1 /* 2 * ext4.h 3 * 4 * Copyright (C) 1992, 1993, 1994, 1995 5 * Remy Card (card@masi.ibp.fr) 6 * Laboratoire MASI - Institut Blaise Pascal 7 * Universite Pierre et Marie Curie (Paris VI) 8 * 9 * from 10 * 11 * linux/include/linux/minix_fs.h 12 * 13 * Copyright (C) 1991, 1992 Linus Torvalds 14 */ 15 16 #ifndef _EXT4_H 17 #define _EXT4_H 18 19 #include <linux/types.h> 20 #include <linux/blkdev.h> 21 #include <linux/magic.h> 22 #include <linux/jbd2.h> 23 #include <linux/quota.h> 24 #include <linux/rwsem.h> 25 #include <linux/rbtree.h> 26 #include <linux/seqlock.h> 27 #include <linux/mutex.h> 28 #include <linux/timer.h> 29 #include <linux/wait.h> 30 #include <linux/blockgroup_lock.h> 31 #include <linux/percpu_counter.h> 32 #ifdef __KERNEL__ 33 #include <linux/compat.h> 34 #endif 35 36 /* 37 * The fourth extended filesystem constants/structures 38 */ 39 40 /* 41 * Define EXT4FS_DEBUG to produce debug messages 42 */ 43 #undef EXT4FS_DEBUG 44 45 /* 46 * Debug code 47 */ 48 #ifdef EXT4FS_DEBUG 49 #define ext4_debug(f, a...) \ 50 do { \ 51 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 52 __FILE__, __LINE__, __func__); \ 53 printk(KERN_DEBUG f, ## a); \ 54 } while (0) 55 #else 56 #define ext4_debug(f, a...) do {} while (0) 57 #endif 58 59 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 60 ext4_error_inode(__func__, (inode), (fmt), ## a) 61 62 #define EXT4_ERROR_FILE(file, fmt, a...) \ 63 ext4_error_file(__func__, (file), (fmt), ## a) 64 65 /* data type for block offset of block group */ 66 typedef int ext4_grpblk_t; 67 68 /* data type for filesystem-wide blocks number */ 69 typedef unsigned long long ext4_fsblk_t; 70 71 /* data type for file logical block number */ 72 typedef __u32 ext4_lblk_t; 73 74 /* data type for block group number */ 75 typedef unsigned int ext4_group_t; 76 77 /* 78 * Flags used in mballoc's allocation_context flags field. 79 * 80 * Also used to show what's going on for debugging purposes when the 81 * flag field is exported via the traceport interface 82 */ 83 84 /* prefer goal again. length */ 85 #define EXT4_MB_HINT_MERGE 0x0001 86 /* blocks already reserved */ 87 #define EXT4_MB_HINT_RESERVED 0x0002 88 /* metadata is being allocated */ 89 #define EXT4_MB_HINT_METADATA 0x0004 90 /* first blocks in the file */ 91 #define EXT4_MB_HINT_FIRST 0x0008 92 /* search for the best chunk */ 93 #define EXT4_MB_HINT_BEST 0x0010 94 /* data is being allocated */ 95 #define EXT4_MB_HINT_DATA 0x0020 96 /* don't preallocate (for tails) */ 97 #define EXT4_MB_HINT_NOPREALLOC 0x0040 98 /* allocate for locality group */ 99 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 100 /* allocate goal blocks or none */ 101 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 102 /* goal is meaningful */ 103 #define EXT4_MB_HINT_TRY_GOAL 0x0200 104 /* blocks already pre-reserved by delayed allocation */ 105 #define EXT4_MB_DELALLOC_RESERVED 0x0400 106 /* We are doing stream allocation */ 107 #define EXT4_MB_STREAM_ALLOC 0x0800 108 109 110 struct ext4_allocation_request { 111 /* target inode for block we're allocating */ 112 struct inode *inode; 113 /* how many blocks we want to allocate */ 114 unsigned int len; 115 /* logical block in target inode */ 116 ext4_lblk_t logical; 117 /* the closest logical allocated block to the left */ 118 ext4_lblk_t lleft; 119 /* the closest logical allocated block to the right */ 120 ext4_lblk_t lright; 121 /* phys. target (a hint) */ 122 ext4_fsblk_t goal; 123 /* phys. block for the closest logical allocated block to the left */ 124 ext4_fsblk_t pleft; 125 /* phys. block for the closest logical allocated block to the right */ 126 ext4_fsblk_t pright; 127 /* flags. see above EXT4_MB_HINT_* */ 128 unsigned int flags; 129 }; 130 131 /* 132 * Logical to physical block mapping, used by ext4_map_blocks() 133 * 134 * This structure is used to pass requests into ext4_map_blocks() as 135 * well as to store the information returned by ext4_map_blocks(). It 136 * takes less room on the stack than a struct buffer_head. 137 */ 138 #define EXT4_MAP_NEW (1 << BH_New) 139 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 140 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 141 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 142 #define EXT4_MAP_UNINIT (1 << BH_Uninit) 143 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 144 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 145 EXT4_MAP_UNINIT) 146 147 struct ext4_map_blocks { 148 ext4_fsblk_t m_pblk; 149 ext4_lblk_t m_lblk; 150 unsigned int m_len; 151 unsigned int m_flags; 152 }; 153 154 /* 155 * For delayed allocation tracking 156 */ 157 struct mpage_da_data { 158 struct inode *inode; 159 sector_t b_blocknr; /* start block number of extent */ 160 size_t b_size; /* size of extent */ 161 unsigned long b_state; /* state of the extent */ 162 unsigned long first_page, next_page; /* extent of pages */ 163 struct writeback_control *wbc; 164 int io_done; 165 int pages_written; 166 int retval; 167 }; 168 #define EXT4_IO_UNWRITTEN 0x1 169 typedef struct ext4_io_end { 170 struct list_head list; /* per-file finished AIO list */ 171 struct inode *inode; /* file being written to */ 172 unsigned int flag; /* unwritten or not */ 173 struct page *page; /* page struct for buffer write */ 174 loff_t offset; /* offset in the file */ 175 ssize_t size; /* size of the extent */ 176 struct work_struct work; /* data work queue */ 177 } ext4_io_end_t; 178 179 /* 180 * Special inodes numbers 181 */ 182 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 183 #define EXT4_ROOT_INO 2 /* Root inode */ 184 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 185 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 186 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 187 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 188 189 /* First non-reserved inode for old ext4 filesystems */ 190 #define EXT4_GOOD_OLD_FIRST_INO 11 191 192 /* 193 * Maximal count of links to a file 194 */ 195 #define EXT4_LINK_MAX 65000 196 197 /* 198 * Macro-instructions used to manage several block sizes 199 */ 200 #define EXT4_MIN_BLOCK_SIZE 1024 201 #define EXT4_MAX_BLOCK_SIZE 65536 202 #define EXT4_MIN_BLOCK_LOG_SIZE 10 203 #ifdef __KERNEL__ 204 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 205 #else 206 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 207 #endif 208 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 209 #ifdef __KERNEL__ 210 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 211 #else 212 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 213 #endif 214 #ifdef __KERNEL__ 215 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 216 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 217 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 218 #else 219 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 220 EXT4_GOOD_OLD_INODE_SIZE : \ 221 (s)->s_inode_size) 222 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 223 EXT4_GOOD_OLD_FIRST_INO : \ 224 (s)->s_first_ino) 225 #endif 226 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 227 228 /* 229 * Structure of a blocks group descriptor 230 */ 231 struct ext4_group_desc 232 { 233 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 234 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 235 __le32 bg_inode_table_lo; /* Inodes table block */ 236 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 237 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 238 __le16 bg_used_dirs_count_lo; /* Directories count */ 239 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 240 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */ 241 __le16 bg_itable_unused_lo; /* Unused inodes count */ 242 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 243 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 244 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 245 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 246 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 247 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 248 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 249 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 250 __u32 bg_reserved2[3]; 251 }; 252 253 /* 254 * Structure of a flex block group info 255 */ 256 257 struct flex_groups { 258 atomic_t free_inodes; 259 atomic_t free_blocks; 260 atomic_t used_dirs; 261 }; 262 263 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 264 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 265 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 266 267 /* 268 * Macro-instructions used to manage group descriptors 269 */ 270 #define EXT4_MIN_DESC_SIZE 32 271 #define EXT4_MIN_DESC_SIZE_64BIT 64 272 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 273 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 274 #ifdef __KERNEL__ 275 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 276 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 277 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 278 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 279 #else 280 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 281 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 282 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 283 #endif 284 285 /* 286 * Constants relative to the data blocks 287 */ 288 #define EXT4_NDIR_BLOCKS 12 289 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 290 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 291 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 292 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 293 294 /* 295 * Inode flags 296 */ 297 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 298 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 299 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 300 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 301 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 302 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 303 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 304 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 305 /* Reserved for compression usage... */ 306 #define EXT4_DIRTY_FL 0x00000100 307 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 308 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 309 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 310 /* End compression flags --- maybe not all used */ 311 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 312 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 313 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 314 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 315 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 316 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 317 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 318 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 319 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 320 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 321 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 322 323 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 324 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */ 325 326 /* Flags that should be inherited by new inodes from their parent. */ 327 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 328 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\ 329 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 330 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 331 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 332 333 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 334 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 335 336 /* Flags that are appropriate for non-directories/regular files. */ 337 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 338 339 /* Mask out flags that are inappropriate for the given type of inode. */ 340 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 341 { 342 if (S_ISDIR(mode)) 343 return flags; 344 else if (S_ISREG(mode)) 345 return flags & EXT4_REG_FLMASK; 346 else 347 return flags & EXT4_OTHER_FLMASK; 348 } 349 350 /* 351 * Inode flags used for atomic set/get 352 */ 353 enum { 354 EXT4_INODE_SECRM = 0, /* Secure deletion */ 355 EXT4_INODE_UNRM = 1, /* Undelete */ 356 EXT4_INODE_COMPR = 2, /* Compress file */ 357 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 358 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 359 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 360 EXT4_INODE_NODUMP = 6, /* do not dump file */ 361 EXT4_INODE_NOATIME = 7, /* do not update atime */ 362 /* Reserved for compression usage... */ 363 EXT4_INODE_DIRTY = 8, 364 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 365 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 366 EXT4_INODE_ECOMPR = 11, /* Compression error */ 367 /* End compression flags --- maybe not all used */ 368 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 369 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 370 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 371 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 372 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 373 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 374 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 375 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 376 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 377 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 378 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 379 }; 380 381 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 382 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \ 383 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \ 384 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); } 385 386 /* 387 * Since it's pretty easy to mix up bit numbers and hex values, and we 388 * can't do a compile-time test for ENUM values, we use a run-time 389 * test to make sure that EXT4_XXX_FL is consistent with respect to 390 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop 391 * out so it won't cost any extra space in the compiled kernel image. 392 * But it's important that these values are the same, since we are 393 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL 394 * must be consistent with the values of FS_XXX_FL defined in 395 * include/linux/fs.h and the on-disk values found in ext2, ext3, and 396 * ext4 filesystems, and of course the values defined in e2fsprogs. 397 * 398 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 399 */ 400 static inline void ext4_check_flag_values(void) 401 { 402 CHECK_FLAG_VALUE(SECRM); 403 CHECK_FLAG_VALUE(UNRM); 404 CHECK_FLAG_VALUE(COMPR); 405 CHECK_FLAG_VALUE(SYNC); 406 CHECK_FLAG_VALUE(IMMUTABLE); 407 CHECK_FLAG_VALUE(APPEND); 408 CHECK_FLAG_VALUE(NODUMP); 409 CHECK_FLAG_VALUE(NOATIME); 410 CHECK_FLAG_VALUE(DIRTY); 411 CHECK_FLAG_VALUE(COMPRBLK); 412 CHECK_FLAG_VALUE(NOCOMPR); 413 CHECK_FLAG_VALUE(ECOMPR); 414 CHECK_FLAG_VALUE(INDEX); 415 CHECK_FLAG_VALUE(IMAGIC); 416 CHECK_FLAG_VALUE(JOURNAL_DATA); 417 CHECK_FLAG_VALUE(NOTAIL); 418 CHECK_FLAG_VALUE(DIRSYNC); 419 CHECK_FLAG_VALUE(TOPDIR); 420 CHECK_FLAG_VALUE(HUGE_FILE); 421 CHECK_FLAG_VALUE(EXTENTS); 422 CHECK_FLAG_VALUE(EA_INODE); 423 CHECK_FLAG_VALUE(EOFBLOCKS); 424 CHECK_FLAG_VALUE(RESERVED); 425 } 426 427 /* Used to pass group descriptor data when online resize is done */ 428 struct ext4_new_group_input { 429 __u32 group; /* Group number for this data */ 430 __u64 block_bitmap; /* Absolute block number of block bitmap */ 431 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 432 __u64 inode_table; /* Absolute block number of inode table start */ 433 __u32 blocks_count; /* Total number of blocks in this group */ 434 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 435 __u16 unused; 436 }; 437 438 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 439 struct compat_ext4_new_group_input { 440 u32 group; 441 compat_u64 block_bitmap; 442 compat_u64 inode_bitmap; 443 compat_u64 inode_table; 444 u32 blocks_count; 445 u16 reserved_blocks; 446 u16 unused; 447 }; 448 #endif 449 450 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 451 struct ext4_new_group_data { 452 __u32 group; 453 __u64 block_bitmap; 454 __u64 inode_bitmap; 455 __u64 inode_table; 456 __u32 blocks_count; 457 __u16 reserved_blocks; 458 __u16 unused; 459 __u32 free_blocks_count; 460 }; 461 462 /* 463 * Flags used by ext4_get_blocks() 464 */ 465 /* Allocate any needed blocks and/or convert an unitialized 466 extent to be an initialized ext4 */ 467 #define EXT4_GET_BLOCKS_CREATE 0x0001 468 /* Request the creation of an unitialized extent */ 469 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 470 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 471 EXT4_GET_BLOCKS_CREATE) 472 /* Caller is from the delayed allocation writeout path, 473 so set the magic i_delalloc_reserve_flag after taking the 474 inode allocation semaphore for */ 475 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 476 /* caller is from the direct IO path, request to creation of an 477 unitialized extents if not allocated, split the uninitialized 478 extent if blocks has been preallocated already*/ 479 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 480 #define EXT4_GET_BLOCKS_CONVERT 0x0010 481 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 482 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 483 /* Convert extent to initialized after IO complete */ 484 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 485 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 486 487 /* 488 * Flags used by ext4_free_blocks 489 */ 490 #define EXT4_FREE_BLOCKS_METADATA 0x0001 491 #define EXT4_FREE_BLOCKS_FORGET 0x0002 492 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 493 494 /* 495 * ioctl commands 496 */ 497 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 498 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 499 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 500 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 501 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 502 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 503 #ifdef CONFIG_JBD2_DEBUG 504 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long) 505 #endif 506 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 507 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 508 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 509 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 510 #define EXT4_IOC_MIGRATE _IO('f', 9) 511 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 512 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 513 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 514 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 515 516 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 517 /* 518 * ioctl commands in 32 bit emulation 519 */ 520 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 521 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 522 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 523 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 524 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 525 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 526 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 527 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 528 #ifdef CONFIG_JBD2_DEBUG 529 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int) 530 #endif 531 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 532 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 533 #endif 534 535 536 /* 537 * Mount options 538 */ 539 struct ext4_mount_options { 540 unsigned long s_mount_opt; 541 uid_t s_resuid; 542 gid_t s_resgid; 543 unsigned long s_commit_interval; 544 u32 s_min_batch_time, s_max_batch_time; 545 #ifdef CONFIG_QUOTA 546 int s_jquota_fmt; 547 char *s_qf_names[MAXQUOTAS]; 548 #endif 549 }; 550 551 /* Max physical block we can addres w/o extents */ 552 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 553 554 /* 555 * Structure of an inode on the disk 556 */ 557 struct ext4_inode { 558 __le16 i_mode; /* File mode */ 559 __le16 i_uid; /* Low 16 bits of Owner Uid */ 560 __le32 i_size_lo; /* Size in bytes */ 561 __le32 i_atime; /* Access time */ 562 __le32 i_ctime; /* Inode Change time */ 563 __le32 i_mtime; /* Modification time */ 564 __le32 i_dtime; /* Deletion Time */ 565 __le16 i_gid; /* Low 16 bits of Group Id */ 566 __le16 i_links_count; /* Links count */ 567 __le32 i_blocks_lo; /* Blocks count */ 568 __le32 i_flags; /* File flags */ 569 union { 570 struct { 571 __le32 l_i_version; 572 } linux1; 573 struct { 574 __u32 h_i_translator; 575 } hurd1; 576 struct { 577 __u32 m_i_reserved1; 578 } masix1; 579 } osd1; /* OS dependent 1 */ 580 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 581 __le32 i_generation; /* File version (for NFS) */ 582 __le32 i_file_acl_lo; /* File ACL */ 583 __le32 i_size_high; 584 __le32 i_obso_faddr; /* Obsoleted fragment address */ 585 union { 586 struct { 587 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 588 __le16 l_i_file_acl_high; 589 __le16 l_i_uid_high; /* these 2 fields */ 590 __le16 l_i_gid_high; /* were reserved2[0] */ 591 __u32 l_i_reserved2; 592 } linux2; 593 struct { 594 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 595 __u16 h_i_mode_high; 596 __u16 h_i_uid_high; 597 __u16 h_i_gid_high; 598 __u32 h_i_author; 599 } hurd2; 600 struct { 601 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 602 __le16 m_i_file_acl_high; 603 __u32 m_i_reserved2[2]; 604 } masix2; 605 } osd2; /* OS dependent 2 */ 606 __le16 i_extra_isize; 607 __le16 i_pad1; 608 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 609 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 610 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 611 __le32 i_crtime; /* File Creation time */ 612 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 613 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 614 }; 615 616 struct move_extent { 617 __u32 reserved; /* should be zero */ 618 __u32 donor_fd; /* donor file descriptor */ 619 __u64 orig_start; /* logical start offset in block for orig */ 620 __u64 donor_start; /* logical start offset in block for donor */ 621 __u64 len; /* block length to be moved */ 622 __u64 moved_len; /* moved block length */ 623 }; 624 625 #define EXT4_EPOCH_BITS 2 626 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 627 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 628 629 /* 630 * Extended fields will fit into an inode if the filesystem was formatted 631 * with large inodes (-I 256 or larger) and there are not currently any EAs 632 * consuming all of the available space. For new inodes we always reserve 633 * enough space for the kernel's known extended fields, but for inodes 634 * created with an old kernel this might not have been the case. None of 635 * the extended inode fields is critical for correct filesystem operation. 636 * This macro checks if a certain field fits in the inode. Note that 637 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 638 */ 639 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 640 ((offsetof(typeof(*ext4_inode), field) + \ 641 sizeof((ext4_inode)->field)) \ 642 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 643 (einode)->i_extra_isize)) \ 644 645 static inline __le32 ext4_encode_extra_time(struct timespec *time) 646 { 647 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 648 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 649 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 650 } 651 652 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 653 { 654 if (sizeof(time->tv_sec) > 4) 655 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 656 << 32; 657 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 658 } 659 660 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 661 do { \ 662 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 663 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 664 (raw_inode)->xtime ## _extra = \ 665 ext4_encode_extra_time(&(inode)->xtime); \ 666 } while (0) 667 668 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 669 do { \ 670 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 671 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 672 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 673 (raw_inode)->xtime ## _extra = \ 674 ext4_encode_extra_time(&(einode)->xtime); \ 675 } while (0) 676 677 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 678 do { \ 679 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 680 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 681 ext4_decode_extra_time(&(inode)->xtime, \ 682 raw_inode->xtime ## _extra); \ 683 } while (0) 684 685 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 686 do { \ 687 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 688 (einode)->xtime.tv_sec = \ 689 (signed)le32_to_cpu((raw_inode)->xtime); \ 690 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 691 ext4_decode_extra_time(&(einode)->xtime, \ 692 raw_inode->xtime ## _extra); \ 693 } while (0) 694 695 #define i_disk_version osd1.linux1.l_i_version 696 697 #if defined(__KERNEL__) || defined(__linux__) 698 #define i_reserved1 osd1.linux1.l_i_reserved1 699 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 700 #define i_blocks_high osd2.linux2.l_i_blocks_high 701 #define i_uid_low i_uid 702 #define i_gid_low i_gid 703 #define i_uid_high osd2.linux2.l_i_uid_high 704 #define i_gid_high osd2.linux2.l_i_gid_high 705 #define i_reserved2 osd2.linux2.l_i_reserved2 706 707 #elif defined(__GNU__) 708 709 #define i_translator osd1.hurd1.h_i_translator 710 #define i_uid_high osd2.hurd2.h_i_uid_high 711 #define i_gid_high osd2.hurd2.h_i_gid_high 712 #define i_author osd2.hurd2.h_i_author 713 714 #elif defined(__masix__) 715 716 #define i_reserved1 osd1.masix1.m_i_reserved1 717 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 718 #define i_reserved2 osd2.masix2.m_i_reserved2 719 720 #endif /* defined(__KERNEL__) || defined(__linux__) */ 721 722 /* 723 * storage for cached extent 724 */ 725 struct ext4_ext_cache { 726 ext4_fsblk_t ec_start; 727 ext4_lblk_t ec_block; 728 __u32 ec_len; /* must be 32bit to return holes */ 729 __u32 ec_type; 730 }; 731 732 /* 733 * fourth extended file system inode data in memory 734 */ 735 struct ext4_inode_info { 736 __le32 i_data[15]; /* unconverted */ 737 __u32 i_dtime; 738 ext4_fsblk_t i_file_acl; 739 740 /* 741 * i_block_group is the number of the block group which contains 742 * this file's inode. Constant across the lifetime of the inode, 743 * it is ued for making block allocation decisions - we try to 744 * place a file's data blocks near its inode block, and new inodes 745 * near to their parent directory's inode. 746 */ 747 ext4_group_t i_block_group; 748 unsigned long i_state_flags; /* Dynamic state flags */ 749 unsigned long i_flags; 750 751 ext4_lblk_t i_dir_start_lookup; 752 #ifdef CONFIG_EXT4_FS_XATTR 753 /* 754 * Extended attributes can be read independently of the main file 755 * data. Taking i_mutex even when reading would cause contention 756 * between readers of EAs and writers of regular file data, so 757 * instead we synchronize on xattr_sem when reading or changing 758 * EAs. 759 */ 760 struct rw_semaphore xattr_sem; 761 #endif 762 763 struct list_head i_orphan; /* unlinked but open inodes */ 764 765 /* 766 * i_disksize keeps track of what the inode size is ON DISK, not 767 * in memory. During truncate, i_size is set to the new size by 768 * the VFS prior to calling ext4_truncate(), but the filesystem won't 769 * set i_disksize to 0 until the truncate is actually under way. 770 * 771 * The intent is that i_disksize always represents the blocks which 772 * are used by this file. This allows recovery to restart truncate 773 * on orphans if we crash during truncate. We actually write i_disksize 774 * into the on-disk inode when writing inodes out, instead of i_size. 775 * 776 * The only time when i_disksize and i_size may be different is when 777 * a truncate is in progress. The only things which change i_disksize 778 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 779 */ 780 loff_t i_disksize; 781 782 /* 783 * i_data_sem is for serialising ext4_truncate() against 784 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 785 * data tree are chopped off during truncate. We can't do that in 786 * ext4 because whenever we perform intermediate commits during 787 * truncate, the inode and all the metadata blocks *must* be in a 788 * consistent state which allows truncation of the orphans to restart 789 * during recovery. Hence we must fix the get_block-vs-truncate race 790 * by other means, so we have i_data_sem. 791 */ 792 struct rw_semaphore i_data_sem; 793 struct inode vfs_inode; 794 struct jbd2_inode jinode; 795 796 struct ext4_ext_cache i_cached_extent; 797 /* 798 * File creation time. Its function is same as that of 799 * struct timespec i_{a,c,m}time in the generic inode. 800 */ 801 struct timespec i_crtime; 802 803 /* mballoc */ 804 struct list_head i_prealloc_list; 805 spinlock_t i_prealloc_lock; 806 807 /* ialloc */ 808 ext4_group_t i_last_alloc_group; 809 810 /* allocation reservation info for delalloc */ 811 unsigned int i_reserved_data_blocks; 812 unsigned int i_reserved_meta_blocks; 813 unsigned int i_allocated_meta_blocks; 814 unsigned short i_delalloc_reserved_flag; 815 sector_t i_da_metadata_calc_last_lblock; 816 int i_da_metadata_calc_len; 817 818 /* on-disk additional length */ 819 __u16 i_extra_isize; 820 821 spinlock_t i_block_reservation_lock; 822 #ifdef CONFIG_QUOTA 823 /* quota space reservation, managed internally by quota code */ 824 qsize_t i_reserved_quota; 825 #endif 826 827 /* completed IOs that might need unwritten extents handling */ 828 struct list_head i_completed_io_list; 829 spinlock_t i_completed_io_lock; 830 /* current io_end structure for async DIO write*/ 831 ext4_io_end_t *cur_aio_dio; 832 833 /* 834 * Transactions that contain inode's metadata needed to complete 835 * fsync and fdatasync, respectively. 836 */ 837 tid_t i_sync_tid; 838 tid_t i_datasync_tid; 839 }; 840 841 /* 842 * File system states 843 */ 844 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 845 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 846 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 847 848 /* 849 * Misc. filesystem flags 850 */ 851 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 852 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 853 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 854 855 /* 856 * Mount flags 857 */ 858 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */ 859 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 860 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 861 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 862 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 863 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 864 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 865 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 866 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 867 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 868 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 869 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 870 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 871 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 872 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 873 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 874 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 875 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 876 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 877 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 878 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 879 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 880 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 881 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 882 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */ 883 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 884 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 885 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 886 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 887 888 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt 889 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt 890 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 891 EXT4_MOUNT_##opt) 892 893 #define ext4_set_bit ext2_set_bit 894 #define ext4_set_bit_atomic ext2_set_bit_atomic 895 #define ext4_clear_bit ext2_clear_bit 896 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 897 #define ext4_test_bit ext2_test_bit 898 #define ext4_find_first_zero_bit ext2_find_first_zero_bit 899 #define ext4_find_next_zero_bit ext2_find_next_zero_bit 900 #define ext4_find_next_bit ext2_find_next_bit 901 902 /* 903 * Maximal mount counts between two filesystem checks 904 */ 905 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 906 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 907 908 /* 909 * Behaviour when detecting errors 910 */ 911 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 912 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 913 #define EXT4_ERRORS_PANIC 3 /* Panic */ 914 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 915 916 /* 917 * Structure of the super block 918 */ 919 struct ext4_super_block { 920 /*00*/ __le32 s_inodes_count; /* Inodes count */ 921 __le32 s_blocks_count_lo; /* Blocks count */ 922 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 923 __le32 s_free_blocks_count_lo; /* Free blocks count */ 924 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 925 __le32 s_first_data_block; /* First Data Block */ 926 __le32 s_log_block_size; /* Block size */ 927 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */ 928 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 929 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */ 930 __le32 s_inodes_per_group; /* # Inodes per group */ 931 __le32 s_mtime; /* Mount time */ 932 /*30*/ __le32 s_wtime; /* Write time */ 933 __le16 s_mnt_count; /* Mount count */ 934 __le16 s_max_mnt_count; /* Maximal mount count */ 935 __le16 s_magic; /* Magic signature */ 936 __le16 s_state; /* File system state */ 937 __le16 s_errors; /* Behaviour when detecting errors */ 938 __le16 s_minor_rev_level; /* minor revision level */ 939 /*40*/ __le32 s_lastcheck; /* time of last check */ 940 __le32 s_checkinterval; /* max. time between checks */ 941 __le32 s_creator_os; /* OS */ 942 __le32 s_rev_level; /* Revision level */ 943 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 944 __le16 s_def_resgid; /* Default gid for reserved blocks */ 945 /* 946 * These fields are for EXT4_DYNAMIC_REV superblocks only. 947 * 948 * Note: the difference between the compatible feature set and 949 * the incompatible feature set is that if there is a bit set 950 * in the incompatible feature set that the kernel doesn't 951 * know about, it should refuse to mount the filesystem. 952 * 953 * e2fsck's requirements are more strict; if it doesn't know 954 * about a feature in either the compatible or incompatible 955 * feature set, it must abort and not try to meddle with 956 * things it doesn't understand... 957 */ 958 __le32 s_first_ino; /* First non-reserved inode */ 959 __le16 s_inode_size; /* size of inode structure */ 960 __le16 s_block_group_nr; /* block group # of this superblock */ 961 __le32 s_feature_compat; /* compatible feature set */ 962 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 963 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 964 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 965 /*78*/ char s_volume_name[16]; /* volume name */ 966 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 967 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 968 /* 969 * Performance hints. Directory preallocation should only 970 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 971 */ 972 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 973 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 974 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 975 /* 976 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 977 */ 978 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 979 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 980 __le32 s_journal_dev; /* device number of journal file */ 981 __le32 s_last_orphan; /* start of list of inodes to delete */ 982 __le32 s_hash_seed[4]; /* HTREE hash seed */ 983 __u8 s_def_hash_version; /* Default hash version to use */ 984 __u8 s_reserved_char_pad; 985 __le16 s_desc_size; /* size of group descriptor */ 986 /*100*/ __le32 s_default_mount_opts; 987 __le32 s_first_meta_bg; /* First metablock block group */ 988 __le32 s_mkfs_time; /* When the filesystem was created */ 989 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 990 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 991 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 992 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 993 __le32 s_free_blocks_count_hi; /* Free blocks count */ 994 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 995 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 996 __le32 s_flags; /* Miscellaneous flags */ 997 __le16 s_raid_stride; /* RAID stride */ 998 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */ 999 __le64 s_mmp_block; /* Block for multi-mount protection */ 1000 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1001 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1002 __u8 s_reserved_char_pad2; 1003 __le16 s_reserved_pad; 1004 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1005 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1006 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1007 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1008 snapshot's future use */ 1009 __le32 s_snapshot_list; /* inode number of the head of the 1010 on-disk snapshot list */ 1011 __u32 s_reserved[155]; /* Padding to the end of the block */ 1012 }; 1013 1014 #ifdef __KERNEL__ 1015 1016 /* 1017 * run-time mount flags 1018 */ 1019 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1020 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1021 1022 /* 1023 * fourth extended-fs super-block data in memory 1024 */ 1025 struct ext4_sb_info { 1026 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1027 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1028 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1029 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1030 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1031 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1032 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1033 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1034 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1035 unsigned long s_overhead_last; /* Last calculated overhead */ 1036 unsigned long s_blocks_last; /* Last seen block count */ 1037 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1038 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1039 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1040 struct buffer_head **s_group_desc; 1041 unsigned int s_mount_opt; 1042 unsigned int s_mount_flags; 1043 ext4_fsblk_t s_sb_block; 1044 uid_t s_resuid; 1045 gid_t s_resgid; 1046 unsigned short s_mount_state; 1047 unsigned short s_pad; 1048 int s_addr_per_block_bits; 1049 int s_desc_per_block_bits; 1050 int s_inode_size; 1051 int s_first_ino; 1052 unsigned int s_inode_readahead_blks; 1053 unsigned int s_inode_goal; 1054 spinlock_t s_next_gen_lock; 1055 u32 s_next_generation; 1056 u32 s_hash_seed[4]; 1057 int s_def_hash_version; 1058 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1059 struct percpu_counter s_freeblocks_counter; 1060 struct percpu_counter s_freeinodes_counter; 1061 struct percpu_counter s_dirs_counter; 1062 struct percpu_counter s_dirtyblocks_counter; 1063 struct blockgroup_lock *s_blockgroup_lock; 1064 struct proc_dir_entry *s_proc; 1065 struct kobject s_kobj; 1066 struct completion s_kobj_unregister; 1067 1068 /* Journaling */ 1069 struct inode *s_journal_inode; 1070 struct journal_s *s_journal; 1071 struct list_head s_orphan; 1072 struct mutex s_orphan_lock; 1073 struct mutex s_resize_lock; 1074 unsigned long s_commit_interval; 1075 u32 s_max_batch_time; 1076 u32 s_min_batch_time; 1077 struct block_device *journal_bdev; 1078 #ifdef CONFIG_JBD2_DEBUG 1079 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */ 1080 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */ 1081 #endif 1082 #ifdef CONFIG_QUOTA 1083 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 1084 int s_jquota_fmt; /* Format of quota to use */ 1085 #endif 1086 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1087 struct rb_root system_blks; 1088 1089 #ifdef EXTENTS_STATS 1090 /* ext4 extents stats */ 1091 unsigned long s_ext_min; 1092 unsigned long s_ext_max; 1093 unsigned long s_depth_max; 1094 spinlock_t s_ext_stats_lock; 1095 unsigned long s_ext_blocks; 1096 unsigned long s_ext_extents; 1097 #endif 1098 1099 /* for buddy allocator */ 1100 struct ext4_group_info ***s_group_info; 1101 struct inode *s_buddy_cache; 1102 long s_blocks_reserved; 1103 spinlock_t s_reserve_lock; 1104 spinlock_t s_md_lock; 1105 tid_t s_last_transaction; 1106 unsigned short *s_mb_offsets; 1107 unsigned int *s_mb_maxs; 1108 1109 /* tunables */ 1110 unsigned long s_stripe; 1111 unsigned int s_mb_stream_request; 1112 unsigned int s_mb_max_to_scan; 1113 unsigned int s_mb_min_to_scan; 1114 unsigned int s_mb_stats; 1115 unsigned int s_mb_order2_reqs; 1116 unsigned int s_mb_group_prealloc; 1117 unsigned int s_max_writeback_mb_bump; 1118 /* where last allocation was done - for stream allocation */ 1119 unsigned long s_mb_last_group; 1120 unsigned long s_mb_last_start; 1121 1122 /* stats for buddy allocator */ 1123 spinlock_t s_mb_pa_lock; 1124 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1125 atomic_t s_bal_success; /* we found long enough chunks */ 1126 atomic_t s_bal_allocated; /* in blocks */ 1127 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1128 atomic_t s_bal_goals; /* goal hits */ 1129 atomic_t s_bal_breaks; /* too long searches */ 1130 atomic_t s_bal_2orders; /* 2^order hits */ 1131 spinlock_t s_bal_lock; 1132 unsigned long s_mb_buddies_generated; 1133 unsigned long long s_mb_generation_time; 1134 atomic_t s_mb_lost_chunks; 1135 atomic_t s_mb_preallocated; 1136 atomic_t s_mb_discarded; 1137 atomic_t s_lock_busy; 1138 1139 /* locality groups */ 1140 struct ext4_locality_group __percpu *s_locality_groups; 1141 1142 /* for write statistics */ 1143 unsigned long s_sectors_written_start; 1144 u64 s_kbytes_written; 1145 1146 unsigned int s_log_groups_per_flex; 1147 struct flex_groups *s_flex_groups; 1148 1149 /* workqueue for dio unwritten */ 1150 struct workqueue_struct *dio_unwritten_wq; 1151 }; 1152 1153 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1154 { 1155 return sb->s_fs_info; 1156 } 1157 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1158 { 1159 return container_of(inode, struct ext4_inode_info, vfs_inode); 1160 } 1161 1162 static inline struct timespec ext4_current_time(struct inode *inode) 1163 { 1164 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1165 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1166 } 1167 1168 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1169 { 1170 return ino == EXT4_ROOT_INO || 1171 ino == EXT4_JOURNAL_INO || 1172 ino == EXT4_RESIZE_INO || 1173 (ino >= EXT4_FIRST_INO(sb) && 1174 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1175 } 1176 1177 /* 1178 * Inode dynamic state flags 1179 */ 1180 enum { 1181 EXT4_STATE_JDATA, /* journaled data exists */ 1182 EXT4_STATE_NEW, /* inode is newly created */ 1183 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1184 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1185 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1186 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1187 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1188 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1189 }; 1190 1191 #define EXT4_INODE_BIT_FNS(name, field) \ 1192 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1193 { \ 1194 return test_bit(bit, &EXT4_I(inode)->i_##field); \ 1195 } \ 1196 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1197 { \ 1198 set_bit(bit, &EXT4_I(inode)->i_##field); \ 1199 } \ 1200 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1201 { \ 1202 clear_bit(bit, &EXT4_I(inode)->i_##field); \ 1203 } 1204 1205 EXT4_INODE_BIT_FNS(flag, flags) 1206 EXT4_INODE_BIT_FNS(state, state_flags) 1207 #else 1208 /* Assume that user mode programs are passing in an ext4fs superblock, not 1209 * a kernel struct super_block. This will allow us to call the feature-test 1210 * macros from user land. */ 1211 #define EXT4_SB(sb) (sb) 1212 #endif 1213 1214 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1215 1216 /* 1217 * Codes for operating systems 1218 */ 1219 #define EXT4_OS_LINUX 0 1220 #define EXT4_OS_HURD 1 1221 #define EXT4_OS_MASIX 2 1222 #define EXT4_OS_FREEBSD 3 1223 #define EXT4_OS_LITES 4 1224 1225 /* 1226 * Revision levels 1227 */ 1228 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1229 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1230 1231 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1232 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1233 1234 #define EXT4_GOOD_OLD_INODE_SIZE 128 1235 1236 /* 1237 * Feature set definitions 1238 */ 1239 1240 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1241 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1242 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1243 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1244 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1245 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1246 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1247 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1248 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1249 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1250 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1251 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1252 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1253 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1254 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1255 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1256 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1257 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1258 1259 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1260 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1261 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1262 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1263 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1264 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1265 1266 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1267 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1268 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1269 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1270 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1271 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1272 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1273 1274 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1275 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1276 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1277 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1278 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1279 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1280 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1281 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1282 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1283 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1284 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1285 1286 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1287 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1288 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1289 EXT4_FEATURE_INCOMPAT_META_BG| \ 1290 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1291 EXT4_FEATURE_INCOMPAT_64BIT| \ 1292 EXT4_FEATURE_INCOMPAT_FLEX_BG) 1293 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1294 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1295 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1296 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1297 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1298 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1299 EXT4_FEATURE_RO_COMPAT_HUGE_FILE) 1300 1301 /* 1302 * Default values for user and/or group using reserved blocks 1303 */ 1304 #define EXT4_DEF_RESUID 0 1305 #define EXT4_DEF_RESGID 0 1306 1307 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1308 1309 /* 1310 * Default mount options 1311 */ 1312 #define EXT4_DEFM_DEBUG 0x0001 1313 #define EXT4_DEFM_BSDGROUPS 0x0002 1314 #define EXT4_DEFM_XATTR_USER 0x0004 1315 #define EXT4_DEFM_ACL 0x0008 1316 #define EXT4_DEFM_UID16 0x0010 1317 #define EXT4_DEFM_JMODE 0x0060 1318 #define EXT4_DEFM_JMODE_DATA 0x0020 1319 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1320 #define EXT4_DEFM_JMODE_WBACK 0x0060 1321 1322 /* 1323 * Default journal batch times 1324 */ 1325 #define EXT4_DEF_MIN_BATCH_TIME 0 1326 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1327 1328 /* 1329 * Minimum number of groups in a flexgroup before we separate out 1330 * directories into the first block group of a flexgroup 1331 */ 1332 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1333 1334 /* 1335 * Structure of a directory entry 1336 */ 1337 #define EXT4_NAME_LEN 255 1338 1339 struct ext4_dir_entry { 1340 __le32 inode; /* Inode number */ 1341 __le16 rec_len; /* Directory entry length */ 1342 __le16 name_len; /* Name length */ 1343 char name[EXT4_NAME_LEN]; /* File name */ 1344 }; 1345 1346 /* 1347 * The new version of the directory entry. Since EXT4 structures are 1348 * stored in intel byte order, and the name_len field could never be 1349 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1350 * file_type field. 1351 */ 1352 struct ext4_dir_entry_2 { 1353 __le32 inode; /* Inode number */ 1354 __le16 rec_len; /* Directory entry length */ 1355 __u8 name_len; /* Name length */ 1356 __u8 file_type; 1357 char name[EXT4_NAME_LEN]; /* File name */ 1358 }; 1359 1360 /* 1361 * Ext4 directory file types. Only the low 3 bits are used. The 1362 * other bits are reserved for now. 1363 */ 1364 #define EXT4_FT_UNKNOWN 0 1365 #define EXT4_FT_REG_FILE 1 1366 #define EXT4_FT_DIR 2 1367 #define EXT4_FT_CHRDEV 3 1368 #define EXT4_FT_BLKDEV 4 1369 #define EXT4_FT_FIFO 5 1370 #define EXT4_FT_SOCK 6 1371 #define EXT4_FT_SYMLINK 7 1372 1373 #define EXT4_FT_MAX 8 1374 1375 /* 1376 * EXT4_DIR_PAD defines the directory entries boundaries 1377 * 1378 * NOTE: It must be a multiple of 4 1379 */ 1380 #define EXT4_DIR_PAD 4 1381 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1382 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1383 ~EXT4_DIR_ROUND) 1384 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1385 1386 /* 1387 * Hash Tree Directory indexing 1388 * (c) Daniel Phillips, 2001 1389 */ 1390 1391 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1392 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1393 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1394 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1395 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1396 1397 /* Legal values for the dx_root hash_version field: */ 1398 1399 #define DX_HASH_LEGACY 0 1400 #define DX_HASH_HALF_MD4 1 1401 #define DX_HASH_TEA 2 1402 #define DX_HASH_LEGACY_UNSIGNED 3 1403 #define DX_HASH_HALF_MD4_UNSIGNED 4 1404 #define DX_HASH_TEA_UNSIGNED 5 1405 1406 #ifdef __KERNEL__ 1407 1408 /* hash info structure used by the directory hash */ 1409 struct dx_hash_info 1410 { 1411 u32 hash; 1412 u32 minor_hash; 1413 int hash_version; 1414 u32 *seed; 1415 }; 1416 1417 #define EXT4_HTREE_EOF 0x7fffffff 1418 1419 /* 1420 * Control parameters used by ext4_htree_next_block 1421 */ 1422 #define HASH_NB_ALWAYS 1 1423 1424 1425 /* 1426 * Describe an inode's exact location on disk and in memory 1427 */ 1428 struct ext4_iloc 1429 { 1430 struct buffer_head *bh; 1431 unsigned long offset; 1432 ext4_group_t block_group; 1433 }; 1434 1435 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1436 { 1437 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1438 } 1439 1440 /* 1441 * This structure is stuffed into the struct file's private_data field 1442 * for directories. It is where we put information so that we can do 1443 * readdir operations in hash tree order. 1444 */ 1445 struct dir_private_info { 1446 struct rb_root root; 1447 struct rb_node *curr_node; 1448 struct fname *extra_fname; 1449 loff_t last_pos; 1450 __u32 curr_hash; 1451 __u32 curr_minor_hash; 1452 __u32 next_hash; 1453 }; 1454 1455 /* calculate the first block number of the group */ 1456 static inline ext4_fsblk_t 1457 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1458 { 1459 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1460 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1461 } 1462 1463 /* 1464 * Special error return code only used by dx_probe() and its callers. 1465 */ 1466 #define ERR_BAD_DX_DIR -75000 1467 1468 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr, 1469 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp); 1470 1471 extern struct proc_dir_entry *ext4_proc_root; 1472 1473 /* 1474 * Function prototypes 1475 */ 1476 1477 /* 1478 * Ok, these declarations are also in <linux/kernel.h> but none of the 1479 * ext4 source programs needs to include it so they are duplicated here. 1480 */ 1481 # define NORET_TYPE /**/ 1482 # define ATTRIB_NORET __attribute__((noreturn)) 1483 # define NORET_AND noreturn, 1484 1485 /* bitmap.c */ 1486 extern unsigned int ext4_count_free(struct buffer_head *, unsigned); 1487 1488 /* balloc.c */ 1489 extern unsigned int ext4_block_group(struct super_block *sb, 1490 ext4_fsblk_t blocknr); 1491 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1492 ext4_fsblk_t blocknr); 1493 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1494 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1495 ext4_group_t group); 1496 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1497 ext4_fsblk_t goal, unsigned long *count, int *errp); 1498 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1499 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1500 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb, 1501 ext4_fsblk_t block, unsigned long count); 1502 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *); 1503 extern void ext4_check_blocks_bitmap(struct super_block *); 1504 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1505 ext4_group_t block_group, 1506 struct buffer_head ** bh); 1507 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1508 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1509 ext4_group_t block_group); 1510 extern unsigned ext4_init_block_bitmap(struct super_block *sb, 1511 struct buffer_head *bh, 1512 ext4_group_t group, 1513 struct ext4_group_desc *desc); 1514 #define ext4_free_blocks_after_init(sb, group, desc) \ 1515 ext4_init_block_bitmap(sb, NULL, group, desc) 1516 1517 /* dir.c */ 1518 extern int ext4_check_dir_entry(const char *, struct inode *, 1519 struct ext4_dir_entry_2 *, 1520 struct buffer_head *, unsigned int); 1521 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1522 __u32 minor_hash, 1523 struct ext4_dir_entry_2 *dirent); 1524 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1525 1526 /* fsync.c */ 1527 extern int ext4_sync_file(struct file *, int); 1528 1529 /* hash.c */ 1530 extern int ext4fs_dirhash(const char *name, int len, struct 1531 dx_hash_info *hinfo); 1532 1533 /* ialloc.c */ 1534 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int, 1535 const struct qstr *qstr, __u32 goal); 1536 extern void ext4_free_inode(handle_t *, struct inode *); 1537 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 1538 extern unsigned long ext4_count_free_inodes(struct super_block *); 1539 extern unsigned long ext4_count_dirs(struct super_block *); 1540 extern void ext4_check_inodes_bitmap(struct super_block *); 1541 extern unsigned ext4_init_inode_bitmap(struct super_block *sb, 1542 struct buffer_head *bh, 1543 ext4_group_t group, 1544 struct ext4_group_desc *desc); 1545 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 1546 1547 /* mballoc.c */ 1548 extern long ext4_mb_stats; 1549 extern long ext4_mb_max_to_scan; 1550 extern int ext4_mb_init(struct super_block *, int); 1551 extern int ext4_mb_release(struct super_block *); 1552 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 1553 struct ext4_allocation_request *, int *); 1554 extern int ext4_mb_reserve_blocks(struct super_block *, int); 1555 extern void ext4_discard_preallocations(struct inode *); 1556 extern int __init init_ext4_mballoc(void); 1557 extern void exit_ext4_mballoc(void); 1558 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 1559 struct buffer_head *bh, ext4_fsblk_t block, 1560 unsigned long count, int flags); 1561 extern int ext4_mb_add_groupinfo(struct super_block *sb, 1562 ext4_group_t i, struct ext4_group_desc *desc); 1563 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t); 1564 extern void ext4_mb_put_buddy_cache_lock(struct super_block *, 1565 ext4_group_t, int); 1566 /* inode.c */ 1567 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 1568 ext4_lblk_t, int, int *); 1569 struct buffer_head *ext4_bread(handle_t *, struct inode *, 1570 ext4_lblk_t, int, int *); 1571 int ext4_get_block(struct inode *inode, sector_t iblock, 1572 struct buffer_head *bh_result, int create); 1573 1574 extern struct inode *ext4_iget(struct super_block *, unsigned long); 1575 extern int ext4_write_inode(struct inode *, struct writeback_control *); 1576 extern int ext4_setattr(struct dentry *, struct iattr *); 1577 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 1578 struct kstat *stat); 1579 extern void ext4_delete_inode(struct inode *); 1580 extern int ext4_sync_inode(handle_t *, struct inode *); 1581 extern void ext4_dirty_inode(struct inode *); 1582 extern int ext4_change_inode_journal_flag(struct inode *, int); 1583 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 1584 extern int ext4_can_truncate(struct inode *inode); 1585 extern void ext4_truncate(struct inode *); 1586 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 1587 extern void ext4_set_inode_flags(struct inode *); 1588 extern void ext4_get_inode_flags(struct ext4_inode_info *); 1589 extern int ext4_alloc_da_blocks(struct inode *inode); 1590 extern void ext4_set_aops(struct inode *inode); 1591 extern int ext4_writepage_trans_blocks(struct inode *); 1592 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks); 1593 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 1594 extern int ext4_block_truncate_page(handle_t *handle, 1595 struct address_space *mapping, loff_t from); 1596 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 1597 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 1598 extern int flush_completed_IO(struct inode *inode); 1599 extern void ext4_da_update_reserve_space(struct inode *inode, 1600 int used, int quota_claim); 1601 /* ioctl.c */ 1602 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 1603 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 1604 1605 /* migrate.c */ 1606 extern int ext4_ext_migrate(struct inode *); 1607 1608 /* namei.c */ 1609 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize); 1610 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize); 1611 extern int ext4_orphan_add(handle_t *, struct inode *); 1612 extern int ext4_orphan_del(handle_t *, struct inode *); 1613 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 1614 __u32 start_minor_hash, __u32 *next_hash); 1615 1616 /* resize.c */ 1617 extern int ext4_group_add(struct super_block *sb, 1618 struct ext4_new_group_data *input); 1619 extern int ext4_group_extend(struct super_block *sb, 1620 struct ext4_super_block *es, 1621 ext4_fsblk_t n_blocks_count); 1622 1623 /* super.c */ 1624 extern void __ext4_error(struct super_block *, const char *, const char *, ...) 1625 __attribute__ ((format (printf, 3, 4))); 1626 #define ext4_error(sb, message...) __ext4_error(sb, __func__, ## message) 1627 extern void ext4_error_inode(const char *, struct inode *, const char *, ...) 1628 __attribute__ ((format (printf, 3, 4))); 1629 extern void ext4_error_file(const char *, struct file *, const char *, ...) 1630 __attribute__ ((format (printf, 3, 4))); 1631 extern void __ext4_std_error(struct super_block *, const char *, int); 1632 extern void ext4_abort(struct super_block *, const char *, const char *, ...) 1633 __attribute__ ((format (printf, 3, 4))); 1634 extern void __ext4_warning(struct super_block *, const char *, 1635 const char *, ...) 1636 __attribute__ ((format (printf, 3, 4))); 1637 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, ## message) 1638 extern void ext4_msg(struct super_block *, const char *, const char *, ...) 1639 __attribute__ ((format (printf, 3, 4))); 1640 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t, 1641 const char *, const char *, ...) 1642 __attribute__ ((format (printf, 4, 5))); 1643 extern void ext4_update_dynamic_rev(struct super_block *sb); 1644 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 1645 __u32 compat); 1646 extern int ext4_update_rocompat_feature(handle_t *handle, 1647 struct super_block *sb, __u32 rocompat); 1648 extern int ext4_update_incompat_feature(handle_t *handle, 1649 struct super_block *sb, __u32 incompat); 1650 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 1651 struct ext4_group_desc *bg); 1652 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 1653 struct ext4_group_desc *bg); 1654 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 1655 struct ext4_group_desc *bg); 1656 extern __u32 ext4_free_blks_count(struct super_block *sb, 1657 struct ext4_group_desc *bg); 1658 extern __u32 ext4_free_inodes_count(struct super_block *sb, 1659 struct ext4_group_desc *bg); 1660 extern __u32 ext4_used_dirs_count(struct super_block *sb, 1661 struct ext4_group_desc *bg); 1662 extern __u32 ext4_itable_unused_count(struct super_block *sb, 1663 struct ext4_group_desc *bg); 1664 extern void ext4_block_bitmap_set(struct super_block *sb, 1665 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1666 extern void ext4_inode_bitmap_set(struct super_block *sb, 1667 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1668 extern void ext4_inode_table_set(struct super_block *sb, 1669 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1670 extern void ext4_free_blks_set(struct super_block *sb, 1671 struct ext4_group_desc *bg, __u32 count); 1672 extern void ext4_free_inodes_set(struct super_block *sb, 1673 struct ext4_group_desc *bg, __u32 count); 1674 extern void ext4_used_dirs_set(struct super_block *sb, 1675 struct ext4_group_desc *bg, __u32 count); 1676 extern void ext4_itable_unused_set(struct super_block *sb, 1677 struct ext4_group_desc *bg, __u32 count); 1678 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group, 1679 struct ext4_group_desc *gdp); 1680 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group, 1681 struct ext4_group_desc *gdp); 1682 1683 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 1684 { 1685 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 1686 le32_to_cpu(es->s_blocks_count_lo); 1687 } 1688 1689 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 1690 { 1691 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 1692 le32_to_cpu(es->s_r_blocks_count_lo); 1693 } 1694 1695 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 1696 { 1697 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 1698 le32_to_cpu(es->s_free_blocks_count_lo); 1699 } 1700 1701 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 1702 ext4_fsblk_t blk) 1703 { 1704 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 1705 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 1706 } 1707 1708 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 1709 ext4_fsblk_t blk) 1710 { 1711 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 1712 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 1713 } 1714 1715 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 1716 ext4_fsblk_t blk) 1717 { 1718 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 1719 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 1720 } 1721 1722 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 1723 { 1724 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 1725 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 1726 le32_to_cpu(raw_inode->i_size_lo); 1727 else 1728 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 1729 } 1730 1731 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 1732 { 1733 raw_inode->i_size_lo = cpu_to_le32(i_size); 1734 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 1735 } 1736 1737 static inline 1738 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 1739 ext4_group_t group) 1740 { 1741 struct ext4_group_info ***grp_info; 1742 long indexv, indexh; 1743 grp_info = EXT4_SB(sb)->s_group_info; 1744 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 1745 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 1746 return grp_info[indexv][indexh]; 1747 } 1748 1749 /* 1750 * Reading s_groups_count requires using smp_rmb() afterwards. See 1751 * the locking protocol documented in the comments of ext4_group_add() 1752 * in resize.c 1753 */ 1754 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 1755 { 1756 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 1757 1758 smp_rmb(); 1759 return ngroups; 1760 } 1761 1762 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 1763 ext4_group_t block_group) 1764 { 1765 return block_group >> sbi->s_log_groups_per_flex; 1766 } 1767 1768 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 1769 { 1770 return 1 << sbi->s_log_groups_per_flex; 1771 } 1772 1773 #define ext4_std_error(sb, errno) \ 1774 do { \ 1775 if ((errno)) \ 1776 __ext4_std_error((sb), __func__, (errno)); \ 1777 } while (0) 1778 1779 #ifdef CONFIG_SMP 1780 /* Each CPU can accumulate percpu_counter_batch blocks in their local 1781 * counters. So we need to make sure we have free blocks more 1782 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 1783 */ 1784 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 1785 #else 1786 #define EXT4_FREEBLOCKS_WATERMARK 0 1787 #endif 1788 1789 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 1790 { 1791 /* 1792 * XXX: replace with spinlock if seen contended -bzzz 1793 */ 1794 down_write(&EXT4_I(inode)->i_data_sem); 1795 if (newsize > EXT4_I(inode)->i_disksize) 1796 EXT4_I(inode)->i_disksize = newsize; 1797 up_write(&EXT4_I(inode)->i_data_sem); 1798 return ; 1799 } 1800 1801 struct ext4_group_info { 1802 unsigned long bb_state; 1803 struct rb_root bb_free_root; 1804 ext4_grpblk_t bb_first_free; /* first free block */ 1805 ext4_grpblk_t bb_free; /* total free blocks */ 1806 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 1807 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 1808 struct list_head bb_prealloc_list; 1809 #ifdef DOUBLE_CHECK 1810 void *bb_bitmap; 1811 #endif 1812 struct rw_semaphore alloc_sem; 1813 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 1814 * regions, index is order. 1815 * bb_counters[3] = 5 means 1816 * 5 free 8-block regions. */ 1817 }; 1818 1819 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 1820 1821 #define EXT4_MB_GRP_NEED_INIT(grp) \ 1822 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 1823 1824 #define EXT4_MAX_CONTENTION 8 1825 #define EXT4_CONTENTION_THRESHOLD 2 1826 1827 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 1828 ext4_group_t group) 1829 { 1830 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 1831 } 1832 1833 /* 1834 * Returns true if the filesystem is busy enough that attempts to 1835 * access the block group locks has run into contention. 1836 */ 1837 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 1838 { 1839 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 1840 } 1841 1842 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 1843 { 1844 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 1845 if (spin_trylock(lock)) 1846 /* 1847 * We're able to grab the lock right away, so drop the 1848 * lock contention counter. 1849 */ 1850 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 1851 else { 1852 /* 1853 * The lock is busy, so bump the contention counter, 1854 * and then wait on the spin lock. 1855 */ 1856 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 1857 EXT4_MAX_CONTENTION); 1858 spin_lock(lock); 1859 } 1860 } 1861 1862 static inline void ext4_unlock_group(struct super_block *sb, 1863 ext4_group_t group) 1864 { 1865 spin_unlock(ext4_group_lock_ptr(sb, group)); 1866 } 1867 1868 static inline void ext4_mark_super_dirty(struct super_block *sb) 1869 { 1870 if (EXT4_SB(sb)->s_journal == NULL) 1871 sb->s_dirt =1; 1872 } 1873 1874 /* 1875 * Inodes and files operations 1876 */ 1877 1878 /* dir.c */ 1879 extern const struct file_operations ext4_dir_operations; 1880 1881 /* file.c */ 1882 extern const struct inode_operations ext4_file_inode_operations; 1883 extern const struct file_operations ext4_file_operations; 1884 1885 /* namei.c */ 1886 extern const struct inode_operations ext4_dir_inode_operations; 1887 extern const struct inode_operations ext4_special_inode_operations; 1888 extern struct dentry *ext4_get_parent(struct dentry *child); 1889 1890 /* symlink.c */ 1891 extern const struct inode_operations ext4_symlink_inode_operations; 1892 extern const struct inode_operations ext4_fast_symlink_inode_operations; 1893 1894 /* block_validity */ 1895 extern void ext4_release_system_zone(struct super_block *sb); 1896 extern int ext4_setup_system_zone(struct super_block *sb); 1897 extern int __init init_ext4_system_zone(void); 1898 extern void exit_ext4_system_zone(void); 1899 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 1900 ext4_fsblk_t start_blk, 1901 unsigned int count); 1902 1903 /* extents.c */ 1904 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 1905 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 1906 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 1907 int chunk); 1908 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 1909 struct ext4_map_blocks *map, int flags); 1910 extern void ext4_ext_truncate(struct inode *); 1911 extern void ext4_ext_init(struct super_block *); 1912 extern void ext4_ext_release(struct super_block *); 1913 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset, 1914 loff_t len); 1915 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 1916 ssize_t len); 1917 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 1918 struct ext4_map_blocks *map, int flags); 1919 extern int ext4_get_blocks(handle_t *handle, struct inode *inode, 1920 sector_t block, unsigned int max_blocks, 1921 struct buffer_head *bh, int flags); 1922 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 1923 __u64 start, __u64 len); 1924 /* move_extent.c */ 1925 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 1926 __u64 start_orig, __u64 start_donor, 1927 __u64 len, __u64 *moved_len); 1928 1929 1930 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */ 1931 enum ext4_state_bits { 1932 BH_Uninit /* blocks are allocated but uninitialized on disk */ 1933 = BH_JBDPrivateStart, 1934 }; 1935 1936 BUFFER_FNS(Uninit, uninit) 1937 TAS_BUFFER_FNS(Uninit, uninit) 1938 1939 /* 1940 * Add new method to test wether block and inode bitmaps are properly 1941 * initialized. With uninit_bg reading the block from disk is not enough 1942 * to mark the bitmap uptodate. We need to also zero-out the bitmap 1943 */ 1944 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 1945 1946 static inline int bitmap_uptodate(struct buffer_head *bh) 1947 { 1948 return (buffer_uptodate(bh) && 1949 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 1950 } 1951 static inline void set_bitmap_uptodate(struct buffer_head *bh) 1952 { 1953 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 1954 } 1955 1956 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 1957 1958 #endif /* __KERNEL__ */ 1959 1960 #endif /* _EXT4_H */ 1961