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