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 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 61 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 62 63 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 64 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 65 66 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 67 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 68 69 /* data type for block offset of block group */ 70 typedef int ext4_grpblk_t; 71 72 /* data type for filesystem-wide blocks number */ 73 typedef unsigned long long ext4_fsblk_t; 74 75 /* data type for file logical block number */ 76 typedef __u32 ext4_lblk_t; 77 78 /* data type for block group number */ 79 typedef unsigned int ext4_group_t; 80 81 /* 82 * Flags used in mballoc's allocation_context flags field. 83 * 84 * Also used to show what's going on for debugging purposes when the 85 * flag field is exported via the traceport interface 86 */ 87 88 /* prefer goal again. length */ 89 #define EXT4_MB_HINT_MERGE 0x0001 90 /* blocks already reserved */ 91 #define EXT4_MB_HINT_RESERVED 0x0002 92 /* metadata is being allocated */ 93 #define EXT4_MB_HINT_METADATA 0x0004 94 /* first blocks in the file */ 95 #define EXT4_MB_HINT_FIRST 0x0008 96 /* search for the best chunk */ 97 #define EXT4_MB_HINT_BEST 0x0010 98 /* data is being allocated */ 99 #define EXT4_MB_HINT_DATA 0x0020 100 /* don't preallocate (for tails) */ 101 #define EXT4_MB_HINT_NOPREALLOC 0x0040 102 /* allocate for locality group */ 103 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 104 /* allocate goal blocks or none */ 105 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 106 /* goal is meaningful */ 107 #define EXT4_MB_HINT_TRY_GOAL 0x0200 108 /* blocks already pre-reserved by delayed allocation */ 109 #define EXT4_MB_DELALLOC_RESERVED 0x0400 110 /* We are doing stream allocation */ 111 #define EXT4_MB_STREAM_ALLOC 0x0800 112 /* Use reserved root blocks if needed */ 113 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 114 115 struct ext4_allocation_request { 116 /* target inode for block we're allocating */ 117 struct inode *inode; 118 /* how many blocks we want to allocate */ 119 unsigned int len; 120 /* logical block in target inode */ 121 ext4_lblk_t logical; 122 /* the closest logical allocated block to the left */ 123 ext4_lblk_t lleft; 124 /* the closest logical allocated block to the right */ 125 ext4_lblk_t lright; 126 /* phys. target (a hint) */ 127 ext4_fsblk_t goal; 128 /* phys. block for the closest logical allocated block to the left */ 129 ext4_fsblk_t pleft; 130 /* phys. block for the closest logical allocated block to the right */ 131 ext4_fsblk_t pright; 132 /* flags. see above EXT4_MB_HINT_* */ 133 unsigned int flags; 134 }; 135 136 /* 137 * Logical to physical block mapping, used by ext4_map_blocks() 138 * 139 * This structure is used to pass requests into ext4_map_blocks() as 140 * well as to store the information returned by ext4_map_blocks(). It 141 * takes less room on the stack than a struct buffer_head. 142 */ 143 #define EXT4_MAP_NEW (1 << BH_New) 144 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 145 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 146 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 147 #define EXT4_MAP_UNINIT (1 << BH_Uninit) 148 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of 149 * ext4_map_blocks wants to know whether or not the underlying cluster has 150 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that 151 * the requested mapping was from previously mapped (or delayed allocated) 152 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster 153 * should never appear on buffer_head's state flags. 154 */ 155 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster) 156 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 157 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 158 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER) 159 160 struct ext4_map_blocks { 161 ext4_fsblk_t m_pblk; 162 ext4_lblk_t m_lblk; 163 unsigned int m_len; 164 unsigned int m_flags; 165 }; 166 167 /* 168 * For delayed allocation tracking 169 */ 170 struct mpage_da_data { 171 struct inode *inode; 172 sector_t b_blocknr; /* start block number of extent */ 173 size_t b_size; /* size of extent */ 174 unsigned long b_state; /* state of the extent */ 175 unsigned long first_page, next_page; /* extent of pages */ 176 struct writeback_control *wbc; 177 int io_done; 178 int pages_written; 179 int retval; 180 }; 181 182 /* 183 * Flags for ext4_io_end->flags 184 */ 185 #define EXT4_IO_END_UNWRITTEN 0x0001 186 #define EXT4_IO_END_ERROR 0x0002 187 #define EXT4_IO_END_QUEUED 0x0004 188 #define EXT4_IO_END_DIRECT 0x0008 189 190 struct ext4_io_page { 191 struct page *p_page; 192 atomic_t p_count; 193 }; 194 195 #define MAX_IO_PAGES 128 196 197 /* 198 * For converting uninitialized extents on a work queue. 199 * 200 * 'page' is only used from the writepage() path; 'pages' is only used for 201 * buffered writes; they are used to keep page references until conversion 202 * takes place. For AIO/DIO, neither field is filled in. 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 struct page *page; /* for writepage() path */ 209 loff_t offset; /* offset in the file */ 210 ssize_t size; /* size of the extent */ 211 struct work_struct work; /* data work queue */ 212 struct kiocb *iocb; /* iocb struct for AIO */ 213 int result; /* error value for AIO */ 214 int num_io_pages; /* for writepages() */ 215 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */ 216 } ext4_io_end_t; 217 218 struct ext4_io_submit { 219 int io_op; 220 struct bio *io_bio; 221 ext4_io_end_t *io_end; 222 struct ext4_io_page *io_page; 223 sector_t io_next_block; 224 }; 225 226 /* 227 * Special inodes numbers 228 */ 229 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 230 #define EXT4_ROOT_INO 2 /* Root inode */ 231 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 232 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 233 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 234 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 235 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 236 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 237 238 /* First non-reserved inode for old ext4 filesystems */ 239 #define EXT4_GOOD_OLD_FIRST_INO 11 240 241 /* 242 * Maximal count of links to a file 243 */ 244 #define EXT4_LINK_MAX 65000 245 246 /* 247 * Macro-instructions used to manage several block sizes 248 */ 249 #define EXT4_MIN_BLOCK_SIZE 1024 250 #define EXT4_MAX_BLOCK_SIZE 65536 251 #define EXT4_MIN_BLOCK_LOG_SIZE 10 252 #define EXT4_MAX_BLOCK_LOG_SIZE 16 253 #ifdef __KERNEL__ 254 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 255 #else 256 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 257 #endif 258 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 259 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 260 EXT4_SB(s)->s_cluster_bits) 261 #ifdef __KERNEL__ 262 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 263 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 264 #else 265 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 266 #endif 267 #ifdef __KERNEL__ 268 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 269 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 270 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 271 #else 272 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 273 EXT4_GOOD_OLD_INODE_SIZE : \ 274 (s)->s_inode_size) 275 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 276 EXT4_GOOD_OLD_FIRST_INO : \ 277 (s)->s_first_ino) 278 #endif 279 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 280 281 /* Translate a block number to a cluster number */ 282 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 283 /* Translate a cluster number to a block number */ 284 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 285 /* Translate # of blks to # of clusters */ 286 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 287 (sbi)->s_cluster_bits) 288 289 /* 290 * Structure of a blocks group descriptor 291 */ 292 struct ext4_group_desc 293 { 294 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 295 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 296 __le32 bg_inode_table_lo; /* Inodes table block */ 297 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 298 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 299 __le16 bg_used_dirs_count_lo; /* Directories count */ 300 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 301 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 302 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 303 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 304 __le16 bg_itable_unused_lo; /* Unused inodes count */ 305 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 306 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 307 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 308 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 309 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 310 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 311 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 312 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 313 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 314 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 315 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 316 __u32 bg_reserved; 317 }; 318 319 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 320 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 321 sizeof(__le16)) 322 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 323 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 324 sizeof(__le16)) 325 326 /* 327 * Structure of a flex block group info 328 */ 329 330 struct flex_groups { 331 atomic_t free_inodes; 332 atomic_t free_clusters; 333 atomic_t used_dirs; 334 }; 335 336 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 337 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 338 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 339 340 /* 341 * Macro-instructions used to manage group descriptors 342 */ 343 #define EXT4_MIN_DESC_SIZE 32 344 #define EXT4_MIN_DESC_SIZE_64BIT 64 345 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 346 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 347 #ifdef __KERNEL__ 348 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 349 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 350 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 351 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 352 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 353 #else 354 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 356 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 357 #endif 358 359 /* 360 * Constants relative to the data blocks 361 */ 362 #define EXT4_NDIR_BLOCKS 12 363 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 364 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 365 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 366 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 367 368 /* 369 * Inode flags 370 */ 371 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 372 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 373 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 374 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 375 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 376 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 377 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 378 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 379 /* Reserved for compression usage... */ 380 #define EXT4_DIRTY_FL 0x00000100 381 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 382 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 383 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 384 /* End compression flags --- maybe not all used */ 385 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 386 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 387 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 388 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 389 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 390 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 391 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 392 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 393 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 394 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 395 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 396 397 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 398 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */ 399 400 /* Flags that should be inherited by new inodes from their parent. */ 401 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 402 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 403 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 404 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 405 406 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 407 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 408 409 /* Flags that are appropriate for non-directories/regular files. */ 410 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 411 412 /* Mask out flags that are inappropriate for the given type of inode. */ 413 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 414 { 415 if (S_ISDIR(mode)) 416 return flags; 417 else if (S_ISREG(mode)) 418 return flags & EXT4_REG_FLMASK; 419 else 420 return flags & EXT4_OTHER_FLMASK; 421 } 422 423 /* 424 * Inode flags used for atomic set/get 425 */ 426 enum { 427 EXT4_INODE_SECRM = 0, /* Secure deletion */ 428 EXT4_INODE_UNRM = 1, /* Undelete */ 429 EXT4_INODE_COMPR = 2, /* Compress file */ 430 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 431 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 432 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 433 EXT4_INODE_NODUMP = 6, /* do not dump file */ 434 EXT4_INODE_NOATIME = 7, /* do not update atime */ 435 /* Reserved for compression usage... */ 436 EXT4_INODE_DIRTY = 8, 437 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 438 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 439 EXT4_INODE_ECOMPR = 11, /* Compression error */ 440 /* End compression flags --- maybe not all used */ 441 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 442 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 443 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 444 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 445 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 446 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 447 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 448 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 449 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 450 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 451 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 452 }; 453 454 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 455 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \ 456 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \ 457 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); } 458 459 /* 460 * Since it's pretty easy to mix up bit numbers and hex values, and we 461 * can't do a compile-time test for ENUM values, we use a run-time 462 * test to make sure that EXT4_XXX_FL is consistent with respect to 463 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop 464 * out so it won't cost any extra space in the compiled kernel image. 465 * But it's important that these values are the same, since we are 466 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL 467 * must be consistent with the values of FS_XXX_FL defined in 468 * include/linux/fs.h and the on-disk values found in ext2, ext3, and 469 * ext4 filesystems, and of course the values defined in e2fsprogs. 470 * 471 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 472 */ 473 static inline void ext4_check_flag_values(void) 474 { 475 CHECK_FLAG_VALUE(SECRM); 476 CHECK_FLAG_VALUE(UNRM); 477 CHECK_FLAG_VALUE(COMPR); 478 CHECK_FLAG_VALUE(SYNC); 479 CHECK_FLAG_VALUE(IMMUTABLE); 480 CHECK_FLAG_VALUE(APPEND); 481 CHECK_FLAG_VALUE(NODUMP); 482 CHECK_FLAG_VALUE(NOATIME); 483 CHECK_FLAG_VALUE(DIRTY); 484 CHECK_FLAG_VALUE(COMPRBLK); 485 CHECK_FLAG_VALUE(NOCOMPR); 486 CHECK_FLAG_VALUE(ECOMPR); 487 CHECK_FLAG_VALUE(INDEX); 488 CHECK_FLAG_VALUE(IMAGIC); 489 CHECK_FLAG_VALUE(JOURNAL_DATA); 490 CHECK_FLAG_VALUE(NOTAIL); 491 CHECK_FLAG_VALUE(DIRSYNC); 492 CHECK_FLAG_VALUE(TOPDIR); 493 CHECK_FLAG_VALUE(HUGE_FILE); 494 CHECK_FLAG_VALUE(EXTENTS); 495 CHECK_FLAG_VALUE(EA_INODE); 496 CHECK_FLAG_VALUE(EOFBLOCKS); 497 CHECK_FLAG_VALUE(RESERVED); 498 } 499 500 /* Used to pass group descriptor data when online resize is done */ 501 struct ext4_new_group_input { 502 __u32 group; /* Group number for this data */ 503 __u64 block_bitmap; /* Absolute block number of block bitmap */ 504 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 505 __u64 inode_table; /* Absolute block number of inode table start */ 506 __u32 blocks_count; /* Total number of blocks in this group */ 507 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 508 __u16 unused; 509 }; 510 511 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 512 struct compat_ext4_new_group_input { 513 u32 group; 514 compat_u64 block_bitmap; 515 compat_u64 inode_bitmap; 516 compat_u64 inode_table; 517 u32 blocks_count; 518 u16 reserved_blocks; 519 u16 unused; 520 }; 521 #endif 522 523 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 524 struct ext4_new_group_data { 525 __u32 group; 526 __u64 block_bitmap; 527 __u64 inode_bitmap; 528 __u64 inode_table; 529 __u32 blocks_count; 530 __u16 reserved_blocks; 531 __u16 unused; 532 __u32 free_blocks_count; 533 }; 534 535 /* Indexes used to index group tables in ext4_new_group_data */ 536 enum { 537 BLOCK_BITMAP = 0, /* block bitmap */ 538 INODE_BITMAP, /* inode bitmap */ 539 INODE_TABLE, /* inode tables */ 540 GROUP_TABLE_COUNT, 541 }; 542 543 /* 544 * Flags used by ext4_map_blocks() 545 */ 546 /* Allocate any needed blocks and/or convert an unitialized 547 extent to be an initialized ext4 */ 548 #define EXT4_GET_BLOCKS_CREATE 0x0001 549 /* Request the creation of an unitialized extent */ 550 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 551 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 552 EXT4_GET_BLOCKS_CREATE) 553 /* Caller is from the delayed allocation writeout path, 554 so set the magic i_delalloc_reserve_flag after taking the 555 inode allocation semaphore for */ 556 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 557 /* caller is from the direct IO path, request to creation of an 558 unitialized extents if not allocated, split the uninitialized 559 extent if blocks has been preallocated already*/ 560 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 561 #define EXT4_GET_BLOCKS_CONVERT 0x0010 562 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 563 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 564 /* Convert extent to initialized after IO complete */ 565 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 566 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 567 /* Punch out blocks of an extent */ 568 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020 569 /* Don't normalize allocation size (used for fallocate) */ 570 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 571 /* Request will not result in inode size update (user for fallocate) */ 572 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 573 /* Do not take i_data_sem locking in ext4_map_blocks */ 574 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 575 576 /* 577 * Flags used by ext4_free_blocks 578 */ 579 #define EXT4_FREE_BLOCKS_METADATA 0x0001 580 #define EXT4_FREE_BLOCKS_FORGET 0x0002 581 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 582 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 583 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 584 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 585 586 /* 587 * Flags used by ext4_discard_partial_page_buffers 588 */ 589 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001 590 591 /* 592 * ioctl commands 593 */ 594 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 595 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 596 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 597 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 598 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 599 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 600 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 601 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 602 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 603 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 604 #define EXT4_IOC_MIGRATE _IO('f', 9) 605 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 606 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 607 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 608 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 609 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 610 611 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 612 /* 613 * ioctl commands in 32 bit emulation 614 */ 615 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 616 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 617 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 618 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 619 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 620 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 621 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 622 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 623 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 624 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 625 #endif 626 627 /* Max physical block we can address w/o extents */ 628 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 629 630 /* 631 * Structure of an inode on the disk 632 */ 633 struct ext4_inode { 634 __le16 i_mode; /* File mode */ 635 __le16 i_uid; /* Low 16 bits of Owner Uid */ 636 __le32 i_size_lo; /* Size in bytes */ 637 __le32 i_atime; /* Access time */ 638 __le32 i_ctime; /* Inode Change time */ 639 __le32 i_mtime; /* Modification time */ 640 __le32 i_dtime; /* Deletion Time */ 641 __le16 i_gid; /* Low 16 bits of Group Id */ 642 __le16 i_links_count; /* Links count */ 643 __le32 i_blocks_lo; /* Blocks count */ 644 __le32 i_flags; /* File flags */ 645 union { 646 struct { 647 __le32 l_i_version; 648 } linux1; 649 struct { 650 __u32 h_i_translator; 651 } hurd1; 652 struct { 653 __u32 m_i_reserved1; 654 } masix1; 655 } osd1; /* OS dependent 1 */ 656 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 657 __le32 i_generation; /* File version (for NFS) */ 658 __le32 i_file_acl_lo; /* File ACL */ 659 __le32 i_size_high; 660 __le32 i_obso_faddr; /* Obsoleted fragment address */ 661 union { 662 struct { 663 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 664 __le16 l_i_file_acl_high; 665 __le16 l_i_uid_high; /* these 2 fields */ 666 __le16 l_i_gid_high; /* were reserved2[0] */ 667 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 668 __le16 l_i_reserved; 669 } linux2; 670 struct { 671 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 672 __u16 h_i_mode_high; 673 __u16 h_i_uid_high; 674 __u16 h_i_gid_high; 675 __u32 h_i_author; 676 } hurd2; 677 struct { 678 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 679 __le16 m_i_file_acl_high; 680 __u32 m_i_reserved2[2]; 681 } masix2; 682 } osd2; /* OS dependent 2 */ 683 __le16 i_extra_isize; 684 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 685 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 686 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 687 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 688 __le32 i_crtime; /* File Creation time */ 689 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 690 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 691 }; 692 693 struct move_extent { 694 __u32 reserved; /* should be zero */ 695 __u32 donor_fd; /* donor file descriptor */ 696 __u64 orig_start; /* logical start offset in block for orig */ 697 __u64 donor_start; /* logical start offset in block for donor */ 698 __u64 len; /* block length to be moved */ 699 __u64 moved_len; /* moved block length */ 700 }; 701 702 #define EXT4_EPOCH_BITS 2 703 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 704 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 705 706 /* 707 * Extended fields will fit into an inode if the filesystem was formatted 708 * with large inodes (-I 256 or larger) and there are not currently any EAs 709 * consuming all of the available space. For new inodes we always reserve 710 * enough space for the kernel's known extended fields, but for inodes 711 * created with an old kernel this might not have been the case. None of 712 * the extended inode fields is critical for correct filesystem operation. 713 * This macro checks if a certain field fits in the inode. Note that 714 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 715 */ 716 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 717 ((offsetof(typeof(*ext4_inode), field) + \ 718 sizeof((ext4_inode)->field)) \ 719 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 720 (einode)->i_extra_isize)) \ 721 722 static inline __le32 ext4_encode_extra_time(struct timespec *time) 723 { 724 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 725 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 726 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 727 } 728 729 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 730 { 731 if (sizeof(time->tv_sec) > 4) 732 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 733 << 32; 734 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 735 } 736 737 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 738 do { \ 739 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 740 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 741 (raw_inode)->xtime ## _extra = \ 742 ext4_encode_extra_time(&(inode)->xtime); \ 743 } while (0) 744 745 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 746 do { \ 747 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 748 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 749 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 750 (raw_inode)->xtime ## _extra = \ 751 ext4_encode_extra_time(&(einode)->xtime); \ 752 } while (0) 753 754 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 755 do { \ 756 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 757 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 758 ext4_decode_extra_time(&(inode)->xtime, \ 759 raw_inode->xtime ## _extra); \ 760 else \ 761 (inode)->xtime.tv_nsec = 0; \ 762 } while (0) 763 764 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 765 do { \ 766 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 767 (einode)->xtime.tv_sec = \ 768 (signed)le32_to_cpu((raw_inode)->xtime); \ 769 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 770 ext4_decode_extra_time(&(einode)->xtime, \ 771 raw_inode->xtime ## _extra); \ 772 else \ 773 (einode)->xtime.tv_nsec = 0; \ 774 } while (0) 775 776 #define i_disk_version osd1.linux1.l_i_version 777 778 #if defined(__KERNEL__) || defined(__linux__) 779 #define i_reserved1 osd1.linux1.l_i_reserved1 780 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 781 #define i_blocks_high osd2.linux2.l_i_blocks_high 782 #define i_uid_low i_uid 783 #define i_gid_low i_gid 784 #define i_uid_high osd2.linux2.l_i_uid_high 785 #define i_gid_high osd2.linux2.l_i_gid_high 786 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 787 788 #elif defined(__GNU__) 789 790 #define i_translator osd1.hurd1.h_i_translator 791 #define i_uid_high osd2.hurd2.h_i_uid_high 792 #define i_gid_high osd2.hurd2.h_i_gid_high 793 #define i_author osd2.hurd2.h_i_author 794 795 #elif defined(__masix__) 796 797 #define i_reserved1 osd1.masix1.m_i_reserved1 798 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 799 #define i_reserved2 osd2.masix2.m_i_reserved2 800 801 #endif /* defined(__KERNEL__) || defined(__linux__) */ 802 803 /* 804 * storage for cached extent 805 * If ec_len == 0, then the cache is invalid. 806 * If ec_start == 0, then the cache represents a gap (null mapping) 807 */ 808 struct ext4_ext_cache { 809 ext4_fsblk_t ec_start; 810 ext4_lblk_t ec_block; 811 __u32 ec_len; /* must be 32bit to return holes */ 812 }; 813 814 #include "extents_status.h" 815 816 /* 817 * fourth extended file system inode data in memory 818 */ 819 struct ext4_inode_info { 820 __le32 i_data[15]; /* unconverted */ 821 __u32 i_dtime; 822 ext4_fsblk_t i_file_acl; 823 824 /* 825 * i_block_group is the number of the block group which contains 826 * this file's inode. Constant across the lifetime of the inode, 827 * it is ued for making block allocation decisions - we try to 828 * place a file's data blocks near its inode block, and new inodes 829 * near to their parent directory's inode. 830 */ 831 ext4_group_t i_block_group; 832 ext4_lblk_t i_dir_start_lookup; 833 #if (BITS_PER_LONG < 64) 834 unsigned long i_state_flags; /* Dynamic state flags */ 835 #endif 836 unsigned long i_flags; 837 838 #ifdef CONFIG_EXT4_FS_XATTR 839 /* 840 * Extended attributes can be read independently of the main file 841 * data. Taking i_mutex even when reading would cause contention 842 * between readers of EAs and writers of regular file data, so 843 * instead we synchronize on xattr_sem when reading or changing 844 * EAs. 845 */ 846 struct rw_semaphore xattr_sem; 847 #endif 848 849 struct list_head i_orphan; /* unlinked but open inodes */ 850 851 /* 852 * i_disksize keeps track of what the inode size is ON DISK, not 853 * in memory. During truncate, i_size is set to the new size by 854 * the VFS prior to calling ext4_truncate(), but the filesystem won't 855 * set i_disksize to 0 until the truncate is actually under way. 856 * 857 * The intent is that i_disksize always represents the blocks which 858 * are used by this file. This allows recovery to restart truncate 859 * on orphans if we crash during truncate. We actually write i_disksize 860 * into the on-disk inode when writing inodes out, instead of i_size. 861 * 862 * The only time when i_disksize and i_size may be different is when 863 * a truncate is in progress. The only things which change i_disksize 864 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 865 */ 866 loff_t i_disksize; 867 868 /* 869 * i_data_sem is for serialising ext4_truncate() against 870 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 871 * data tree are chopped off during truncate. We can't do that in 872 * ext4 because whenever we perform intermediate commits during 873 * truncate, the inode and all the metadata blocks *must* be in a 874 * consistent state which allows truncation of the orphans to restart 875 * during recovery. Hence we must fix the get_block-vs-truncate race 876 * by other means, so we have i_data_sem. 877 */ 878 struct rw_semaphore i_data_sem; 879 struct inode vfs_inode; 880 struct jbd2_inode *jinode; 881 882 struct ext4_ext_cache i_cached_extent; 883 /* 884 * File creation time. Its function is same as that of 885 * struct timespec i_{a,c,m}time in the generic inode. 886 */ 887 struct timespec i_crtime; 888 889 /* mballoc */ 890 struct list_head i_prealloc_list; 891 spinlock_t i_prealloc_lock; 892 893 /* extents status tree */ 894 struct ext4_es_tree i_es_tree; 895 rwlock_t i_es_lock; 896 897 /* ialloc */ 898 ext4_group_t i_last_alloc_group; 899 900 /* allocation reservation info for delalloc */ 901 /* In case of bigalloc, these refer to clusters rather than blocks */ 902 unsigned int i_reserved_data_blocks; 903 unsigned int i_reserved_meta_blocks; 904 unsigned int i_allocated_meta_blocks; 905 ext4_lblk_t i_da_metadata_calc_last_lblock; 906 int i_da_metadata_calc_len; 907 908 /* on-disk additional length */ 909 __u16 i_extra_isize; 910 911 #ifdef CONFIG_QUOTA 912 /* quota space reservation, managed internally by quota code */ 913 qsize_t i_reserved_quota; 914 #endif 915 916 /* completed IOs that might need unwritten extents handling */ 917 struct list_head i_completed_io_list; 918 spinlock_t i_completed_io_lock; 919 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 920 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 921 922 spinlock_t i_block_reservation_lock; 923 924 /* 925 * Transactions that contain inode's metadata needed to complete 926 * fsync and fdatasync, respectively. 927 */ 928 tid_t i_sync_tid; 929 tid_t i_datasync_tid; 930 931 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 932 __u32 i_csum_seed; 933 }; 934 935 /* 936 * File system states 937 */ 938 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 939 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 940 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 941 942 /* 943 * Misc. filesystem flags 944 */ 945 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 946 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 947 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 948 949 /* 950 * Mount flags 951 */ 952 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 953 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 954 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 955 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 956 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 957 #define EXT4_MOUNT_ERRORS_MASK 0x00070 958 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 959 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 960 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 961 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 962 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 963 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 964 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 965 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 966 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 967 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 968 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 969 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 970 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 971 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 972 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 973 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 974 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 975 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 976 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */ 977 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 978 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 979 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 980 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 981 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 982 983 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 984 specified delalloc */ 985 986 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 987 ~EXT4_MOUNT_##opt 988 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 989 EXT4_MOUNT_##opt 990 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 991 EXT4_MOUNT_##opt) 992 993 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 994 ~EXT4_MOUNT2_##opt 995 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 996 EXT4_MOUNT2_##opt 997 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 998 EXT4_MOUNT2_##opt) 999 1000 #define ext4_test_and_set_bit __test_and_set_bit_le 1001 #define ext4_set_bit __set_bit_le 1002 #define ext4_set_bit_atomic ext2_set_bit_atomic 1003 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1004 #define ext4_clear_bit __clear_bit_le 1005 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1006 #define ext4_test_bit test_bit_le 1007 #define ext4_find_next_zero_bit find_next_zero_bit_le 1008 #define ext4_find_next_bit find_next_bit_le 1009 1010 extern void ext4_set_bits(void *bm, int cur, int len); 1011 1012 /* 1013 * Maximal mount counts between two filesystem checks 1014 */ 1015 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1016 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1017 1018 /* 1019 * Behaviour when detecting errors 1020 */ 1021 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1022 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1023 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1024 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1025 1026 /* Metadata checksum algorithm codes */ 1027 #define EXT4_CRC32C_CHKSUM 1 1028 1029 /* 1030 * Structure of the super block 1031 */ 1032 struct ext4_super_block { 1033 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1034 __le32 s_blocks_count_lo; /* Blocks count */ 1035 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1036 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1037 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1038 __le32 s_first_data_block; /* First Data Block */ 1039 __le32 s_log_block_size; /* Block size */ 1040 __le32 s_log_cluster_size; /* Allocation cluster size */ 1041 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1042 __le32 s_clusters_per_group; /* # Clusters per group */ 1043 __le32 s_inodes_per_group; /* # Inodes per group */ 1044 __le32 s_mtime; /* Mount time */ 1045 /*30*/ __le32 s_wtime; /* Write time */ 1046 __le16 s_mnt_count; /* Mount count */ 1047 __le16 s_max_mnt_count; /* Maximal mount count */ 1048 __le16 s_magic; /* Magic signature */ 1049 __le16 s_state; /* File system state */ 1050 __le16 s_errors; /* Behaviour when detecting errors */ 1051 __le16 s_minor_rev_level; /* minor revision level */ 1052 /*40*/ __le32 s_lastcheck; /* time of last check */ 1053 __le32 s_checkinterval; /* max. time between checks */ 1054 __le32 s_creator_os; /* OS */ 1055 __le32 s_rev_level; /* Revision level */ 1056 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1057 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1058 /* 1059 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1060 * 1061 * Note: the difference between the compatible feature set and 1062 * the incompatible feature set is that if there is a bit set 1063 * in the incompatible feature set that the kernel doesn't 1064 * know about, it should refuse to mount the filesystem. 1065 * 1066 * e2fsck's requirements are more strict; if it doesn't know 1067 * about a feature in either the compatible or incompatible 1068 * feature set, it must abort and not try to meddle with 1069 * things it doesn't understand... 1070 */ 1071 __le32 s_first_ino; /* First non-reserved inode */ 1072 __le16 s_inode_size; /* size of inode structure */ 1073 __le16 s_block_group_nr; /* block group # of this superblock */ 1074 __le32 s_feature_compat; /* compatible feature set */ 1075 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1076 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1077 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1078 /*78*/ char s_volume_name[16]; /* volume name */ 1079 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1080 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1081 /* 1082 * Performance hints. Directory preallocation should only 1083 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1084 */ 1085 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1086 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1087 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1088 /* 1089 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1090 */ 1091 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1092 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1093 __le32 s_journal_dev; /* device number of journal file */ 1094 __le32 s_last_orphan; /* start of list of inodes to delete */ 1095 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1096 __u8 s_def_hash_version; /* Default hash version to use */ 1097 __u8 s_jnl_backup_type; 1098 __le16 s_desc_size; /* size of group descriptor */ 1099 /*100*/ __le32 s_default_mount_opts; 1100 __le32 s_first_meta_bg; /* First metablock block group */ 1101 __le32 s_mkfs_time; /* When the filesystem was created */ 1102 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1103 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1104 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1105 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1106 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1107 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1108 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1109 __le32 s_flags; /* Miscellaneous flags */ 1110 __le16 s_raid_stride; /* RAID stride */ 1111 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1112 __le64 s_mmp_block; /* Block for multi-mount protection */ 1113 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1114 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1115 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1116 __le16 s_reserved_pad; 1117 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1118 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1119 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1120 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1121 snapshot's future use */ 1122 __le32 s_snapshot_list; /* inode number of the head of the 1123 on-disk snapshot list */ 1124 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1125 __le32 s_error_count; /* number of fs errors */ 1126 __le32 s_first_error_time; /* first time an error happened */ 1127 __le32 s_first_error_ino; /* inode involved in first error */ 1128 __le64 s_first_error_block; /* block involved of first error */ 1129 __u8 s_first_error_func[32]; /* function where the error happened */ 1130 __le32 s_first_error_line; /* line number where error happened */ 1131 __le32 s_last_error_time; /* most recent time of an error */ 1132 __le32 s_last_error_ino; /* inode involved in last error */ 1133 __le32 s_last_error_line; /* line number where error happened */ 1134 __le64 s_last_error_block; /* block involved of last error */ 1135 __u8 s_last_error_func[32]; /* function where the error happened */ 1136 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1137 __u8 s_mount_opts[64]; 1138 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1139 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1140 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1141 __le32 s_reserved[108]; /* Padding to the end of the block */ 1142 __le32 s_checksum; /* crc32c(superblock) */ 1143 }; 1144 1145 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1146 1147 #ifdef __KERNEL__ 1148 1149 /* 1150 * run-time mount flags 1151 */ 1152 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1153 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1154 1155 /* 1156 * fourth extended-fs super-block data in memory 1157 */ 1158 struct ext4_sb_info { 1159 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1160 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1161 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1162 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1163 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1164 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1165 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1166 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1167 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1168 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1169 unsigned long s_overhead; /* # of fs overhead clusters */ 1170 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1171 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1172 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1173 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1174 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1175 struct buffer_head **s_group_desc; 1176 unsigned int s_mount_opt; 1177 unsigned int s_mount_opt2; 1178 unsigned int s_mount_flags; 1179 unsigned int s_def_mount_opt; 1180 ext4_fsblk_t s_sb_block; 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 1309 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1310 { 1311 return sb->s_fs_info; 1312 } 1313 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1314 { 1315 return container_of(inode, struct ext4_inode_info, vfs_inode); 1316 } 1317 1318 static inline struct timespec ext4_current_time(struct inode *inode) 1319 { 1320 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1321 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1322 } 1323 1324 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1325 { 1326 return ino == EXT4_ROOT_INO || 1327 ino == EXT4_USR_QUOTA_INO || 1328 ino == EXT4_GRP_QUOTA_INO || 1329 ino == EXT4_JOURNAL_INO || 1330 ino == EXT4_RESIZE_INO || 1331 (ino >= EXT4_FIRST_INO(sb) && 1332 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1333 } 1334 1335 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1336 struct ext4_io_end *io_end) 1337 { 1338 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1339 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1340 atomic_inc(&EXT4_I(inode)->i_unwritten); 1341 } 1342 } 1343 1344 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1345 { 1346 return inode->i_private; 1347 } 1348 1349 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1350 { 1351 inode->i_private = io; 1352 } 1353 1354 /* 1355 * Inode dynamic state flags 1356 */ 1357 enum { 1358 EXT4_STATE_JDATA, /* journaled data exists */ 1359 EXT4_STATE_NEW, /* inode is newly created */ 1360 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1361 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1362 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1363 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1364 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1365 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1366 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */ 1367 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1368 nolocking */ 1369 }; 1370 1371 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1372 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1373 { \ 1374 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1375 } \ 1376 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1377 { \ 1378 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1379 } \ 1380 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1381 { \ 1382 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1383 } 1384 1385 EXT4_INODE_BIT_FNS(flag, flags, 0) 1386 #if (BITS_PER_LONG < 64) 1387 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1388 1389 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1390 { 1391 (ei)->i_state_flags = 0; 1392 } 1393 #else 1394 EXT4_INODE_BIT_FNS(state, flags, 32) 1395 1396 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1397 { 1398 /* We depend on the fact that callers will set i_flags */ 1399 } 1400 #endif 1401 #else 1402 /* Assume that user mode programs are passing in an ext4fs superblock, not 1403 * a kernel struct super_block. This will allow us to call the feature-test 1404 * macros from user land. */ 1405 #define EXT4_SB(sb) (sb) 1406 #endif 1407 1408 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1409 1410 /* 1411 * Codes for operating systems 1412 */ 1413 #define EXT4_OS_LINUX 0 1414 #define EXT4_OS_HURD 1 1415 #define EXT4_OS_MASIX 2 1416 #define EXT4_OS_FREEBSD 3 1417 #define EXT4_OS_LITES 4 1418 1419 /* 1420 * Revision levels 1421 */ 1422 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1423 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1424 1425 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1426 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1427 1428 #define EXT4_GOOD_OLD_INODE_SIZE 128 1429 1430 /* 1431 * Feature set definitions 1432 */ 1433 1434 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1435 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1436 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1437 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1438 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1439 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1440 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1441 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1442 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1443 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1444 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1445 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1446 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1447 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1448 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1449 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1450 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1451 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1452 1453 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1454 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1455 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1456 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1457 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1458 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1459 1460 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1461 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1462 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1463 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1464 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1465 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1466 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1467 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1468 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1469 /* 1470 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1471 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1472 * all other data structures' checksums. However, the METADATA_CSUM and 1473 * GDT_CSUM bits are mutually exclusive. 1474 */ 1475 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1476 1477 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1478 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1479 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1480 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1481 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1482 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1483 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1484 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1485 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1486 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1487 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1488 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1489 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1490 #define EXT4_FEATURE_INCOMPAT_INLINEDATA 0x8000 /* data in inode */ 1491 1492 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1493 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1494 EXT4_FEATURE_INCOMPAT_META_BG) 1495 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1496 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1497 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1498 1499 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1500 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1501 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1502 EXT4_FEATURE_INCOMPAT_META_BG) 1503 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1504 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1505 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1506 1507 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1508 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1509 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1510 EXT4_FEATURE_INCOMPAT_META_BG| \ 1511 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1512 EXT4_FEATURE_INCOMPAT_64BIT| \ 1513 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1514 EXT4_FEATURE_INCOMPAT_MMP) 1515 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1516 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1517 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1518 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1519 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1520 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1521 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1522 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1523 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1524 EXT4_FEATURE_RO_COMPAT_QUOTA) 1525 1526 /* 1527 * Default values for user and/or group using reserved blocks 1528 */ 1529 #define EXT4_DEF_RESUID 0 1530 #define EXT4_DEF_RESGID 0 1531 1532 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1533 1534 /* 1535 * Default mount options 1536 */ 1537 #define EXT4_DEFM_DEBUG 0x0001 1538 #define EXT4_DEFM_BSDGROUPS 0x0002 1539 #define EXT4_DEFM_XATTR_USER 0x0004 1540 #define EXT4_DEFM_ACL 0x0008 1541 #define EXT4_DEFM_UID16 0x0010 1542 #define EXT4_DEFM_JMODE 0x0060 1543 #define EXT4_DEFM_JMODE_DATA 0x0020 1544 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1545 #define EXT4_DEFM_JMODE_WBACK 0x0060 1546 #define EXT4_DEFM_NOBARRIER 0x0100 1547 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1548 #define EXT4_DEFM_DISCARD 0x0400 1549 #define EXT4_DEFM_NODELALLOC 0x0800 1550 1551 /* 1552 * Default journal batch times 1553 */ 1554 #define EXT4_DEF_MIN_BATCH_TIME 0 1555 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1556 1557 /* 1558 * Minimum number of groups in a flexgroup before we separate out 1559 * directories into the first block group of a flexgroup 1560 */ 1561 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1562 1563 /* 1564 * Structure of a directory entry 1565 */ 1566 #define EXT4_NAME_LEN 255 1567 1568 struct ext4_dir_entry { 1569 __le32 inode; /* Inode number */ 1570 __le16 rec_len; /* Directory entry length */ 1571 __le16 name_len; /* Name length */ 1572 char name[EXT4_NAME_LEN]; /* File name */ 1573 }; 1574 1575 /* 1576 * The new version of the directory entry. Since EXT4 structures are 1577 * stored in intel byte order, and the name_len field could never be 1578 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1579 * file_type field. 1580 */ 1581 struct ext4_dir_entry_2 { 1582 __le32 inode; /* Inode number */ 1583 __le16 rec_len; /* Directory entry length */ 1584 __u8 name_len; /* Name length */ 1585 __u8 file_type; 1586 char name[EXT4_NAME_LEN]; /* File name */ 1587 }; 1588 1589 /* 1590 * This is a bogus directory entry at the end of each leaf block that 1591 * records checksums. 1592 */ 1593 struct ext4_dir_entry_tail { 1594 __le32 det_reserved_zero1; /* Pretend to be unused */ 1595 __le16 det_rec_len; /* 12 */ 1596 __u8 det_reserved_zero2; /* Zero name length */ 1597 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1598 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1599 }; 1600 1601 /* 1602 * Ext4 directory file types. Only the low 3 bits are used. The 1603 * other bits are reserved for now. 1604 */ 1605 #define EXT4_FT_UNKNOWN 0 1606 #define EXT4_FT_REG_FILE 1 1607 #define EXT4_FT_DIR 2 1608 #define EXT4_FT_CHRDEV 3 1609 #define EXT4_FT_BLKDEV 4 1610 #define EXT4_FT_FIFO 5 1611 #define EXT4_FT_SOCK 6 1612 #define EXT4_FT_SYMLINK 7 1613 1614 #define EXT4_FT_MAX 8 1615 1616 #define EXT4_FT_DIR_CSUM 0xDE 1617 1618 /* 1619 * EXT4_DIR_PAD defines the directory entries boundaries 1620 * 1621 * NOTE: It must be a multiple of 4 1622 */ 1623 #define EXT4_DIR_PAD 4 1624 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1625 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1626 ~EXT4_DIR_ROUND) 1627 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1628 1629 /* 1630 * If we ever get support for fs block sizes > page_size, we'll need 1631 * to remove the #if statements in the next two functions... 1632 */ 1633 static inline unsigned int 1634 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1635 { 1636 unsigned len = le16_to_cpu(dlen); 1637 1638 #if (PAGE_CACHE_SIZE >= 65536) 1639 if (len == EXT4_MAX_REC_LEN || len == 0) 1640 return blocksize; 1641 return (len & 65532) | ((len & 3) << 16); 1642 #else 1643 return len; 1644 #endif 1645 } 1646 1647 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1648 { 1649 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1650 BUG(); 1651 #if (PAGE_CACHE_SIZE >= 65536) 1652 if (len < 65536) 1653 return cpu_to_le16(len); 1654 if (len == blocksize) { 1655 if (blocksize == 65536) 1656 return cpu_to_le16(EXT4_MAX_REC_LEN); 1657 else 1658 return cpu_to_le16(0); 1659 } 1660 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1661 #else 1662 return cpu_to_le16(len); 1663 #endif 1664 } 1665 1666 /* 1667 * Hash Tree Directory indexing 1668 * (c) Daniel Phillips, 2001 1669 */ 1670 1671 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1672 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1673 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1674 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1675 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1676 1677 /* Legal values for the dx_root hash_version field: */ 1678 1679 #define DX_HASH_LEGACY 0 1680 #define DX_HASH_HALF_MD4 1 1681 #define DX_HASH_TEA 2 1682 #define DX_HASH_LEGACY_UNSIGNED 3 1683 #define DX_HASH_HALF_MD4_UNSIGNED 4 1684 #define DX_HASH_TEA_UNSIGNED 5 1685 1686 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1687 const void *address, unsigned int length) 1688 { 1689 struct { 1690 struct shash_desc shash; 1691 char ctx[4]; 1692 } desc; 1693 int err; 1694 1695 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1696 1697 desc.shash.tfm = sbi->s_chksum_driver; 1698 desc.shash.flags = 0; 1699 *(u32 *)desc.ctx = crc; 1700 1701 err = crypto_shash_update(&desc.shash, address, length); 1702 BUG_ON(err); 1703 1704 return *(u32 *)desc.ctx; 1705 } 1706 1707 #ifdef __KERNEL__ 1708 1709 /* hash info structure used by the directory hash */ 1710 struct dx_hash_info 1711 { 1712 u32 hash; 1713 u32 minor_hash; 1714 int hash_version; 1715 u32 *seed; 1716 }; 1717 1718 1719 /* 32 and 64 bit signed EOF for dx directories */ 1720 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1721 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1722 1723 1724 /* 1725 * Control parameters used by ext4_htree_next_block 1726 */ 1727 #define HASH_NB_ALWAYS 1 1728 1729 1730 /* 1731 * Describe an inode's exact location on disk and in memory 1732 */ 1733 struct ext4_iloc 1734 { 1735 struct buffer_head *bh; 1736 unsigned long offset; 1737 ext4_group_t block_group; 1738 }; 1739 1740 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1741 { 1742 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1743 } 1744 1745 /* 1746 * This structure is stuffed into the struct file's private_data field 1747 * for directories. It is where we put information so that we can do 1748 * readdir operations in hash tree order. 1749 */ 1750 struct dir_private_info { 1751 struct rb_root root; 1752 struct rb_node *curr_node; 1753 struct fname *extra_fname; 1754 loff_t last_pos; 1755 __u32 curr_hash; 1756 __u32 curr_minor_hash; 1757 __u32 next_hash; 1758 }; 1759 1760 /* calculate the first block number of the group */ 1761 static inline ext4_fsblk_t 1762 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1763 { 1764 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1765 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1766 } 1767 1768 /* 1769 * Special error return code only used by dx_probe() and its callers. 1770 */ 1771 #define ERR_BAD_DX_DIR -75000 1772 1773 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr, 1774 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp); 1775 1776 /* 1777 * Timeout and state flag for lazy initialization inode thread. 1778 */ 1779 #define EXT4_DEF_LI_WAIT_MULT 10 1780 #define EXT4_DEF_LI_MAX_START_DELAY 5 1781 #define EXT4_LAZYINIT_QUIT 0x0001 1782 #define EXT4_LAZYINIT_RUNNING 0x0002 1783 1784 /* 1785 * Lazy inode table initialization info 1786 */ 1787 struct ext4_lazy_init { 1788 unsigned long li_state; 1789 struct list_head li_request_list; 1790 struct mutex li_list_mtx; 1791 }; 1792 1793 struct ext4_li_request { 1794 struct super_block *lr_super; 1795 struct ext4_sb_info *lr_sbi; 1796 ext4_group_t lr_next_group; 1797 struct list_head lr_request; 1798 unsigned long lr_next_sched; 1799 unsigned long lr_timeout; 1800 }; 1801 1802 struct ext4_features { 1803 struct kobject f_kobj; 1804 struct completion f_kobj_unregister; 1805 }; 1806 1807 /* 1808 * This structure will be used for multiple mount protection. It will be 1809 * written into the block number saved in the s_mmp_block field in the 1810 * superblock. Programs that check MMP should assume that if 1811 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1812 * to use the filesystem, regardless of how old the timestamp is. 1813 */ 1814 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1815 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1816 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1817 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1818 1819 struct mmp_struct { 1820 __le32 mmp_magic; /* Magic number for MMP */ 1821 __le32 mmp_seq; /* Sequence no. updated periodically */ 1822 1823 /* 1824 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1825 * purposes and do not affect the correctness of the algorithm 1826 */ 1827 __le64 mmp_time; /* Time last updated */ 1828 char mmp_nodename[64]; /* Node which last updated MMP block */ 1829 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1830 1831 /* 1832 * mmp_check_interval is used to verify if the MMP block has been 1833 * updated on the block device. The value is updated based on the 1834 * maximum time to write the MMP block during an update cycle. 1835 */ 1836 __le16 mmp_check_interval; 1837 1838 __le16 mmp_pad1; 1839 __le32 mmp_pad2[226]; 1840 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1841 }; 1842 1843 /* arguments passed to the mmp thread */ 1844 struct mmpd_data { 1845 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1846 struct super_block *sb; /* super block of the fs */ 1847 }; 1848 1849 /* 1850 * Check interval multiplier 1851 * The MMP block is written every update interval and initially checked every 1852 * update interval x the multiplier (the value is then adapted based on the 1853 * write latency). The reason is that writes can be delayed under load and we 1854 * don't want readers to incorrectly assume that the filesystem is no longer 1855 * in use. 1856 */ 1857 #define EXT4_MMP_CHECK_MULT 2UL 1858 1859 /* 1860 * Minimum interval for MMP checking in seconds. 1861 */ 1862 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1863 1864 /* 1865 * Maximum interval for MMP checking in seconds. 1866 */ 1867 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1868 1869 /* 1870 * Function prototypes 1871 */ 1872 1873 /* 1874 * Ok, these declarations are also in <linux/kernel.h> but none of the 1875 * ext4 source programs needs to include it so they are duplicated here. 1876 */ 1877 # define NORET_TYPE /**/ 1878 # define ATTRIB_NORET __attribute__((noreturn)) 1879 # define NORET_AND noreturn, 1880 1881 /* bitmap.c */ 1882 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1883 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1884 struct ext4_group_desc *gdp, 1885 struct buffer_head *bh, int sz); 1886 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1887 struct ext4_group_desc *gdp, 1888 struct buffer_head *bh, int sz); 1889 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1890 struct ext4_group_desc *gdp, 1891 struct buffer_head *bh); 1892 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1893 struct ext4_group_desc *gdp, 1894 struct buffer_head *bh); 1895 1896 /* balloc.c */ 1897 extern void ext4_validate_block_bitmap(struct super_block *sb, 1898 struct ext4_group_desc *desc, 1899 unsigned int block_group, 1900 struct buffer_head *bh); 1901 extern unsigned int ext4_block_group(struct super_block *sb, 1902 ext4_fsblk_t blocknr); 1903 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1904 ext4_fsblk_t blocknr); 1905 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1906 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1907 ext4_group_t group); 1908 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1909 ext4_fsblk_t goal, 1910 unsigned int flags, 1911 unsigned long *count, 1912 int *errp); 1913 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1914 s64 nclusters, unsigned int flags); 1915 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1916 extern void ext4_check_blocks_bitmap(struct super_block *); 1917 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1918 ext4_group_t block_group, 1919 struct buffer_head ** bh); 1920 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1921 1922 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1923 ext4_group_t block_group); 1924 extern int ext4_wait_block_bitmap(struct super_block *sb, 1925 ext4_group_t block_group, 1926 struct buffer_head *bh); 1927 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1928 ext4_group_t block_group); 1929 extern void ext4_init_block_bitmap(struct super_block *sb, 1930 struct buffer_head *bh, 1931 ext4_group_t group, 1932 struct ext4_group_desc *desc); 1933 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1934 ext4_group_t block_group, 1935 struct ext4_group_desc *gdp); 1936 extern unsigned ext4_num_overhead_clusters(struct super_block *sb, 1937 ext4_group_t block_group, 1938 struct ext4_group_desc *gdp); 1939 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1940 1941 /* dir.c */ 1942 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1943 struct file *, 1944 struct ext4_dir_entry_2 *, 1945 struct buffer_head *, unsigned int); 1946 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \ 1947 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 1948 (de), (bh), (offset))) 1949 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1950 __u32 minor_hash, 1951 struct ext4_dir_entry_2 *dirent); 1952 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1953 1954 /* fsync.c */ 1955 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 1956 extern int ext4_flush_unwritten_io(struct inode *); 1957 1958 /* hash.c */ 1959 extern int ext4fs_dirhash(const char *name, int len, struct 1960 dx_hash_info *hinfo); 1961 1962 /* ialloc.c */ 1963 extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t, 1964 const struct qstr *qstr, __u32 goal, 1965 uid_t *owner); 1966 extern void ext4_free_inode(handle_t *, struct inode *); 1967 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 1968 extern unsigned long ext4_count_free_inodes(struct super_block *); 1969 extern unsigned long ext4_count_dirs(struct super_block *); 1970 extern void ext4_check_inodes_bitmap(struct super_block *); 1971 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 1972 extern int ext4_init_inode_table(struct super_block *sb, 1973 ext4_group_t group, int barrier); 1974 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 1975 1976 /* mballoc.c */ 1977 extern long ext4_mb_stats; 1978 extern long ext4_mb_max_to_scan; 1979 extern int ext4_mb_init(struct super_block *); 1980 extern int ext4_mb_release(struct super_block *); 1981 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 1982 struct ext4_allocation_request *, int *); 1983 extern int ext4_mb_reserve_blocks(struct super_block *, int); 1984 extern void ext4_discard_preallocations(struct inode *); 1985 extern int __init ext4_init_mballoc(void); 1986 extern void ext4_exit_mballoc(void); 1987 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 1988 struct buffer_head *bh, ext4_fsblk_t block, 1989 unsigned long count, int flags); 1990 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 1991 ext4_group_t ngroups); 1992 extern int ext4_mb_add_groupinfo(struct super_block *sb, 1993 ext4_group_t i, struct ext4_group_desc *desc); 1994 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 1995 ext4_fsblk_t block, unsigned long count); 1996 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 1997 1998 /* inode.c */ 1999 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 2000 ext4_lblk_t, int, int *); 2001 struct buffer_head *ext4_bread(handle_t *, struct inode *, 2002 ext4_lblk_t, int, int *); 2003 int ext4_get_block(struct inode *inode, sector_t iblock, 2004 struct buffer_head *bh_result, int create); 2005 2006 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2007 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2008 extern int ext4_setattr(struct dentry *, struct iattr *); 2009 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2010 struct kstat *stat); 2011 extern void ext4_evict_inode(struct inode *); 2012 extern void ext4_clear_inode(struct inode *); 2013 extern int ext4_sync_inode(handle_t *, struct inode *); 2014 extern void ext4_dirty_inode(struct inode *, int); 2015 extern int ext4_change_inode_journal_flag(struct inode *, int); 2016 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2017 extern int ext4_can_truncate(struct inode *inode); 2018 extern void ext4_truncate(struct inode *); 2019 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2020 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2021 extern void ext4_set_inode_flags(struct inode *); 2022 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2023 extern int ext4_alloc_da_blocks(struct inode *inode); 2024 extern void ext4_set_aops(struct inode *inode); 2025 extern int ext4_writepage_trans_blocks(struct inode *); 2026 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2027 extern int ext4_discard_partial_page_buffers(handle_t *handle, 2028 struct address_space *mapping, loff_t from, 2029 loff_t length, int flags); 2030 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2031 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2032 extern void ext4_da_update_reserve_space(struct inode *inode, 2033 int used, int quota_claim); 2034 2035 /* indirect.c */ 2036 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2037 struct ext4_map_blocks *map, int flags); 2038 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2039 const struct iovec *iov, loff_t offset, 2040 unsigned long nr_segs); 2041 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2042 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk); 2043 extern void ext4_ind_truncate(struct inode *inode); 2044 2045 /* ioctl.c */ 2046 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2047 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2048 2049 /* migrate.c */ 2050 extern int ext4_ext_migrate(struct inode *); 2051 2052 /* namei.c */ 2053 extern int ext4_dirent_csum_verify(struct inode *inode, 2054 struct ext4_dir_entry *dirent); 2055 extern int ext4_orphan_add(handle_t *, struct inode *); 2056 extern int ext4_orphan_del(handle_t *, struct inode *); 2057 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2058 __u32 start_minor_hash, __u32 *next_hash); 2059 2060 /* resize.c */ 2061 extern int ext4_group_add(struct super_block *sb, 2062 struct ext4_new_group_data *input); 2063 extern int ext4_group_extend(struct super_block *sb, 2064 struct ext4_super_block *es, 2065 ext4_fsblk_t n_blocks_count); 2066 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2067 2068 /* super.c */ 2069 extern int ext4_calculate_overhead(struct super_block *sb); 2070 extern int ext4_superblock_csum_verify(struct super_block *sb, 2071 struct ext4_super_block *es); 2072 extern void ext4_superblock_csum_set(struct super_block *sb); 2073 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2074 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2075 extern void ext4_kvfree(void *ptr); 2076 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2077 ext4_group_t ngroup); 2078 extern __printf(4, 5) 2079 void __ext4_error(struct super_block *, const char *, unsigned int, 2080 const char *, ...); 2081 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \ 2082 __LINE__, ## message) 2083 extern __printf(5, 6) 2084 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2085 const char *, ...); 2086 extern __printf(5, 6) 2087 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2088 const char *, ...); 2089 extern void __ext4_std_error(struct super_block *, const char *, 2090 unsigned int, int); 2091 extern __printf(4, 5) 2092 void __ext4_abort(struct super_block *, const char *, unsigned int, 2093 const char *, ...); 2094 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \ 2095 __LINE__, ## message) 2096 extern __printf(4, 5) 2097 void __ext4_warning(struct super_block *, const char *, unsigned int, 2098 const char *, ...); 2099 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \ 2100 __LINE__, ## message) 2101 extern __printf(3, 4) 2102 void ext4_msg(struct super_block *, const char *, const char *, ...); 2103 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2104 const char *, unsigned int, const char *); 2105 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \ 2106 __LINE__, msg) 2107 extern __printf(7, 8) 2108 void __ext4_grp_locked_error(const char *, unsigned int, 2109 struct super_block *, ext4_group_t, 2110 unsigned long, ext4_fsblk_t, 2111 const char *, ...); 2112 #define ext4_grp_locked_error(sb, grp, message...) \ 2113 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message) 2114 extern void ext4_update_dynamic_rev(struct super_block *sb); 2115 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2116 __u32 compat); 2117 extern int ext4_update_rocompat_feature(handle_t *handle, 2118 struct super_block *sb, __u32 rocompat); 2119 extern int ext4_update_incompat_feature(handle_t *handle, 2120 struct super_block *sb, __u32 incompat); 2121 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2122 struct ext4_group_desc *bg); 2123 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2124 struct ext4_group_desc *bg); 2125 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2126 struct ext4_group_desc *bg); 2127 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2128 struct ext4_group_desc *bg); 2129 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2130 struct ext4_group_desc *bg); 2131 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2132 struct ext4_group_desc *bg); 2133 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2134 struct ext4_group_desc *bg); 2135 extern void ext4_block_bitmap_set(struct super_block *sb, 2136 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2137 extern void ext4_inode_bitmap_set(struct super_block *sb, 2138 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2139 extern void ext4_inode_table_set(struct super_block *sb, 2140 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2141 extern void ext4_free_group_clusters_set(struct super_block *sb, 2142 struct ext4_group_desc *bg, 2143 __u32 count); 2144 extern void ext4_free_inodes_set(struct super_block *sb, 2145 struct ext4_group_desc *bg, __u32 count); 2146 extern void ext4_used_dirs_set(struct super_block *sb, 2147 struct ext4_group_desc *bg, __u32 count); 2148 extern void ext4_itable_unused_set(struct super_block *sb, 2149 struct ext4_group_desc *bg, __u32 count); 2150 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2151 struct ext4_group_desc *gdp); 2152 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2153 struct ext4_group_desc *gdp); 2154 2155 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2156 { 2157 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2158 EXT4_FEATURE_RO_COMPAT_GDT_CSUM | 2159 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM); 2160 } 2161 2162 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2163 { 2164 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2165 le32_to_cpu(es->s_blocks_count_lo); 2166 } 2167 2168 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2169 { 2170 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2171 le32_to_cpu(es->s_r_blocks_count_lo); 2172 } 2173 2174 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2175 { 2176 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2177 le32_to_cpu(es->s_free_blocks_count_lo); 2178 } 2179 2180 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2181 ext4_fsblk_t blk) 2182 { 2183 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2184 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2185 } 2186 2187 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2188 ext4_fsblk_t blk) 2189 { 2190 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2191 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2192 } 2193 2194 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2195 ext4_fsblk_t blk) 2196 { 2197 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2198 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2199 } 2200 2201 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2202 { 2203 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2204 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2205 le32_to_cpu(raw_inode->i_size_lo); 2206 else 2207 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2208 } 2209 2210 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2211 { 2212 raw_inode->i_size_lo = cpu_to_le32(i_size); 2213 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2214 } 2215 2216 static inline 2217 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2218 ext4_group_t group) 2219 { 2220 struct ext4_group_info ***grp_info; 2221 long indexv, indexh; 2222 grp_info = EXT4_SB(sb)->s_group_info; 2223 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2224 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2225 return grp_info[indexv][indexh]; 2226 } 2227 2228 /* 2229 * Reading s_groups_count requires using smp_rmb() afterwards. See 2230 * the locking protocol documented in the comments of ext4_group_add() 2231 * in resize.c 2232 */ 2233 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2234 { 2235 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2236 2237 smp_rmb(); 2238 return ngroups; 2239 } 2240 2241 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2242 ext4_group_t block_group) 2243 { 2244 return block_group >> sbi->s_log_groups_per_flex; 2245 } 2246 2247 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2248 { 2249 return 1 << sbi->s_log_groups_per_flex; 2250 } 2251 2252 #define ext4_std_error(sb, errno) \ 2253 do { \ 2254 if ((errno)) \ 2255 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2256 } while (0) 2257 2258 #ifdef CONFIG_SMP 2259 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2260 * counters. So we need to make sure we have free clusters more 2261 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2262 */ 2263 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2264 #else 2265 #define EXT4_FREECLUSTERS_WATERMARK 0 2266 #endif 2267 2268 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2269 { 2270 /* 2271 * XXX: replace with spinlock if seen contended -bzzz 2272 */ 2273 down_write(&EXT4_I(inode)->i_data_sem); 2274 if (newsize > EXT4_I(inode)->i_disksize) 2275 EXT4_I(inode)->i_disksize = newsize; 2276 up_write(&EXT4_I(inode)->i_data_sem); 2277 return ; 2278 } 2279 2280 struct ext4_group_info { 2281 unsigned long bb_state; 2282 struct rb_root bb_free_root; 2283 ext4_grpblk_t bb_first_free; /* first free block */ 2284 ext4_grpblk_t bb_free; /* total free blocks */ 2285 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2286 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2287 struct list_head bb_prealloc_list; 2288 #ifdef DOUBLE_CHECK 2289 void *bb_bitmap; 2290 #endif 2291 struct rw_semaphore alloc_sem; 2292 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2293 * regions, index is order. 2294 * bb_counters[3] = 5 means 2295 * 5 free 8-block regions. */ 2296 }; 2297 2298 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2299 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2300 2301 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2302 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2303 2304 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2305 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2306 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2307 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2308 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2309 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2310 2311 #define EXT4_MAX_CONTENTION 8 2312 #define EXT4_CONTENTION_THRESHOLD 2 2313 2314 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2315 ext4_group_t group) 2316 { 2317 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2318 } 2319 2320 /* 2321 * Returns true if the filesystem is busy enough that attempts to 2322 * access the block group locks has run into contention. 2323 */ 2324 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2325 { 2326 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2327 } 2328 2329 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2330 { 2331 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2332 if (spin_trylock(lock)) 2333 /* 2334 * We're able to grab the lock right away, so drop the 2335 * lock contention counter. 2336 */ 2337 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2338 else { 2339 /* 2340 * The lock is busy, so bump the contention counter, 2341 * and then wait on the spin lock. 2342 */ 2343 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2344 EXT4_MAX_CONTENTION); 2345 spin_lock(lock); 2346 } 2347 } 2348 2349 static inline void ext4_unlock_group(struct super_block *sb, 2350 ext4_group_t group) 2351 { 2352 spin_unlock(ext4_group_lock_ptr(sb, group)); 2353 } 2354 2355 /* 2356 * Block validity checking 2357 */ 2358 #define ext4_check_indirect_blockref(inode, bh) \ 2359 ext4_check_blockref(__func__, __LINE__, inode, \ 2360 (__le32 *)(bh)->b_data, \ 2361 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2362 2363 #define ext4_ind_check_inode(inode) \ 2364 ext4_check_blockref(__func__, __LINE__, inode, \ 2365 EXT4_I(inode)->i_data, \ 2366 EXT4_NDIR_BLOCKS) 2367 2368 /* 2369 * Inodes and files operations 2370 */ 2371 2372 /* dir.c */ 2373 extern const struct file_operations ext4_dir_operations; 2374 2375 /* file.c */ 2376 extern const struct inode_operations ext4_file_inode_operations; 2377 extern const struct file_operations ext4_file_operations; 2378 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2379 extern void ext4_unwritten_wait(struct inode *inode); 2380 2381 /* namei.c */ 2382 extern const struct inode_operations ext4_dir_inode_operations; 2383 extern const struct inode_operations ext4_special_inode_operations; 2384 extern struct dentry *ext4_get_parent(struct dentry *child); 2385 2386 /* symlink.c */ 2387 extern const struct inode_operations ext4_symlink_inode_operations; 2388 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2389 2390 /* block_validity */ 2391 extern void ext4_release_system_zone(struct super_block *sb); 2392 extern int ext4_setup_system_zone(struct super_block *sb); 2393 extern int __init ext4_init_system_zone(void); 2394 extern void ext4_exit_system_zone(void); 2395 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2396 ext4_fsblk_t start_blk, 2397 unsigned int count); 2398 extern int ext4_check_blockref(const char *, unsigned int, 2399 struct inode *, __le32 *, unsigned int); 2400 2401 /* extents.c */ 2402 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2403 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2404 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 2405 int chunk); 2406 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2407 struct ext4_map_blocks *map, int flags); 2408 extern void ext4_ext_truncate(struct inode *); 2409 extern int ext4_ext_punch_hole(struct file *file, loff_t offset, 2410 loff_t length); 2411 extern void ext4_ext_init(struct super_block *); 2412 extern void ext4_ext_release(struct super_block *); 2413 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2414 loff_t len); 2415 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 2416 ssize_t len); 2417 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2418 struct ext4_map_blocks *map, int flags); 2419 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2420 __u64 start, __u64 len); 2421 /* move_extent.c */ 2422 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2423 __u64 start_orig, __u64 start_donor, 2424 __u64 len, __u64 *moved_len); 2425 2426 /* page-io.c */ 2427 extern int __init ext4_init_pageio(void); 2428 extern void ext4_add_complete_io(ext4_io_end_t *io_end); 2429 extern void ext4_exit_pageio(void); 2430 extern void ext4_ioend_wait(struct inode *); 2431 extern void ext4_free_io_end(ext4_io_end_t *io); 2432 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2433 extern void ext4_io_submit(struct ext4_io_submit *io); 2434 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2435 struct page *page, 2436 int len, 2437 struct writeback_control *wbc); 2438 2439 /* mmp.c */ 2440 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2441 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp); 2442 extern int ext4_mmp_csum_verify(struct super_block *sb, 2443 struct mmp_struct *mmp); 2444 2445 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */ 2446 enum ext4_state_bits { 2447 BH_Uninit /* blocks are allocated but uninitialized on disk */ 2448 = BH_JBDPrivateStart, 2449 BH_AllocFromCluster, /* allocated blocks were part of already 2450 * allocated cluster. Note that this flag will 2451 * never, ever appear in a buffer_head's state 2452 * flag. See EXT4_MAP_FROM_CLUSTER to see where 2453 * this is used. */ 2454 }; 2455 2456 BUFFER_FNS(Uninit, uninit) 2457 TAS_BUFFER_FNS(Uninit, uninit) 2458 2459 /* 2460 * Add new method to test wether block and inode bitmaps are properly 2461 * initialized. With uninit_bg reading the block from disk is not enough 2462 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2463 */ 2464 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2465 2466 static inline int bitmap_uptodate(struct buffer_head *bh) 2467 { 2468 return (buffer_uptodate(bh) && 2469 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2470 } 2471 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2472 { 2473 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2474 } 2475 2476 /* 2477 * Disable DIO read nolock optimization, so new dioreaders will be forced 2478 * to grab i_mutex 2479 */ 2480 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2481 { 2482 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2483 smp_mb(); 2484 } 2485 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2486 { 2487 smp_mb(); 2488 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2489 } 2490 2491 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2492 2493 /* For ioend & aio unwritten conversion wait queues */ 2494 #define EXT4_WQ_HASH_SZ 37 2495 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2496 EXT4_WQ_HASH_SZ]) 2497 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2498 EXT4_WQ_HASH_SZ]) 2499 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2500 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2501 2502 #define EXT4_RESIZING 0 2503 extern int ext4_resize_begin(struct super_block *sb); 2504 extern void ext4_resize_end(struct super_block *sb); 2505 2506 #endif /* __KERNEL__ */ 2507 2508 #include "ext4_extents.h" 2509 2510 #endif /* _EXT4_H */ 2511