1 /* 2 * ext4.h 3 * 4 * Copyright (C) 1992, 1993, 1994, 1995 5 * Remy Card (card@masi.ibp.fr) 6 * Laboratoire MASI - Institut Blaise Pascal 7 * Universite Pierre et Marie Curie (Paris VI) 8 * 9 * from 10 * 11 * linux/include/linux/minix_fs.h 12 * 13 * Copyright (C) 1991, 1992 Linus Torvalds 14 */ 15 16 #ifndef _EXT4_H 17 #define _EXT4_H 18 19 #include <linux/types.h> 20 #include <linux/blkdev.h> 21 #include <linux/magic.h> 22 #include <linux/jbd2.h> 23 #include <linux/quota.h> 24 #include <linux/rwsem.h> 25 #include <linux/rbtree.h> 26 #include <linux/seqlock.h> 27 #include <linux/mutex.h> 28 #include <linux/timer.h> 29 #include <linux/wait.h> 30 #include <linux/blockgroup_lock.h> 31 #include <linux/percpu_counter.h> 32 #include <crypto/hash.h> 33 #ifdef __KERNEL__ 34 #include <linux/compat.h> 35 #endif 36 37 /* 38 * The fourth extended filesystem constants/structures 39 */ 40 41 /* 42 * Define EXT4FS_DEBUG to produce debug messages 43 */ 44 #undef EXT4FS_DEBUG 45 46 /* 47 * Debug code 48 */ 49 #ifdef EXT4FS_DEBUG 50 #define ext4_debug(f, a...) \ 51 do { \ 52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 53 __FILE__, __LINE__, __func__); \ 54 printk(KERN_DEBUG f, ## a); \ 55 } while (0) 56 #else 57 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 58 #endif 59 60 /* 61 * Turn on EXT_DEBUG to get lots of info about extents operations. 62 */ 63 #define EXT_DEBUG__ 64 #ifdef EXT_DEBUG 65 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 66 #else 67 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 68 #endif 69 70 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 71 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 72 73 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 75 76 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 78 79 /* data type for block offset of block group */ 80 typedef int ext4_grpblk_t; 81 82 /* data type for filesystem-wide blocks number */ 83 typedef unsigned long long ext4_fsblk_t; 84 85 /* data type for file logical block number */ 86 typedef __u32 ext4_lblk_t; 87 88 /* data type for block group number */ 89 typedef unsigned int ext4_group_t; 90 91 /* 92 * Flags used in mballoc's allocation_context flags field. 93 * 94 * Also used to show what's going on for debugging purposes when the 95 * flag field is exported via the traceport interface 96 */ 97 98 /* prefer goal again. length */ 99 #define EXT4_MB_HINT_MERGE 0x0001 100 /* blocks already reserved */ 101 #define EXT4_MB_HINT_RESERVED 0x0002 102 /* metadata is being allocated */ 103 #define EXT4_MB_HINT_METADATA 0x0004 104 /* first blocks in the file */ 105 #define EXT4_MB_HINT_FIRST 0x0008 106 /* search for the best chunk */ 107 #define EXT4_MB_HINT_BEST 0x0010 108 /* data is being allocated */ 109 #define EXT4_MB_HINT_DATA 0x0020 110 /* don't preallocate (for tails) */ 111 #define EXT4_MB_HINT_NOPREALLOC 0x0040 112 /* allocate for locality group */ 113 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 114 /* allocate goal blocks or none */ 115 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 116 /* goal is meaningful */ 117 #define EXT4_MB_HINT_TRY_GOAL 0x0200 118 /* blocks already pre-reserved by delayed allocation */ 119 #define EXT4_MB_DELALLOC_RESERVED 0x0400 120 /* We are doing stream allocation */ 121 #define EXT4_MB_STREAM_ALLOC 0x0800 122 /* Use reserved root blocks if needed */ 123 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 124 /* Use blocks from reserved pool */ 125 #define EXT4_MB_USE_RESERVED 0x2000 126 127 struct ext4_allocation_request { 128 /* target inode for block we're allocating */ 129 struct inode *inode; 130 /* how many blocks we want to allocate */ 131 unsigned int len; 132 /* logical block in target inode */ 133 ext4_lblk_t logical; 134 /* the closest logical allocated block to the left */ 135 ext4_lblk_t lleft; 136 /* the closest logical allocated block to the right */ 137 ext4_lblk_t lright; 138 /* phys. target (a hint) */ 139 ext4_fsblk_t goal; 140 /* phys. block for the closest logical allocated block to the left */ 141 ext4_fsblk_t pleft; 142 /* phys. block for the closest logical allocated block to the right */ 143 ext4_fsblk_t pright; 144 /* flags. see above EXT4_MB_HINT_* */ 145 unsigned int flags; 146 }; 147 148 /* 149 * Logical to physical block mapping, used by ext4_map_blocks() 150 * 151 * This structure is used to pass requests into ext4_map_blocks() as 152 * well as to store the information returned by ext4_map_blocks(). It 153 * takes less room on the stack than a struct buffer_head. 154 */ 155 #define EXT4_MAP_NEW (1 << BH_New) 156 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 157 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 158 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 159 #define EXT4_MAP_UNINIT (1 << BH_Uninit) 160 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of 161 * ext4_map_blocks wants to know whether or not the underlying cluster has 162 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that 163 * the requested mapping was from previously mapped (or delayed allocated) 164 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster 165 * should never appear on buffer_head's state flags. 166 */ 167 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster) 168 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 169 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 170 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER) 171 172 struct ext4_map_blocks { 173 ext4_fsblk_t m_pblk; 174 ext4_lblk_t m_lblk; 175 unsigned int m_len; 176 unsigned int m_flags; 177 }; 178 179 /* 180 * Flags for ext4_io_end->flags 181 */ 182 #define EXT4_IO_END_UNWRITTEN 0x0001 183 #define EXT4_IO_END_ERROR 0x0002 184 #define EXT4_IO_END_DIRECT 0x0004 185 186 /* 187 * For converting uninitialized extents on a work queue. 188 */ 189 typedef struct ext4_io_end { 190 struct list_head list; /* per-file finished IO list */ 191 struct inode *inode; /* file being written to */ 192 unsigned int flag; /* unwritten or not */ 193 loff_t offset; /* offset in the file */ 194 ssize_t size; /* size of the extent */ 195 struct kiocb *iocb; /* iocb struct for AIO */ 196 int result; /* error value for AIO */ 197 atomic_t count; /* reference counter */ 198 } ext4_io_end_t; 199 200 struct ext4_io_submit { 201 int io_op; 202 struct bio *io_bio; 203 ext4_io_end_t *io_end; 204 sector_t io_next_block; 205 }; 206 207 /* 208 * Special inodes numbers 209 */ 210 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 211 #define EXT4_ROOT_INO 2 /* Root inode */ 212 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 213 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 214 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 215 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 216 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 217 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 218 219 /* First non-reserved inode for old ext4 filesystems */ 220 #define EXT4_GOOD_OLD_FIRST_INO 11 221 222 /* 223 * Maximal count of links to a file 224 */ 225 #define EXT4_LINK_MAX 65000 226 227 /* 228 * Macro-instructions used to manage several block sizes 229 */ 230 #define EXT4_MIN_BLOCK_SIZE 1024 231 #define EXT4_MAX_BLOCK_SIZE 65536 232 #define EXT4_MIN_BLOCK_LOG_SIZE 10 233 #define EXT4_MAX_BLOCK_LOG_SIZE 16 234 #ifdef __KERNEL__ 235 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 236 #else 237 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 238 #endif 239 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 240 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 241 EXT4_SB(s)->s_cluster_bits) 242 #ifdef __KERNEL__ 243 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 244 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 245 #else 246 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 247 #endif 248 #ifdef __KERNEL__ 249 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 250 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 251 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 252 #else 253 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 254 EXT4_GOOD_OLD_INODE_SIZE : \ 255 (s)->s_inode_size) 256 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 257 EXT4_GOOD_OLD_FIRST_INO : \ 258 (s)->s_first_ino) 259 #endif 260 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 261 262 /* Translate a block number to a cluster number */ 263 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 264 /* Translate a cluster number to a block number */ 265 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 266 /* Translate # of blks to # of clusters */ 267 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 268 (sbi)->s_cluster_bits) 269 270 /* 271 * Structure of a blocks group descriptor 272 */ 273 struct ext4_group_desc 274 { 275 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 276 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 277 __le32 bg_inode_table_lo; /* Inodes table block */ 278 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 279 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 280 __le16 bg_used_dirs_count_lo; /* Directories count */ 281 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 282 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 283 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 284 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 285 __le16 bg_itable_unused_lo; /* Unused inodes count */ 286 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 287 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 288 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 289 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 290 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 291 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 292 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 293 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 294 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 295 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 296 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 297 __u32 bg_reserved; 298 }; 299 300 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 301 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 302 sizeof(__le16)) 303 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 304 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 305 sizeof(__le16)) 306 307 /* 308 * Structure of a flex block group info 309 */ 310 311 struct flex_groups { 312 atomic64_t free_clusters; 313 atomic_t free_inodes; 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_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 377 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 378 379 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 380 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */ 381 382 /* Flags that should be inherited by new inodes from their parent. */ 383 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 384 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 385 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 386 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 387 388 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 389 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 390 391 /* Flags that are appropriate for non-directories/regular files. */ 392 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 393 394 /* Mask out flags that are inappropriate for the given type of inode. */ 395 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 396 { 397 if (S_ISDIR(mode)) 398 return flags; 399 else if (S_ISREG(mode)) 400 return flags & EXT4_REG_FLMASK; 401 else 402 return flags & EXT4_OTHER_FLMASK; 403 } 404 405 /* 406 * Inode flags used for atomic set/get 407 */ 408 enum { 409 EXT4_INODE_SECRM = 0, /* Secure deletion */ 410 EXT4_INODE_UNRM = 1, /* Undelete */ 411 EXT4_INODE_COMPR = 2, /* Compress file */ 412 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 413 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 414 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 415 EXT4_INODE_NODUMP = 6, /* do not dump file */ 416 EXT4_INODE_NOATIME = 7, /* do not update atime */ 417 /* Reserved for compression usage... */ 418 EXT4_INODE_DIRTY = 8, 419 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 420 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 421 EXT4_INODE_ECOMPR = 11, /* Compression error */ 422 /* End compression flags --- maybe not all used */ 423 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 424 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 425 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 426 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 427 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 428 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 429 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 430 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 431 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 432 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 433 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 434 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 435 }; 436 437 /* 438 * Since it's pretty easy to mix up bit numbers and hex values, we use a 439 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 440 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 441 * any extra space in the compiled kernel image, otherwise, the build will fail. 442 * It's important that these values are the same, since we are using 443 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 444 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 445 * values found in ext2, ext3 and ext4 filesystems, and of course the values 446 * defined in e2fsprogs. 447 * 448 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 449 */ 450 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 451 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 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(INLINE_DATA); 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 * finally doing the actual allocation of delayed blocks */ 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 /* Eventual metadata allocation (due to growing extent tree) 548 * should not fail, so try to use reserved blocks for that.*/ 549 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 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 /* Do not take i_data_sem locking in ext4_map_blocks */ 555 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 556 /* Do not put hole in extent cache */ 557 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200 558 559 /* 560 * Flags used by ext4_free_blocks 561 */ 562 #define EXT4_FREE_BLOCKS_METADATA 0x0001 563 #define EXT4_FREE_BLOCKS_FORGET 0x0002 564 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 565 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 566 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 567 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 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 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 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 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 647 __le16 l_i_reserved; 648 } linux2; 649 struct { 650 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 651 __u16 h_i_mode_high; 652 __u16 h_i_uid_high; 653 __u16 h_i_gid_high; 654 __u32 h_i_author; 655 } hurd2; 656 struct { 657 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 658 __le16 m_i_file_acl_high; 659 __u32 m_i_reserved2[2]; 660 } masix2; 661 } osd2; /* OS dependent 2 */ 662 __le16 i_extra_isize; 663 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 664 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 665 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 666 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 667 __le32 i_crtime; /* File Creation time */ 668 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 669 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 670 }; 671 672 struct move_extent { 673 __u32 reserved; /* should be zero */ 674 __u32 donor_fd; /* donor file descriptor */ 675 __u64 orig_start; /* logical start offset in block for orig */ 676 __u64 donor_start; /* logical start offset in block for donor */ 677 __u64 len; /* block length to be moved */ 678 __u64 moved_len; /* moved block length */ 679 }; 680 681 #define EXT4_EPOCH_BITS 2 682 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 683 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 684 685 /* 686 * Extended fields will fit into an inode if the filesystem was formatted 687 * with large inodes (-I 256 or larger) and there are not currently any EAs 688 * consuming all of the available space. For new inodes we always reserve 689 * enough space for the kernel's known extended fields, but for inodes 690 * created with an old kernel this might not have been the case. None of 691 * the extended inode fields is critical for correct filesystem operation. 692 * This macro checks if a certain field fits in the inode. Note that 693 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 694 */ 695 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 696 ((offsetof(typeof(*ext4_inode), field) + \ 697 sizeof((ext4_inode)->field)) \ 698 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 699 (einode)->i_extra_isize)) \ 700 701 static inline __le32 ext4_encode_extra_time(struct timespec *time) 702 { 703 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 704 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 705 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 706 } 707 708 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 709 { 710 if (sizeof(time->tv_sec) > 4) 711 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 712 << 32; 713 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 714 } 715 716 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 717 do { \ 718 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 719 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 720 (raw_inode)->xtime ## _extra = \ 721 ext4_encode_extra_time(&(inode)->xtime); \ 722 } while (0) 723 724 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 725 do { \ 726 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 727 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 728 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 729 (raw_inode)->xtime ## _extra = \ 730 ext4_encode_extra_time(&(einode)->xtime); \ 731 } while (0) 732 733 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 734 do { \ 735 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 736 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 737 ext4_decode_extra_time(&(inode)->xtime, \ 738 raw_inode->xtime ## _extra); \ 739 else \ 740 (inode)->xtime.tv_nsec = 0; \ 741 } while (0) 742 743 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 744 do { \ 745 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 746 (einode)->xtime.tv_sec = \ 747 (signed)le32_to_cpu((raw_inode)->xtime); \ 748 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 749 ext4_decode_extra_time(&(einode)->xtime, \ 750 raw_inode->xtime ## _extra); \ 751 else \ 752 (einode)->xtime.tv_nsec = 0; \ 753 } while (0) 754 755 #define i_disk_version osd1.linux1.l_i_version 756 757 #if defined(__KERNEL__) || defined(__linux__) 758 #define i_reserved1 osd1.linux1.l_i_reserved1 759 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 760 #define i_blocks_high osd2.linux2.l_i_blocks_high 761 #define i_uid_low i_uid 762 #define i_gid_low i_gid 763 #define i_uid_high osd2.linux2.l_i_uid_high 764 #define i_gid_high osd2.linux2.l_i_gid_high 765 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 766 767 #elif defined(__GNU__) 768 769 #define i_translator osd1.hurd1.h_i_translator 770 #define i_uid_high osd2.hurd2.h_i_uid_high 771 #define i_gid_high osd2.hurd2.h_i_gid_high 772 #define i_author osd2.hurd2.h_i_author 773 774 #elif defined(__masix__) 775 776 #define i_reserved1 osd1.masix1.m_i_reserved1 777 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 778 #define i_reserved2 osd2.masix2.m_i_reserved2 779 780 #endif /* defined(__KERNEL__) || defined(__linux__) */ 781 782 #include "extents_status.h" 783 784 /* 785 * fourth extended file system inode data in memory 786 */ 787 struct ext4_inode_info { 788 __le32 i_data[15]; /* unconverted */ 789 __u32 i_dtime; 790 ext4_fsblk_t i_file_acl; 791 792 /* 793 * i_block_group is the number of the block group which contains 794 * this file's inode. Constant across the lifetime of the inode, 795 * it is ued for making block allocation decisions - we try to 796 * place a file's data blocks near its inode block, and new inodes 797 * near to their parent directory's inode. 798 */ 799 ext4_group_t i_block_group; 800 ext4_lblk_t i_dir_start_lookup; 801 #if (BITS_PER_LONG < 64) 802 unsigned long i_state_flags; /* Dynamic state flags */ 803 #endif 804 unsigned long i_flags; 805 806 /* 807 * Extended attributes can be read independently of the main file 808 * data. Taking i_mutex even when reading would cause contention 809 * between readers of EAs and writers of regular file data, so 810 * instead we synchronize on xattr_sem when reading or changing 811 * EAs. 812 */ 813 struct rw_semaphore xattr_sem; 814 815 struct list_head i_orphan; /* unlinked but open inodes */ 816 817 /* 818 * i_disksize keeps track of what the inode size is ON DISK, not 819 * in memory. During truncate, i_size is set to the new size by 820 * the VFS prior to calling ext4_truncate(), but the filesystem won't 821 * set i_disksize to 0 until the truncate is actually under way. 822 * 823 * The intent is that i_disksize always represents the blocks which 824 * are used by this file. This allows recovery to restart truncate 825 * on orphans if we crash during truncate. We actually write i_disksize 826 * into the on-disk inode when writing inodes out, instead of i_size. 827 * 828 * The only time when i_disksize and i_size may be different is when 829 * a truncate is in progress. The only things which change i_disksize 830 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 831 */ 832 loff_t i_disksize; 833 834 /* 835 * i_data_sem is for serialising ext4_truncate() against 836 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 837 * data tree are chopped off during truncate. We can't do that in 838 * ext4 because whenever we perform intermediate commits during 839 * truncate, the inode and all the metadata blocks *must* be in a 840 * consistent state which allows truncation of the orphans to restart 841 * during recovery. Hence we must fix the get_block-vs-truncate race 842 * by other means, so we have i_data_sem. 843 */ 844 struct rw_semaphore i_data_sem; 845 struct inode vfs_inode; 846 struct jbd2_inode *jinode; 847 848 /* 849 * File creation time. Its function is same as that of 850 * struct timespec i_{a,c,m}time in the generic inode. 851 */ 852 struct timespec i_crtime; 853 854 /* mballoc */ 855 struct list_head i_prealloc_list; 856 spinlock_t i_prealloc_lock; 857 858 /* extents status tree */ 859 struct ext4_es_tree i_es_tree; 860 rwlock_t i_es_lock; 861 struct list_head i_es_lru; 862 unsigned int i_es_lru_nr; /* protected by i_es_lock */ 863 864 /* ialloc */ 865 ext4_group_t i_last_alloc_group; 866 867 /* allocation reservation info for delalloc */ 868 /* In case of bigalloc, these refer to clusters rather than blocks */ 869 unsigned int i_reserved_data_blocks; 870 unsigned int i_reserved_meta_blocks; 871 unsigned int i_allocated_meta_blocks; 872 ext4_lblk_t i_da_metadata_calc_last_lblock; 873 int i_da_metadata_calc_len; 874 875 /* on-disk additional length */ 876 __u16 i_extra_isize; 877 878 /* Indicate the inline data space. */ 879 u16 i_inline_off; 880 u16 i_inline_size; 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 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 892 struct work_struct i_unwritten_work; /* deferred extent conversion */ 893 894 spinlock_t i_block_reservation_lock; 895 896 /* 897 * Transactions that contain inode's metadata needed to complete 898 * fsync and fdatasync, respectively. 899 */ 900 tid_t i_sync_tid; 901 tid_t i_datasync_tid; 902 903 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 904 __u32 i_csum_seed; 905 }; 906 907 /* 908 * File system states 909 */ 910 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 911 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 912 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 913 914 /* 915 * Misc. filesystem flags 916 */ 917 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 918 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 919 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 920 921 /* 922 * Mount flags set via mount options or defaults 923 */ 924 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 925 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 926 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 927 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 928 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 929 #define EXT4_MOUNT_ERRORS_MASK 0x00070 930 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 931 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 932 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 933 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 934 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 935 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 936 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 937 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 938 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 939 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 940 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 941 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 942 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 943 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 944 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 945 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 946 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 947 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 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 /* 955 * Mount flags set either automatically (could not be set by mount option) 956 * based on per file system feature or property or in special cases such as 957 * distinguishing between explicit mount option definition and default. 958 */ 959 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 960 specified delalloc */ 961 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 962 size of blocksize * 8 963 blocks */ 964 965 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 966 ~EXT4_MOUNT_##opt 967 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 968 EXT4_MOUNT_##opt 969 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 970 EXT4_MOUNT_##opt) 971 972 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 973 ~EXT4_MOUNT2_##opt 974 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 975 EXT4_MOUNT2_##opt 976 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 977 EXT4_MOUNT2_##opt) 978 979 #define ext4_test_and_set_bit __test_and_set_bit_le 980 #define ext4_set_bit __set_bit_le 981 #define ext4_set_bit_atomic ext2_set_bit_atomic 982 #define ext4_test_and_clear_bit __test_and_clear_bit_le 983 #define ext4_clear_bit __clear_bit_le 984 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 985 #define ext4_test_bit test_bit_le 986 #define ext4_find_next_zero_bit find_next_zero_bit_le 987 #define ext4_find_next_bit find_next_bit_le 988 989 extern void ext4_set_bits(void *bm, int cur, int len); 990 991 /* 992 * Maximal mount counts between two filesystem checks 993 */ 994 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 995 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 996 997 /* 998 * Behaviour when detecting errors 999 */ 1000 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1001 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1002 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1003 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1004 1005 /* Metadata checksum algorithm codes */ 1006 #define EXT4_CRC32C_CHKSUM 1 1007 1008 /* 1009 * Structure of the super block 1010 */ 1011 struct ext4_super_block { 1012 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1013 __le32 s_blocks_count_lo; /* Blocks count */ 1014 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1015 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1016 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1017 __le32 s_first_data_block; /* First Data Block */ 1018 __le32 s_log_block_size; /* Block size */ 1019 __le32 s_log_cluster_size; /* Allocation cluster size */ 1020 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1021 __le32 s_clusters_per_group; /* # Clusters per group */ 1022 __le32 s_inodes_per_group; /* # Inodes per group */ 1023 __le32 s_mtime; /* Mount time */ 1024 /*30*/ __le32 s_wtime; /* Write time */ 1025 __le16 s_mnt_count; /* Mount count */ 1026 __le16 s_max_mnt_count; /* Maximal mount count */ 1027 __le16 s_magic; /* Magic signature */ 1028 __le16 s_state; /* File system state */ 1029 __le16 s_errors; /* Behaviour when detecting errors */ 1030 __le16 s_minor_rev_level; /* minor revision level */ 1031 /*40*/ __le32 s_lastcheck; /* time of last check */ 1032 __le32 s_checkinterval; /* max. time between checks */ 1033 __le32 s_creator_os; /* OS */ 1034 __le32 s_rev_level; /* Revision level */ 1035 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1036 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1037 /* 1038 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1039 * 1040 * Note: the difference between the compatible feature set and 1041 * the incompatible feature set is that if there is a bit set 1042 * in the incompatible feature set that the kernel doesn't 1043 * know about, it should refuse to mount the filesystem. 1044 * 1045 * e2fsck's requirements are more strict; if it doesn't know 1046 * about a feature in either the compatible or incompatible 1047 * feature set, it must abort and not try to meddle with 1048 * things it doesn't understand... 1049 */ 1050 __le32 s_first_ino; /* First non-reserved inode */ 1051 __le16 s_inode_size; /* size of inode structure */ 1052 __le16 s_block_group_nr; /* block group # of this superblock */ 1053 __le32 s_feature_compat; /* compatible feature set */ 1054 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1055 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1056 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1057 /*78*/ char s_volume_name[16]; /* volume name */ 1058 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1059 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1060 /* 1061 * Performance hints. Directory preallocation should only 1062 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1063 */ 1064 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1065 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1066 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1067 /* 1068 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1069 */ 1070 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1071 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1072 __le32 s_journal_dev; /* device number of journal file */ 1073 __le32 s_last_orphan; /* start of list of inodes to delete */ 1074 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1075 __u8 s_def_hash_version; /* Default hash version to use */ 1076 __u8 s_jnl_backup_type; 1077 __le16 s_desc_size; /* size of group descriptor */ 1078 /*100*/ __le32 s_default_mount_opts; 1079 __le32 s_first_meta_bg; /* First metablock block group */ 1080 __le32 s_mkfs_time; /* When the filesystem was created */ 1081 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1082 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1083 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1084 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1085 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1086 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1087 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1088 __le32 s_flags; /* Miscellaneous flags */ 1089 __le16 s_raid_stride; /* RAID stride */ 1090 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1091 __le64 s_mmp_block; /* Block for multi-mount protection */ 1092 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1093 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1094 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1095 __le16 s_reserved_pad; 1096 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1097 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1098 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1099 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1100 snapshot's future use */ 1101 __le32 s_snapshot_list; /* inode number of the head of the 1102 on-disk snapshot list */ 1103 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1104 __le32 s_error_count; /* number of fs errors */ 1105 __le32 s_first_error_time; /* first time an error happened */ 1106 __le32 s_first_error_ino; /* inode involved in first error */ 1107 __le64 s_first_error_block; /* block involved of first error */ 1108 __u8 s_first_error_func[32]; /* function where the error happened */ 1109 __le32 s_first_error_line; /* line number where error happened */ 1110 __le32 s_last_error_time; /* most recent time of an error */ 1111 __le32 s_last_error_ino; /* inode involved in last error */ 1112 __le32 s_last_error_line; /* line number where error happened */ 1113 __le64 s_last_error_block; /* block involved of last error */ 1114 __u8 s_last_error_func[32]; /* function where the error happened */ 1115 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1116 __u8 s_mount_opts[64]; 1117 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1118 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1119 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1120 __le32 s_reserved[108]; /* Padding to the end of the block */ 1121 __le32 s_checksum; /* crc32c(superblock) */ 1122 }; 1123 1124 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1125 1126 #ifdef __KERNEL__ 1127 1128 /* 1129 * run-time mount flags 1130 */ 1131 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1132 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1133 1134 /* 1135 * fourth extended-fs super-block data in memory 1136 */ 1137 struct ext4_sb_info { 1138 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1139 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1140 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1141 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1142 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1143 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1144 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1145 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1146 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1147 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1148 unsigned long s_overhead; /* # of fs overhead clusters */ 1149 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1150 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1151 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1152 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1153 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1154 struct buffer_head **s_group_desc; 1155 unsigned int s_mount_opt; 1156 unsigned int s_mount_opt2; 1157 unsigned int s_mount_flags; 1158 unsigned int s_def_mount_opt; 1159 ext4_fsblk_t s_sb_block; 1160 atomic64_t s_resv_clusters; 1161 kuid_t s_resuid; 1162 kgid_t s_resgid; 1163 unsigned short s_mount_state; 1164 unsigned short s_pad; 1165 int s_addr_per_block_bits; 1166 int s_desc_per_block_bits; 1167 int s_inode_size; 1168 int s_first_ino; 1169 unsigned int s_inode_readahead_blks; 1170 unsigned int s_inode_goal; 1171 spinlock_t s_next_gen_lock; 1172 u32 s_next_generation; 1173 u32 s_hash_seed[4]; 1174 int s_def_hash_version; 1175 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1176 struct percpu_counter s_freeclusters_counter; 1177 struct percpu_counter s_freeinodes_counter; 1178 struct percpu_counter s_dirs_counter; 1179 struct percpu_counter s_dirtyclusters_counter; 1180 struct blockgroup_lock *s_blockgroup_lock; 1181 struct proc_dir_entry *s_proc; 1182 struct kobject s_kobj; 1183 struct completion s_kobj_unregister; 1184 struct super_block *s_sb; 1185 1186 /* Journaling */ 1187 struct journal_s *s_journal; 1188 struct list_head s_orphan; 1189 struct mutex s_orphan_lock; 1190 unsigned long s_resize_flags; /* Flags indicating if there 1191 is a resizer */ 1192 unsigned long s_commit_interval; 1193 u32 s_max_batch_time; 1194 u32 s_min_batch_time; 1195 struct block_device *journal_bdev; 1196 #ifdef CONFIG_QUOTA 1197 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 1198 int s_jquota_fmt; /* Format of quota to use */ 1199 #endif 1200 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1201 struct rb_root system_blks; 1202 1203 #ifdef EXTENTS_STATS 1204 /* ext4 extents stats */ 1205 unsigned long s_ext_min; 1206 unsigned long s_ext_max; 1207 unsigned long s_depth_max; 1208 spinlock_t s_ext_stats_lock; 1209 unsigned long s_ext_blocks; 1210 unsigned long s_ext_extents; 1211 #endif 1212 1213 /* for buddy allocator */ 1214 struct ext4_group_info ***s_group_info; 1215 struct inode *s_buddy_cache; 1216 spinlock_t s_md_lock; 1217 unsigned short *s_mb_offsets; 1218 unsigned int *s_mb_maxs; 1219 unsigned int s_group_info_size; 1220 1221 /* tunables */ 1222 unsigned long s_stripe; 1223 unsigned int s_mb_stream_request; 1224 unsigned int s_mb_max_to_scan; 1225 unsigned int s_mb_min_to_scan; 1226 unsigned int s_mb_stats; 1227 unsigned int s_mb_order2_reqs; 1228 unsigned int s_mb_group_prealloc; 1229 unsigned int s_max_dir_size_kb; 1230 /* where last allocation was done - for stream allocation */ 1231 unsigned long s_mb_last_group; 1232 unsigned long s_mb_last_start; 1233 1234 /* stats for buddy allocator */ 1235 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1236 atomic_t s_bal_success; /* we found long enough chunks */ 1237 atomic_t s_bal_allocated; /* in blocks */ 1238 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1239 atomic_t s_bal_goals; /* goal hits */ 1240 atomic_t s_bal_breaks; /* too long searches */ 1241 atomic_t s_bal_2orders; /* 2^order hits */ 1242 spinlock_t s_bal_lock; 1243 unsigned long s_mb_buddies_generated; 1244 unsigned long long s_mb_generation_time; 1245 atomic_t s_mb_lost_chunks; 1246 atomic_t s_mb_preallocated; 1247 atomic_t s_mb_discarded; 1248 atomic_t s_lock_busy; 1249 1250 /* locality groups */ 1251 struct ext4_locality_group __percpu *s_locality_groups; 1252 1253 /* for write statistics */ 1254 unsigned long s_sectors_written_start; 1255 u64 s_kbytes_written; 1256 1257 /* the size of zero-out chunk */ 1258 unsigned int s_extent_max_zeroout_kb; 1259 1260 unsigned int s_log_groups_per_flex; 1261 struct flex_groups *s_flex_groups; 1262 ext4_group_t s_flex_groups_allocated; 1263 1264 /* workqueue for dio unwritten */ 1265 struct workqueue_struct *dio_unwritten_wq; 1266 1267 /* timer for periodic error stats printing */ 1268 struct timer_list s_err_report; 1269 1270 /* Lazy inode table initialization info */ 1271 struct ext4_li_request *s_li_request; 1272 /* Wait multiplier for lazy initialization thread */ 1273 unsigned int s_li_wait_mult; 1274 1275 /* Kernel thread for multiple mount protection */ 1276 struct task_struct *s_mmp_tsk; 1277 1278 /* record the last minlen when FITRIM is called. */ 1279 atomic_t s_last_trim_minblks; 1280 1281 /* Reference to checksum algorithm driver via cryptoapi */ 1282 struct crypto_shash *s_chksum_driver; 1283 1284 /* Precomputed FS UUID checksum for seeding other checksums */ 1285 __u32 s_csum_seed; 1286 1287 /* Reclaim extents from extent status tree */ 1288 struct shrinker s_es_shrinker; 1289 struct list_head s_es_lru; 1290 struct percpu_counter s_extent_cache_cnt; 1291 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp; 1292 }; 1293 1294 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1295 { 1296 return sb->s_fs_info; 1297 } 1298 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1299 { 1300 return container_of(inode, struct ext4_inode_info, vfs_inode); 1301 } 1302 1303 static inline struct timespec ext4_current_time(struct inode *inode) 1304 { 1305 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1306 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1307 } 1308 1309 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1310 { 1311 return ino == EXT4_ROOT_INO || 1312 ino == EXT4_USR_QUOTA_INO || 1313 ino == EXT4_GRP_QUOTA_INO || 1314 ino == EXT4_BOOT_LOADER_INO || 1315 ino == EXT4_JOURNAL_INO || 1316 ino == EXT4_RESIZE_INO || 1317 (ino >= EXT4_FIRST_INO(sb) && 1318 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1319 } 1320 1321 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1322 struct ext4_io_end *io_end) 1323 { 1324 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1325 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1326 atomic_inc(&EXT4_I(inode)->i_unwritten); 1327 } 1328 } 1329 1330 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1331 { 1332 return inode->i_private; 1333 } 1334 1335 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1336 { 1337 inode->i_private = io; 1338 } 1339 1340 /* 1341 * Inode dynamic state flags 1342 */ 1343 enum { 1344 EXT4_STATE_JDATA, /* journaled data exists */ 1345 EXT4_STATE_NEW, /* inode is newly created */ 1346 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1347 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1348 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1349 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1350 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1351 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1352 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */ 1353 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1354 nolocking */ 1355 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1356 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */ 1357 }; 1358 1359 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1360 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1361 { \ 1362 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1363 } \ 1364 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1365 { \ 1366 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1367 } \ 1368 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1369 { \ 1370 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1371 } 1372 1373 EXT4_INODE_BIT_FNS(flag, flags, 0) 1374 #if (BITS_PER_LONG < 64) 1375 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1376 1377 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1378 { 1379 (ei)->i_state_flags = 0; 1380 } 1381 #else 1382 EXT4_INODE_BIT_FNS(state, flags, 32) 1383 1384 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1385 { 1386 /* We depend on the fact that callers will set i_flags */ 1387 } 1388 #endif 1389 #else 1390 /* Assume that user mode programs are passing in an ext4fs superblock, not 1391 * a kernel struct super_block. This will allow us to call the feature-test 1392 * macros from user land. */ 1393 #define EXT4_SB(sb) (sb) 1394 #endif 1395 1396 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1397 1398 /* 1399 * Codes for operating systems 1400 */ 1401 #define EXT4_OS_LINUX 0 1402 #define EXT4_OS_HURD 1 1403 #define EXT4_OS_MASIX 2 1404 #define EXT4_OS_FREEBSD 3 1405 #define EXT4_OS_LITES 4 1406 1407 /* 1408 * Revision levels 1409 */ 1410 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1411 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1412 1413 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1414 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1415 1416 #define EXT4_GOOD_OLD_INODE_SIZE 128 1417 1418 /* 1419 * Feature set definitions 1420 */ 1421 1422 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1423 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1424 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1425 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1426 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1427 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1428 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1429 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1430 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1431 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1432 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1433 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1434 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1435 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1436 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1437 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1438 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1439 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1440 1441 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1442 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1443 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1444 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1445 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1446 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1447 1448 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1449 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1450 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1451 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1452 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1453 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1454 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1455 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1456 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1457 /* 1458 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1459 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1460 * all other data structures' checksums. However, the METADATA_CSUM and 1461 * GDT_CSUM bits are mutually exclusive. 1462 */ 1463 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1464 1465 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1466 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1467 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1468 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1469 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1470 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1471 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1472 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1473 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1474 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1475 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1476 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1477 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1478 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1479 1480 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1481 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1482 EXT4_FEATURE_INCOMPAT_META_BG) 1483 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1484 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1485 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1486 1487 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1488 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1489 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1490 EXT4_FEATURE_INCOMPAT_META_BG) 1491 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1492 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1493 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1494 1495 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1496 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1497 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1498 EXT4_FEATURE_INCOMPAT_META_BG| \ 1499 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1500 EXT4_FEATURE_INCOMPAT_64BIT| \ 1501 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1502 EXT4_FEATURE_INCOMPAT_MMP | \ 1503 EXT4_FEATURE_INCOMPAT_INLINE_DATA) 1504 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1505 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1506 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1507 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1508 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1509 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1510 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1511 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1512 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1513 EXT4_FEATURE_RO_COMPAT_QUOTA) 1514 1515 /* 1516 * Default values for user and/or group using reserved blocks 1517 */ 1518 #define EXT4_DEF_RESUID 0 1519 #define EXT4_DEF_RESGID 0 1520 1521 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1522 1523 /* 1524 * Default mount options 1525 */ 1526 #define EXT4_DEFM_DEBUG 0x0001 1527 #define EXT4_DEFM_BSDGROUPS 0x0002 1528 #define EXT4_DEFM_XATTR_USER 0x0004 1529 #define EXT4_DEFM_ACL 0x0008 1530 #define EXT4_DEFM_UID16 0x0010 1531 #define EXT4_DEFM_JMODE 0x0060 1532 #define EXT4_DEFM_JMODE_DATA 0x0020 1533 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1534 #define EXT4_DEFM_JMODE_WBACK 0x0060 1535 #define EXT4_DEFM_NOBARRIER 0x0100 1536 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1537 #define EXT4_DEFM_DISCARD 0x0400 1538 #define EXT4_DEFM_NODELALLOC 0x0800 1539 1540 /* 1541 * Default journal batch times 1542 */ 1543 #define EXT4_DEF_MIN_BATCH_TIME 0 1544 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1545 1546 /* 1547 * Minimum number of groups in a flexgroup before we separate out 1548 * directories into the first block group of a flexgroup 1549 */ 1550 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1551 1552 /* 1553 * Structure of a directory entry 1554 */ 1555 #define EXT4_NAME_LEN 255 1556 1557 struct ext4_dir_entry { 1558 __le32 inode; /* Inode number */ 1559 __le16 rec_len; /* Directory entry length */ 1560 __le16 name_len; /* Name length */ 1561 char name[EXT4_NAME_LEN]; /* File name */ 1562 }; 1563 1564 /* 1565 * The new version of the directory entry. Since EXT4 structures are 1566 * stored in intel byte order, and the name_len field could never be 1567 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1568 * file_type field. 1569 */ 1570 struct ext4_dir_entry_2 { 1571 __le32 inode; /* Inode number */ 1572 __le16 rec_len; /* Directory entry length */ 1573 __u8 name_len; /* Name length */ 1574 __u8 file_type; 1575 char name[EXT4_NAME_LEN]; /* File name */ 1576 }; 1577 1578 /* 1579 * This is a bogus directory entry at the end of each leaf block that 1580 * records checksums. 1581 */ 1582 struct ext4_dir_entry_tail { 1583 __le32 det_reserved_zero1; /* Pretend to be unused */ 1584 __le16 det_rec_len; /* 12 */ 1585 __u8 det_reserved_zero2; /* Zero name length */ 1586 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1587 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1588 }; 1589 1590 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1591 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1592 ((blocksize) - \ 1593 sizeof(struct ext4_dir_entry_tail)))) 1594 1595 /* 1596 * Ext4 directory file types. Only the low 3 bits are used. The 1597 * other bits are reserved for now. 1598 */ 1599 #define EXT4_FT_UNKNOWN 0 1600 #define EXT4_FT_REG_FILE 1 1601 #define EXT4_FT_DIR 2 1602 #define EXT4_FT_CHRDEV 3 1603 #define EXT4_FT_BLKDEV 4 1604 #define EXT4_FT_FIFO 5 1605 #define EXT4_FT_SOCK 6 1606 #define EXT4_FT_SYMLINK 7 1607 1608 #define EXT4_FT_MAX 8 1609 1610 #define EXT4_FT_DIR_CSUM 0xDE 1611 1612 /* 1613 * EXT4_DIR_PAD defines the directory entries boundaries 1614 * 1615 * NOTE: It must be a multiple of 4 1616 */ 1617 #define EXT4_DIR_PAD 4 1618 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1619 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1620 ~EXT4_DIR_ROUND) 1621 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1622 1623 /* 1624 * If we ever get support for fs block sizes > page_size, we'll need 1625 * to remove the #if statements in the next two functions... 1626 */ 1627 static inline unsigned int 1628 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1629 { 1630 unsigned len = le16_to_cpu(dlen); 1631 1632 #if (PAGE_CACHE_SIZE >= 65536) 1633 if (len == EXT4_MAX_REC_LEN || len == 0) 1634 return blocksize; 1635 return (len & 65532) | ((len & 3) << 16); 1636 #else 1637 return len; 1638 #endif 1639 } 1640 1641 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1642 { 1643 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1644 BUG(); 1645 #if (PAGE_CACHE_SIZE >= 65536) 1646 if (len < 65536) 1647 return cpu_to_le16(len); 1648 if (len == blocksize) { 1649 if (blocksize == 65536) 1650 return cpu_to_le16(EXT4_MAX_REC_LEN); 1651 else 1652 return cpu_to_le16(0); 1653 } 1654 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1655 #else 1656 return cpu_to_le16(len); 1657 #endif 1658 } 1659 1660 /* 1661 * Hash Tree Directory indexing 1662 * (c) Daniel Phillips, 2001 1663 */ 1664 1665 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1666 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1667 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1668 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1669 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1670 1671 /* Legal values for the dx_root hash_version field: */ 1672 1673 #define DX_HASH_LEGACY 0 1674 #define DX_HASH_HALF_MD4 1 1675 #define DX_HASH_TEA 2 1676 #define DX_HASH_LEGACY_UNSIGNED 3 1677 #define DX_HASH_HALF_MD4_UNSIGNED 4 1678 #define DX_HASH_TEA_UNSIGNED 5 1679 1680 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1681 const void *address, unsigned int length) 1682 { 1683 struct { 1684 struct shash_desc shash; 1685 char ctx[4]; 1686 } desc; 1687 int err; 1688 1689 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1690 1691 desc.shash.tfm = sbi->s_chksum_driver; 1692 desc.shash.flags = 0; 1693 *(u32 *)desc.ctx = crc; 1694 1695 err = crypto_shash_update(&desc.shash, address, length); 1696 BUG_ON(err); 1697 1698 return *(u32 *)desc.ctx; 1699 } 1700 1701 #ifdef __KERNEL__ 1702 1703 /* hash info structure used by the directory hash */ 1704 struct dx_hash_info 1705 { 1706 u32 hash; 1707 u32 minor_hash; 1708 int hash_version; 1709 u32 *seed; 1710 }; 1711 1712 1713 /* 32 and 64 bit signed EOF for dx directories */ 1714 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1715 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1716 1717 1718 /* 1719 * Control parameters used by ext4_htree_next_block 1720 */ 1721 #define HASH_NB_ALWAYS 1 1722 1723 1724 /* 1725 * Describe an inode's exact location on disk and in memory 1726 */ 1727 struct ext4_iloc 1728 { 1729 struct buffer_head *bh; 1730 unsigned long offset; 1731 ext4_group_t block_group; 1732 }; 1733 1734 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1735 { 1736 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1737 } 1738 1739 /* 1740 * This structure is stuffed into the struct file's private_data field 1741 * for directories. It is where we put information so that we can do 1742 * readdir operations in hash tree order. 1743 */ 1744 struct dir_private_info { 1745 struct rb_root root; 1746 struct rb_node *curr_node; 1747 struct fname *extra_fname; 1748 loff_t last_pos; 1749 __u32 curr_hash; 1750 __u32 curr_minor_hash; 1751 __u32 next_hash; 1752 }; 1753 1754 /* calculate the first block number of the group */ 1755 static inline ext4_fsblk_t 1756 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1757 { 1758 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1759 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1760 } 1761 1762 /* 1763 * Special error return code only used by dx_probe() and its callers. 1764 */ 1765 #define ERR_BAD_DX_DIR -75000 1766 1767 /* 1768 * Timeout and state flag for lazy initialization inode thread. 1769 */ 1770 #define EXT4_DEF_LI_WAIT_MULT 10 1771 #define EXT4_DEF_LI_MAX_START_DELAY 5 1772 #define EXT4_LAZYINIT_QUIT 0x0001 1773 #define EXT4_LAZYINIT_RUNNING 0x0002 1774 1775 /* 1776 * Lazy inode table initialization info 1777 */ 1778 struct ext4_lazy_init { 1779 unsigned long li_state; 1780 struct list_head li_request_list; 1781 struct mutex li_list_mtx; 1782 }; 1783 1784 struct ext4_li_request { 1785 struct super_block *lr_super; 1786 struct ext4_sb_info *lr_sbi; 1787 ext4_group_t lr_next_group; 1788 struct list_head lr_request; 1789 unsigned long lr_next_sched; 1790 unsigned long lr_timeout; 1791 }; 1792 1793 struct ext4_features { 1794 struct kobject f_kobj; 1795 struct completion f_kobj_unregister; 1796 }; 1797 1798 /* 1799 * This structure will be used for multiple mount protection. It will be 1800 * written into the block number saved in the s_mmp_block field in the 1801 * superblock. Programs that check MMP should assume that if 1802 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1803 * to use the filesystem, regardless of how old the timestamp is. 1804 */ 1805 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1806 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1807 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1808 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1809 1810 struct mmp_struct { 1811 __le32 mmp_magic; /* Magic number for MMP */ 1812 __le32 mmp_seq; /* Sequence no. updated periodically */ 1813 1814 /* 1815 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1816 * purposes and do not affect the correctness of the algorithm 1817 */ 1818 __le64 mmp_time; /* Time last updated */ 1819 char mmp_nodename[64]; /* Node which last updated MMP block */ 1820 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1821 1822 /* 1823 * mmp_check_interval is used to verify if the MMP block has been 1824 * updated on the block device. The value is updated based on the 1825 * maximum time to write the MMP block during an update cycle. 1826 */ 1827 __le16 mmp_check_interval; 1828 1829 __le16 mmp_pad1; 1830 __le32 mmp_pad2[226]; 1831 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1832 }; 1833 1834 /* arguments passed to the mmp thread */ 1835 struct mmpd_data { 1836 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1837 struct super_block *sb; /* super block of the fs */ 1838 }; 1839 1840 /* 1841 * Check interval multiplier 1842 * The MMP block is written every update interval and initially checked every 1843 * update interval x the multiplier (the value is then adapted based on the 1844 * write latency). The reason is that writes can be delayed under load and we 1845 * don't want readers to incorrectly assume that the filesystem is no longer 1846 * in use. 1847 */ 1848 #define EXT4_MMP_CHECK_MULT 2UL 1849 1850 /* 1851 * Minimum interval for MMP checking in seconds. 1852 */ 1853 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1854 1855 /* 1856 * Maximum interval for MMP checking in seconds. 1857 */ 1858 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1859 1860 /* 1861 * Function prototypes 1862 */ 1863 1864 /* 1865 * Ok, these declarations are also in <linux/kernel.h> but none of the 1866 * ext4 source programs needs to include it so they are duplicated here. 1867 */ 1868 # define NORET_TYPE /**/ 1869 # define ATTRIB_NORET __attribute__((noreturn)) 1870 # define NORET_AND noreturn, 1871 1872 /* bitmap.c */ 1873 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1874 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1875 struct ext4_group_desc *gdp, 1876 struct buffer_head *bh, int sz); 1877 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1878 struct ext4_group_desc *gdp, 1879 struct buffer_head *bh, int sz); 1880 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1881 struct ext4_group_desc *gdp, 1882 struct buffer_head *bh); 1883 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1884 struct ext4_group_desc *gdp, 1885 struct buffer_head *bh); 1886 1887 /* balloc.c */ 1888 extern void ext4_get_group_no_and_offset(struct super_block *sb, 1889 ext4_fsblk_t blocknr, 1890 ext4_group_t *blockgrpp, 1891 ext4_grpblk_t *offsetp); 1892 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 1893 ext4_fsblk_t block); 1894 1895 extern void ext4_validate_block_bitmap(struct super_block *sb, 1896 struct ext4_group_desc *desc, 1897 unsigned int block_group, 1898 struct buffer_head *bh); 1899 extern unsigned int ext4_block_group(struct super_block *sb, 1900 ext4_fsblk_t blocknr); 1901 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1902 ext4_fsblk_t blocknr); 1903 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1904 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1905 ext4_group_t group); 1906 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1907 ext4_fsblk_t goal, 1908 unsigned int flags, 1909 unsigned long *count, 1910 int *errp); 1911 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1912 s64 nclusters, unsigned int flags); 1913 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1914 extern void ext4_check_blocks_bitmap(struct super_block *); 1915 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1916 ext4_group_t block_group, 1917 struct buffer_head ** bh); 1918 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1919 1920 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1921 ext4_group_t block_group); 1922 extern int ext4_wait_block_bitmap(struct super_block *sb, 1923 ext4_group_t block_group, 1924 struct buffer_head *bh); 1925 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1926 ext4_group_t block_group); 1927 extern void ext4_init_block_bitmap(struct super_block *sb, 1928 struct buffer_head *bh, 1929 ext4_group_t group, 1930 struct ext4_group_desc *desc); 1931 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1932 ext4_group_t block_group, 1933 struct ext4_group_desc *gdp); 1934 extern unsigned ext4_num_overhead_clusters(struct super_block *sb, 1935 ext4_group_t block_group, 1936 struct ext4_group_desc *gdp); 1937 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1938 1939 /* dir.c */ 1940 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1941 struct file *, 1942 struct ext4_dir_entry_2 *, 1943 struct buffer_head *, char *, int, 1944 unsigned int); 1945 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 1946 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 1947 (de), (bh), (buf), (size), (offset))) 1948 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1949 __u32 minor_hash, 1950 struct ext4_dir_entry_2 *dirent); 1951 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1952 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 1953 struct buffer_head *bh, 1954 void *buf, int buf_size, 1955 const char *name, int namelen, 1956 struct ext4_dir_entry_2 **dest_de); 1957 void ext4_insert_dentry(struct inode *inode, 1958 struct ext4_dir_entry_2 *de, 1959 int buf_size, 1960 const char *name, int namelen); 1961 static inline void ext4_update_dx_flag(struct inode *inode) 1962 { 1963 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb, 1964 EXT4_FEATURE_COMPAT_DIR_INDEX)) 1965 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 1966 } 1967 static unsigned char ext4_filetype_table[] = { 1968 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 1969 }; 1970 1971 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 1972 { 1973 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) || 1974 (filetype >= EXT4_FT_MAX)) 1975 return DT_UNKNOWN; 1976 1977 return ext4_filetype_table[filetype]; 1978 } 1979 1980 /* fsync.c */ 1981 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 1982 extern int ext4_flush_unwritten_io(struct inode *); 1983 1984 /* hash.c */ 1985 extern int ext4fs_dirhash(const char *name, int len, struct 1986 dx_hash_info *hinfo); 1987 1988 /* ialloc.c */ 1989 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 1990 const struct qstr *qstr, __u32 goal, 1991 uid_t *owner, int handle_type, 1992 unsigned int line_no, int nblocks); 1993 1994 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 1995 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 1996 0, 0, 0) 1997 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 1998 type, nblocks) \ 1999 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2000 (type), __LINE__, (nblocks)) 2001 2002 2003 extern void ext4_free_inode(handle_t *, struct inode *); 2004 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2005 extern unsigned long ext4_count_free_inodes(struct super_block *); 2006 extern unsigned long ext4_count_dirs(struct super_block *); 2007 extern void ext4_check_inodes_bitmap(struct super_block *); 2008 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2009 extern int ext4_init_inode_table(struct super_block *sb, 2010 ext4_group_t group, int barrier); 2011 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2012 2013 /* mballoc.c */ 2014 extern long ext4_mb_stats; 2015 extern long ext4_mb_max_to_scan; 2016 extern int ext4_mb_init(struct super_block *); 2017 extern int ext4_mb_release(struct super_block *); 2018 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2019 struct ext4_allocation_request *, int *); 2020 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2021 extern void ext4_discard_preallocations(struct inode *); 2022 extern int __init ext4_init_mballoc(void); 2023 extern void ext4_exit_mballoc(void); 2024 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2025 struct buffer_head *bh, ext4_fsblk_t block, 2026 unsigned long count, int flags); 2027 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2028 ext4_group_t ngroups); 2029 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2030 ext4_group_t i, struct ext4_group_desc *desc); 2031 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2032 ext4_fsblk_t block, unsigned long count); 2033 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2034 2035 /* inode.c */ 2036 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 2037 ext4_lblk_t, int, int *); 2038 struct buffer_head *ext4_bread(handle_t *, struct inode *, 2039 ext4_lblk_t, int, int *); 2040 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2041 struct buffer_head *bh_result, int create); 2042 int ext4_get_block(struct inode *inode, sector_t iblock, 2043 struct buffer_head *bh_result, int create); 2044 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2045 struct buffer_head *bh, int create); 2046 int ext4_walk_page_buffers(handle_t *handle, 2047 struct buffer_head *head, 2048 unsigned from, 2049 unsigned to, 2050 int *partial, 2051 int (*fn)(handle_t *handle, 2052 struct buffer_head *bh)); 2053 int do_journal_get_write_access(handle_t *handle, 2054 struct buffer_head *bh); 2055 #define FALL_BACK_TO_NONDELALLOC 1 2056 #define CONVERT_INLINE_DATA 2 2057 2058 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2059 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2060 extern int ext4_setattr(struct dentry *, struct iattr *); 2061 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2062 struct kstat *stat); 2063 extern void ext4_evict_inode(struct inode *); 2064 extern void ext4_clear_inode(struct inode *); 2065 extern int ext4_sync_inode(handle_t *, struct inode *); 2066 extern void ext4_dirty_inode(struct inode *, int); 2067 extern int ext4_change_inode_journal_flag(struct inode *, int); 2068 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2069 extern int ext4_can_truncate(struct inode *inode); 2070 extern void ext4_truncate(struct inode *); 2071 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2072 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2073 extern void ext4_set_inode_flags(struct inode *); 2074 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2075 extern int ext4_alloc_da_blocks(struct inode *inode); 2076 extern void ext4_set_aops(struct inode *inode); 2077 extern int ext4_writepage_trans_blocks(struct inode *); 2078 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2079 extern int ext4_block_truncate_page(handle_t *handle, 2080 struct address_space *mapping, loff_t from); 2081 extern int ext4_block_zero_page_range(handle_t *handle, 2082 struct address_space *mapping, loff_t from, loff_t length); 2083 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2084 loff_t lstart, loff_t lend); 2085 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2086 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2087 extern void ext4_da_update_reserve_space(struct inode *inode, 2088 int used, int quota_claim); 2089 2090 /* indirect.c */ 2091 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2092 struct ext4_map_blocks *map, int flags); 2093 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2094 const struct iovec *iov, loff_t offset, 2095 unsigned long nr_segs); 2096 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2097 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2098 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2099 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode, 2100 ext4_lblk_t first, ext4_lblk_t stop); 2101 2102 /* ioctl.c */ 2103 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2104 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2105 2106 /* migrate.c */ 2107 extern int ext4_ext_migrate(struct inode *); 2108 extern int ext4_ind_migrate(struct inode *inode); 2109 2110 /* namei.c */ 2111 extern int ext4_dirent_csum_verify(struct inode *inode, 2112 struct ext4_dir_entry *dirent); 2113 extern int ext4_orphan_add(handle_t *, struct inode *); 2114 extern int ext4_orphan_del(handle_t *, struct inode *); 2115 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2116 __u32 start_minor_hash, __u32 *next_hash); 2117 extern int search_dir(struct buffer_head *bh, 2118 char *search_buf, 2119 int buf_size, 2120 struct inode *dir, 2121 const struct qstr *d_name, 2122 unsigned int offset, 2123 struct ext4_dir_entry_2 **res_dir); 2124 extern int ext4_generic_delete_entry(handle_t *handle, 2125 struct inode *dir, 2126 struct ext4_dir_entry_2 *de_del, 2127 struct buffer_head *bh, 2128 void *entry_buf, 2129 int buf_size, 2130 int csum_size); 2131 2132 /* resize.c */ 2133 extern int ext4_group_add(struct super_block *sb, 2134 struct ext4_new_group_data *input); 2135 extern int ext4_group_extend(struct super_block *sb, 2136 struct ext4_super_block *es, 2137 ext4_fsblk_t n_blocks_count); 2138 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2139 2140 /* super.c */ 2141 extern int ext4_calculate_overhead(struct super_block *sb); 2142 extern int ext4_superblock_csum_verify(struct super_block *sb, 2143 struct ext4_super_block *es); 2144 extern void ext4_superblock_csum_set(struct super_block *sb); 2145 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2146 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2147 extern void ext4_kvfree(void *ptr); 2148 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2149 ext4_group_t ngroup); 2150 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2151 char nbuf[16]); 2152 extern __printf(4, 5) 2153 void __ext4_error(struct super_block *, const char *, unsigned int, 2154 const char *, ...); 2155 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \ 2156 __LINE__, ## message) 2157 extern __printf(5, 6) 2158 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2159 const char *, ...); 2160 extern __printf(5, 6) 2161 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2162 const char *, ...); 2163 extern void __ext4_std_error(struct super_block *, const char *, 2164 unsigned int, int); 2165 extern __printf(4, 5) 2166 void __ext4_abort(struct super_block *, const char *, unsigned int, 2167 const char *, ...); 2168 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \ 2169 __LINE__, ## message) 2170 extern __printf(4, 5) 2171 void __ext4_warning(struct super_block *, const char *, unsigned int, 2172 const char *, ...); 2173 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \ 2174 __LINE__, ## message) 2175 extern __printf(3, 4) 2176 void ext4_msg(struct super_block *, const char *, const char *, ...); 2177 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2178 const char *, unsigned int, const char *); 2179 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \ 2180 __LINE__, msg) 2181 extern __printf(7, 8) 2182 void __ext4_grp_locked_error(const char *, unsigned int, 2183 struct super_block *, ext4_group_t, 2184 unsigned long, ext4_fsblk_t, 2185 const char *, ...); 2186 #define ext4_grp_locked_error(sb, grp, message...) \ 2187 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message) 2188 extern void ext4_update_dynamic_rev(struct super_block *sb); 2189 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2190 __u32 compat); 2191 extern int ext4_update_rocompat_feature(handle_t *handle, 2192 struct super_block *sb, __u32 rocompat); 2193 extern int ext4_update_incompat_feature(handle_t *handle, 2194 struct super_block *sb, __u32 incompat); 2195 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2196 struct ext4_group_desc *bg); 2197 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2198 struct ext4_group_desc *bg); 2199 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2200 struct ext4_group_desc *bg); 2201 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2202 struct ext4_group_desc *bg); 2203 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2204 struct ext4_group_desc *bg); 2205 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2206 struct ext4_group_desc *bg); 2207 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2208 struct ext4_group_desc *bg); 2209 extern void ext4_block_bitmap_set(struct super_block *sb, 2210 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2211 extern void ext4_inode_bitmap_set(struct super_block *sb, 2212 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2213 extern void ext4_inode_table_set(struct super_block *sb, 2214 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2215 extern void ext4_free_group_clusters_set(struct super_block *sb, 2216 struct ext4_group_desc *bg, 2217 __u32 count); 2218 extern void ext4_free_inodes_set(struct super_block *sb, 2219 struct ext4_group_desc *bg, __u32 count); 2220 extern void ext4_used_dirs_set(struct super_block *sb, 2221 struct ext4_group_desc *bg, __u32 count); 2222 extern void ext4_itable_unused_set(struct super_block *sb, 2223 struct ext4_group_desc *bg, __u32 count); 2224 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2225 struct ext4_group_desc *gdp); 2226 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2227 struct ext4_group_desc *gdp); 2228 extern int ext4_register_li_request(struct super_block *sb, 2229 ext4_group_t first_not_zeroed); 2230 2231 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2232 { 2233 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2234 EXT4_FEATURE_RO_COMPAT_GDT_CSUM | 2235 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM); 2236 } 2237 2238 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2239 { 2240 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2241 le32_to_cpu(es->s_blocks_count_lo); 2242 } 2243 2244 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2245 { 2246 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2247 le32_to_cpu(es->s_r_blocks_count_lo); 2248 } 2249 2250 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2251 { 2252 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2253 le32_to_cpu(es->s_free_blocks_count_lo); 2254 } 2255 2256 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2257 ext4_fsblk_t blk) 2258 { 2259 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2260 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2261 } 2262 2263 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2264 ext4_fsblk_t blk) 2265 { 2266 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2267 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2268 } 2269 2270 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2271 ext4_fsblk_t blk) 2272 { 2273 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2274 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2275 } 2276 2277 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2278 { 2279 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2280 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2281 le32_to_cpu(raw_inode->i_size_lo); 2282 else 2283 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2284 } 2285 2286 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2287 { 2288 raw_inode->i_size_lo = cpu_to_le32(i_size); 2289 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2290 } 2291 2292 static inline 2293 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2294 ext4_group_t group) 2295 { 2296 struct ext4_group_info ***grp_info; 2297 long indexv, indexh; 2298 grp_info = EXT4_SB(sb)->s_group_info; 2299 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2300 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2301 return grp_info[indexv][indexh]; 2302 } 2303 2304 /* 2305 * Reading s_groups_count requires using smp_rmb() afterwards. See 2306 * the locking protocol documented in the comments of ext4_group_add() 2307 * in resize.c 2308 */ 2309 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2310 { 2311 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2312 2313 smp_rmb(); 2314 return ngroups; 2315 } 2316 2317 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2318 ext4_group_t block_group) 2319 { 2320 return block_group >> sbi->s_log_groups_per_flex; 2321 } 2322 2323 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2324 { 2325 return 1 << sbi->s_log_groups_per_flex; 2326 } 2327 2328 #define ext4_std_error(sb, errno) \ 2329 do { \ 2330 if ((errno)) \ 2331 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2332 } while (0) 2333 2334 #ifdef CONFIG_SMP 2335 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2336 * counters. So we need to make sure we have free clusters more 2337 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2338 */ 2339 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2340 #else 2341 #define EXT4_FREECLUSTERS_WATERMARK 0 2342 #endif 2343 2344 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2345 { 2346 /* 2347 * XXX: replace with spinlock if seen contended -bzzz 2348 */ 2349 down_write(&EXT4_I(inode)->i_data_sem); 2350 if (newsize > EXT4_I(inode)->i_disksize) 2351 EXT4_I(inode)->i_disksize = newsize; 2352 up_write(&EXT4_I(inode)->i_data_sem); 2353 return ; 2354 } 2355 2356 struct ext4_group_info { 2357 unsigned long bb_state; 2358 struct rb_root bb_free_root; 2359 ext4_grpblk_t bb_first_free; /* first free block */ 2360 ext4_grpblk_t bb_free; /* total free blocks */ 2361 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2362 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2363 struct list_head bb_prealloc_list; 2364 #ifdef DOUBLE_CHECK 2365 void *bb_bitmap; 2366 #endif 2367 struct rw_semaphore alloc_sem; 2368 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2369 * regions, index is order. 2370 * bb_counters[3] = 5 means 2371 * 5 free 8-block regions. */ 2372 }; 2373 2374 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2375 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2376 2377 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2378 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2379 2380 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2381 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2382 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2383 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2384 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2385 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2386 2387 #define EXT4_MAX_CONTENTION 8 2388 #define EXT4_CONTENTION_THRESHOLD 2 2389 2390 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2391 ext4_group_t group) 2392 { 2393 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2394 } 2395 2396 /* 2397 * Returns true if the filesystem is busy enough that attempts to 2398 * access the block group locks has run into contention. 2399 */ 2400 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2401 { 2402 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2403 } 2404 2405 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2406 { 2407 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2408 if (spin_trylock(lock)) 2409 /* 2410 * We're able to grab the lock right away, so drop the 2411 * lock contention counter. 2412 */ 2413 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2414 else { 2415 /* 2416 * The lock is busy, so bump the contention counter, 2417 * and then wait on the spin lock. 2418 */ 2419 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2420 EXT4_MAX_CONTENTION); 2421 spin_lock(lock); 2422 } 2423 } 2424 2425 static inline void ext4_unlock_group(struct super_block *sb, 2426 ext4_group_t group) 2427 { 2428 spin_unlock(ext4_group_lock_ptr(sb, group)); 2429 } 2430 2431 /* 2432 * Block validity checking 2433 */ 2434 #define ext4_check_indirect_blockref(inode, bh) \ 2435 ext4_check_blockref(__func__, __LINE__, inode, \ 2436 (__le32 *)(bh)->b_data, \ 2437 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2438 2439 #define ext4_ind_check_inode(inode) \ 2440 ext4_check_blockref(__func__, __LINE__, inode, \ 2441 EXT4_I(inode)->i_data, \ 2442 EXT4_NDIR_BLOCKS) 2443 2444 /* 2445 * Inodes and files operations 2446 */ 2447 2448 /* dir.c */ 2449 extern const struct file_operations ext4_dir_operations; 2450 2451 /* file.c */ 2452 extern const struct inode_operations ext4_file_inode_operations; 2453 extern const struct file_operations ext4_file_operations; 2454 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2455 extern void ext4_unwritten_wait(struct inode *inode); 2456 2457 /* inline.c */ 2458 extern int ext4_has_inline_data(struct inode *inode); 2459 extern int ext4_get_inline_size(struct inode *inode); 2460 extern int ext4_get_max_inline_size(struct inode *inode); 2461 extern int ext4_find_inline_data_nolock(struct inode *inode); 2462 extern void ext4_write_inline_data(struct inode *inode, 2463 struct ext4_iloc *iloc, 2464 void *buffer, loff_t pos, 2465 unsigned int len); 2466 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode, 2467 unsigned int len); 2468 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2469 unsigned int len); 2470 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2471 2472 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2473 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2474 struct inode *inode, 2475 loff_t pos, unsigned len, 2476 unsigned flags, 2477 struct page **pagep); 2478 extern int ext4_write_inline_data_end(struct inode *inode, 2479 loff_t pos, unsigned len, 2480 unsigned copied, 2481 struct page *page); 2482 extern struct buffer_head * 2483 ext4_journalled_write_inline_data(struct inode *inode, 2484 unsigned len, 2485 struct page *page); 2486 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2487 struct inode *inode, 2488 loff_t pos, unsigned len, 2489 unsigned flags, 2490 struct page **pagep, 2491 void **fsdata); 2492 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2493 unsigned len, unsigned copied, 2494 struct page *page); 2495 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry, 2496 struct inode *inode); 2497 extern int ext4_try_create_inline_dir(handle_t *handle, 2498 struct inode *parent, 2499 struct inode *inode); 2500 extern int ext4_read_inline_dir(struct file *filp, 2501 void *dirent, filldir_t filldir, 2502 int *has_inline_data); 2503 extern int htree_inlinedir_to_tree(struct file *dir_file, 2504 struct inode *dir, ext4_lblk_t block, 2505 struct dx_hash_info *hinfo, 2506 __u32 start_hash, __u32 start_minor_hash, 2507 int *has_inline_data); 2508 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2509 const struct qstr *d_name, 2510 struct ext4_dir_entry_2 **res_dir, 2511 int *has_inline_data); 2512 extern int ext4_delete_inline_entry(handle_t *handle, 2513 struct inode *dir, 2514 struct ext4_dir_entry_2 *de_del, 2515 struct buffer_head *bh, 2516 int *has_inline_data); 2517 extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 2518 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 2519 struct ext4_dir_entry_2 **parent_de, 2520 int *retval); 2521 extern int ext4_inline_data_fiemap(struct inode *inode, 2522 struct fiemap_extent_info *fieinfo, 2523 int *has_inline); 2524 extern int ext4_try_to_evict_inline_data(handle_t *handle, 2525 struct inode *inode, 2526 int needed); 2527 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 2528 2529 extern int ext4_convert_inline_data(struct inode *inode); 2530 2531 /* namei.c */ 2532 extern const struct inode_operations ext4_dir_inode_operations; 2533 extern const struct inode_operations ext4_special_inode_operations; 2534 extern struct dentry *ext4_get_parent(struct dentry *child); 2535 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 2536 struct ext4_dir_entry_2 *de, 2537 int blocksize, int csum_size, 2538 unsigned int parent_ino, int dotdot_real_len); 2539 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 2540 unsigned int blocksize); 2541 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 2542 struct inode *inode, 2543 struct buffer_head *bh); 2544 #define S_SHIFT 12 2545 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 2546 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 2547 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 2548 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 2549 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 2550 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 2551 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 2552 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 2553 }; 2554 2555 static inline void ext4_set_de_type(struct super_block *sb, 2556 struct ext4_dir_entry_2 *de, 2557 umode_t mode) { 2558 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE)) 2559 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 2560 } 2561 2562 2563 /* symlink.c */ 2564 extern const struct inode_operations ext4_symlink_inode_operations; 2565 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2566 2567 /* block_validity */ 2568 extern void ext4_release_system_zone(struct super_block *sb); 2569 extern int ext4_setup_system_zone(struct super_block *sb); 2570 extern int __init ext4_init_system_zone(void); 2571 extern void ext4_exit_system_zone(void); 2572 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2573 ext4_fsblk_t start_blk, 2574 unsigned int count); 2575 extern int ext4_check_blockref(const char *, unsigned int, 2576 struct inode *, __le32 *, unsigned int); 2577 2578 /* extents.c */ 2579 struct ext4_ext_path; 2580 struct ext4_extent; 2581 2582 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2583 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2584 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 2585 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2586 struct ext4_map_blocks *map, int flags); 2587 extern void ext4_ext_truncate(handle_t *, struct inode *); 2588 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 2589 ext4_lblk_t end); 2590 extern void ext4_ext_init(struct super_block *); 2591 extern void ext4_ext_release(struct super_block *); 2592 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2593 loff_t len); 2594 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 2595 ssize_t len); 2596 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2597 struct ext4_map_blocks *map, int flags); 2598 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 2599 ext4_lblk_t lblocks); 2600 extern int ext4_extent_tree_init(handle_t *, struct inode *); 2601 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 2602 int num, 2603 struct ext4_ext_path *path); 2604 extern int ext4_can_extents_be_merged(struct inode *inode, 2605 struct ext4_extent *ex1, 2606 struct ext4_extent *ex2); 2607 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 2608 struct ext4_ext_path *, 2609 struct ext4_extent *, int); 2610 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t, 2611 struct ext4_ext_path *); 2612 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 2613 extern int ext4_ext_check_inode(struct inode *inode); 2614 extern int ext4_find_delalloc_range(struct inode *inode, 2615 ext4_lblk_t lblk_start, 2616 ext4_lblk_t lblk_end); 2617 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 2618 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2619 __u64 start, __u64 len); 2620 2621 2622 /* move_extent.c */ 2623 extern void ext4_double_down_write_data_sem(struct inode *first, 2624 struct inode *second); 2625 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 2626 struct inode *donor_inode); 2627 void ext4_inode_double_lock(struct inode *inode1, struct inode *inode2); 2628 void ext4_inode_double_unlock(struct inode *inode1, struct inode *inode2); 2629 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2630 __u64 start_orig, __u64 start_donor, 2631 __u64 len, __u64 *moved_len); 2632 2633 /* page-io.c */ 2634 extern int __init ext4_init_pageio(void); 2635 extern void ext4_exit_pageio(void); 2636 extern void ext4_ioend_shutdown(struct inode *); 2637 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2638 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 2639 extern int ext4_put_io_end(ext4_io_end_t *io_end); 2640 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 2641 extern void ext4_io_submit_init(struct ext4_io_submit *io, 2642 struct writeback_control *wbc); 2643 extern void ext4_end_io_work(struct work_struct *work); 2644 extern void ext4_io_submit(struct ext4_io_submit *io); 2645 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2646 struct page *page, 2647 int len, 2648 struct writeback_control *wbc); 2649 2650 /* mmp.c */ 2651 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2652 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp); 2653 extern int ext4_mmp_csum_verify(struct super_block *sb, 2654 struct mmp_struct *mmp); 2655 2656 /* 2657 * Note that these flags will never ever appear in a buffer_head's state flag. 2658 * See EXT4_MAP_... to see where this is used. 2659 */ 2660 enum ext4_state_bits { 2661 BH_Uninit /* blocks are allocated but uninitialized on disk */ 2662 = BH_JBDPrivateStart, 2663 BH_AllocFromCluster, /* allocated blocks were part of already 2664 * allocated cluster. */ 2665 }; 2666 2667 /* 2668 * Add new method to test whether block and inode bitmaps are properly 2669 * initialized. With uninit_bg reading the block from disk is not enough 2670 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2671 */ 2672 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2673 2674 static inline int bitmap_uptodate(struct buffer_head *bh) 2675 { 2676 return (buffer_uptodate(bh) && 2677 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2678 } 2679 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2680 { 2681 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2682 } 2683 2684 /* 2685 * Disable DIO read nolock optimization, so new dioreaders will be forced 2686 * to grab i_mutex 2687 */ 2688 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2689 { 2690 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2691 smp_mb(); 2692 } 2693 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2694 { 2695 smp_mb(); 2696 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2697 } 2698 2699 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2700 2701 /* For ioend & aio unwritten conversion wait queues */ 2702 #define EXT4_WQ_HASH_SZ 37 2703 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2704 EXT4_WQ_HASH_SZ]) 2705 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2706 EXT4_WQ_HASH_SZ]) 2707 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2708 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2709 2710 #define EXT4_RESIZING 0 2711 extern int ext4_resize_begin(struct super_block *sb); 2712 extern void ext4_resize_end(struct super_block *sb); 2713 2714 #endif /* __KERNEL__ */ 2715 2716 #endif /* _EXT4_H */ 2717