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 * For delayed allocation tracking 181 */ 182 struct mpage_da_data { 183 struct inode *inode; 184 sector_t b_blocknr; /* start block number of extent */ 185 size_t b_size; /* size of extent */ 186 unsigned long b_state; /* state of the extent */ 187 unsigned long first_page, next_page; /* extent of pages */ 188 struct writeback_control *wbc; 189 int io_done; 190 int pages_written; 191 int retval; 192 }; 193 194 /* 195 * Flags for ext4_io_end->flags 196 */ 197 #define EXT4_IO_END_UNWRITTEN 0x0001 198 #define EXT4_IO_END_ERROR 0x0002 199 #define EXT4_IO_END_DIRECT 0x0004 200 201 /* 202 * For converting uninitialized extents on a work queue. 203 */ 204 typedef struct ext4_io_end { 205 struct list_head list; /* per-file finished IO list */ 206 struct inode *inode; /* file being written to */ 207 unsigned int flag; /* unwritten or not */ 208 loff_t offset; /* offset in the file */ 209 ssize_t size; /* size of the extent */ 210 struct kiocb *iocb; /* iocb struct for AIO */ 211 int result; /* error value for AIO */ 212 atomic_t count; /* reference counter */ 213 } ext4_io_end_t; 214 215 struct ext4_io_submit { 216 int io_op; 217 struct bio *io_bio; 218 ext4_io_end_t *io_end; 219 sector_t io_next_block; 220 }; 221 222 /* 223 * Special inodes numbers 224 */ 225 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 226 #define EXT4_ROOT_INO 2 /* Root inode */ 227 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 228 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 229 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 230 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 231 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 232 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 233 234 /* First non-reserved inode for old ext4 filesystems */ 235 #define EXT4_GOOD_OLD_FIRST_INO 11 236 237 /* 238 * Maximal count of links to a file 239 */ 240 #define EXT4_LINK_MAX 65000 241 242 /* 243 * Macro-instructions used to manage several block sizes 244 */ 245 #define EXT4_MIN_BLOCK_SIZE 1024 246 #define EXT4_MAX_BLOCK_SIZE 65536 247 #define EXT4_MIN_BLOCK_LOG_SIZE 10 248 #define EXT4_MAX_BLOCK_LOG_SIZE 16 249 #ifdef __KERNEL__ 250 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 251 #else 252 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 253 #endif 254 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 255 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 256 EXT4_SB(s)->s_cluster_bits) 257 #ifdef __KERNEL__ 258 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 259 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 260 #else 261 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 262 #endif 263 #ifdef __KERNEL__ 264 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 265 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 266 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 267 #else 268 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 269 EXT4_GOOD_OLD_INODE_SIZE : \ 270 (s)->s_inode_size) 271 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 272 EXT4_GOOD_OLD_FIRST_INO : \ 273 (s)->s_first_ino) 274 #endif 275 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 276 277 /* Translate a block number to a cluster number */ 278 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 279 /* Translate a cluster number to a block number */ 280 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 281 /* Translate # of blks to # of clusters */ 282 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 283 (sbi)->s_cluster_bits) 284 285 /* 286 * Structure of a blocks group descriptor 287 */ 288 struct ext4_group_desc 289 { 290 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 291 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 292 __le32 bg_inode_table_lo; /* Inodes table block */ 293 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 294 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 295 __le16 bg_used_dirs_count_lo; /* Directories count */ 296 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 297 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 298 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 299 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 300 __le16 bg_itable_unused_lo; /* Unused inodes count */ 301 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 302 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 303 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 304 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 305 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 306 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 307 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 308 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 309 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 310 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 311 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 312 __u32 bg_reserved; 313 }; 314 315 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 316 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 317 sizeof(__le16)) 318 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 319 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 320 sizeof(__le16)) 321 322 /* 323 * Structure of a flex block group info 324 */ 325 326 struct flex_groups { 327 atomic64_t free_clusters; 328 atomic_t free_inodes; 329 atomic_t used_dirs; 330 }; 331 332 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 333 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 334 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 335 336 /* 337 * Macro-instructions used to manage group descriptors 338 */ 339 #define EXT4_MIN_DESC_SIZE 32 340 #define EXT4_MIN_DESC_SIZE_64BIT 64 341 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 342 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 343 #ifdef __KERNEL__ 344 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 345 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 346 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 347 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 348 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 349 #else 350 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 351 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 352 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 353 #endif 354 355 /* 356 * Constants relative to the data blocks 357 */ 358 #define EXT4_NDIR_BLOCKS 12 359 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 360 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 361 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 362 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 363 364 /* 365 * Inode flags 366 */ 367 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 368 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 369 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 370 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 371 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 372 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 373 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 374 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 375 /* Reserved for compression usage... */ 376 #define EXT4_DIRTY_FL 0x00000100 377 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 378 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 379 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 380 /* End compression flags --- maybe not all used */ 381 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 382 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 383 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 384 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 385 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 386 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 387 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 388 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 389 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 390 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 391 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 392 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 393 394 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 395 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */ 396 397 /* Flags that should be inherited by new inodes from their parent. */ 398 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 399 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 400 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 401 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 402 403 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 404 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 405 406 /* Flags that are appropriate for non-directories/regular files. */ 407 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 408 409 /* Mask out flags that are inappropriate for the given type of inode. */ 410 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 411 { 412 if (S_ISDIR(mode)) 413 return flags; 414 else if (S_ISREG(mode)) 415 return flags & EXT4_REG_FLMASK; 416 else 417 return flags & EXT4_OTHER_FLMASK; 418 } 419 420 /* 421 * Inode flags used for atomic set/get 422 */ 423 enum { 424 EXT4_INODE_SECRM = 0, /* Secure deletion */ 425 EXT4_INODE_UNRM = 1, /* Undelete */ 426 EXT4_INODE_COMPR = 2, /* Compress file */ 427 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 428 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 429 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 430 EXT4_INODE_NODUMP = 6, /* do not dump file */ 431 EXT4_INODE_NOATIME = 7, /* do not update atime */ 432 /* Reserved for compression usage... */ 433 EXT4_INODE_DIRTY = 8, 434 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 435 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 436 EXT4_INODE_ECOMPR = 11, /* Compression error */ 437 /* End compression flags --- maybe not all used */ 438 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 439 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 440 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 441 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 442 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 443 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 444 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 445 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 446 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 447 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 448 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 449 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 450 }; 451 452 /* 453 * Since it's pretty easy to mix up bit numbers and hex values, we use a 454 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 455 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 456 * any extra space in the compiled kernel image, otherwise, the build will fail. 457 * It's important that these values are the same, since we are using 458 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 459 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 460 * values found in ext2, ext3 and ext4 filesystems, and of course the values 461 * defined in e2fsprogs. 462 * 463 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 464 */ 465 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 466 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 467 468 static inline void ext4_check_flag_values(void) 469 { 470 CHECK_FLAG_VALUE(SECRM); 471 CHECK_FLAG_VALUE(UNRM); 472 CHECK_FLAG_VALUE(COMPR); 473 CHECK_FLAG_VALUE(SYNC); 474 CHECK_FLAG_VALUE(IMMUTABLE); 475 CHECK_FLAG_VALUE(APPEND); 476 CHECK_FLAG_VALUE(NODUMP); 477 CHECK_FLAG_VALUE(NOATIME); 478 CHECK_FLAG_VALUE(DIRTY); 479 CHECK_FLAG_VALUE(COMPRBLK); 480 CHECK_FLAG_VALUE(NOCOMPR); 481 CHECK_FLAG_VALUE(ECOMPR); 482 CHECK_FLAG_VALUE(INDEX); 483 CHECK_FLAG_VALUE(IMAGIC); 484 CHECK_FLAG_VALUE(JOURNAL_DATA); 485 CHECK_FLAG_VALUE(NOTAIL); 486 CHECK_FLAG_VALUE(DIRSYNC); 487 CHECK_FLAG_VALUE(TOPDIR); 488 CHECK_FLAG_VALUE(HUGE_FILE); 489 CHECK_FLAG_VALUE(EXTENTS); 490 CHECK_FLAG_VALUE(EA_INODE); 491 CHECK_FLAG_VALUE(EOFBLOCKS); 492 CHECK_FLAG_VALUE(INLINE_DATA); 493 CHECK_FLAG_VALUE(RESERVED); 494 } 495 496 /* Used to pass group descriptor data when online resize is done */ 497 struct ext4_new_group_input { 498 __u32 group; /* Group number for this data */ 499 __u64 block_bitmap; /* Absolute block number of block bitmap */ 500 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 501 __u64 inode_table; /* Absolute block number of inode table start */ 502 __u32 blocks_count; /* Total number of blocks in this group */ 503 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 504 __u16 unused; 505 }; 506 507 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 508 struct compat_ext4_new_group_input { 509 u32 group; 510 compat_u64 block_bitmap; 511 compat_u64 inode_bitmap; 512 compat_u64 inode_table; 513 u32 blocks_count; 514 u16 reserved_blocks; 515 u16 unused; 516 }; 517 #endif 518 519 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 520 struct ext4_new_group_data { 521 __u32 group; 522 __u64 block_bitmap; 523 __u64 inode_bitmap; 524 __u64 inode_table; 525 __u32 blocks_count; 526 __u16 reserved_blocks; 527 __u16 unused; 528 __u32 free_blocks_count; 529 }; 530 531 /* Indexes used to index group tables in ext4_new_group_data */ 532 enum { 533 BLOCK_BITMAP = 0, /* block bitmap */ 534 INODE_BITMAP, /* inode bitmap */ 535 INODE_TABLE, /* inode tables */ 536 GROUP_TABLE_COUNT, 537 }; 538 539 /* 540 * Flags used by ext4_map_blocks() 541 */ 542 /* Allocate any needed blocks and/or convert an unitialized 543 extent to be an initialized ext4 */ 544 #define EXT4_GET_BLOCKS_CREATE 0x0001 545 /* Request the creation of an unitialized extent */ 546 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 547 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 548 EXT4_GET_BLOCKS_CREATE) 549 /* Caller is from the delayed allocation writeout path 550 * finally doing the actual allocation of delayed blocks */ 551 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 552 /* caller is from the direct IO path, request to creation of an 553 unitialized extents if not allocated, split the uninitialized 554 extent if blocks has been preallocated already*/ 555 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 556 #define EXT4_GET_BLOCKS_CONVERT 0x0010 557 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 558 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 559 /* Convert extent to initialized after IO complete */ 560 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 561 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 562 /* Eventual metadata allocation (due to growing extent tree) 563 * should not fail, so try to use reserved blocks for that.*/ 564 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 565 /* Don't normalize allocation size (used for fallocate) */ 566 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 567 /* Request will not result in inode size update (user for fallocate) */ 568 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 569 /* Do not take i_data_sem locking in ext4_map_blocks */ 570 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 571 /* Do not put hole in extent cache */ 572 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200 573 574 /* 575 * Flags used by ext4_free_blocks 576 */ 577 #define EXT4_FREE_BLOCKS_METADATA 0x0001 578 #define EXT4_FREE_BLOCKS_FORGET 0x0002 579 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 580 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 581 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 582 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 583 584 /* 585 * Flags used by ext4_discard_partial_page_buffers 586 */ 587 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001 588 589 /* 590 * ioctl commands 591 */ 592 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 593 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 594 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 595 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 596 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 597 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 598 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 599 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 600 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 601 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 602 #define EXT4_IOC_MIGRATE _IO('f', 9) 603 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 604 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 605 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 606 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 607 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 608 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 609 610 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 611 /* 612 * ioctl commands in 32 bit emulation 613 */ 614 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 615 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 616 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 617 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 618 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 619 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 620 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 621 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 622 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 623 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 624 #endif 625 626 /* Max physical block we can address w/o extents */ 627 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 628 629 /* 630 * Structure of an inode on the disk 631 */ 632 struct ext4_inode { 633 __le16 i_mode; /* File mode */ 634 __le16 i_uid; /* Low 16 bits of Owner Uid */ 635 __le32 i_size_lo; /* Size in bytes */ 636 __le32 i_atime; /* Access time */ 637 __le32 i_ctime; /* Inode Change time */ 638 __le32 i_mtime; /* Modification time */ 639 __le32 i_dtime; /* Deletion Time */ 640 __le16 i_gid; /* Low 16 bits of Group Id */ 641 __le16 i_links_count; /* Links count */ 642 __le32 i_blocks_lo; /* Blocks count */ 643 __le32 i_flags; /* File flags */ 644 union { 645 struct { 646 __le32 l_i_version; 647 } linux1; 648 struct { 649 __u32 h_i_translator; 650 } hurd1; 651 struct { 652 __u32 m_i_reserved1; 653 } masix1; 654 } osd1; /* OS dependent 1 */ 655 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 656 __le32 i_generation; /* File version (for NFS) */ 657 __le32 i_file_acl_lo; /* File ACL */ 658 __le32 i_size_high; 659 __le32 i_obso_faddr; /* Obsoleted fragment address */ 660 union { 661 struct { 662 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 663 __le16 l_i_file_acl_high; 664 __le16 l_i_uid_high; /* these 2 fields */ 665 __le16 l_i_gid_high; /* were reserved2[0] */ 666 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 667 __le16 l_i_reserved; 668 } linux2; 669 struct { 670 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 671 __u16 h_i_mode_high; 672 __u16 h_i_uid_high; 673 __u16 h_i_gid_high; 674 __u32 h_i_author; 675 } hurd2; 676 struct { 677 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 678 __le16 m_i_file_acl_high; 679 __u32 m_i_reserved2[2]; 680 } masix2; 681 } osd2; /* OS dependent 2 */ 682 __le16 i_extra_isize; 683 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 684 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 685 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 686 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 687 __le32 i_crtime; /* File Creation time */ 688 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 689 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 690 }; 691 692 struct move_extent { 693 __u32 reserved; /* should be zero */ 694 __u32 donor_fd; /* donor file descriptor */ 695 __u64 orig_start; /* logical start offset in block for orig */ 696 __u64 donor_start; /* logical start offset in block for donor */ 697 __u64 len; /* block length to be moved */ 698 __u64 moved_len; /* moved block length */ 699 }; 700 701 #define EXT4_EPOCH_BITS 2 702 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 703 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 704 705 /* 706 * Extended fields will fit into an inode if the filesystem was formatted 707 * with large inodes (-I 256 or larger) and there are not currently any EAs 708 * consuming all of the available space. For new inodes we always reserve 709 * enough space for the kernel's known extended fields, but for inodes 710 * created with an old kernel this might not have been the case. None of 711 * the extended inode fields is critical for correct filesystem operation. 712 * This macro checks if a certain field fits in the inode. Note that 713 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 714 */ 715 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 716 ((offsetof(typeof(*ext4_inode), field) + \ 717 sizeof((ext4_inode)->field)) \ 718 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 719 (einode)->i_extra_isize)) \ 720 721 static inline __le32 ext4_encode_extra_time(struct timespec *time) 722 { 723 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 724 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 725 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 726 } 727 728 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 729 { 730 if (sizeof(time->tv_sec) > 4) 731 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 732 << 32; 733 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 734 } 735 736 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 737 do { \ 738 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 739 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 740 (raw_inode)->xtime ## _extra = \ 741 ext4_encode_extra_time(&(inode)->xtime); \ 742 } while (0) 743 744 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 745 do { \ 746 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 747 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 748 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 749 (raw_inode)->xtime ## _extra = \ 750 ext4_encode_extra_time(&(einode)->xtime); \ 751 } while (0) 752 753 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 754 do { \ 755 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 756 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 757 ext4_decode_extra_time(&(inode)->xtime, \ 758 raw_inode->xtime ## _extra); \ 759 else \ 760 (inode)->xtime.tv_nsec = 0; \ 761 } while (0) 762 763 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 764 do { \ 765 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 766 (einode)->xtime.tv_sec = \ 767 (signed)le32_to_cpu((raw_inode)->xtime); \ 768 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 769 ext4_decode_extra_time(&(einode)->xtime, \ 770 raw_inode->xtime ## _extra); \ 771 else \ 772 (einode)->xtime.tv_nsec = 0; \ 773 } while (0) 774 775 #define i_disk_version osd1.linux1.l_i_version 776 777 #if defined(__KERNEL__) || defined(__linux__) 778 #define i_reserved1 osd1.linux1.l_i_reserved1 779 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 780 #define i_blocks_high osd2.linux2.l_i_blocks_high 781 #define i_uid_low i_uid 782 #define i_gid_low i_gid 783 #define i_uid_high osd2.linux2.l_i_uid_high 784 #define i_gid_high osd2.linux2.l_i_gid_high 785 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 786 787 #elif defined(__GNU__) 788 789 #define i_translator osd1.hurd1.h_i_translator 790 #define i_uid_high osd2.hurd2.h_i_uid_high 791 #define i_gid_high osd2.hurd2.h_i_gid_high 792 #define i_author osd2.hurd2.h_i_author 793 794 #elif defined(__masix__) 795 796 #define i_reserved1 osd1.masix1.m_i_reserved1 797 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 798 #define i_reserved2 osd2.masix2.m_i_reserved2 799 800 #endif /* defined(__KERNEL__) || defined(__linux__) */ 801 802 #include "extents_status.h" 803 804 /* 805 * fourth extended file system inode data in memory 806 */ 807 struct ext4_inode_info { 808 __le32 i_data[15]; /* unconverted */ 809 __u32 i_dtime; 810 ext4_fsblk_t i_file_acl; 811 812 /* 813 * i_block_group is the number of the block group which contains 814 * this file's inode. Constant across the lifetime of the inode, 815 * it is ued for making block allocation decisions - we try to 816 * place a file's data blocks near its inode block, and new inodes 817 * near to their parent directory's inode. 818 */ 819 ext4_group_t i_block_group; 820 ext4_lblk_t i_dir_start_lookup; 821 #if (BITS_PER_LONG < 64) 822 unsigned long i_state_flags; /* Dynamic state flags */ 823 #endif 824 unsigned long i_flags; 825 826 /* 827 * Extended attributes can be read independently of the main file 828 * data. Taking i_mutex even when reading would cause contention 829 * between readers of EAs and writers of regular file data, so 830 * instead we synchronize on xattr_sem when reading or changing 831 * EAs. 832 */ 833 struct rw_semaphore xattr_sem; 834 835 struct list_head i_orphan; /* unlinked but open inodes */ 836 837 /* 838 * i_disksize keeps track of what the inode size is ON DISK, not 839 * in memory. During truncate, i_size is set to the new size by 840 * the VFS prior to calling ext4_truncate(), but the filesystem won't 841 * set i_disksize to 0 until the truncate is actually under way. 842 * 843 * The intent is that i_disksize always represents the blocks which 844 * are used by this file. This allows recovery to restart truncate 845 * on orphans if we crash during truncate. We actually write i_disksize 846 * into the on-disk inode when writing inodes out, instead of i_size. 847 * 848 * The only time when i_disksize and i_size may be different is when 849 * a truncate is in progress. The only things which change i_disksize 850 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 851 */ 852 loff_t i_disksize; 853 854 /* 855 * i_data_sem is for serialising ext4_truncate() against 856 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 857 * data tree are chopped off during truncate. We can't do that in 858 * ext4 because whenever we perform intermediate commits during 859 * truncate, the inode and all the metadata blocks *must* be in a 860 * consistent state which allows truncation of the orphans to restart 861 * during recovery. Hence we must fix the get_block-vs-truncate race 862 * by other means, so we have i_data_sem. 863 */ 864 struct rw_semaphore i_data_sem; 865 struct inode vfs_inode; 866 struct jbd2_inode *jinode; 867 868 /* 869 * File creation time. Its function is same as that of 870 * struct timespec i_{a,c,m}time in the generic inode. 871 */ 872 struct timespec i_crtime; 873 874 /* mballoc */ 875 struct list_head i_prealloc_list; 876 spinlock_t i_prealloc_lock; 877 878 /* extents status tree */ 879 struct ext4_es_tree i_es_tree; 880 rwlock_t i_es_lock; 881 struct list_head i_es_lru; 882 unsigned int i_es_lru_nr; /* protected by i_es_lock */ 883 884 /* ialloc */ 885 ext4_group_t i_last_alloc_group; 886 887 /* allocation reservation info for delalloc */ 888 /* In case of bigalloc, these refer to clusters rather than blocks */ 889 unsigned int i_reserved_data_blocks; 890 unsigned int i_reserved_meta_blocks; 891 unsigned int i_allocated_meta_blocks; 892 ext4_lblk_t i_da_metadata_calc_last_lblock; 893 int i_da_metadata_calc_len; 894 895 /* on-disk additional length */ 896 __u16 i_extra_isize; 897 898 /* Indicate the inline data space. */ 899 u16 i_inline_off; 900 u16 i_inline_size; 901 902 #ifdef CONFIG_QUOTA 903 /* quota space reservation, managed internally by quota code */ 904 qsize_t i_reserved_quota; 905 #endif 906 907 /* completed IOs that might need unwritten extents handling */ 908 struct list_head i_completed_io_list; 909 spinlock_t i_completed_io_lock; 910 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 911 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 912 struct work_struct i_unwritten_work; /* deferred extent conversion */ 913 914 spinlock_t i_block_reservation_lock; 915 916 /* 917 * Transactions that contain inode's metadata needed to complete 918 * fsync and fdatasync, respectively. 919 */ 920 tid_t i_sync_tid; 921 tid_t i_datasync_tid; 922 923 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 924 __u32 i_csum_seed; 925 }; 926 927 /* 928 * File system states 929 */ 930 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 931 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 932 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 933 934 /* 935 * Misc. filesystem flags 936 */ 937 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 938 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 939 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 940 941 /* 942 * Mount flags set via mount options or defaults 943 */ 944 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 945 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 946 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 947 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 948 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 949 #define EXT4_MOUNT_ERRORS_MASK 0x00070 950 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 951 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 952 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 953 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 954 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 955 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 956 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 957 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 958 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 959 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 960 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 961 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 962 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 963 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 964 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 965 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 966 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 967 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 968 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 969 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 970 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 971 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 972 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 973 974 /* 975 * Mount flags set either automatically (could not be set by mount option) 976 * based on per file system feature or property or in special cases such as 977 * distinguishing between explicit mount option definition and default. 978 */ 979 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 980 specified delalloc */ 981 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 982 size of blocksize * 8 983 blocks */ 984 985 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 986 ~EXT4_MOUNT_##opt 987 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 988 EXT4_MOUNT_##opt 989 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 990 EXT4_MOUNT_##opt) 991 992 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 993 ~EXT4_MOUNT2_##opt 994 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 995 EXT4_MOUNT2_##opt 996 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 997 EXT4_MOUNT2_##opt) 998 999 #define ext4_test_and_set_bit __test_and_set_bit_le 1000 #define ext4_set_bit __set_bit_le 1001 #define ext4_set_bit_atomic ext2_set_bit_atomic 1002 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1003 #define ext4_clear_bit __clear_bit_le 1004 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1005 #define ext4_test_bit test_bit_le 1006 #define ext4_find_next_zero_bit find_next_zero_bit_le 1007 #define ext4_find_next_bit find_next_bit_le 1008 1009 extern void ext4_set_bits(void *bm, int cur, int len); 1010 1011 /* 1012 * Maximal mount counts between two filesystem checks 1013 */ 1014 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1015 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1016 1017 /* 1018 * Behaviour when detecting errors 1019 */ 1020 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1021 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1022 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1023 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1024 1025 /* Metadata checksum algorithm codes */ 1026 #define EXT4_CRC32C_CHKSUM 1 1027 1028 /* 1029 * Structure of the super block 1030 */ 1031 struct ext4_super_block { 1032 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1033 __le32 s_blocks_count_lo; /* Blocks count */ 1034 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1035 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1036 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1037 __le32 s_first_data_block; /* First Data Block */ 1038 __le32 s_log_block_size; /* Block size */ 1039 __le32 s_log_cluster_size; /* Allocation cluster size */ 1040 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1041 __le32 s_clusters_per_group; /* # Clusters per group */ 1042 __le32 s_inodes_per_group; /* # Inodes per group */ 1043 __le32 s_mtime; /* Mount time */ 1044 /*30*/ __le32 s_wtime; /* Write time */ 1045 __le16 s_mnt_count; /* Mount count */ 1046 __le16 s_max_mnt_count; /* Maximal mount count */ 1047 __le16 s_magic; /* Magic signature */ 1048 __le16 s_state; /* File system state */ 1049 __le16 s_errors; /* Behaviour when detecting errors */ 1050 __le16 s_minor_rev_level; /* minor revision level */ 1051 /*40*/ __le32 s_lastcheck; /* time of last check */ 1052 __le32 s_checkinterval; /* max. time between checks */ 1053 __le32 s_creator_os; /* OS */ 1054 __le32 s_rev_level; /* Revision level */ 1055 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1056 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1057 /* 1058 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1059 * 1060 * Note: the difference between the compatible feature set and 1061 * the incompatible feature set is that if there is a bit set 1062 * in the incompatible feature set that the kernel doesn't 1063 * know about, it should refuse to mount the filesystem. 1064 * 1065 * e2fsck's requirements are more strict; if it doesn't know 1066 * about a feature in either the compatible or incompatible 1067 * feature set, it must abort and not try to meddle with 1068 * things it doesn't understand... 1069 */ 1070 __le32 s_first_ino; /* First non-reserved inode */ 1071 __le16 s_inode_size; /* size of inode structure */ 1072 __le16 s_block_group_nr; /* block group # of this superblock */ 1073 __le32 s_feature_compat; /* compatible feature set */ 1074 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1075 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1076 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1077 /*78*/ char s_volume_name[16]; /* volume name */ 1078 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1079 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1080 /* 1081 * Performance hints. Directory preallocation should only 1082 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1083 */ 1084 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1085 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1086 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1087 /* 1088 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1089 */ 1090 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1091 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1092 __le32 s_journal_dev; /* device number of journal file */ 1093 __le32 s_last_orphan; /* start of list of inodes to delete */ 1094 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1095 __u8 s_def_hash_version; /* Default hash version to use */ 1096 __u8 s_jnl_backup_type; 1097 __le16 s_desc_size; /* size of group descriptor */ 1098 /*100*/ __le32 s_default_mount_opts; 1099 __le32 s_first_meta_bg; /* First metablock block group */ 1100 __le32 s_mkfs_time; /* When the filesystem was created */ 1101 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1102 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1103 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1104 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1105 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1106 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1107 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1108 __le32 s_flags; /* Miscellaneous flags */ 1109 __le16 s_raid_stride; /* RAID stride */ 1110 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1111 __le64 s_mmp_block; /* Block for multi-mount protection */ 1112 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1113 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1114 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1115 __le16 s_reserved_pad; 1116 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1117 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1118 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1119 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1120 snapshot's future use */ 1121 __le32 s_snapshot_list; /* inode number of the head of the 1122 on-disk snapshot list */ 1123 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1124 __le32 s_error_count; /* number of fs errors */ 1125 __le32 s_first_error_time; /* first time an error happened */ 1126 __le32 s_first_error_ino; /* inode involved in first error */ 1127 __le64 s_first_error_block; /* block involved of first error */ 1128 __u8 s_first_error_func[32]; /* function where the error happened */ 1129 __le32 s_first_error_line; /* line number where error happened */ 1130 __le32 s_last_error_time; /* most recent time of an error */ 1131 __le32 s_last_error_ino; /* inode involved in last error */ 1132 __le32 s_last_error_line; /* line number where error happened */ 1133 __le64 s_last_error_block; /* block involved of last error */ 1134 __u8 s_last_error_func[32]; /* function where the error happened */ 1135 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1136 __u8 s_mount_opts[64]; 1137 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1138 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1139 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1140 __le32 s_reserved[108]; /* Padding to the end of the block */ 1141 __le32 s_checksum; /* crc32c(superblock) */ 1142 }; 1143 1144 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1145 1146 #ifdef __KERNEL__ 1147 1148 /* 1149 * run-time mount flags 1150 */ 1151 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1152 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1153 1154 /* 1155 * fourth extended-fs super-block data in memory 1156 */ 1157 struct ext4_sb_info { 1158 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1159 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1160 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1161 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1162 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1163 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1164 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1165 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1166 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1167 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1168 unsigned long s_overhead; /* # of fs overhead clusters */ 1169 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1170 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1171 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1172 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1173 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1174 struct buffer_head **s_group_desc; 1175 unsigned int s_mount_opt; 1176 unsigned int s_mount_opt2; 1177 unsigned int s_mount_flags; 1178 unsigned int s_def_mount_opt; 1179 ext4_fsblk_t s_sb_block; 1180 atomic64_t s_resv_clusters; 1181 kuid_t s_resuid; 1182 kgid_t s_resgid; 1183 unsigned short s_mount_state; 1184 unsigned short s_pad; 1185 int s_addr_per_block_bits; 1186 int s_desc_per_block_bits; 1187 int s_inode_size; 1188 int s_first_ino; 1189 unsigned int s_inode_readahead_blks; 1190 unsigned int s_inode_goal; 1191 spinlock_t s_next_gen_lock; 1192 u32 s_next_generation; 1193 u32 s_hash_seed[4]; 1194 int s_def_hash_version; 1195 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1196 struct percpu_counter s_freeclusters_counter; 1197 struct percpu_counter s_freeinodes_counter; 1198 struct percpu_counter s_dirs_counter; 1199 struct percpu_counter s_dirtyclusters_counter; 1200 struct blockgroup_lock *s_blockgroup_lock; 1201 struct proc_dir_entry *s_proc; 1202 struct kobject s_kobj; 1203 struct completion s_kobj_unregister; 1204 struct super_block *s_sb; 1205 1206 /* Journaling */ 1207 struct journal_s *s_journal; 1208 struct list_head s_orphan; 1209 struct mutex s_orphan_lock; 1210 unsigned long s_resize_flags; /* Flags indicating if there 1211 is a resizer */ 1212 unsigned long s_commit_interval; 1213 u32 s_max_batch_time; 1214 u32 s_min_batch_time; 1215 struct block_device *journal_bdev; 1216 #ifdef CONFIG_QUOTA 1217 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 1218 int s_jquota_fmt; /* Format of quota to use */ 1219 #endif 1220 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1221 struct rb_root system_blks; 1222 1223 #ifdef EXTENTS_STATS 1224 /* ext4 extents stats */ 1225 unsigned long s_ext_min; 1226 unsigned long s_ext_max; 1227 unsigned long s_depth_max; 1228 spinlock_t s_ext_stats_lock; 1229 unsigned long s_ext_blocks; 1230 unsigned long s_ext_extents; 1231 #endif 1232 1233 /* for buddy allocator */ 1234 struct ext4_group_info ***s_group_info; 1235 struct inode *s_buddy_cache; 1236 spinlock_t s_md_lock; 1237 unsigned short *s_mb_offsets; 1238 unsigned int *s_mb_maxs; 1239 unsigned int s_group_info_size; 1240 1241 /* tunables */ 1242 unsigned long s_stripe; 1243 unsigned int s_mb_stream_request; 1244 unsigned int s_mb_max_to_scan; 1245 unsigned int s_mb_min_to_scan; 1246 unsigned int s_mb_stats; 1247 unsigned int s_mb_order2_reqs; 1248 unsigned int s_mb_group_prealloc; 1249 unsigned int s_max_writeback_mb_bump; 1250 unsigned int s_max_dir_size_kb; 1251 /* where last allocation was done - for stream allocation */ 1252 unsigned long s_mb_last_group; 1253 unsigned long s_mb_last_start; 1254 1255 /* stats for buddy allocator */ 1256 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1257 atomic_t s_bal_success; /* we found long enough chunks */ 1258 atomic_t s_bal_allocated; /* in blocks */ 1259 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1260 atomic_t s_bal_goals; /* goal hits */ 1261 atomic_t s_bal_breaks; /* too long searches */ 1262 atomic_t s_bal_2orders; /* 2^order hits */ 1263 spinlock_t s_bal_lock; 1264 unsigned long s_mb_buddies_generated; 1265 unsigned long long s_mb_generation_time; 1266 atomic_t s_mb_lost_chunks; 1267 atomic_t s_mb_preallocated; 1268 atomic_t s_mb_discarded; 1269 atomic_t s_lock_busy; 1270 1271 /* locality groups */ 1272 struct ext4_locality_group __percpu *s_locality_groups; 1273 1274 /* for write statistics */ 1275 unsigned long s_sectors_written_start; 1276 u64 s_kbytes_written; 1277 1278 /* the size of zero-out chunk */ 1279 unsigned int s_extent_max_zeroout_kb; 1280 1281 unsigned int s_log_groups_per_flex; 1282 struct flex_groups *s_flex_groups; 1283 ext4_group_t s_flex_groups_allocated; 1284 1285 /* workqueue for dio unwritten */ 1286 struct workqueue_struct *dio_unwritten_wq; 1287 1288 /* timer for periodic error stats printing */ 1289 struct timer_list s_err_report; 1290 1291 /* Lazy inode table initialization info */ 1292 struct ext4_li_request *s_li_request; 1293 /* Wait multiplier for lazy initialization thread */ 1294 unsigned int s_li_wait_mult; 1295 1296 /* Kernel thread for multiple mount protection */ 1297 struct task_struct *s_mmp_tsk; 1298 1299 /* record the last minlen when FITRIM is called. */ 1300 atomic_t s_last_trim_minblks; 1301 1302 /* Reference to checksum algorithm driver via cryptoapi */ 1303 struct crypto_shash *s_chksum_driver; 1304 1305 /* Precomputed FS UUID checksum for seeding other checksums */ 1306 __u32 s_csum_seed; 1307 1308 /* Reclaim extents from extent status tree */ 1309 struct shrinker s_es_shrinker; 1310 struct list_head s_es_lru; 1311 struct percpu_counter s_extent_cache_cnt; 1312 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp; 1313 }; 1314 1315 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1316 { 1317 return sb->s_fs_info; 1318 } 1319 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1320 { 1321 return container_of(inode, struct ext4_inode_info, vfs_inode); 1322 } 1323 1324 static inline struct timespec ext4_current_time(struct inode *inode) 1325 { 1326 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1327 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1328 } 1329 1330 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1331 { 1332 return ino == EXT4_ROOT_INO || 1333 ino == EXT4_USR_QUOTA_INO || 1334 ino == EXT4_GRP_QUOTA_INO || 1335 ino == EXT4_BOOT_LOADER_INO || 1336 ino == EXT4_JOURNAL_INO || 1337 ino == EXT4_RESIZE_INO || 1338 (ino >= EXT4_FIRST_INO(sb) && 1339 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1340 } 1341 1342 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1343 struct ext4_io_end *io_end) 1344 { 1345 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1346 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1347 atomic_inc(&EXT4_I(inode)->i_unwritten); 1348 } 1349 } 1350 1351 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1352 { 1353 return inode->i_private; 1354 } 1355 1356 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1357 { 1358 inode->i_private = io; 1359 } 1360 1361 /* 1362 * Inode dynamic state flags 1363 */ 1364 enum { 1365 EXT4_STATE_JDATA, /* journaled data exists */ 1366 EXT4_STATE_NEW, /* inode is newly created */ 1367 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1368 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1369 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1370 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1371 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1372 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1373 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */ 1374 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1375 nolocking */ 1376 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1377 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */ 1378 }; 1379 1380 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1381 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1382 { \ 1383 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1384 } \ 1385 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1386 { \ 1387 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1388 } \ 1389 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1390 { \ 1391 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1392 } 1393 1394 EXT4_INODE_BIT_FNS(flag, flags, 0) 1395 #if (BITS_PER_LONG < 64) 1396 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1397 1398 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1399 { 1400 (ei)->i_state_flags = 0; 1401 } 1402 #else 1403 EXT4_INODE_BIT_FNS(state, flags, 32) 1404 1405 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1406 { 1407 /* We depend on the fact that callers will set i_flags */ 1408 } 1409 #endif 1410 #else 1411 /* Assume that user mode programs are passing in an ext4fs superblock, not 1412 * a kernel struct super_block. This will allow us to call the feature-test 1413 * macros from user land. */ 1414 #define EXT4_SB(sb) (sb) 1415 #endif 1416 1417 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1418 1419 /* 1420 * Codes for operating systems 1421 */ 1422 #define EXT4_OS_LINUX 0 1423 #define EXT4_OS_HURD 1 1424 #define EXT4_OS_MASIX 2 1425 #define EXT4_OS_FREEBSD 3 1426 #define EXT4_OS_LITES 4 1427 1428 /* 1429 * Revision levels 1430 */ 1431 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1432 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1433 1434 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1435 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1436 1437 #define EXT4_GOOD_OLD_INODE_SIZE 128 1438 1439 /* 1440 * Feature set definitions 1441 */ 1442 1443 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1444 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1445 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1446 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1447 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1448 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1449 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1450 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1451 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1452 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1453 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1454 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1455 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1456 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1457 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1458 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1459 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1460 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1461 1462 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1463 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1464 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1465 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1466 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1467 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1468 1469 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1470 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1471 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1472 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1473 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1474 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1475 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1476 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1477 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1478 /* 1479 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1480 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1481 * all other data structures' checksums. However, the METADATA_CSUM and 1482 * GDT_CSUM bits are mutually exclusive. 1483 */ 1484 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1485 1486 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1487 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1488 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1489 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1490 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1491 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1492 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1493 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1494 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1495 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1496 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1497 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1498 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1499 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1500 1501 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1502 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1503 EXT4_FEATURE_INCOMPAT_META_BG) 1504 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1505 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1506 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1507 1508 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1509 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1510 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1511 EXT4_FEATURE_INCOMPAT_META_BG) 1512 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1513 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1514 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1515 1516 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1517 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1518 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1519 EXT4_FEATURE_INCOMPAT_META_BG| \ 1520 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1521 EXT4_FEATURE_INCOMPAT_64BIT| \ 1522 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1523 EXT4_FEATURE_INCOMPAT_MMP | \ 1524 EXT4_FEATURE_INCOMPAT_INLINE_DATA) 1525 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1526 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1527 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1528 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1529 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1530 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1531 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1532 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1533 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1534 EXT4_FEATURE_RO_COMPAT_QUOTA) 1535 1536 /* 1537 * Default values for user and/or group using reserved blocks 1538 */ 1539 #define EXT4_DEF_RESUID 0 1540 #define EXT4_DEF_RESGID 0 1541 1542 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1543 1544 /* 1545 * Default mount options 1546 */ 1547 #define EXT4_DEFM_DEBUG 0x0001 1548 #define EXT4_DEFM_BSDGROUPS 0x0002 1549 #define EXT4_DEFM_XATTR_USER 0x0004 1550 #define EXT4_DEFM_ACL 0x0008 1551 #define EXT4_DEFM_UID16 0x0010 1552 #define EXT4_DEFM_JMODE 0x0060 1553 #define EXT4_DEFM_JMODE_DATA 0x0020 1554 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1555 #define EXT4_DEFM_JMODE_WBACK 0x0060 1556 #define EXT4_DEFM_NOBARRIER 0x0100 1557 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1558 #define EXT4_DEFM_DISCARD 0x0400 1559 #define EXT4_DEFM_NODELALLOC 0x0800 1560 1561 /* 1562 * Default journal batch times 1563 */ 1564 #define EXT4_DEF_MIN_BATCH_TIME 0 1565 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1566 1567 /* 1568 * Minimum number of groups in a flexgroup before we separate out 1569 * directories into the first block group of a flexgroup 1570 */ 1571 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1572 1573 /* 1574 * Structure of a directory entry 1575 */ 1576 #define EXT4_NAME_LEN 255 1577 1578 struct ext4_dir_entry { 1579 __le32 inode; /* Inode number */ 1580 __le16 rec_len; /* Directory entry length */ 1581 __le16 name_len; /* Name length */ 1582 char name[EXT4_NAME_LEN]; /* File name */ 1583 }; 1584 1585 /* 1586 * The new version of the directory entry. Since EXT4 structures are 1587 * stored in intel byte order, and the name_len field could never be 1588 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1589 * file_type field. 1590 */ 1591 struct ext4_dir_entry_2 { 1592 __le32 inode; /* Inode number */ 1593 __le16 rec_len; /* Directory entry length */ 1594 __u8 name_len; /* Name length */ 1595 __u8 file_type; 1596 char name[EXT4_NAME_LEN]; /* File name */ 1597 }; 1598 1599 /* 1600 * This is a bogus directory entry at the end of each leaf block that 1601 * records checksums. 1602 */ 1603 struct ext4_dir_entry_tail { 1604 __le32 det_reserved_zero1; /* Pretend to be unused */ 1605 __le16 det_rec_len; /* 12 */ 1606 __u8 det_reserved_zero2; /* Zero name length */ 1607 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1608 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1609 }; 1610 1611 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1612 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1613 ((blocksize) - \ 1614 sizeof(struct ext4_dir_entry_tail)))) 1615 1616 /* 1617 * Ext4 directory file types. Only the low 3 bits are used. The 1618 * other bits are reserved for now. 1619 */ 1620 #define EXT4_FT_UNKNOWN 0 1621 #define EXT4_FT_REG_FILE 1 1622 #define EXT4_FT_DIR 2 1623 #define EXT4_FT_CHRDEV 3 1624 #define EXT4_FT_BLKDEV 4 1625 #define EXT4_FT_FIFO 5 1626 #define EXT4_FT_SOCK 6 1627 #define EXT4_FT_SYMLINK 7 1628 1629 #define EXT4_FT_MAX 8 1630 1631 #define EXT4_FT_DIR_CSUM 0xDE 1632 1633 /* 1634 * EXT4_DIR_PAD defines the directory entries boundaries 1635 * 1636 * NOTE: It must be a multiple of 4 1637 */ 1638 #define EXT4_DIR_PAD 4 1639 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1640 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1641 ~EXT4_DIR_ROUND) 1642 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1643 1644 /* 1645 * If we ever get support for fs block sizes > page_size, we'll need 1646 * to remove the #if statements in the next two functions... 1647 */ 1648 static inline unsigned int 1649 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1650 { 1651 unsigned len = le16_to_cpu(dlen); 1652 1653 #if (PAGE_CACHE_SIZE >= 65536) 1654 if (len == EXT4_MAX_REC_LEN || len == 0) 1655 return blocksize; 1656 return (len & 65532) | ((len & 3) << 16); 1657 #else 1658 return len; 1659 #endif 1660 } 1661 1662 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1663 { 1664 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1665 BUG(); 1666 #if (PAGE_CACHE_SIZE >= 65536) 1667 if (len < 65536) 1668 return cpu_to_le16(len); 1669 if (len == blocksize) { 1670 if (blocksize == 65536) 1671 return cpu_to_le16(EXT4_MAX_REC_LEN); 1672 else 1673 return cpu_to_le16(0); 1674 } 1675 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1676 #else 1677 return cpu_to_le16(len); 1678 #endif 1679 } 1680 1681 /* 1682 * Hash Tree Directory indexing 1683 * (c) Daniel Phillips, 2001 1684 */ 1685 1686 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1687 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1688 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1689 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1690 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1691 1692 /* Legal values for the dx_root hash_version field: */ 1693 1694 #define DX_HASH_LEGACY 0 1695 #define DX_HASH_HALF_MD4 1 1696 #define DX_HASH_TEA 2 1697 #define DX_HASH_LEGACY_UNSIGNED 3 1698 #define DX_HASH_HALF_MD4_UNSIGNED 4 1699 #define DX_HASH_TEA_UNSIGNED 5 1700 1701 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1702 const void *address, unsigned int length) 1703 { 1704 struct { 1705 struct shash_desc shash; 1706 char ctx[4]; 1707 } desc; 1708 int err; 1709 1710 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1711 1712 desc.shash.tfm = sbi->s_chksum_driver; 1713 desc.shash.flags = 0; 1714 *(u32 *)desc.ctx = crc; 1715 1716 err = crypto_shash_update(&desc.shash, address, length); 1717 BUG_ON(err); 1718 1719 return *(u32 *)desc.ctx; 1720 } 1721 1722 #ifdef __KERNEL__ 1723 1724 /* hash info structure used by the directory hash */ 1725 struct dx_hash_info 1726 { 1727 u32 hash; 1728 u32 minor_hash; 1729 int hash_version; 1730 u32 *seed; 1731 }; 1732 1733 1734 /* 32 and 64 bit signed EOF for dx directories */ 1735 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1736 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1737 1738 1739 /* 1740 * Control parameters used by ext4_htree_next_block 1741 */ 1742 #define HASH_NB_ALWAYS 1 1743 1744 1745 /* 1746 * Describe an inode's exact location on disk and in memory 1747 */ 1748 struct ext4_iloc 1749 { 1750 struct buffer_head *bh; 1751 unsigned long offset; 1752 ext4_group_t block_group; 1753 }; 1754 1755 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1756 { 1757 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1758 } 1759 1760 /* 1761 * This structure is stuffed into the struct file's private_data field 1762 * for directories. It is where we put information so that we can do 1763 * readdir operations in hash tree order. 1764 */ 1765 struct dir_private_info { 1766 struct rb_root root; 1767 struct rb_node *curr_node; 1768 struct fname *extra_fname; 1769 loff_t last_pos; 1770 __u32 curr_hash; 1771 __u32 curr_minor_hash; 1772 __u32 next_hash; 1773 }; 1774 1775 /* calculate the first block number of the group */ 1776 static inline ext4_fsblk_t 1777 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1778 { 1779 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1780 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1781 } 1782 1783 /* 1784 * Special error return code only used by dx_probe() and its callers. 1785 */ 1786 #define ERR_BAD_DX_DIR -75000 1787 1788 /* 1789 * Timeout and state flag for lazy initialization inode thread. 1790 */ 1791 #define EXT4_DEF_LI_WAIT_MULT 10 1792 #define EXT4_DEF_LI_MAX_START_DELAY 5 1793 #define EXT4_LAZYINIT_QUIT 0x0001 1794 #define EXT4_LAZYINIT_RUNNING 0x0002 1795 1796 /* 1797 * Lazy inode table initialization info 1798 */ 1799 struct ext4_lazy_init { 1800 unsigned long li_state; 1801 struct list_head li_request_list; 1802 struct mutex li_list_mtx; 1803 }; 1804 1805 struct ext4_li_request { 1806 struct super_block *lr_super; 1807 struct ext4_sb_info *lr_sbi; 1808 ext4_group_t lr_next_group; 1809 struct list_head lr_request; 1810 unsigned long lr_next_sched; 1811 unsigned long lr_timeout; 1812 }; 1813 1814 struct ext4_features { 1815 struct kobject f_kobj; 1816 struct completion f_kobj_unregister; 1817 }; 1818 1819 /* 1820 * This structure will be used for multiple mount protection. It will be 1821 * written into the block number saved in the s_mmp_block field in the 1822 * superblock. Programs that check MMP should assume that if 1823 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1824 * to use the filesystem, regardless of how old the timestamp is. 1825 */ 1826 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1827 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1828 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1829 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1830 1831 struct mmp_struct { 1832 __le32 mmp_magic; /* Magic number for MMP */ 1833 __le32 mmp_seq; /* Sequence no. updated periodically */ 1834 1835 /* 1836 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1837 * purposes and do not affect the correctness of the algorithm 1838 */ 1839 __le64 mmp_time; /* Time last updated */ 1840 char mmp_nodename[64]; /* Node which last updated MMP block */ 1841 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1842 1843 /* 1844 * mmp_check_interval is used to verify if the MMP block has been 1845 * updated on the block device. The value is updated based on the 1846 * maximum time to write the MMP block during an update cycle. 1847 */ 1848 __le16 mmp_check_interval; 1849 1850 __le16 mmp_pad1; 1851 __le32 mmp_pad2[226]; 1852 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1853 }; 1854 1855 /* arguments passed to the mmp thread */ 1856 struct mmpd_data { 1857 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1858 struct super_block *sb; /* super block of the fs */ 1859 }; 1860 1861 /* 1862 * Check interval multiplier 1863 * The MMP block is written every update interval and initially checked every 1864 * update interval x the multiplier (the value is then adapted based on the 1865 * write latency). The reason is that writes can be delayed under load and we 1866 * don't want readers to incorrectly assume that the filesystem is no longer 1867 * in use. 1868 */ 1869 #define EXT4_MMP_CHECK_MULT 2UL 1870 1871 /* 1872 * Minimum interval for MMP checking in seconds. 1873 */ 1874 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1875 1876 /* 1877 * Maximum interval for MMP checking in seconds. 1878 */ 1879 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1880 1881 /* 1882 * Function prototypes 1883 */ 1884 1885 /* 1886 * Ok, these declarations are also in <linux/kernel.h> but none of the 1887 * ext4 source programs needs to include it so they are duplicated here. 1888 */ 1889 # define NORET_TYPE /**/ 1890 # define ATTRIB_NORET __attribute__((noreturn)) 1891 # define NORET_AND noreturn, 1892 1893 /* bitmap.c */ 1894 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1895 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1896 struct ext4_group_desc *gdp, 1897 struct buffer_head *bh, int sz); 1898 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1899 struct ext4_group_desc *gdp, 1900 struct buffer_head *bh, int sz); 1901 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1902 struct ext4_group_desc *gdp, 1903 struct buffer_head *bh); 1904 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1905 struct ext4_group_desc *gdp, 1906 struct buffer_head *bh); 1907 1908 /* balloc.c */ 1909 extern void ext4_get_group_no_and_offset(struct super_block *sb, 1910 ext4_fsblk_t blocknr, 1911 ext4_group_t *blockgrpp, 1912 ext4_grpblk_t *offsetp); 1913 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 1914 ext4_fsblk_t block); 1915 1916 extern void ext4_validate_block_bitmap(struct super_block *sb, 1917 struct ext4_group_desc *desc, 1918 unsigned int block_group, 1919 struct buffer_head *bh); 1920 extern unsigned int ext4_block_group(struct super_block *sb, 1921 ext4_fsblk_t blocknr); 1922 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1923 ext4_fsblk_t blocknr); 1924 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1925 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1926 ext4_group_t group); 1927 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1928 ext4_fsblk_t goal, 1929 unsigned int flags, 1930 unsigned long *count, 1931 int *errp); 1932 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1933 s64 nclusters, unsigned int flags); 1934 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1935 extern void ext4_check_blocks_bitmap(struct super_block *); 1936 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1937 ext4_group_t block_group, 1938 struct buffer_head ** bh); 1939 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1940 1941 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1942 ext4_group_t block_group); 1943 extern int ext4_wait_block_bitmap(struct super_block *sb, 1944 ext4_group_t block_group, 1945 struct buffer_head *bh); 1946 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1947 ext4_group_t block_group); 1948 extern void ext4_init_block_bitmap(struct super_block *sb, 1949 struct buffer_head *bh, 1950 ext4_group_t group, 1951 struct ext4_group_desc *desc); 1952 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1953 ext4_group_t block_group, 1954 struct ext4_group_desc *gdp); 1955 extern unsigned ext4_num_overhead_clusters(struct super_block *sb, 1956 ext4_group_t block_group, 1957 struct ext4_group_desc *gdp); 1958 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1959 1960 /* dir.c */ 1961 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1962 struct file *, 1963 struct ext4_dir_entry_2 *, 1964 struct buffer_head *, char *, int, 1965 unsigned int); 1966 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 1967 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 1968 (de), (bh), (buf), (size), (offset))) 1969 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1970 __u32 minor_hash, 1971 struct ext4_dir_entry_2 *dirent); 1972 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1973 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 1974 struct buffer_head *bh, 1975 void *buf, int buf_size, 1976 const char *name, int namelen, 1977 struct ext4_dir_entry_2 **dest_de); 1978 void ext4_insert_dentry(struct inode *inode, 1979 struct ext4_dir_entry_2 *de, 1980 int buf_size, 1981 const char *name, int namelen); 1982 static inline void ext4_update_dx_flag(struct inode *inode) 1983 { 1984 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb, 1985 EXT4_FEATURE_COMPAT_DIR_INDEX)) 1986 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 1987 } 1988 static unsigned char ext4_filetype_table[] = { 1989 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 1990 }; 1991 1992 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 1993 { 1994 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) || 1995 (filetype >= EXT4_FT_MAX)) 1996 return DT_UNKNOWN; 1997 1998 return ext4_filetype_table[filetype]; 1999 } 2000 2001 /* fsync.c */ 2002 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2003 extern int ext4_flush_unwritten_io(struct inode *); 2004 2005 /* hash.c */ 2006 extern int ext4fs_dirhash(const char *name, int len, struct 2007 dx_hash_info *hinfo); 2008 2009 /* ialloc.c */ 2010 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2011 const struct qstr *qstr, __u32 goal, 2012 uid_t *owner, int handle_type, 2013 unsigned int line_no, int nblocks); 2014 2015 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2016 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2017 0, 0, 0) 2018 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2019 type, nblocks) \ 2020 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2021 (type), __LINE__, (nblocks)) 2022 2023 2024 extern void ext4_free_inode(handle_t *, struct inode *); 2025 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2026 extern unsigned long ext4_count_free_inodes(struct super_block *); 2027 extern unsigned long ext4_count_dirs(struct super_block *); 2028 extern void ext4_check_inodes_bitmap(struct super_block *); 2029 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2030 extern int ext4_init_inode_table(struct super_block *sb, 2031 ext4_group_t group, int barrier); 2032 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2033 2034 /* mballoc.c */ 2035 extern long ext4_mb_stats; 2036 extern long ext4_mb_max_to_scan; 2037 extern int ext4_mb_init(struct super_block *); 2038 extern int ext4_mb_release(struct super_block *); 2039 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2040 struct ext4_allocation_request *, int *); 2041 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2042 extern void ext4_discard_preallocations(struct inode *); 2043 extern int __init ext4_init_mballoc(void); 2044 extern void ext4_exit_mballoc(void); 2045 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2046 struct buffer_head *bh, ext4_fsblk_t block, 2047 unsigned long count, int flags); 2048 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2049 ext4_group_t ngroups); 2050 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2051 ext4_group_t i, struct ext4_group_desc *desc); 2052 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2053 ext4_fsblk_t block, unsigned long count); 2054 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2055 2056 /* inode.c */ 2057 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 2058 ext4_lblk_t, int, int *); 2059 struct buffer_head *ext4_bread(handle_t *, struct inode *, 2060 ext4_lblk_t, int, int *); 2061 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2062 struct buffer_head *bh_result, int create); 2063 int ext4_get_block(struct inode *inode, sector_t iblock, 2064 struct buffer_head *bh_result, int create); 2065 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2066 struct buffer_head *bh, int create); 2067 int ext4_walk_page_buffers(handle_t *handle, 2068 struct buffer_head *head, 2069 unsigned from, 2070 unsigned to, 2071 int *partial, 2072 int (*fn)(handle_t *handle, 2073 struct buffer_head *bh)); 2074 int do_journal_get_write_access(handle_t *handle, 2075 struct buffer_head *bh); 2076 #define FALL_BACK_TO_NONDELALLOC 1 2077 #define CONVERT_INLINE_DATA 2 2078 2079 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2080 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2081 extern int ext4_setattr(struct dentry *, struct iattr *); 2082 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2083 struct kstat *stat); 2084 extern void ext4_evict_inode(struct inode *); 2085 extern void ext4_clear_inode(struct inode *); 2086 extern int ext4_sync_inode(handle_t *, struct inode *); 2087 extern void ext4_dirty_inode(struct inode *, int); 2088 extern int ext4_change_inode_journal_flag(struct inode *, int); 2089 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2090 extern int ext4_can_truncate(struct inode *inode); 2091 extern void ext4_truncate(struct inode *); 2092 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2093 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2094 extern void ext4_set_inode_flags(struct inode *); 2095 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2096 extern int ext4_alloc_da_blocks(struct inode *inode); 2097 extern void ext4_set_aops(struct inode *inode); 2098 extern int ext4_writepage_trans_blocks(struct inode *); 2099 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2100 extern int ext4_discard_partial_page_buffers(handle_t *handle, 2101 struct address_space *mapping, loff_t from, 2102 loff_t length, int flags); 2103 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2104 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2105 extern void ext4_da_update_reserve_space(struct inode *inode, 2106 int used, int quota_claim); 2107 2108 /* indirect.c */ 2109 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2110 struct ext4_map_blocks *map, int flags); 2111 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2112 const struct iovec *iov, loff_t offset, 2113 unsigned long nr_segs); 2114 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2115 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk); 2116 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2117 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode, 2118 ext4_lblk_t first, ext4_lblk_t stop); 2119 2120 /* ioctl.c */ 2121 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2122 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2123 2124 /* migrate.c */ 2125 extern int ext4_ext_migrate(struct inode *); 2126 extern int ext4_ind_migrate(struct inode *inode); 2127 2128 /* namei.c */ 2129 extern int ext4_dirent_csum_verify(struct inode *inode, 2130 struct ext4_dir_entry *dirent); 2131 extern int ext4_orphan_add(handle_t *, struct inode *); 2132 extern int ext4_orphan_del(handle_t *, struct inode *); 2133 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2134 __u32 start_minor_hash, __u32 *next_hash); 2135 extern int search_dir(struct buffer_head *bh, 2136 char *search_buf, 2137 int buf_size, 2138 struct inode *dir, 2139 const struct qstr *d_name, 2140 unsigned int offset, 2141 struct ext4_dir_entry_2 **res_dir); 2142 extern int ext4_generic_delete_entry(handle_t *handle, 2143 struct inode *dir, 2144 struct ext4_dir_entry_2 *de_del, 2145 struct buffer_head *bh, 2146 void *entry_buf, 2147 int buf_size, 2148 int csum_size); 2149 2150 /* resize.c */ 2151 extern int ext4_group_add(struct super_block *sb, 2152 struct ext4_new_group_data *input); 2153 extern int ext4_group_extend(struct super_block *sb, 2154 struct ext4_super_block *es, 2155 ext4_fsblk_t n_blocks_count); 2156 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2157 2158 /* super.c */ 2159 extern int ext4_calculate_overhead(struct super_block *sb); 2160 extern int ext4_superblock_csum_verify(struct super_block *sb, 2161 struct ext4_super_block *es); 2162 extern void ext4_superblock_csum_set(struct super_block *sb); 2163 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2164 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2165 extern void ext4_kvfree(void *ptr); 2166 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2167 ext4_group_t ngroup); 2168 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2169 char nbuf[16]); 2170 extern __printf(4, 5) 2171 void __ext4_error(struct super_block *, const char *, unsigned int, 2172 const char *, ...); 2173 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \ 2174 __LINE__, ## message) 2175 extern __printf(5, 6) 2176 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2177 const char *, ...); 2178 extern __printf(5, 6) 2179 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2180 const char *, ...); 2181 extern void __ext4_std_error(struct super_block *, const char *, 2182 unsigned int, int); 2183 extern __printf(4, 5) 2184 void __ext4_abort(struct super_block *, const char *, unsigned int, 2185 const char *, ...); 2186 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \ 2187 __LINE__, ## message) 2188 extern __printf(4, 5) 2189 void __ext4_warning(struct super_block *, const char *, unsigned int, 2190 const char *, ...); 2191 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \ 2192 __LINE__, ## message) 2193 extern __printf(3, 4) 2194 void ext4_msg(struct super_block *, const char *, const char *, ...); 2195 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2196 const char *, unsigned int, const char *); 2197 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \ 2198 __LINE__, msg) 2199 extern __printf(7, 8) 2200 void __ext4_grp_locked_error(const char *, unsigned int, 2201 struct super_block *, ext4_group_t, 2202 unsigned long, ext4_fsblk_t, 2203 const char *, ...); 2204 #define ext4_grp_locked_error(sb, grp, message...) \ 2205 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message) 2206 extern void ext4_update_dynamic_rev(struct super_block *sb); 2207 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2208 __u32 compat); 2209 extern int ext4_update_rocompat_feature(handle_t *handle, 2210 struct super_block *sb, __u32 rocompat); 2211 extern int ext4_update_incompat_feature(handle_t *handle, 2212 struct super_block *sb, __u32 incompat); 2213 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2214 struct ext4_group_desc *bg); 2215 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2216 struct ext4_group_desc *bg); 2217 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2218 struct ext4_group_desc *bg); 2219 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2220 struct ext4_group_desc *bg); 2221 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2222 struct ext4_group_desc *bg); 2223 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2224 struct ext4_group_desc *bg); 2225 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2226 struct ext4_group_desc *bg); 2227 extern void ext4_block_bitmap_set(struct super_block *sb, 2228 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2229 extern void ext4_inode_bitmap_set(struct super_block *sb, 2230 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2231 extern void ext4_inode_table_set(struct super_block *sb, 2232 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2233 extern void ext4_free_group_clusters_set(struct super_block *sb, 2234 struct ext4_group_desc *bg, 2235 __u32 count); 2236 extern void ext4_free_inodes_set(struct super_block *sb, 2237 struct ext4_group_desc *bg, __u32 count); 2238 extern void ext4_used_dirs_set(struct super_block *sb, 2239 struct ext4_group_desc *bg, __u32 count); 2240 extern void ext4_itable_unused_set(struct super_block *sb, 2241 struct ext4_group_desc *bg, __u32 count); 2242 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2243 struct ext4_group_desc *gdp); 2244 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2245 struct ext4_group_desc *gdp); 2246 extern int ext4_register_li_request(struct super_block *sb, 2247 ext4_group_t first_not_zeroed); 2248 2249 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2250 { 2251 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2252 EXT4_FEATURE_RO_COMPAT_GDT_CSUM | 2253 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM); 2254 } 2255 2256 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2257 { 2258 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2259 le32_to_cpu(es->s_blocks_count_lo); 2260 } 2261 2262 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2263 { 2264 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2265 le32_to_cpu(es->s_r_blocks_count_lo); 2266 } 2267 2268 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2269 { 2270 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2271 le32_to_cpu(es->s_free_blocks_count_lo); 2272 } 2273 2274 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2275 ext4_fsblk_t blk) 2276 { 2277 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2278 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2279 } 2280 2281 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2282 ext4_fsblk_t blk) 2283 { 2284 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2285 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2286 } 2287 2288 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2289 ext4_fsblk_t blk) 2290 { 2291 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2292 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2293 } 2294 2295 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2296 { 2297 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2298 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2299 le32_to_cpu(raw_inode->i_size_lo); 2300 else 2301 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2302 } 2303 2304 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2305 { 2306 raw_inode->i_size_lo = cpu_to_le32(i_size); 2307 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2308 } 2309 2310 static inline 2311 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2312 ext4_group_t group) 2313 { 2314 struct ext4_group_info ***grp_info; 2315 long indexv, indexh; 2316 grp_info = EXT4_SB(sb)->s_group_info; 2317 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2318 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2319 return grp_info[indexv][indexh]; 2320 } 2321 2322 /* 2323 * Reading s_groups_count requires using smp_rmb() afterwards. See 2324 * the locking protocol documented in the comments of ext4_group_add() 2325 * in resize.c 2326 */ 2327 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2328 { 2329 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2330 2331 smp_rmb(); 2332 return ngroups; 2333 } 2334 2335 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2336 ext4_group_t block_group) 2337 { 2338 return block_group >> sbi->s_log_groups_per_flex; 2339 } 2340 2341 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2342 { 2343 return 1 << sbi->s_log_groups_per_flex; 2344 } 2345 2346 #define ext4_std_error(sb, errno) \ 2347 do { \ 2348 if ((errno)) \ 2349 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2350 } while (0) 2351 2352 #ifdef CONFIG_SMP 2353 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2354 * counters. So we need to make sure we have free clusters more 2355 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2356 */ 2357 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2358 #else 2359 #define EXT4_FREECLUSTERS_WATERMARK 0 2360 #endif 2361 2362 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2363 { 2364 /* 2365 * XXX: replace with spinlock if seen contended -bzzz 2366 */ 2367 down_write(&EXT4_I(inode)->i_data_sem); 2368 if (newsize > EXT4_I(inode)->i_disksize) 2369 EXT4_I(inode)->i_disksize = newsize; 2370 up_write(&EXT4_I(inode)->i_data_sem); 2371 return ; 2372 } 2373 2374 struct ext4_group_info { 2375 unsigned long bb_state; 2376 struct rb_root bb_free_root; 2377 ext4_grpblk_t bb_first_free; /* first free block */ 2378 ext4_grpblk_t bb_free; /* total free blocks */ 2379 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2380 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2381 struct list_head bb_prealloc_list; 2382 #ifdef DOUBLE_CHECK 2383 void *bb_bitmap; 2384 #endif 2385 struct rw_semaphore alloc_sem; 2386 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2387 * regions, index is order. 2388 * bb_counters[3] = 5 means 2389 * 5 free 8-block regions. */ 2390 }; 2391 2392 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2393 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2394 2395 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2396 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2397 2398 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2399 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2400 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2401 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2402 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2403 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2404 2405 #define EXT4_MAX_CONTENTION 8 2406 #define EXT4_CONTENTION_THRESHOLD 2 2407 2408 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2409 ext4_group_t group) 2410 { 2411 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2412 } 2413 2414 /* 2415 * Returns true if the filesystem is busy enough that attempts to 2416 * access the block group locks has run into contention. 2417 */ 2418 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2419 { 2420 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2421 } 2422 2423 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2424 { 2425 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2426 if (spin_trylock(lock)) 2427 /* 2428 * We're able to grab the lock right away, so drop the 2429 * lock contention counter. 2430 */ 2431 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2432 else { 2433 /* 2434 * The lock is busy, so bump the contention counter, 2435 * and then wait on the spin lock. 2436 */ 2437 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2438 EXT4_MAX_CONTENTION); 2439 spin_lock(lock); 2440 } 2441 } 2442 2443 static inline void ext4_unlock_group(struct super_block *sb, 2444 ext4_group_t group) 2445 { 2446 spin_unlock(ext4_group_lock_ptr(sb, group)); 2447 } 2448 2449 /* 2450 * Block validity checking 2451 */ 2452 #define ext4_check_indirect_blockref(inode, bh) \ 2453 ext4_check_blockref(__func__, __LINE__, inode, \ 2454 (__le32 *)(bh)->b_data, \ 2455 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2456 2457 #define ext4_ind_check_inode(inode) \ 2458 ext4_check_blockref(__func__, __LINE__, inode, \ 2459 EXT4_I(inode)->i_data, \ 2460 EXT4_NDIR_BLOCKS) 2461 2462 /* 2463 * Inodes and files operations 2464 */ 2465 2466 /* dir.c */ 2467 extern const struct file_operations ext4_dir_operations; 2468 2469 /* file.c */ 2470 extern const struct inode_operations ext4_file_inode_operations; 2471 extern const struct file_operations ext4_file_operations; 2472 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2473 extern void ext4_unwritten_wait(struct inode *inode); 2474 2475 /* inline.c */ 2476 extern int ext4_has_inline_data(struct inode *inode); 2477 extern int ext4_get_inline_size(struct inode *inode); 2478 extern int ext4_get_max_inline_size(struct inode *inode); 2479 extern int ext4_find_inline_data_nolock(struct inode *inode); 2480 extern void ext4_write_inline_data(struct inode *inode, 2481 struct ext4_iloc *iloc, 2482 void *buffer, loff_t pos, 2483 unsigned int len); 2484 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode, 2485 unsigned int len); 2486 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2487 unsigned int len); 2488 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2489 2490 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2491 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2492 struct inode *inode, 2493 loff_t pos, unsigned len, 2494 unsigned flags, 2495 struct page **pagep); 2496 extern int ext4_write_inline_data_end(struct inode *inode, 2497 loff_t pos, unsigned len, 2498 unsigned copied, 2499 struct page *page); 2500 extern struct buffer_head * 2501 ext4_journalled_write_inline_data(struct inode *inode, 2502 unsigned len, 2503 struct page *page); 2504 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2505 struct inode *inode, 2506 loff_t pos, unsigned len, 2507 unsigned flags, 2508 struct page **pagep, 2509 void **fsdata); 2510 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2511 unsigned len, unsigned copied, 2512 struct page *page); 2513 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry, 2514 struct inode *inode); 2515 extern int ext4_try_create_inline_dir(handle_t *handle, 2516 struct inode *parent, 2517 struct inode *inode); 2518 extern int ext4_read_inline_dir(struct file *filp, 2519 void *dirent, filldir_t filldir, 2520 int *has_inline_data); 2521 extern int htree_inlinedir_to_tree(struct file *dir_file, 2522 struct inode *dir, ext4_lblk_t block, 2523 struct dx_hash_info *hinfo, 2524 __u32 start_hash, __u32 start_minor_hash, 2525 int *has_inline_data); 2526 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2527 const struct qstr *d_name, 2528 struct ext4_dir_entry_2 **res_dir, 2529 int *has_inline_data); 2530 extern int ext4_delete_inline_entry(handle_t *handle, 2531 struct inode *dir, 2532 struct ext4_dir_entry_2 *de_del, 2533 struct buffer_head *bh, 2534 int *has_inline_data); 2535 extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 2536 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 2537 struct ext4_dir_entry_2 **parent_de, 2538 int *retval); 2539 extern int ext4_inline_data_fiemap(struct inode *inode, 2540 struct fiemap_extent_info *fieinfo, 2541 int *has_inline); 2542 extern int ext4_try_to_evict_inline_data(handle_t *handle, 2543 struct inode *inode, 2544 int needed); 2545 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 2546 2547 extern int ext4_convert_inline_data(struct inode *inode); 2548 2549 /* namei.c */ 2550 extern const struct inode_operations ext4_dir_inode_operations; 2551 extern const struct inode_operations ext4_special_inode_operations; 2552 extern struct dentry *ext4_get_parent(struct dentry *child); 2553 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 2554 struct ext4_dir_entry_2 *de, 2555 int blocksize, int csum_size, 2556 unsigned int parent_ino, int dotdot_real_len); 2557 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 2558 unsigned int blocksize); 2559 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 2560 struct inode *inode, 2561 struct buffer_head *bh); 2562 #define S_SHIFT 12 2563 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 2564 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 2565 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 2566 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 2567 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 2568 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 2569 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 2570 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 2571 }; 2572 2573 static inline void ext4_set_de_type(struct super_block *sb, 2574 struct ext4_dir_entry_2 *de, 2575 umode_t mode) { 2576 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE)) 2577 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 2578 } 2579 2580 2581 /* symlink.c */ 2582 extern const struct inode_operations ext4_symlink_inode_operations; 2583 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2584 2585 /* block_validity */ 2586 extern void ext4_release_system_zone(struct super_block *sb); 2587 extern int ext4_setup_system_zone(struct super_block *sb); 2588 extern int __init ext4_init_system_zone(void); 2589 extern void ext4_exit_system_zone(void); 2590 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2591 ext4_fsblk_t start_blk, 2592 unsigned int count); 2593 extern int ext4_check_blockref(const char *, unsigned int, 2594 struct inode *, __le32 *, unsigned int); 2595 2596 /* extents.c */ 2597 struct ext4_ext_path; 2598 struct ext4_extent; 2599 2600 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2601 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2602 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 2603 int chunk); 2604 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2605 struct ext4_map_blocks *map, int flags); 2606 extern void ext4_ext_truncate(handle_t *, struct inode *); 2607 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 2608 ext4_lblk_t end); 2609 extern void ext4_ext_init(struct super_block *); 2610 extern void ext4_ext_release(struct super_block *); 2611 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2612 loff_t len); 2613 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 2614 ssize_t len); 2615 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2616 struct ext4_map_blocks *map, int flags); 2617 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 2618 ext4_lblk_t lblocks); 2619 extern int ext4_extent_tree_init(handle_t *, struct inode *); 2620 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 2621 int num, 2622 struct ext4_ext_path *path); 2623 extern int ext4_can_extents_be_merged(struct inode *inode, 2624 struct ext4_extent *ex1, 2625 struct ext4_extent *ex2); 2626 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 2627 struct ext4_ext_path *, 2628 struct ext4_extent *, int); 2629 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t, 2630 struct ext4_ext_path *); 2631 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 2632 extern int ext4_ext_check_inode(struct inode *inode); 2633 extern int ext4_find_delalloc_range(struct inode *inode, 2634 ext4_lblk_t lblk_start, 2635 ext4_lblk_t lblk_end); 2636 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 2637 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2638 __u64 start, __u64 len); 2639 2640 2641 /* move_extent.c */ 2642 extern void ext4_double_down_write_data_sem(struct inode *first, 2643 struct inode *second); 2644 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 2645 struct inode *donor_inode); 2646 void ext4_inode_double_lock(struct inode *inode1, struct inode *inode2); 2647 void ext4_inode_double_unlock(struct inode *inode1, struct inode *inode2); 2648 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2649 __u64 start_orig, __u64 start_donor, 2650 __u64 len, __u64 *moved_len); 2651 2652 /* page-io.c */ 2653 extern int __init ext4_init_pageio(void); 2654 extern void ext4_exit_pageio(void); 2655 extern void ext4_ioend_shutdown(struct inode *); 2656 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2657 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 2658 extern int ext4_put_io_end(ext4_io_end_t *io_end); 2659 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 2660 extern void ext4_io_submit_init(struct ext4_io_submit *io, 2661 struct writeback_control *wbc); 2662 extern void ext4_end_io_work(struct work_struct *work); 2663 extern void ext4_io_submit(struct ext4_io_submit *io); 2664 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2665 struct page *page, 2666 int len, 2667 struct writeback_control *wbc); 2668 2669 /* mmp.c */ 2670 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2671 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp); 2672 extern int ext4_mmp_csum_verify(struct super_block *sb, 2673 struct mmp_struct *mmp); 2674 2675 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */ 2676 enum ext4_state_bits { 2677 BH_Uninit /* blocks are allocated but uninitialized on disk */ 2678 = BH_JBDPrivateStart, 2679 BH_AllocFromCluster, /* allocated blocks were part of already 2680 * allocated cluster. Note that this flag will 2681 * never, ever appear in a buffer_head's state 2682 * flag. See EXT4_MAP_FROM_CLUSTER to see where 2683 * this is used. */ 2684 }; 2685 2686 BUFFER_FNS(Uninit, uninit) 2687 TAS_BUFFER_FNS(Uninit, uninit) 2688 2689 /* 2690 * Add new method to test whether block and inode bitmaps are properly 2691 * initialized. With uninit_bg reading the block from disk is not enough 2692 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2693 */ 2694 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2695 2696 static inline int bitmap_uptodate(struct buffer_head *bh) 2697 { 2698 return (buffer_uptodate(bh) && 2699 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2700 } 2701 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2702 { 2703 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2704 } 2705 2706 /* 2707 * Disable DIO read nolock optimization, so new dioreaders will be forced 2708 * to grab i_mutex 2709 */ 2710 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2711 { 2712 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2713 smp_mb(); 2714 } 2715 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2716 { 2717 smp_mb(); 2718 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2719 } 2720 2721 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2722 2723 /* For ioend & aio unwritten conversion wait queues */ 2724 #define EXT4_WQ_HASH_SZ 37 2725 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2726 EXT4_WQ_HASH_SZ]) 2727 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2728 EXT4_WQ_HASH_SZ]) 2729 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2730 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2731 2732 #define EXT4_RESIZING 0 2733 extern int ext4_resize_begin(struct super_block *sb); 2734 extern void ext4_resize_end(struct super_block *sb); 2735 2736 #endif /* __KERNEL__ */ 2737 2738 #endif /* _EXT4_H */ 2739