather than blocks */ 874 unsigned int i_reserved_data_blocks; 875 unsigned int i_reserved_meta_blocks; 876 unsigned int i_allocated_meta_blocks; 877 ext4_lblk_t i_da_metadata_calc_last_lblock; 878 int i_da_metadata_calc_len; 879 880 /* on-disk additional length */ 881 __u16 i_extra_isize; 882 883 #ifdef CONFIG_QUOTA 884 /* quota space reservation, managed internally by quota code */ 885 qsize_t i_reserved_quota; 886 #endif 887 888 /* completed IOs that might need unwritten extents handling */ 889 struct list_head i_completed_io_list; 890 spinlock_t i_completed_io_lock; 891 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 892 /* current io_end structure for async DIO write*/ 893 ext4_io_end_t *cur_aio_dio; 894 atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */ 895 896 spinlock_t i_block_reservation_lock; 897 898 /* 899 * Transactions that contain inode's metadata needed to complete 900 * fsync and fdatasync, respectively. 901 */ 902 tid_t i_sync_tid; 903 tid_t i_datasync_tid; 904 }; 905 906 /* 907 * File system states 908 */ 909 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 910 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 911 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 912 913 /* 914 * Misc. filesystem flags 915 */ 916 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 917 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 918 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 919 920 /* 921 * Mount flags 922 */ 923 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 924 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 925 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 926 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 927 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 928 #define EXT4_MOUNT_ERRORS_MASK 0x00070 929 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 930 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 931 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 932 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 933 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 934 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 935 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 936 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 937 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 938 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 939 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 940 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 941 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 942 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 943 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 944 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 945 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 946 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 947 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */ 948 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 949 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 950 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 951 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 952 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 953 954 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 955 specified delalloc */ 956 957 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 958 ~EXT4_MOUNT_##opt 959 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 960 EXT4_MOUNT_##opt 961 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 962 EXT4_MOUNT_##opt) 963 964 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 965 ~EXT4_MOUNT2_##opt 966 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 967 EXT4_MOUNT2_##opt 968 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 969 EXT4_MOUNT2_##opt) 970 971 #define ext4_test_and_set_bit __test_and_set_bit_le 972 #define ext4_set_bit __set_bit_le 973 #define ext4_set_bit_atomic ext2_set_bit_atomic 974 #define ext4_test_and_clear_bit __test_and_clear_bit_le 975 #define ext4_clear_bit __clear_bit_le 976 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 977 #define ext4_test_bit test_bit_le 978 #define ext4_find_next_zero_bit find_next_zero_bit_le 979 #define ext4_find_next_bit find_next_bit_le 980 981 extern void ext4_set_bits(void *bm, int cur, int len); 982 983 /* 984 * Maximal mount counts between two filesystem checks 985 */ 986 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 987 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 988 989 /* 990 * Behaviour when detecting errors 991 */ 992 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 993 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 994 #define EXT4_ERRORS_PANIC 3 /* Panic */ 995 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 996 997 /* 998 * Structure of the super block 999 */ 1000 struct ext4_super_block { 1001 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1002 __le32 s_blocks_count_lo; /* Blocks count */ 1003 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1004 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1005 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1006 __le32 s_first_data_block; /* First Data Block */ 1007 __le32 s_log_block_size; /* Block size */ 1008 __le32 s_log_cluster_size; /* Allocation cluster size */ 1009 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1010 __le32 s_clusters_per_group; /* # Clusters per group */ 1011 __le32 s_inodes_per_group; /* # Inodes per group */ 1012 __le32 s_mtime; /* Mount time */ 1013 /*30*/ __le32 s_wtime; /* Write time */ 1014 __le16 s_mnt_count; /* Mount count */ 1015 __le16 s_max_mnt_count; /* Maximal mount count */ 1016 __le16 s_magic; /* Magic signature */ 1017 __le16 s_state; /* File system state */ 1018 __le16 s_errors; /* Behaviour when detecting errors */ 1019 __le16 s_minor_rev_level; /* minor revision level */ 1020 /*40*/ __le32 s_lastcheck; /* time of last check */ 1021 __le32 s_checkinterval; /* max. time between checks */ 1022 __le32 s_creator_os; /* OS */ 1023 __le32 s_rev_level; /* Revision level */ 1024 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1025 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1026 /* 1027 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1028 * 1029 * Note: the difference between the compatible feature set and 1030 * the incompatible feature set is that if there is a bit set 1031 * in the incompatible feature set that the kernel doesn't 1032 * know about, it should refuse to mount the filesystem. 1033 * 1034 * e2fsck's requirements are more strict; if it doesn't know 1035 * about a feature in either the compatible or incompatible 1036 * feature set, it must abort and not try to meddle with 1037 * things it doesn't understand... 1038 */ 1039 __le32 s_first_ino; /* First non-reserved inode */ 1040 __le16 s_inode_size; /* size of inode structure */ 1041 __le16 s_block_group_nr; /* block group # of this superblock */ 1042 __le32 s_feature_compat; /* compatible feature set */ 1043 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1044 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1045 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1046 /*78*/ char s_volume_name[16]; /* volume name */ 1047 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1048 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1049 /* 1050 * Performance hints. Directory preallocation should only 1051 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1052 */ 1053 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1054 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1055 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1056 /* 1057 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1058 */ 1059 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1060 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1061 __le32 s_journal_dev; /* device number of journal file */ 1062 __le32 s_last_orphan; /* start of list of inodes to delete */ 1063 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1064 __u8 s_def_hash_version; /* Default hash version to use */ 1065 __u8 s_jnl_backup_type; 1066 __le16 s_desc_size; /* size of group descriptor */ 1067 /*100*/ __le32 s_default_mount_opts; 1068 __le32 s_first_meta_bg; /* First metablock block group */ 1069 __le32 s_mkfs_time; /* When the filesystem was created */ 1070 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1071 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1072 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1073 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1074 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1075 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1076 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1077 __le32 s_flags; /* Miscellaneous flags */ 1078 __le16 s_raid_stride; /* RAID stride */ 1079 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1080 __le64 s_mmp_block; /* Block for multi-mount protection */ 1081 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1082 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1083 __u8 s_reserved_char_pad; 1084 __le16 s_reserved_pad; 1085 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1086 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1087 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1088 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1089 snapshot's future use */ 1090 __le32 s_snapshot_list; /* inode number of the head of the 1091 on-disk snapshot list */ 1092 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1093 __le32 s_error_count; /* number of fs errors */ 1094 __le32 s_first_error_time; /* first time an error happened */ 1095 __le32 s_first_error_ino; /* inode involved in first error */ 1096 __le64 s_first_error_block; /* block involved of first error */ 1097 __u8 s_first_error_func[32]; /* function where the error happened */ 1098 __le32 s_first_error_line; /* line number where error happened */ 1099 __le32 s_last_error_time; /* most recent time of an error */ 1100 __le32 s_last_error_ino; /* inode involved in last error */ 1101 __le32 s_last_error_line; /* line number where error happened */ 1102 __le64 s_last_error_block; /* block involved of last error */ 1103 __u8 s_last_error_func[32]; /* function where the error happened */ 1104 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1105 __u8 s_mount_opts[64]; 1106 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1107 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1108 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1109 __le32 s_reserved[109]; /* Padding to the end of the block */ 1110 }; 1111 1112 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1113 1114 #ifdef __KERNEL__ 1115 1116 /* 1117 * run-time mount flags 1118 */ 1119 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1120 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1121 1122 /* 1123 * fourth extended-fs super-block data in memory 1124 */ 1125 struct ext4_sb_info { 1126 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1127 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1128 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1129 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1130 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1131 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1132 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1133 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1134 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1135 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1136 unsigned long s_overhead_last; /* Last calculated overhead */ 1137 unsigned long s_blocks_last; /* Last seen block count */ 1138 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1139 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1140 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1141 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1142 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1143 struct buffer_head **s_group_desc; 1144 unsigned int s_mount_opt; 1145 unsigned int s_mount_opt2; 1146 unsigned int s_mount_flags; 1147 unsigned int s_def_mount_opt; 1148 ext4_fsblk_t s_sb_block; 1149 uid_t s_resuid; 1150 gid_t s_resgid; 1151 unsigned short s_mount_state; 1152 unsigned short s_pad; 1153 int s_addr_per_block_bits; 1154 int s_desc_per_block_bits; 1155 int s_inode_size; 1156 int s_first_ino; 1157 unsigned int s_inode_readahead_blks; 1158 unsigned int s_inode_goal; 1159 spinlock_t s_next_gen_lock; 1160 u32 s_next_generation; 1161 u32 s_hash_seed[4]; 1162 int s_def_hash_version; 1163 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1164 struct percpu_counter s_freeclusters_counter; 1165 struct percpu_counter s_freeinodes_counter; 1166 struct percpu_counter s_dirs_counter; 1167 struct percpu_counter s_dirtyclusters_counter; 1168 struct blockgroup_lock *s_blockgroup_lock; 1169 struct proc_dir_entry *s_proc; 1170 struct kobject s_kobj; 1171 struct completion s_kobj_unregister; 1172 1173 /* Journaling */ 1174 struct journal_s *s_journal; 1175 struct list_head s_orphan; 1176 struct mutex s_orphan_lock; 1177 unsigned long s_resize_flags; /* Flags indicating if there 1178 is a resizer */ 1179 unsigned long s_commit_interval; 1180 u32 s_max_batch_time; 1181 u32 s_min_batch_time; 1182 struct block_device *journal_bdev; 1183 #ifdef CONFIG_QUOTA 1184 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 1185 int s_jquota_fmt; /* Format of quota to use */ 1186 #endif 1187 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1188 struct rb_root system_blks; 1189 1190 #ifdef EXTENTS_STATS 1191 /* ext4 extents stats */ 1192 unsigned long s_ext_min; 1193 unsigned long s_ext_max; 1194 unsigned long s_depth_max; 1195 spinlock_t s_ext_stats_lock; 1196 unsigned long s_ext_blocks; 1197 unsigned long s_ext_extents; 1198 #endif 1199 /* ext4 extent cache stats */ 1200 unsigned long extent_cache_hits; 1201 unsigned long extent_cache_misses; 1202 1203 /* for buddy allocator */ 1204 struct ext4_group_info ***s_group_info; 1205 struct inode *s_buddy_cache; 1206 spinlock_t s_md_lock; 1207 unsigned short *s_mb_offsets; 1208 unsigned int *s_mb_maxs; 1209 1210 /* tunables */ 1211 unsigned long s_stripe; 1212 unsigned int s_mb_stream_request; 1213 unsigned int s_mb_max_to_scan; 1214 unsigned int s_mb_min_to_scan; 1215 unsigned int s_mb_stats; 1216 unsigned int s_mb_order2_reqs; 1217 unsigned int s_mb_group_prealloc; 1218 unsigned int s_max_writeback_mb_bump; 1219 /* where last allocation was done - for stream allocation */ 1220 unsigned long s_mb_last_group; 1221 unsigned long s_mb_last_start; 1222 1223 /* stats for buddy allocator */ 1224 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1225 atomic_t s_bal_success; /* we found long enough chunks */ 1226 atomic_t s_bal_allocated; /* in blocks */ 1227 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1228 atomic_t s_bal_goals; /* goal hits */ 1229 atomic_t s_bal_breaks; /* too long searches */ 1230 atomic_t s_bal_2orders; /* 2^order hits */ 1231 spinlock_t s_bal_lock; 1232 unsigned long s_mb_buddies_generated; 1233 unsigned long long s_mb_generation_time; 1234 atomic_t s_mb_lost_chunks; 1235 atomic_t s_mb_preallocated; 1236 atomic_t s_mb_discarded; 1237 atomic_t s_lock_busy; 1238 1239 /* locality groups */ 1240 struct ext4_locality_group __percpu *s_locality_groups; 1241 1242 /* for write statistics */ 1243 unsigned long s_sectors_written_start; 1244 u64 s_kbytes_written; 1245 1246 unsigned int s_log_groups_per_flex; 1247 struct flex_groups *s_flex_groups; 1248 1249 /* workqueue for dio unwritten */ 1250 struct workqueue_struct *dio_unwritten_wq; 1251 1252 /* timer for periodic error stats printing */ 1253 struct timer_list s_err_report; 1254 1255 /* Lazy inode table initialization info */ 1256 struct ext4_li_request *s_li_request; 1257 /* Wait multiplier for lazy initialization thread */ 1258 unsigned int s_li_wait_mult; 1259 1260 /* Kernel thread for multiple mount protection */ 1261 struct task_struct *s_mmp_tsk; 1262 1263 /* record the last minlen when FITRIM is called. */ 1264 atomic_t s_last_trim_minblks; 1265 }; 1266 1267 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1268 { 1269 return sb->s_fs_info; 1270 } 1271 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1272 { 1273 return container_of(inode, struct ext4_inode_info, vfs_inode); 1274 } 1275 1276 static inline struct timespec ext4_current_time(struct inode *inode) 1277 { 1278 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1279 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1280 } 1281 1282 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1283 { 1284 return ino == EXT4_ROOT_INO || 1285 ino == EXT4_JOURNAL_INO || 1286 ino == EXT4_RESIZE_INO || 1287 (ino >= EXT4_FIRST_INO(sb) && 1288 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1289 } 1290 1291 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1292 struct ext4_io_end *io_end) 1293 { 1294 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1295 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1296 atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); 1297 } 1298 } 1299 1300 /* 1301 * Inode dynamic state flags 1302 */ 1303 enum { 1304 EXT4_STATE_JDATA, /* journaled data exists */ 1305 EXT4_STATE_NEW, /* inode is newly created */ 1306 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1307 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1308 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1309 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1310 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1311 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1312 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */ 1313 }; 1314 1315 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1316 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1317 { \ 1318 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1319 } \ 1320 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1321 { \ 1322 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1323 } \ 1324 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1325 { \ 1326 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1327 } 1328 1329 EXT4_INODE_BIT_FNS(flag, flags, 0) 1330 #if (BITS_PER_LONG < 64) 1331 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1332 1333 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1334 { 1335 (ei)->i_state_flags = 0; 1336 } 1337 #else 1338 EXT4_INODE_BIT_FNS(state, flags, 32) 1339 1340 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1341 { 1342 /* We depend on the fact that callers will set i_flags */ 1343 } 1344 #endif 1345 #else 1346 /* Assume that user mode programs are passing in an ext4fs superblock, not 1347 * a kernel struct super_block. This will allow us to call the feature-test 1348 * macros from user land. */ 1349 #define EXT4_SB(sb) (sb) 1350 #endif 1351 1352 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1353 1354 /* 1355 * Codes for operating systems 1356 */ 1357 #define EXT4_OS_LINUX 0 1358 #define EXT4_OS_HURD 1 1359 #define EXT4_OS_MASIX 2 1360 #define EXT4_OS_FREEBSD 3 1361 #define EXT4_OS_LITES 4 1362 1363 /* 1364 * Revision levels 1365 */ 1366 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1367 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1368 1369 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1370 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1371 1372 #define EXT4_GOOD_OLD_INODE_SIZE 128 1373 1374 /* 1375 * Feature set definitions 1376 */ 1377 1378 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1379 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1380 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1381 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1382 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1383 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1384 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1385 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1386 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1387 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1388 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1389 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1390 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1391 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1392 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1393 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1394 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1395 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1396 1397 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1398 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1399 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1400 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1401 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1402 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1403 1404 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1405 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1406 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1407 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1408 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1409 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1410 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1411 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1412 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1413 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1414 1415 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1416 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1417 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1418 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1419 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1420 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1421 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1422 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1423 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1424 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1425 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1426 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1427 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1428 #define EXT4_FEATURE_INCOMPAT_INLINEDATA 0x8000 /* data in inode */ 1429 1430 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1431 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1432 EXT4_FEATURE_INCOMPAT_META_BG) 1433 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1434 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1435 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1436 1437 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1438 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1439 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1440 EXT4_FEATURE_INCOMPAT_META_BG) 1441 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1442 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1443 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1444 1445 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1446 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1447 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1448 EXT4_FEATURE_INCOMPAT_META_BG| \ 1449 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1450 EXT4_FEATURE_INCOMPAT_64BIT| \ 1451 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1452 EXT4_FEATURE_INCOMPAT_MMP) 1453 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1454 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1455 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1456 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1457 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1458 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1459 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1460 EXT4_FEATURE_RO_COMPAT_BIGALLOC) 1461 1462 /* 1463 * Default values for user and/or group using reserved blocks 1464 */ 1465 #define EXT4_DEF_RESUID 0 1466 #define EXT4_DEF_RESGID 0 1467 1468 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1469 1470 /* 1471 * Default mount options 1472 */ 1473 #define EXT4_DEFM_DEBUG 0x0001 1474 #define EXT4_DEFM_BSDGROUPS 0x0002 1475 #define EXT4_DEFM_XATTR_USER 0x0004 1476 #define EXT4_DEFM_ACL 0x0008 1477 #define EXT4_DEFM_UID16 0x0010 1478 #define EXT4_DEFM_JMODE 0x0060 1479 #define EXT4_DEFM_JMODE_DATA 0x0020 1480 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1481 #define EXT4_DEFM_JMODE_WBACK 0x0060 1482 #define EXT4_DEFM_NOBARRIER 0x0100 1483 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1484 #define EXT4_DEFM_DISCARD 0x0400 1485 #define EXT4_DEFM_NODELALLOC 0x0800 1486 1487 /* 1488 * Default journal batch times 1489 */ 1490 #define EXT4_DEF_MIN_BATCH_TIME 0 1491 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1492 1493 /* 1494 * Minimum number of groups in a flexgroup before we separate out 1495 * directories into the first block group of a flexgroup 1496 */ 1497 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1498 1499 /* 1500 * Structure of a directory entry 1501 */ 1502 #define EXT4_NAME_LEN 255 1503 1504 struct ext4_dir_entry { 1505 __le32 inode; /* Inode number */ 1506 __le16 rec_len; /* Directory entry length */ 1507 __le16 name_len; /* Name length */ 1508 char name[EXT4_NAME_LEN]; /* File name */ 1509 }; 1510 1511 /* 1512 * The new version of the directory entry. Since EXT4 structures are 1513 * stored in intel byte order, and the name_len field could never be 1514 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1515 * file_type field. 1516 */ 1517 struct ext4_dir_entry_2 { 1518 __le32 inode; /* Inode number */ 1519 __le16 rec_len; /* Directory entry length */ 1520 __u8 name_len; /* Name length */ 1521 __u8 file_type; 1522 char name[EXT4_NAME_LEN]; /* File name */ 1523 }; 1524 1525 /* 1526 * Ext4 directory file types. Only the low 3 bits are used. The 1527 * other bits are reserved for now. 1528 */ 1529 #define EXT4_FT_UNKNOWN 0 1530 #define EXT4_FT_REG_FILE 1 1531 #define EXT4_FT_DIR 2 1532 #define EXT4_FT_CHRDEV 3 1533 #define EXT4_FT_BLKDEV 4 1534 #define EXT4_FT_FIFO 5 1535 #define EXT4_FT_SOCK 6 1536 #define EXT4_FT_SYMLINK 7 1537 1538 #define EXT4_FT_MAX 8 1539 1540 /* 1541 * EXT4_DIR_PAD defines the directory entries boundaries 1542 * 1543 * NOTE: It must be a multiple of 4 1544 */ 1545 #define EXT4_DIR_PAD 4 1546 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1547 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1548 ~EXT4_DIR_ROUND) 1549 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1550 1551 /* 1552 * If we ever get support for fs block sizes > page_size, we'll need 1553 * to remove the #if statements in the next two functions... 1554 */ 1555 static inline unsigned int 1556 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1557 { 1558 unsigned len = le16_to_cpu(dlen); 1559 1560 #if (PAGE_CACHE_SIZE >= 65536) 1561 if (len == EXT4_MAX_REC_LEN || len == 0) 1562 return blocksize; 1563 return (len & 65532) | ((len & 3) << 16); 1564 #else 1565 return len; 1566 #endif 1567 } 1568 1569 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1570 { 1571 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1572 BUG(); 1573 #if (PAGE_CACHE_SIZE >= 65536) 1574 if (len < 65536) 1575 return cpu_to_le16(len); 1576 if (len == blocksize) { 1577 if (blocksize == 65536) 1578 return cpu_to_le16(EXT4_MAX_REC_LEN); 1579 else 1580 return cpu_to_le16(0); 1581 } 1582 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1583 #else 1584 return cpu_to_le16(len); 1585 #endif 1586 } 1587 1588 /* 1589 * Hash Tree Directory indexing 1590 * (c) Daniel Phillips, 2001 1591 */ 1592 1593 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1594 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1595 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1596 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1597 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1598 1599 /* Legal values for the dx_root hash_version field: */ 1600 1601 #define DX_HASH_LEGACY 0 1602 #define DX_HASH_HALF_MD4 1 1603 #define DX_HASH_TEA 2 1604 #define DX_HASH_LEGACY_UNSIGNED 3 1605 #define DX_HASH_HALF_MD4_UNSIGNED 4 1606 #define DX_HASH_TEA_UNSIGNED 5 1607 1608 #ifdef __KERNEL__ 1609 1610 /* hash info structure used by the directory hash */ 1611 struct dx_hash_info 1612 { 1613 u32 hash; 1614 u32 minor_hash; 1615 int hash_version; 1616 u32 *seed; 1617 }; 1618 1619 #define EXT4_HTREE_EOF 0x7fffffff 1620 1621 /* 1622 * Control parameters used by ext4_htree_next_block 1623 */ 1624 #define HASH_NB_ALWAYS 1 1625 1626 1627 /* 1628 * Describe an inode's exact location on disk and in memory 1629 */ 1630 struct ext4_iloc 1631 { 1632 struct buffer_head *bh; 1633 unsigned long offset; 1634 ext4_group_t block_group; 1635 }; 1636 1637 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1638 { 1639 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1640 } 1641 1642 /* 1643 * This structure is stuffed into the struct file's private_data field 1644 * for directories. It is where we put information so that we can do 1645 * readdir operations in hash tree order. 1646 */ 1647 struct dir_private_info { 1648 struct rb_root root; 1649 struct rb_node *curr_node; 1650 struct fname *extra_fname; 1651 loff_t last_pos; 1652 __u32 curr_hash; 1653 __u32 curr_minor_hash; 1654 __u32 next_hash; 1655 }; 1656 1657 /* calculate the first block number of the group */ 1658 static inline ext4_fsblk_t 1659 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1660 { 1661 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1662 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1663 } 1664 1665 /* 1666 * Special error return code only used by dx_probe() and its callers. 1667 */ 1668 #define ERR_BAD_DX_DIR -75000 1669 1670 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr, 1671 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp); 1672 1673 /* 1674 * Timeout and state flag for lazy initialization inode thread. 1675 */ 1676 #define EXT4_DEF_LI_WAIT_MULT 10 1677 #define EXT4_DEF_LI_MAX_START_DELAY 5 1678 #define EXT4_LAZYINIT_QUIT 0x0001 1679 #define EXT4_LAZYINIT_RUNNING 0x0002 1680 1681 /* 1682 * Lazy inode table initialization info 1683 */ 1684 struct ext4_lazy_init { 1685 unsigned long li_state; 1686 struct list_head li_request_list; 1687 struct mutex li_list_mtx; 1688 }; 1689 1690 struct ext4_li_request { 1691 struct super_block *lr_super; 1692 struct ext4_sb_info *lr_sbi; 1693 ext4_group_t lr_next_group; 1694 struct list_head lr_request; 1695 unsigned long lr_next_sched; 1696 unsigned long lr_timeout; 1697 }; 1698 1699 struct ext4_features { 1700 struct kobject f_kobj; 1701 struct completion f_kobj_unregister; 1702 }; 1703 1704 /* 1705 * This structure will be used for multiple mount protection. It will be 1706 * written into the block number saved in the s_mmp_block field in the 1707 * superblock. Programs that check MMP should assume that if 1708 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1709 * to use the filesystem, regardless of how old the timestamp is. 1710 */ 1711 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1712 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1713 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1714 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1715 1716 struct mmp_struct { 1717 __le32 mmp_magic; /* Magic number for MMP */ 1718 __le32 mmp_seq; /* Sequence no. updated periodically */ 1719 1720 /* 1721 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1722 * purposes and do not affect the correctness of the algorithm 1723 */ 1724 __le64 mmp_time; /* Time last updated */ 1725 char mmp_nodename[64]; /* Node which last updated MMP block */ 1726 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1727 1728 /* 1729 * mmp_check_interval is used to verify if the MMP block has been 1730 * updated on the block device. The value is updated based on the 1731 * maximum time to write the MMP block during an update cycle. 1732 */ 1733 __le16 mmp_check_interval; 1734 1735 __le16 mmp_pad1; 1736 __le32 mmp_pad2[227]; 1737 }; 1738 1739 /* arguments passed to the mmp thread */ 1740 struct mmpd_data { 1741 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1742 struct super_block *sb; /* super block of the fs */ 1743 }; 1744 1745 /* 1746 * Check interval multiplier 1747 * The MMP block is written every update interval and initially checked every 1748 * update interval x the multiplier (the value is then adapted based on the 1749 * write latency). The reason is that writes can be delayed under load and we 1750 * don't want readers to incorrectly assume that the filesystem is no longer 1751 * in use. 1752 */ 1753 #define EXT4_MMP_CHECK_MULT 2UL 1754 1755 /* 1756 * Minimum interval for MMP checking in seconds. 1757 */ 1758 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1759 1760 /* 1761 * Maximum interval for MMP checking in seconds. 1762 */ 1763 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1764 1765 /* 1766 * Function prototypes 1767 */ 1768 1769 /* 1770 * Ok, these declarations are also in <linux/kernel.h> but none of the 1771 * ext4 source programs needs to include it so they are duplicated here. 1772 */ 1773 # define NORET_TYPE /**/ 1774 # define ATTRIB_NORET __attribute__((noreturn)) 1775 # define NORET_AND noreturn, 1776 1777 /* bitmap.c */ 1778 extern unsigned int ext4_count_free(struct buffer_head *, unsigned); 1779 1780 /* balloc.c */ 1781 extern unsigned int ext4_block_group(struct super_block *sb, 1782 ext4_fsblk_t blocknr); 1783 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1784 ext4_fsblk_t blocknr); 1785 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1786 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1787 ext4_group_t group); 1788 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1789 ext4_fsblk_t goal, 1790 unsigned int flags, 1791 unsigned long *count, 1792 int *errp); 1793 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1794 s64 nclusters, unsigned int flags); 1795 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1796 extern void ext4_check_blocks_bitmap(struct super_block *); 1797 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1798 ext4_group_t block_group, 1799 struct buffer_head ** bh); 1800 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1801 1802 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1803 ext4_group_t block_group); 1804 extern int ext4_wait_block_bitmap(struct super_block *sb, 1805 ext4_group_t block_group, 1806 struct buffer_head *bh); 1807 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1808 ext4_group_t block_group); 1809 extern void ext4_init_block_bitmap(struct super_block *sb, 1810 struct buffer_head *bh, 1811 ext4_group_t group, 1812 struct ext4_group_desc *desc); 1813 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1814 ext4_group_t block_group, 1815 struct ext4_group_desc *gdp); 1816 extern unsigned ext4_num_overhead_clusters(struct super_block *sb, 1817 ext4_group_t block_group, 1818 struct ext4_group_desc *gdp); 1819 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1820 1821 /* dir.c */ 1822 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1823 struct file *, 1824 struct ext4_dir_entry_2 *, 1825 struct buffer_head *, unsigned int); 1826 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \ 1827 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 1828 (de), (bh), (offset))) 1829 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1830 __u32 minor_hash, 1831 struct ext4_dir_entry_2 *dirent); 1832 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1833 1834 /* fsync.c */ 1835 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 1836 extern int ext4_flush_completed_IO(struct inode *); 1837 1838 /* hash.c */ 1839 extern int ext4fs_dirhash(const char *name, int len, struct 1840 dx_hash_info *hinfo); 1841 1842 /* ialloc.c */ 1843 extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t, 1844 const struct qstr *qstr, __u32 goal, 1845 uid_t *owner); 1846 extern void ext4_free_inode(handle_t *, struct inode *); 1847 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 1848 extern unsigned long ext4_count_free_inodes(struct super_block *); 1849 extern unsigned long ext4_count_dirs(struct super_block *); 1850 extern void ext4_check_inodes_bitmap(struct super_block *); 1851 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 1852 extern int ext4_init_inode_table(struct super_block *sb, 1853 ext4_group_t group, int barrier); 1854 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 1855 1856 /* mballoc.c */ 1857 extern long ext4_mb_stats; 1858 extern long ext4_mb_max_to_scan; 1859 extern int ext4_mb_init(struct super_block *, int); 1860 extern int ext4_mb_release(struct super_block *); 1861 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 1862 struct ext4_allocation_request *, int *); 1863 extern int ext4_mb_reserve_blocks(struct super_block *, int); 1864 extern void ext4_discard_preallocations(struct inode *); 1865 extern int __init ext4_init_mballoc(void); 1866 extern void ext4_exit_mballoc(void); 1867 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 1868 struct buffer_head *bh, ext4_fsblk_t block, 1869 unsigned long count, int flags); 1870 extern int ext4_mb_add_groupinfo(struct super_block *sb, 1871 ext4_group_t i, struct ext4_group_desc *desc); 1872 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 1873 ext4_fsblk_t block, unsigned long count); 1874 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 1875 1876 /* inode.c */ 1877 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 1878 ext4_lblk_t, int, int *); 1879 struct buffer_head *ext4_bread(handle_t *, struct inode *, 1880 ext4_lblk_t, int, int *); 1881 int ext4_get_block(struct inode *inode, sector_t iblock, 1882 struct buffer_head *bh_result, int create); 1883 1884 extern struct inode *ext4_iget(struct super_block *, unsigned long); 1885 extern int ext4_write_inode(struct inode *, struct writeback_control *); 1886 extern int ext4_setattr(struct dentry *, struct iattr *); 1887 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 1888 struct kstat *stat); 1889 extern void ext4_evict_inode(struct inode *); 1890 extern void ext4_clear_inode(struct inode *); 1891 extern int ext4_sync_inode(handle_t *, struct inode *); 1892 extern void ext4_dirty_inode(struct inode *, int); 1893 extern int ext4_change_inode_journal_flag(struct inode *, int); 1894 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 1895 extern int ext4_can_truncate(struct inode *inode); 1896 extern void ext4_truncate(struct inode *); 1897 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 1898 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 1899 extern void ext4_set_inode_flags(struct inode *); 1900 extern void ext4_get_inode_flags(struct ext4_inode_info *); 1901 extern int ext4_alloc_da_blocks(struct inode *inode); 1902 extern void ext4_set_aops(struct inode *inode); 1903 extern int ext4_writepage_trans_blocks(struct inode *); 1904 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 1905 extern int ext4_discard_partial_page_buffers(handle_t *handle, 1906 struct address_space *mapping, loff_t from, 1907 loff_t length, int flags); 1908 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 1909 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 1910 extern void ext4_da_update_reserve_space(struct inode *inode, 1911 int used, int quota_claim); 1912 1913 /* indirect.c */ 1914 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 1915 struct ext4_map_blocks *map, int flags); 1916 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 1917 const struct iovec *iov, loff_t offset, 1918 unsigned long nr_segs); 1919 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 1920 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk); 1921 extern void ext4_ind_truncate(struct inode *inode); 1922 1923 /* ioctl.c */ 1924 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 1925 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 1926 1927 /* migrate.c */ 1928 extern int ext4_ext_migrate(struct inode *); 1929 1930 /* namei.c */ 1931 extern int ext4_orphan_add(handle_t *, struct inode *); 1932 extern int ext4_orphan_del(handle_t *, struct inode *); 1933 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 1934 __u32 start_minor_hash, __u32 *next_hash); 1935 1936 /* resize.c */ 1937 extern int ext4_group_add(struct super_block *sb, 1938 struct ext4_new_group_data *input); 1939 extern int ext4_group_extend(struct super_block *sb, 1940 struct ext4_super_block *es, 1941 ext4_fsblk_t n_blocks_count); 1942 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 1943 1944 /* super.c */ 1945 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 1946 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 1947 extern void ext4_kvfree(void *ptr); 1948 extern __printf(4, 5) 1949 void __ext4_error(struct super_block *, const char *, unsigned int, 1950 const char *, ...); 1951 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \ 1952 __LINE__, ## message) 1953 extern __printf(5, 6) 1954 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 1955 const char *, ...); 1956 extern __printf(5, 6) 1957 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 1958 const char *, ...); 1959 extern void __ext4_std_error(struct super_block *, const char *, 1960 unsigned int, int); 1961 extern __printf(4, 5) 1962 void __ext4_abort(struct super_block *, const char *, unsigned int, 1963 const char *, ...); 1964 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \ 1965 __LINE__, ## message) 1966 extern __printf(4, 5) 1967 void __ext4_warning(struct super_block *, const char *, unsigned int, 1968 const char *, ...); 1969 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \ 1970 __LINE__, ## message) 1971 extern __printf(3, 4) 1972 void ext4_msg(struct super_block *, const char *, const char *, ...); 1973 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 1974 const char *, unsigned int, const char *); 1975 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \ 1976 __LINE__, msg) 1977 extern __printf(7, 8) 1978 void __ext4_grp_locked_error(const char *, unsigned int, 1979 struct super_block *, ext4_group_t, 1980 unsigned long, ext4_fsblk_t, 1981 const char *, ...); 1982 #define ext4_grp_locked_error(sb, grp, message...) \ 1983 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message) 1984 extern void ext4_update_dynamic_rev(struct super_block *sb); 1985 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 1986 __u32 compat); 1987 extern int ext4_update_rocompat_feature(handle_t *handle, 1988 struct super_block *sb, __u32 rocompat); 1989 extern int ext4_update_incompat_feature(handle_t *handle, 1990 struct super_block *sb, __u32 incompat); 1991 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 1992 struct ext4_group_desc *bg); 1993 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 1994 struct ext4_group_desc *bg); 1995 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 1996 struct ext4_group_desc *bg); 1997 extern __u32 ext4_free_group_clusters(struct super_block *sb, 1998 struct ext4_group_desc *bg); 1999 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2000 struct ext4_group_desc *bg); 2001 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2002 struct ext4_group_desc *bg); 2003 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2004 struct ext4_group_desc *bg); 2005 extern void ext4_block_bitmap_set(struct super_block *sb, 2006 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2007 extern void ext4_inode_bitmap_set(struct super_block *sb, 2008 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2009 extern void ext4_inode_table_set(struct super_block *sb, 2010 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2011 extern void ext4_free_group_clusters_set(struct super_block *sb, 2012 struct ext4_group_desc *bg, 2013 __u32 count); 2014 extern void ext4_free_inodes_set(struct super_block *sb, 2015 struct ext4_group_desc *bg, __u32 count); 2016 extern void ext4_used_dirs_set(struct super_block *sb, 2017 struct ext4_group_desc *bg, __u32 count); 2018 extern void ext4_itable_unused_set(struct super_block *sb, 2019 struct ext4_group_desc *bg, __u32 count); 2020 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group, 2021 struct ext4_group_desc *gdp); 2022 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group, 2023 struct ext4_group_desc *gdp); 2024 2025 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2026 { 2027 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2028 le32_to_cpu(es->s_blocks_count_lo); 2029 } 2030 2031 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2032 { 2033 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2034 le32_to_cpu(es->s_r_blocks_count_lo); 2035 } 2036 2037 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2038 { 2039 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2040 le32_to_cpu(es->s_free_blocks_count_lo); 2041 } 2042 2043 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2044 ext4_fsblk_t blk) 2045 { 2046 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2047 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2048 } 2049 2050 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2051 ext4_fsblk_t blk) 2052 { 2053 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2054 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2055 } 2056 2057 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2058 ext4_fsblk_t blk) 2059 { 2060 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2061 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2062 } 2063 2064 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2065 { 2066 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2067 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2068 le32_to_cpu(raw_inode->i_size_lo); 2069 else 2070 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2071 } 2072 2073 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2074 { 2075 raw_inode->i_size_lo = cpu_to_le32(i_size); 2076 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2077 } 2078 2079 static inline 2080 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2081 ext4_group_t group) 2082 { 2083 struct ext4_group_info ***grp_info; 2084 long indexv, indexh; 2085 grp_info = EXT4_SB(sb)->s_group_info; 2086 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2087 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2088 return grp_info[indexv][indexh]; 2089 } 2090 2091 /* 2092 * Reading s_groups_count requires using smp_rmb() afterwards. See 2093 * the locking protocol documented in the comments of ext4_group_add() 2094 * in resize.c 2095 */ 2096 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2097 { 2098 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2099 2100 smp_rmb(); 2101 return ngroups; 2102 } 2103 2104 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2105 ext4_group_t block_group) 2106 { 2107 return block_group >> sbi->s_log_groups_per_flex; 2108 } 2109 2110 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2111 { 2112 return 1 << sbi->s_log_groups_per_flex; 2113 } 2114 2115 #define ext4_std_error(sb, errno) \ 2116 do { \ 2117 if ((errno)) \ 2118 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2119 } while (0) 2120 2121 #ifdef CONFIG_SMP 2122 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2123 * counters. So we need to make sure we have free clusters more 2124 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2125 */ 2126 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2127 #else 2128 #define EXT4_FREECLUSTERS_WATERMARK 0 2129 #endif 2130 2131 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2132 { 2133 /* 2134 * XXX: replace with spinlock if seen contended -bzzz 2135 */ 2136 down_write(&EXT4_I(inode)->i_data_sem); 2137 if (newsize > EXT4_I(inode)->i_disksize) 2138 EXT4_I(inode)->i_disksize = newsize; 2139 up_write(&EXT4_I(inode)->i_data_sem); 2140 return ; 2141 } 2142 2143 struct ext4_group_info { 2144 unsigned long bb_state; 2145 struct rb_root bb_free_root; 2146 ext4_grpblk_t bb_first_free; /* first free block */ 2147 ext4_grpblk_t bb_free; /* total free blocks */ 2148 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2149 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2150 struct list_head bb_prealloc_list; 2151 #ifdef DOUBLE_CHECK 2152 void *bb_bitmap; 2153 #endif 2154 struct rw_semaphore alloc_sem; 2155 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2156 * regions, index is order. 2157 * bb_counters[3] = 5 means 2158 * 5 free 8-block regions. */ 2159 }; 2160 2161 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2162 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2163 2164 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2165 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2166 2167 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2168 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2169 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2170 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2171 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2172 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2173 2174 #define EXT4_MAX_CONTENTION 8 2175 #define EXT4_CONTENTION_THRESHOLD 2 2176 2177 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2178 ext4_group_t group) 2179 { 2180 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2181 } 2182 2183 /* 2184 * Returns true if the filesystem is busy enough that attempts to 2185 * access the block group locks has run into contention. 2186 */ 2187 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2188 { 2189 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2190 } 2191 2192 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2193 { 2194 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2195 if (spin_trylock(lock)) 2196 /* 2197 * We're able to grab the lock right away, so drop the 2198 * lock contention counter. 2199 */ 2200 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2201 else { 2202 /* 2203 * The lock is busy, so bump the contention counter, 2204 * and then wait on the spin lock. 2205 */ 2206 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2207 EXT4_MAX_CONTENTION); 2208 spin_lock(lock); 2209 } 2210 } 2211 2212 static inline void ext4_unlock_group(struct super_block *sb, 2213 ext4_group_t group) 2214 { 2215 spin_unlock(ext4_group_lock_ptr(sb, group)); 2216 } 2217 2218 static inline void ext4_mark_super_dirty(struct super_block *sb) 2219 { 2220 if (EXT4_SB(sb)->s_journal == NULL) 2221 sb->s_dirt =1; 2222 } 2223 2224 /* 2225 * Block validity checking 2226 */ 2227 #define ext4_check_indirect_blockref(inode, bh) \ 2228 ext4_check_blockref(__func__, __LINE__, inode, \ 2229 (__le32 *)(bh)->b_data, \ 2230 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2231 2232 #define ext4_ind_check_inode(inode) \ 2233 ext4_check_blockref(__func__, __LINE__, inode, \ 2234 EXT4_I(inode)->i_data, \ 2235 EXT4_NDIR_BLOCKS) 2236 2237 /* 2238 * Inodes and files operations 2239 */ 2240 2241 /* dir.c */ 2242 extern const struct file_operations ext4_dir_operations; 2243 2244 /* file.c */ 2245 extern const struct inode_operations ext4_file_inode_operations; 2246 extern const struct file_operations ext4_file_operations; 2247 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2248 2249 /* namei.c */ 2250 extern const struct inode_operations ext4_dir_inode_operations; 2251 extern const struct inode_operations ext4_special_inode_operations; 2252 extern struct dentry *ext4_get_parent(struct dentry *child); 2253 2254 /* symlink.c */ 2255 extern const struct inode_operations ext4_symlink_inode_operations; 2256 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2257 2258 /* block_validity */ 2259 extern void ext4_release_system_zone(struct super_block *sb); 2260 extern int ext4_setup_system_zone(struct super_block *sb); 2261 extern int __init ext4_init_system_zone(void); 2262 extern void ext4_exit_system_zone(void); 2263 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2264 ext4_fsblk_t start_blk, 2265 unsigned int count); 2266 extern int ext4_check_blockref(const char *, unsigned int, 2267 struct inode *, __le32 *, unsigned int); 2268 2269 /* extents.c */ 2270 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2271 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2272 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 2273 int chunk); 2274 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2275 struct ext4_map_blocks *map, int flags); 2276 extern void ext4_ext_truncate(struct inode *); 2277 extern int ext4_ext_punch_hole(struct file *file, loff_t offset, 2278 loff_t length); 2279 extern void ext4_ext_init(struct super_block *); 2280 extern void ext4_ext_release(struct super_block *); 2281 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2282 loff_t len); 2283 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 2284 ssize_t len); 2285 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2286 struct ext4_map_blocks *map, int flags); 2287 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2288 __u64 start, __u64 len); 2289 /* move_extent.c */ 2290 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2291 __u64 start_orig, __u64 start_donor, 2292 __u64 len, __u64 *moved_len); 2293 2294 /* page-io.c */ 2295 extern int __init ext4_init_pageio(void); 2296 extern void ext4_exit_pageio(void); 2297 extern void ext4_ioend_wait(struct inode *); 2298 extern void ext4_free_io_end(ext4_io_end_t *io); 2299 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2300 extern int ext4_end_io_nolock(ext4_io_end_t *io); 2301 extern void ext4_io_submit(struct ext4_io_submit *io); 2302 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2303 struct page *page, 2304 int len, 2305 struct writeback_control *wbc); 2306 2307 /* mmp.c */ 2308 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2309 2310 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */ 2311 enum ext4_state_bits { 2312 BH_Uninit /* blocks are allocated but uninitialized on disk */ 2313 = BH_JBDPrivateStart, 2314 BH_AllocFromCluster, /* allocated blocks were part of already 2315 * allocated cluster. Note that this flag will 2316 * never, ever appear in a buffer_head's state 2317 * flag. See EXT4_MAP_FROM_CLUSTER to see where 2318 * this is used. */ 2319 BH_Da_Mapped, /* Delayed allocated block that now has a mapping. This 2320 * flag is set when ext4_map_blocks is called on a 2321 * delayed allocated block to get its real mapping. */ 2322 }; 2323 2324 BUFFER_FNS(Uninit, uninit) 2325 TAS_BUFFER_FNS(Uninit, uninit) 2326 BUFFER_FNS(Da_Mapped, da_mapped) 2327 2328 /* 2329 * Add new method to test wether block and inode bitmaps are properly 2330 * initialized. With uninit_bg reading the block from disk is not enough 2331 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2332 */ 2333 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2334 2335 static inline int bitmap_uptodate(struct buffer_head *bh) 2336 { 2337 return (buffer_uptodate(bh) && 2338 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2339 } 2340 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2341 { 2342 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2343 } 2344 2345 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2346 2347 /* For ioend & aio unwritten conversion wait queues */ 2348 #define EXT4_WQ_HASH_SZ 37 2349 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2350 EXT4_WQ_HASH_SZ]) 2351 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2352 EXT4_WQ_HASH_SZ]) 2353 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2354 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2355 2356 #define EXT4_RESIZING 0 2357 extern int ext4_resize_begin(struct super_block *sb); 2358 extern void ext4_resize_end(struct super_block *sb); 2359 2360 #endif /* __KERNEL__ */ 2361 2362 #include "ext4_extents.h" 2363 2364 #endif /* _EXT4_H */ 2365