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/version.h> 30 #include <linux/wait.h> 31 #include <linux/blockgroup_lock.h> 32 #include <linux/percpu_counter.h> 33 #include <linux/ratelimit.h> 34 #include <crypto/hash.h> 35 #include <linux/falloc.h> 36 #ifdef __KERNEL__ 37 #include <linux/compat.h> 38 #endif 39 40 /* 41 * The fourth extended filesystem constants/structures 42 */ 43 44 /* 45 * Define EXT4FS_DEBUG to produce debug messages 46 */ 47 #undef EXT4FS_DEBUG 48 49 /* 50 * Debug code 51 */ 52 #ifdef EXT4FS_DEBUG 53 #define ext4_debug(f, a...) \ 54 do { \ 55 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 56 __FILE__, __LINE__, __func__); \ 57 printk(KERN_DEBUG f, ## a); \ 58 } while (0) 59 #else 60 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 61 #endif 62 63 /* 64 * Turn on EXT_DEBUG to get lots of info about extents operations. 65 */ 66 #define EXT_DEBUG__ 67 #ifdef EXT_DEBUG 68 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 69 #else 70 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 71 #endif 72 73 /* data type for block offset of block group */ 74 typedef int ext4_grpblk_t; 75 76 /* data type for filesystem-wide blocks number */ 77 typedef unsigned long long ext4_fsblk_t; 78 79 /* data type for file logical block number */ 80 typedef __u32 ext4_lblk_t; 81 82 /* data type for block group number */ 83 typedef unsigned int ext4_group_t; 84 85 enum SHIFT_DIRECTION { 86 SHIFT_LEFT = 0, 87 SHIFT_RIGHT, 88 }; 89 90 /* 91 * Flags used in mballoc's allocation_context flags field. 92 * 93 * Also used to show what's going on for debugging purposes when the 94 * flag field is exported via the traceport interface 95 */ 96 97 /* prefer goal again. length */ 98 #define EXT4_MB_HINT_MERGE 0x0001 99 /* blocks already reserved */ 100 #define EXT4_MB_HINT_RESERVED 0x0002 101 /* metadata is being allocated */ 102 #define EXT4_MB_HINT_METADATA 0x0004 103 /* first blocks in the file */ 104 #define EXT4_MB_HINT_FIRST 0x0008 105 /* search for the best chunk */ 106 #define EXT4_MB_HINT_BEST 0x0010 107 /* data is being allocated */ 108 #define EXT4_MB_HINT_DATA 0x0020 109 /* don't preallocate (for tails) */ 110 #define EXT4_MB_HINT_NOPREALLOC 0x0040 111 /* allocate for locality group */ 112 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 113 /* allocate goal blocks or none */ 114 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 115 /* goal is meaningful */ 116 #define EXT4_MB_HINT_TRY_GOAL 0x0200 117 /* blocks already pre-reserved by delayed allocation */ 118 #define EXT4_MB_DELALLOC_RESERVED 0x0400 119 /* We are doing stream allocation */ 120 #define EXT4_MB_STREAM_ALLOC 0x0800 121 /* Use reserved root blocks if needed */ 122 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 123 /* Use blocks from reserved pool */ 124 #define EXT4_MB_USE_RESERVED 0x2000 125 126 struct ext4_allocation_request { 127 /* target inode for block we're allocating */ 128 struct inode *inode; 129 /* how many blocks we want to allocate */ 130 unsigned int len; 131 /* logical block in target inode */ 132 ext4_lblk_t logical; 133 /* the closest logical allocated block to the left */ 134 ext4_lblk_t lleft; 135 /* the closest logical allocated block to the right */ 136 ext4_lblk_t lright; 137 /* phys. target (a hint) */ 138 ext4_fsblk_t goal; 139 /* phys. block for the closest logical allocated block to the left */ 140 ext4_fsblk_t pleft; 141 /* phys. block for the closest logical allocated block to the right */ 142 ext4_fsblk_t pright; 143 /* flags. see above EXT4_MB_HINT_* */ 144 unsigned int flags; 145 }; 146 147 /* 148 * Logical to physical block mapping, used by ext4_map_blocks() 149 * 150 * This structure is used to pass requests into ext4_map_blocks() as 151 * well as to store the information returned by ext4_map_blocks(). It 152 * takes less room on the stack than a struct buffer_head. 153 */ 154 #define EXT4_MAP_NEW (1 << BH_New) 155 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 156 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 157 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 158 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 159 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 160 161 struct ext4_map_blocks { 162 ext4_fsblk_t m_pblk; 163 ext4_lblk_t m_lblk; 164 unsigned int m_len; 165 unsigned int m_flags; 166 }; 167 168 /* 169 * Flags for ext4_io_end->flags 170 */ 171 #define EXT4_IO_END_UNWRITTEN 0x0001 172 173 /* 174 * For converting unwritten extents on a work queue. 'handle' is used for 175 * buffered writeback. 176 */ 177 typedef struct ext4_io_end { 178 struct list_head list; /* per-file finished IO list */ 179 handle_t *handle; /* handle reserved for extent 180 * conversion */ 181 struct inode *inode; /* file being written to */ 182 struct bio *bio; /* Linked list of completed 183 * bios covering the extent */ 184 unsigned int flag; /* unwritten or not */ 185 loff_t offset; /* offset in the file */ 186 ssize_t size; /* size of the extent */ 187 atomic_t count; /* reference counter */ 188 } ext4_io_end_t; 189 190 struct ext4_io_submit { 191 struct writeback_control *io_wbc; 192 struct bio *io_bio; 193 ext4_io_end_t *io_end; 194 sector_t io_next_block; 195 }; 196 197 /* 198 * Special inodes numbers 199 */ 200 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 201 #define EXT4_ROOT_INO 2 /* Root inode */ 202 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 203 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 204 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 205 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 206 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 207 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 208 209 /* First non-reserved inode for old ext4 filesystems */ 210 #define EXT4_GOOD_OLD_FIRST_INO 11 211 212 /* 213 * Maximal count of links to a file 214 */ 215 #define EXT4_LINK_MAX 65000 216 217 /* 218 * Macro-instructions used to manage several block sizes 219 */ 220 #define EXT4_MIN_BLOCK_SIZE 1024 221 #define EXT4_MAX_BLOCK_SIZE 65536 222 #define EXT4_MIN_BLOCK_LOG_SIZE 10 223 #define EXT4_MAX_BLOCK_LOG_SIZE 16 224 #ifdef __KERNEL__ 225 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 226 #else 227 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 228 #endif 229 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 230 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 231 EXT4_SB(s)->s_cluster_bits) 232 #ifdef __KERNEL__ 233 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 234 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 235 #else 236 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 237 #endif 238 #ifdef __KERNEL__ 239 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 240 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 241 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 242 #else 243 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 244 EXT4_GOOD_OLD_INODE_SIZE : \ 245 (s)->s_inode_size) 246 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 247 EXT4_GOOD_OLD_FIRST_INO : \ 248 (s)->s_first_ino) 249 #endif 250 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 251 252 /* Translate a block number to a cluster number */ 253 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 254 /* Translate a cluster number to a block number */ 255 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 256 /* Translate # of blks to # of clusters */ 257 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 258 (sbi)->s_cluster_bits) 259 /* Mask out the low bits to get the starting block of the cluster */ 260 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 261 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 262 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 263 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 264 /* Get the cluster offset */ 265 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 266 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 267 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 268 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 269 270 /* 271 * Structure of a blocks group descriptor 272 */ 273 struct ext4_group_desc 274 { 275 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 276 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 277 __le32 bg_inode_table_lo; /* Inodes table block */ 278 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 279 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 280 __le16 bg_used_dirs_count_lo; /* Directories count */ 281 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 282 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 283 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 284 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 285 __le16 bg_itable_unused_lo; /* Unused inodes count */ 286 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 287 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 288 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 289 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 290 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 291 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 292 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 293 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 294 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 295 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 296 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 297 __u32 bg_reserved; 298 }; 299 300 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 301 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 302 sizeof(__le16)) 303 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 304 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 305 sizeof(__le16)) 306 307 /* 308 * Structure of a flex block group info 309 */ 310 311 struct flex_groups { 312 atomic64_t free_clusters; 313 atomic_t free_inodes; 314 atomic_t used_dirs; 315 }; 316 317 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 318 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 319 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 320 321 /* 322 * Macro-instructions used to manage group descriptors 323 */ 324 #define EXT4_MIN_DESC_SIZE 32 325 #define EXT4_MIN_DESC_SIZE_64BIT 64 326 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 327 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 328 #ifdef __KERNEL__ 329 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 330 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 331 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 332 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 333 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 334 #else 335 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 336 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 337 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 338 #endif 339 340 /* 341 * Constants relative to the data blocks 342 */ 343 #define EXT4_NDIR_BLOCKS 12 344 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 345 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 346 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 347 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 348 349 /* 350 * Inode flags 351 */ 352 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 353 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 354 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 355 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 356 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 357 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 358 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 359 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 360 /* Reserved for compression usage... */ 361 #define EXT4_DIRTY_FL 0x00000100 362 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 363 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 364 /* nb: was previously EXT2_ECOMPR_FL */ 365 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 366 /* End compression flags --- maybe not all used */ 367 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 368 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 369 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 370 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 371 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 372 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 373 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 374 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 375 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 376 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 377 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 378 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 379 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 380 381 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 382 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */ 383 384 /* Flags that should be inherited by new inodes from their parent. */ 385 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 386 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 387 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 388 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 389 390 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 391 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 392 393 /* Flags that are appropriate for non-directories/regular files. */ 394 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 395 396 /* Mask out flags that are inappropriate for the given type of inode. */ 397 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 398 { 399 if (S_ISDIR(mode)) 400 return flags; 401 else if (S_ISREG(mode)) 402 return flags & EXT4_REG_FLMASK; 403 else 404 return flags & EXT4_OTHER_FLMASK; 405 } 406 407 /* 408 * Inode flags used for atomic set/get 409 */ 410 enum { 411 EXT4_INODE_SECRM = 0, /* Secure deletion */ 412 EXT4_INODE_UNRM = 1, /* Undelete */ 413 EXT4_INODE_COMPR = 2, /* Compress file */ 414 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 415 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 416 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 417 EXT4_INODE_NODUMP = 6, /* do not dump file */ 418 EXT4_INODE_NOATIME = 7, /* do not update atime */ 419 /* Reserved for compression usage... */ 420 EXT4_INODE_DIRTY = 8, 421 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 422 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 423 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 424 /* End compression flags --- maybe not all used */ 425 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 426 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 427 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 428 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 429 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 430 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 431 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 432 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 433 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 434 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 435 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 436 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 437 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 438 }; 439 440 /* 441 * Since it's pretty easy to mix up bit numbers and hex values, we use a 442 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 443 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 444 * any extra space in the compiled kernel image, otherwise, the build will fail. 445 * It's important that these values are the same, since we are using 446 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 447 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 448 * values found in ext2, ext3 and ext4 filesystems, and of course the values 449 * defined in e2fsprogs. 450 * 451 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 452 */ 453 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 454 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 455 456 static inline void ext4_check_flag_values(void) 457 { 458 CHECK_FLAG_VALUE(SECRM); 459 CHECK_FLAG_VALUE(UNRM); 460 CHECK_FLAG_VALUE(COMPR); 461 CHECK_FLAG_VALUE(SYNC); 462 CHECK_FLAG_VALUE(IMMUTABLE); 463 CHECK_FLAG_VALUE(APPEND); 464 CHECK_FLAG_VALUE(NODUMP); 465 CHECK_FLAG_VALUE(NOATIME); 466 CHECK_FLAG_VALUE(DIRTY); 467 CHECK_FLAG_VALUE(COMPRBLK); 468 CHECK_FLAG_VALUE(NOCOMPR); 469 CHECK_FLAG_VALUE(ENCRYPT); 470 CHECK_FLAG_VALUE(INDEX); 471 CHECK_FLAG_VALUE(IMAGIC); 472 CHECK_FLAG_VALUE(JOURNAL_DATA); 473 CHECK_FLAG_VALUE(NOTAIL); 474 CHECK_FLAG_VALUE(DIRSYNC); 475 CHECK_FLAG_VALUE(TOPDIR); 476 CHECK_FLAG_VALUE(HUGE_FILE); 477 CHECK_FLAG_VALUE(EXTENTS); 478 CHECK_FLAG_VALUE(EA_INODE); 479 CHECK_FLAG_VALUE(EOFBLOCKS); 480 CHECK_FLAG_VALUE(INLINE_DATA); 481 CHECK_FLAG_VALUE(PROJINHERIT); 482 CHECK_FLAG_VALUE(RESERVED); 483 } 484 485 /* Used to pass group descriptor data when online resize is done */ 486 struct ext4_new_group_input { 487 __u32 group; /* Group number for this data */ 488 __u64 block_bitmap; /* Absolute block number of block bitmap */ 489 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 490 __u64 inode_table; /* Absolute block number of inode table start */ 491 __u32 blocks_count; /* Total number of blocks in this group */ 492 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 493 __u16 unused; 494 }; 495 496 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 497 struct compat_ext4_new_group_input { 498 u32 group; 499 compat_u64 block_bitmap; 500 compat_u64 inode_bitmap; 501 compat_u64 inode_table; 502 u32 blocks_count; 503 u16 reserved_blocks; 504 u16 unused; 505 }; 506 #endif 507 508 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 509 struct ext4_new_group_data { 510 __u32 group; 511 __u64 block_bitmap; 512 __u64 inode_bitmap; 513 __u64 inode_table; 514 __u32 blocks_count; 515 __u16 reserved_blocks; 516 __u16 unused; 517 __u32 free_blocks_count; 518 }; 519 520 /* Indexes used to index group tables in ext4_new_group_data */ 521 enum { 522 BLOCK_BITMAP = 0, /* block bitmap */ 523 INODE_BITMAP, /* inode bitmap */ 524 INODE_TABLE, /* inode tables */ 525 GROUP_TABLE_COUNT, 526 }; 527 528 /* 529 * Flags used by ext4_map_blocks() 530 */ 531 /* Allocate any needed blocks and/or convert an unwritten 532 extent to be an initialized ext4 */ 533 #define EXT4_GET_BLOCKS_CREATE 0x0001 534 /* Request the creation of an unwritten extent */ 535 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 536 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 537 EXT4_GET_BLOCKS_CREATE) 538 /* Caller is from the delayed allocation writeout path 539 * finally doing the actual allocation of delayed blocks */ 540 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 541 /* caller is from the direct IO path, request to creation of an 542 unwritten extents if not allocated, split the unwritten 543 extent if blocks has been preallocated already*/ 544 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 545 #define EXT4_GET_BLOCKS_CONVERT 0x0010 546 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 547 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 548 /* Convert extent to initialized after IO complete */ 549 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 550 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 551 /* Eventual metadata allocation (due to growing extent tree) 552 * should not fail, so try to use reserved blocks for that.*/ 553 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 554 /* Don't normalize allocation size (used for fallocate) */ 555 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 556 /* Request will not result in inode size update (user for fallocate) */ 557 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 558 /* Do not take i_data_sem locking in ext4_map_blocks */ 559 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 560 /* Convert written extents to unwritten */ 561 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200 562 563 /* 564 * The bit position of these flags must not overlap with any of the 565 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 566 * read_extent_tree_block(), ext4_split_extent_at(), 567 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 568 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 569 * caching the extents when reading from the extent tree while a 570 * truncate or punch hole operation is in progress. 571 */ 572 #define EXT4_EX_NOCACHE 0x40000000 573 #define EXT4_EX_FORCE_CACHE 0x20000000 574 575 /* 576 * Flags used by ext4_free_blocks 577 */ 578 #define EXT4_FREE_BLOCKS_METADATA 0x0001 579 #define EXT4_FREE_BLOCKS_FORGET 0x0002 580 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 581 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 582 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 583 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 584 585 /* Encryption algorithms */ 586 #define EXT4_ENCRYPTION_MODE_INVALID 0 587 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1 588 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2 589 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3 590 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4 591 592 #include "ext4_crypto.h" 593 594 /* 595 * ioctl commands 596 */ 597 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 598 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 599 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 600 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 601 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 602 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 603 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 604 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 605 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 606 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 607 #define EXT4_IOC_MIGRATE _IO('f', 9) 608 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 609 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 610 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 611 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 612 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 613 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 614 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 615 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy) 616 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16]) 617 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy) 618 619 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 620 /* 621 * ioctl commands in 32 bit emulation 622 */ 623 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 624 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 625 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 626 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 627 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 628 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 629 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 630 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 631 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 632 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 633 #endif 634 635 /* Max physical block we can address w/o extents */ 636 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 637 638 /* 639 * Structure of an inode on the disk 640 */ 641 struct ext4_inode { 642 __le16 i_mode; /* File mode */ 643 __le16 i_uid; /* Low 16 bits of Owner Uid */ 644 __le32 i_size_lo; /* Size in bytes */ 645 __le32 i_atime; /* Access time */ 646 __le32 i_ctime; /* Inode Change time */ 647 __le32 i_mtime; /* Modification time */ 648 __le32 i_dtime; /* Deletion Time */ 649 __le16 i_gid; /* Low 16 bits of Group Id */ 650 __le16 i_links_count; /* Links count */ 651 __le32 i_blocks_lo; /* Blocks count */ 652 __le32 i_flags; /* File flags */ 653 union { 654 struct { 655 __le32 l_i_version; 656 } linux1; 657 struct { 658 __u32 h_i_translator; 659 } hurd1; 660 struct { 661 __u32 m_i_reserved1; 662 } masix1; 663 } osd1; /* OS dependent 1 */ 664 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 665 __le32 i_generation; /* File version (for NFS) */ 666 __le32 i_file_acl_lo; /* File ACL */ 667 __le32 i_size_high; 668 __le32 i_obso_faddr; /* Obsoleted fragment address */ 669 union { 670 struct { 671 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 672 __le16 l_i_file_acl_high; 673 __le16 l_i_uid_high; /* these 2 fields */ 674 __le16 l_i_gid_high; /* were reserved2[0] */ 675 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 676 __le16 l_i_reserved; 677 } linux2; 678 struct { 679 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 680 __u16 h_i_mode_high; 681 __u16 h_i_uid_high; 682 __u16 h_i_gid_high; 683 __u32 h_i_author; 684 } hurd2; 685 struct { 686 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 687 __le16 m_i_file_acl_high; 688 __u32 m_i_reserved2[2]; 689 } masix2; 690 } osd2; /* OS dependent 2 */ 691 __le16 i_extra_isize; 692 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 693 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 694 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 695 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 696 __le32 i_crtime; /* File Creation time */ 697 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 698 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 699 __le32 i_projid; /* Project ID */ 700 }; 701 702 struct move_extent { 703 __u32 reserved; /* should be zero */ 704 __u32 donor_fd; /* donor file descriptor */ 705 __u64 orig_start; /* logical start offset in block for orig */ 706 __u64 donor_start; /* logical start offset in block for donor */ 707 __u64 len; /* block length to be moved */ 708 __u64 moved_len; /* moved block length */ 709 }; 710 711 #define EXT4_EPOCH_BITS 2 712 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 713 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 714 715 /* 716 * Extended fields will fit into an inode if the filesystem was formatted 717 * with large inodes (-I 256 or larger) and there are not currently any EAs 718 * consuming all of the available space. For new inodes we always reserve 719 * enough space for the kernel's known extended fields, but for inodes 720 * created with an old kernel this might not have been the case. None of 721 * the extended inode fields is critical for correct filesystem operation. 722 * This macro checks if a certain field fits in the inode. Note that 723 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 724 */ 725 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 726 ((offsetof(typeof(*ext4_inode), field) + \ 727 sizeof((ext4_inode)->field)) \ 728 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 729 (einode)->i_extra_isize)) \ 730 731 /* 732 * We use an encoding that preserves the times for extra epoch "00": 733 * 734 * extra msb of adjust for signed 735 * epoch 32-bit 32-bit tv_sec to 736 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 737 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 738 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 739 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 740 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 741 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 742 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 743 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 744 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 745 * 746 * Note that previous versions of the kernel on 64-bit systems would 747 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 748 * 1970. e2fsck will correct this, assuming that it is run on the 749 * affected filesystem before 2242. 750 */ 751 752 static inline __le32 ext4_encode_extra_time(struct timespec *time) 753 { 754 u32 extra = sizeof(time->tv_sec) > 4 ? 755 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0; 756 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 757 } 758 759 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 760 { 761 if (unlikely(sizeof(time->tv_sec) > 4 && 762 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) { 763 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0) 764 /* Handle legacy encoding of pre-1970 dates with epoch 765 * bits 1,1. We assume that by kernel version 4.20, 766 * everyone will have run fsck over the affected 767 * filesystems to correct the problem. (This 768 * backwards compatibility may be removed before this 769 * time, at the discretion of the ext4 developers.) 770 */ 771 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 772 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 773 extra_bits = 0; 774 time->tv_sec += extra_bits << 32; 775 #else 776 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 777 #endif 778 } 779 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 780 } 781 782 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 783 do { \ 784 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 785 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 786 (raw_inode)->xtime ## _extra = \ 787 ext4_encode_extra_time(&(inode)->xtime); \ 788 } while (0) 789 790 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 791 do { \ 792 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 793 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 794 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 795 (raw_inode)->xtime ## _extra = \ 796 ext4_encode_extra_time(&(einode)->xtime); \ 797 } while (0) 798 799 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 800 do { \ 801 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 802 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 803 ext4_decode_extra_time(&(inode)->xtime, \ 804 raw_inode->xtime ## _extra); \ 805 else \ 806 (inode)->xtime.tv_nsec = 0; \ 807 } while (0) 808 809 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 810 do { \ 811 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 812 (einode)->xtime.tv_sec = \ 813 (signed)le32_to_cpu((raw_inode)->xtime); \ 814 else \ 815 (einode)->xtime.tv_sec = 0; \ 816 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 817 ext4_decode_extra_time(&(einode)->xtime, \ 818 raw_inode->xtime ## _extra); \ 819 else \ 820 (einode)->xtime.tv_nsec = 0; \ 821 } while (0) 822 823 #define i_disk_version osd1.linux1.l_i_version 824 825 #if defined(__KERNEL__) || defined(__linux__) 826 #define i_reserved1 osd1.linux1.l_i_reserved1 827 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 828 #define i_blocks_high osd2.linux2.l_i_blocks_high 829 #define i_uid_low i_uid 830 #define i_gid_low i_gid 831 #define i_uid_high osd2.linux2.l_i_uid_high 832 #define i_gid_high osd2.linux2.l_i_gid_high 833 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 834 835 #elif defined(__GNU__) 836 837 #define i_translator osd1.hurd1.h_i_translator 838 #define i_uid_high osd2.hurd2.h_i_uid_high 839 #define i_gid_high osd2.hurd2.h_i_gid_high 840 #define i_author osd2.hurd2.h_i_author 841 842 #elif defined(__masix__) 843 844 #define i_reserved1 osd1.masix1.m_i_reserved1 845 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 846 #define i_reserved2 osd2.masix2.m_i_reserved2 847 848 #endif /* defined(__KERNEL__) || defined(__linux__) */ 849 850 #include "extents_status.h" 851 852 /* 853 * fourth extended file system inode data in memory 854 */ 855 struct ext4_inode_info { 856 __le32 i_data[15]; /* unconverted */ 857 __u32 i_dtime; 858 ext4_fsblk_t i_file_acl; 859 860 /* 861 * i_block_group is the number of the block group which contains 862 * this file's inode. Constant across the lifetime of the inode, 863 * it is ued for making block allocation decisions - we try to 864 * place a file's data blocks near its inode block, and new inodes 865 * near to their parent directory's inode. 866 */ 867 ext4_group_t i_block_group; 868 ext4_lblk_t i_dir_start_lookup; 869 #if (BITS_PER_LONG < 64) 870 unsigned long i_state_flags; /* Dynamic state flags */ 871 #endif 872 unsigned long i_flags; 873 874 /* 875 * Extended attributes can be read independently of the main file 876 * data. Taking i_mutex even when reading would cause contention 877 * between readers of EAs and writers of regular file data, so 878 * instead we synchronize on xattr_sem when reading or changing 879 * EAs. 880 */ 881 struct rw_semaphore xattr_sem; 882 883 struct list_head i_orphan; /* unlinked but open inodes */ 884 885 /* 886 * i_disksize keeps track of what the inode size is ON DISK, not 887 * in memory. During truncate, i_size is set to the new size by 888 * the VFS prior to calling ext4_truncate(), but the filesystem won't 889 * set i_disksize to 0 until the truncate is actually under way. 890 * 891 * The intent is that i_disksize always represents the blocks which 892 * are used by this file. This allows recovery to restart truncate 893 * on orphans if we crash during truncate. We actually write i_disksize 894 * into the on-disk inode when writing inodes out, instead of i_size. 895 * 896 * The only time when i_disksize and i_size may be different is when 897 * a truncate is in progress. The only things which change i_disksize 898 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 899 */ 900 loff_t i_disksize; 901 902 /* 903 * i_data_sem is for serialising ext4_truncate() against 904 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 905 * data tree are chopped off during truncate. We can't do that in 906 * ext4 because whenever we perform intermediate commits during 907 * truncate, the inode and all the metadata blocks *must* be in a 908 * consistent state which allows truncation of the orphans to restart 909 * during recovery. Hence we must fix the get_block-vs-truncate race 910 * by other means, so we have i_data_sem. 911 */ 912 struct rw_semaphore i_data_sem; 913 /* 914 * i_mmap_sem is for serializing page faults with truncate / punch hole 915 * operations. We have to make sure that new page cannot be faulted in 916 * a section of the inode that is being punched. We cannot easily use 917 * i_data_sem for this since we need protection for the whole punch 918 * operation and i_data_sem ranks below transaction start so we have 919 * to occasionally drop it. 920 */ 921 struct rw_semaphore i_mmap_sem; 922 struct inode vfs_inode; 923 struct jbd2_inode *jinode; 924 925 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 926 927 /* 928 * File creation time. Its function is same as that of 929 * struct timespec i_{a,c,m}time in the generic inode. 930 */ 931 struct timespec i_crtime; 932 933 /* mballoc */ 934 struct list_head i_prealloc_list; 935 spinlock_t i_prealloc_lock; 936 937 /* extents status tree */ 938 struct ext4_es_tree i_es_tree; 939 rwlock_t i_es_lock; 940 struct list_head i_es_list; 941 unsigned int i_es_all_nr; /* protected by i_es_lock */ 942 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 943 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 944 extents to shrink. Protected by 945 i_es_lock */ 946 947 /* ialloc */ 948 ext4_group_t i_last_alloc_group; 949 950 /* allocation reservation info for delalloc */ 951 /* In case of bigalloc, these refer to clusters rather than blocks */ 952 unsigned int i_reserved_data_blocks; 953 unsigned int i_reserved_meta_blocks; 954 unsigned int i_allocated_meta_blocks; 955 ext4_lblk_t i_da_metadata_calc_last_lblock; 956 int i_da_metadata_calc_len; 957 958 /* on-disk additional length */ 959 __u16 i_extra_isize; 960 961 /* Indicate the inline data space. */ 962 u16 i_inline_off; 963 u16 i_inline_size; 964 965 #ifdef CONFIG_QUOTA 966 /* quota space reservation, managed internally by quota code */ 967 qsize_t i_reserved_quota; 968 #endif 969 970 /* Lock protecting lists below */ 971 spinlock_t i_completed_io_lock; 972 /* 973 * Completed IOs that need unwritten extents handling and have 974 * transaction reserved 975 */ 976 struct list_head i_rsv_conversion_list; 977 /* 978 * Completed IOs that need unwritten extents handling and don't have 979 * transaction reserved 980 */ 981 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 982 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 983 struct work_struct i_rsv_conversion_work; 984 985 spinlock_t i_block_reservation_lock; 986 987 /* 988 * Transactions that contain inode's metadata needed to complete 989 * fsync and fdatasync, respectively. 990 */ 991 tid_t i_sync_tid; 992 tid_t i_datasync_tid; 993 994 #ifdef CONFIG_QUOTA 995 struct dquot *i_dquot[MAXQUOTAS]; 996 #endif 997 998 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 999 __u32 i_csum_seed; 1000 1001 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1002 /* Encryption params */ 1003 struct ext4_crypt_info *i_crypt_info; 1004 #endif 1005 }; 1006 1007 /* 1008 * File system states 1009 */ 1010 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1011 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1012 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1013 1014 /* 1015 * Misc. filesystem flags 1016 */ 1017 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1018 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1019 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1020 1021 /* 1022 * Mount flags set via mount options or defaults 1023 */ 1024 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1025 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1026 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1027 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1028 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1029 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1030 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1031 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1032 #ifdef CONFIG_FS_DAX 1033 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1034 #else 1035 #define EXT4_MOUNT_DAX 0 1036 #endif 1037 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1038 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1039 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1040 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1041 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1042 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1043 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1044 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1045 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1046 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1047 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 1048 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 1049 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 1050 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1051 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1052 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1053 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1054 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1055 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1056 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1057 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1058 1059 /* 1060 * Mount flags set either automatically (could not be set by mount option) 1061 * based on per file system feature or property or in special cases such as 1062 * distinguishing between explicit mount option definition and default. 1063 */ 1064 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1065 specified delalloc */ 1066 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1067 size of blocksize * 8 1068 blocks */ 1069 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1070 file systems */ 1071 1072 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1073 specified journal checksum */ 1074 1075 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1076 ~EXT4_MOUNT_##opt 1077 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1078 EXT4_MOUNT_##opt 1079 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1080 EXT4_MOUNT_##opt) 1081 1082 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1083 ~EXT4_MOUNT2_##opt 1084 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1085 EXT4_MOUNT2_##opt 1086 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1087 EXT4_MOUNT2_##opt) 1088 1089 #define ext4_test_and_set_bit __test_and_set_bit_le 1090 #define ext4_set_bit __set_bit_le 1091 #define ext4_set_bit_atomic ext2_set_bit_atomic 1092 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1093 #define ext4_clear_bit __clear_bit_le 1094 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1095 #define ext4_test_bit test_bit_le 1096 #define ext4_find_next_zero_bit find_next_zero_bit_le 1097 #define ext4_find_next_bit find_next_bit_le 1098 1099 extern void ext4_set_bits(void *bm, int cur, int len); 1100 1101 /* 1102 * Maximal mount counts between two filesystem checks 1103 */ 1104 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1105 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1106 1107 /* 1108 * Behaviour when detecting errors 1109 */ 1110 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1111 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1112 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1113 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1114 1115 /* Metadata checksum algorithm codes */ 1116 #define EXT4_CRC32C_CHKSUM 1 1117 1118 /* 1119 * Structure of the super block 1120 */ 1121 struct ext4_super_block { 1122 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1123 __le32 s_blocks_count_lo; /* Blocks count */ 1124 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1125 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1126 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1127 __le32 s_first_data_block; /* First Data Block */ 1128 __le32 s_log_block_size; /* Block size */ 1129 __le32 s_log_cluster_size; /* Allocation cluster size */ 1130 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1131 __le32 s_clusters_per_group; /* # Clusters per group */ 1132 __le32 s_inodes_per_group; /* # Inodes per group */ 1133 __le32 s_mtime; /* Mount time */ 1134 /*30*/ __le32 s_wtime; /* Write time */ 1135 __le16 s_mnt_count; /* Mount count */ 1136 __le16 s_max_mnt_count; /* Maximal mount count */ 1137 __le16 s_magic; /* Magic signature */ 1138 __le16 s_state; /* File system state */ 1139 __le16 s_errors; /* Behaviour when detecting errors */ 1140 __le16 s_minor_rev_level; /* minor revision level */ 1141 /*40*/ __le32 s_lastcheck; /* time of last check */ 1142 __le32 s_checkinterval; /* max. time between checks */ 1143 __le32 s_creator_os; /* OS */ 1144 __le32 s_rev_level; /* Revision level */ 1145 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1146 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1147 /* 1148 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1149 * 1150 * Note: the difference between the compatible feature set and 1151 * the incompatible feature set is that if there is a bit set 1152 * in the incompatible feature set that the kernel doesn't 1153 * know about, it should refuse to mount the filesystem. 1154 * 1155 * e2fsck's requirements are more strict; if it doesn't know 1156 * about a feature in either the compatible or incompatible 1157 * feature set, it must abort and not try to meddle with 1158 * things it doesn't understand... 1159 */ 1160 __le32 s_first_ino; /* First non-reserved inode */ 1161 __le16 s_inode_size; /* size of inode structure */ 1162 __le16 s_block_group_nr; /* block group # of this superblock */ 1163 __le32 s_feature_compat; /* compatible feature set */ 1164 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1165 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1166 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1167 /*78*/ char s_volume_name[16]; /* volume name */ 1168 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1169 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1170 /* 1171 * Performance hints. Directory preallocation should only 1172 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1173 */ 1174 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1175 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1176 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1177 /* 1178 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1179 */ 1180 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1181 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1182 __le32 s_journal_dev; /* device number of journal file */ 1183 __le32 s_last_orphan; /* start of list of inodes to delete */ 1184 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1185 __u8 s_def_hash_version; /* Default hash version to use */ 1186 __u8 s_jnl_backup_type; 1187 __le16 s_desc_size; /* size of group descriptor */ 1188 /*100*/ __le32 s_default_mount_opts; 1189 __le32 s_first_meta_bg; /* First metablock block group */ 1190 __le32 s_mkfs_time; /* When the filesystem was created */ 1191 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1192 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1193 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1194 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1195 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1196 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1197 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1198 __le32 s_flags; /* Miscellaneous flags */ 1199 __le16 s_raid_stride; /* RAID stride */ 1200 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1201 __le64 s_mmp_block; /* Block for multi-mount protection */ 1202 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1203 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1204 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1205 __u8 s_encryption_level; /* versioning level for encryption */ 1206 __u8 s_reserved_pad; /* Padding to next 32bits */ 1207 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1208 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1209 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1210 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1211 snapshot's future use */ 1212 __le32 s_snapshot_list; /* inode number of the head of the 1213 on-disk snapshot list */ 1214 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1215 __le32 s_error_count; /* number of fs errors */ 1216 __le32 s_first_error_time; /* first time an error happened */ 1217 __le32 s_first_error_ino; /* inode involved in first error */ 1218 __le64 s_first_error_block; /* block involved of first error */ 1219 __u8 s_first_error_func[32]; /* function where the error happened */ 1220 __le32 s_first_error_line; /* line number where error happened */ 1221 __le32 s_last_error_time; /* most recent time of an error */ 1222 __le32 s_last_error_ino; /* inode involved in last error */ 1223 __le32 s_last_error_line; /* line number where error happened */ 1224 __le64 s_last_error_block; /* block involved of last error */ 1225 __u8 s_last_error_func[32]; /* function where the error happened */ 1226 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1227 __u8 s_mount_opts[64]; 1228 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1229 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1230 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1231 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1232 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1233 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1234 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1235 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1236 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1237 __le32 s_reserved[98]; /* Padding to the end of the block */ 1238 __le32 s_checksum; /* crc32c(superblock) */ 1239 }; 1240 1241 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1242 1243 #ifdef __KERNEL__ 1244 1245 /* 1246 * run-time mount flags 1247 */ 1248 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1249 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1250 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1251 1252 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1253 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1254 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1255 #else 1256 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1257 #endif 1258 1259 /* Number of quota types we support */ 1260 #define EXT4_MAXQUOTAS 2 1261 1262 /* 1263 * fourth extended-fs super-block data in memory 1264 */ 1265 struct ext4_sb_info { 1266 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1267 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1268 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1269 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1270 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1271 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1272 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1273 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1274 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1275 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1276 unsigned long s_overhead; /* # of fs overhead clusters */ 1277 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1278 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1279 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1280 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1281 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1282 struct buffer_head **s_group_desc; 1283 unsigned int s_mount_opt; 1284 unsigned int s_mount_opt2; 1285 unsigned int s_mount_flags; 1286 unsigned int s_def_mount_opt; 1287 ext4_fsblk_t s_sb_block; 1288 atomic64_t s_resv_clusters; 1289 kuid_t s_resuid; 1290 kgid_t s_resgid; 1291 unsigned short s_mount_state; 1292 unsigned short s_pad; 1293 int s_addr_per_block_bits; 1294 int s_desc_per_block_bits; 1295 int s_inode_size; 1296 int s_first_ino; 1297 unsigned int s_inode_readahead_blks; 1298 unsigned int s_inode_goal; 1299 spinlock_t s_next_gen_lock; 1300 u32 s_next_generation; 1301 u32 s_hash_seed[4]; 1302 int s_def_hash_version; 1303 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1304 struct percpu_counter s_freeclusters_counter; 1305 struct percpu_counter s_freeinodes_counter; 1306 struct percpu_counter s_dirs_counter; 1307 struct percpu_counter s_dirtyclusters_counter; 1308 struct blockgroup_lock *s_blockgroup_lock; 1309 struct proc_dir_entry *s_proc; 1310 struct kobject s_kobj; 1311 struct completion s_kobj_unregister; 1312 struct super_block *s_sb; 1313 1314 /* Journaling */ 1315 struct journal_s *s_journal; 1316 struct list_head s_orphan; 1317 struct mutex s_orphan_lock; 1318 unsigned long s_resize_flags; /* Flags indicating if there 1319 is a resizer */ 1320 unsigned long s_commit_interval; 1321 u32 s_max_batch_time; 1322 u32 s_min_batch_time; 1323 struct block_device *journal_bdev; 1324 #ifdef CONFIG_QUOTA 1325 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1326 int s_jquota_fmt; /* Format of quota to use */ 1327 #endif 1328 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1329 struct rb_root system_blks; 1330 1331 #ifdef EXTENTS_STATS 1332 /* ext4 extents stats */ 1333 unsigned long s_ext_min; 1334 unsigned long s_ext_max; 1335 unsigned long s_depth_max; 1336 spinlock_t s_ext_stats_lock; 1337 unsigned long s_ext_blocks; 1338 unsigned long s_ext_extents; 1339 #endif 1340 1341 /* for buddy allocator */ 1342 struct ext4_group_info ***s_group_info; 1343 struct inode *s_buddy_cache; 1344 spinlock_t s_md_lock; 1345 unsigned short *s_mb_offsets; 1346 unsigned int *s_mb_maxs; 1347 unsigned int s_group_info_size; 1348 1349 /* tunables */ 1350 unsigned long s_stripe; 1351 unsigned int s_mb_stream_request; 1352 unsigned int s_mb_max_to_scan; 1353 unsigned int s_mb_min_to_scan; 1354 unsigned int s_mb_stats; 1355 unsigned int s_mb_order2_reqs; 1356 unsigned int s_mb_group_prealloc; 1357 unsigned int s_max_dir_size_kb; 1358 /* where last allocation was done - for stream allocation */ 1359 unsigned long s_mb_last_group; 1360 unsigned long s_mb_last_start; 1361 1362 /* stats for buddy allocator */ 1363 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1364 atomic_t s_bal_success; /* we found long enough chunks */ 1365 atomic_t s_bal_allocated; /* in blocks */ 1366 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1367 atomic_t s_bal_goals; /* goal hits */ 1368 atomic_t s_bal_breaks; /* too long searches */ 1369 atomic_t s_bal_2orders; /* 2^order hits */ 1370 spinlock_t s_bal_lock; 1371 unsigned long s_mb_buddies_generated; 1372 unsigned long long s_mb_generation_time; 1373 atomic_t s_mb_lost_chunks; 1374 atomic_t s_mb_preallocated; 1375 atomic_t s_mb_discarded; 1376 atomic_t s_lock_busy; 1377 1378 /* locality groups */ 1379 struct ext4_locality_group __percpu *s_locality_groups; 1380 1381 /* for write statistics */ 1382 unsigned long s_sectors_written_start; 1383 u64 s_kbytes_written; 1384 1385 /* the size of zero-out chunk */ 1386 unsigned int s_extent_max_zeroout_kb; 1387 1388 unsigned int s_log_groups_per_flex; 1389 struct flex_groups *s_flex_groups; 1390 ext4_group_t s_flex_groups_allocated; 1391 1392 /* workqueue for reserved extent conversions (buffered io) */ 1393 struct workqueue_struct *rsv_conversion_wq; 1394 1395 /* timer for periodic error stats printing */ 1396 struct timer_list s_err_report; 1397 1398 /* Lazy inode table initialization info */ 1399 struct ext4_li_request *s_li_request; 1400 /* Wait multiplier for lazy initialization thread */ 1401 unsigned int s_li_wait_mult; 1402 1403 /* Kernel thread for multiple mount protection */ 1404 struct task_struct *s_mmp_tsk; 1405 1406 /* record the last minlen when FITRIM is called. */ 1407 atomic_t s_last_trim_minblks; 1408 1409 /* Reference to checksum algorithm driver via cryptoapi */ 1410 struct crypto_shash *s_chksum_driver; 1411 1412 /* Precomputed FS UUID checksum for seeding other checksums */ 1413 __u32 s_csum_seed; 1414 1415 /* Reclaim extents from extent status tree */ 1416 struct shrinker s_es_shrinker; 1417 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1418 long s_es_nr_inode; 1419 struct ext4_es_stats s_es_stats; 1420 struct mb_cache *s_mb_cache; 1421 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1422 1423 /* Ratelimit ext4 messages. */ 1424 struct ratelimit_state s_err_ratelimit_state; 1425 struct ratelimit_state s_warning_ratelimit_state; 1426 struct ratelimit_state s_msg_ratelimit_state; 1427 }; 1428 1429 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1430 { 1431 return sb->s_fs_info; 1432 } 1433 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1434 { 1435 return container_of(inode, struct ext4_inode_info, vfs_inode); 1436 } 1437 1438 static inline struct timespec ext4_current_time(struct inode *inode) 1439 { 1440 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1441 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1442 } 1443 1444 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1445 { 1446 return ino == EXT4_ROOT_INO || 1447 ino == EXT4_USR_QUOTA_INO || 1448 ino == EXT4_GRP_QUOTA_INO || 1449 ino == EXT4_BOOT_LOADER_INO || 1450 ino == EXT4_JOURNAL_INO || 1451 ino == EXT4_RESIZE_INO || 1452 (ino >= EXT4_FIRST_INO(sb) && 1453 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1454 } 1455 1456 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1457 struct ext4_io_end *io_end) 1458 { 1459 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1460 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1461 atomic_inc(&EXT4_I(inode)->i_unwritten); 1462 } 1463 } 1464 1465 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1466 { 1467 return inode->i_private; 1468 } 1469 1470 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1471 { 1472 inode->i_private = io; 1473 } 1474 1475 /* 1476 * Inode dynamic state flags 1477 */ 1478 enum { 1479 EXT4_STATE_JDATA, /* journaled data exists */ 1480 EXT4_STATE_NEW, /* inode is newly created */ 1481 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1482 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1483 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1484 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1485 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1486 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1487 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1488 nolocking */ 1489 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1490 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */ 1491 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1492 }; 1493 1494 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1495 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1496 { \ 1497 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1498 } \ 1499 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1500 { \ 1501 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1502 } \ 1503 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1504 { \ 1505 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1506 } 1507 1508 /* Add these declarations here only so that these functions can be 1509 * found by name. Otherwise, they are very hard to locate. */ 1510 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1511 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1512 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1513 EXT4_INODE_BIT_FNS(flag, flags, 0) 1514 1515 /* Add these declarations here only so that these functions can be 1516 * found by name. Otherwise, they are very hard to locate. */ 1517 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1518 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1519 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1520 #if (BITS_PER_LONG < 64) 1521 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1522 1523 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1524 { 1525 (ei)->i_state_flags = 0; 1526 } 1527 #else 1528 EXT4_INODE_BIT_FNS(state, flags, 32) 1529 1530 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1531 { 1532 /* We depend on the fact that callers will set i_flags */ 1533 } 1534 #endif 1535 #else 1536 /* Assume that user mode programs are passing in an ext4fs superblock, not 1537 * a kernel struct super_block. This will allow us to call the feature-test 1538 * macros from user land. */ 1539 #define EXT4_SB(sb) (sb) 1540 #endif 1541 1542 /* 1543 * Returns true if the inode is inode is encrypted 1544 */ 1545 static inline int ext4_encrypted_inode(struct inode *inode) 1546 { 1547 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1548 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 1549 #else 1550 return 0; 1551 #endif 1552 } 1553 1554 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1555 1556 /* 1557 * Codes for operating systems 1558 */ 1559 #define EXT4_OS_LINUX 0 1560 #define EXT4_OS_HURD 1 1561 #define EXT4_OS_MASIX 2 1562 #define EXT4_OS_FREEBSD 3 1563 #define EXT4_OS_LITES 4 1564 1565 /* 1566 * Revision levels 1567 */ 1568 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1569 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1570 1571 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1572 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1573 1574 #define EXT4_GOOD_OLD_INODE_SIZE 128 1575 1576 /* 1577 * Feature set definitions 1578 */ 1579 1580 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */ 1581 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1582 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1583 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1584 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1585 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1586 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1587 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1588 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1589 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1590 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1591 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1592 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1593 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1594 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1595 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1596 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1597 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1598 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1599 1600 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1601 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1602 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1603 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1604 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1605 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1606 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1607 1608 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1609 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1610 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1611 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1612 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1613 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1614 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1615 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1616 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1617 /* 1618 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1619 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1620 * all other data structures' checksums. However, the METADATA_CSUM and 1621 * GDT_CSUM bits are mutually exclusive. 1622 */ 1623 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1624 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1625 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1626 1627 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1628 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1629 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1630 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1631 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1632 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1633 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1634 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1635 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1636 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1637 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1638 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1639 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1640 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1641 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1642 1643 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1644 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1645 { \ 1646 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1647 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1648 } \ 1649 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1650 { \ 1651 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1652 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1653 } \ 1654 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1655 { \ 1656 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1657 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1658 } 1659 1660 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1661 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1662 { \ 1663 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1664 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1665 } \ 1666 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1667 { \ 1668 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1669 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1670 } \ 1671 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1672 { \ 1673 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1674 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1675 } 1676 1677 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1678 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1679 { \ 1680 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1681 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1682 } \ 1683 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1684 { \ 1685 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1686 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1687 } \ 1688 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1689 { \ 1690 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1691 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1692 } 1693 1694 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1695 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1696 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1697 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1698 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1699 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1700 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1701 1702 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1703 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1704 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1705 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1706 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1707 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1708 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1709 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1710 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1711 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1712 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1713 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1714 1715 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1716 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1717 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1718 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1719 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1720 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1721 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1722 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1723 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1724 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1725 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1726 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1727 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1728 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1729 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1730 1731 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1732 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1733 EXT4_FEATURE_INCOMPAT_META_BG) 1734 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1735 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1736 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1737 1738 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1739 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1740 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1741 EXT4_FEATURE_INCOMPAT_META_BG) 1742 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1743 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1744 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1745 1746 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1747 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1748 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1749 EXT4_FEATURE_INCOMPAT_META_BG| \ 1750 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1751 EXT4_FEATURE_INCOMPAT_64BIT| \ 1752 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1753 EXT4_FEATURE_INCOMPAT_MMP | \ 1754 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1755 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1756 EXT4_FEATURE_INCOMPAT_CSUM_SEED) 1757 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1758 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1759 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1760 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1761 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1762 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1763 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1764 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1765 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1766 EXT4_FEATURE_RO_COMPAT_QUOTA) 1767 1768 #define EXTN_FEATURE_FUNCS(ver) \ 1769 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1770 { \ 1771 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1772 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1773 } \ 1774 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1775 { \ 1776 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1777 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1778 } \ 1779 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1780 { \ 1781 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1782 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1783 } 1784 1785 EXTN_FEATURE_FUNCS(2) 1786 EXTN_FEATURE_FUNCS(3) 1787 EXTN_FEATURE_FUNCS(4) 1788 1789 static inline bool ext4_has_compat_features(struct super_block *sb) 1790 { 1791 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1792 } 1793 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1794 { 1795 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1796 } 1797 static inline bool ext4_has_incompat_features(struct super_block *sb) 1798 { 1799 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1800 } 1801 1802 /* 1803 * Default values for user and/or group using reserved blocks 1804 */ 1805 #define EXT4_DEF_RESUID 0 1806 #define EXT4_DEF_RESGID 0 1807 1808 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1809 1810 /* 1811 * Default mount options 1812 */ 1813 #define EXT4_DEFM_DEBUG 0x0001 1814 #define EXT4_DEFM_BSDGROUPS 0x0002 1815 #define EXT4_DEFM_XATTR_USER 0x0004 1816 #define EXT4_DEFM_ACL 0x0008 1817 #define EXT4_DEFM_UID16 0x0010 1818 #define EXT4_DEFM_JMODE 0x0060 1819 #define EXT4_DEFM_JMODE_DATA 0x0020 1820 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1821 #define EXT4_DEFM_JMODE_WBACK 0x0060 1822 #define EXT4_DEFM_NOBARRIER 0x0100 1823 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1824 #define EXT4_DEFM_DISCARD 0x0400 1825 #define EXT4_DEFM_NODELALLOC 0x0800 1826 1827 /* 1828 * Default journal batch times 1829 */ 1830 #define EXT4_DEF_MIN_BATCH_TIME 0 1831 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1832 1833 /* 1834 * Minimum number of groups in a flexgroup before we separate out 1835 * directories into the first block group of a flexgroup 1836 */ 1837 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1838 1839 /* 1840 * Structure of a directory entry 1841 */ 1842 #define EXT4_NAME_LEN 255 1843 1844 struct ext4_dir_entry { 1845 __le32 inode; /* Inode number */ 1846 __le16 rec_len; /* Directory entry length */ 1847 __le16 name_len; /* Name length */ 1848 char name[EXT4_NAME_LEN]; /* File name */ 1849 }; 1850 1851 /* 1852 * The new version of the directory entry. Since EXT4 structures are 1853 * stored in intel byte order, and the name_len field could never be 1854 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1855 * file_type field. 1856 */ 1857 struct ext4_dir_entry_2 { 1858 __le32 inode; /* Inode number */ 1859 __le16 rec_len; /* Directory entry length */ 1860 __u8 name_len; /* Name length */ 1861 __u8 file_type; 1862 char name[EXT4_NAME_LEN]; /* File name */ 1863 }; 1864 1865 /* 1866 * This is a bogus directory entry at the end of each leaf block that 1867 * records checksums. 1868 */ 1869 struct ext4_dir_entry_tail { 1870 __le32 det_reserved_zero1; /* Pretend to be unused */ 1871 __le16 det_rec_len; /* 12 */ 1872 __u8 det_reserved_zero2; /* Zero name length */ 1873 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1874 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1875 }; 1876 1877 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1878 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1879 ((blocksize) - \ 1880 sizeof(struct ext4_dir_entry_tail)))) 1881 1882 /* 1883 * Ext4 directory file types. Only the low 3 bits are used. The 1884 * other bits are reserved for now. 1885 */ 1886 #define EXT4_FT_UNKNOWN 0 1887 #define EXT4_FT_REG_FILE 1 1888 #define EXT4_FT_DIR 2 1889 #define EXT4_FT_CHRDEV 3 1890 #define EXT4_FT_BLKDEV 4 1891 #define EXT4_FT_FIFO 5 1892 #define EXT4_FT_SOCK 6 1893 #define EXT4_FT_SYMLINK 7 1894 1895 #define EXT4_FT_MAX 8 1896 1897 #define EXT4_FT_DIR_CSUM 0xDE 1898 1899 /* 1900 * EXT4_DIR_PAD defines the directory entries boundaries 1901 * 1902 * NOTE: It must be a multiple of 4 1903 */ 1904 #define EXT4_DIR_PAD 4 1905 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1906 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1907 ~EXT4_DIR_ROUND) 1908 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1909 1910 /* 1911 * If we ever get support for fs block sizes > page_size, we'll need 1912 * to remove the #if statements in the next two functions... 1913 */ 1914 static inline unsigned int 1915 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1916 { 1917 unsigned len = le16_to_cpu(dlen); 1918 1919 #if (PAGE_CACHE_SIZE >= 65536) 1920 if (len == EXT4_MAX_REC_LEN || len == 0) 1921 return blocksize; 1922 return (len & 65532) | ((len & 3) << 16); 1923 #else 1924 return len; 1925 #endif 1926 } 1927 1928 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1929 { 1930 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1931 BUG(); 1932 #if (PAGE_CACHE_SIZE >= 65536) 1933 if (len < 65536) 1934 return cpu_to_le16(len); 1935 if (len == blocksize) { 1936 if (blocksize == 65536) 1937 return cpu_to_le16(EXT4_MAX_REC_LEN); 1938 else 1939 return cpu_to_le16(0); 1940 } 1941 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1942 #else 1943 return cpu_to_le16(len); 1944 #endif 1945 } 1946 1947 /* 1948 * Hash Tree Directory indexing 1949 * (c) Daniel Phillips, 2001 1950 */ 1951 1952 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1953 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1954 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1955 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1956 1957 /* Legal values for the dx_root hash_version field: */ 1958 1959 #define DX_HASH_LEGACY 0 1960 #define DX_HASH_HALF_MD4 1 1961 #define DX_HASH_TEA 2 1962 #define DX_HASH_LEGACY_UNSIGNED 3 1963 #define DX_HASH_HALF_MD4_UNSIGNED 4 1964 #define DX_HASH_TEA_UNSIGNED 5 1965 1966 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1967 const void *address, unsigned int length) 1968 { 1969 struct { 1970 struct shash_desc shash; 1971 char ctx[4]; 1972 } desc; 1973 int err; 1974 1975 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1976 1977 desc.shash.tfm = sbi->s_chksum_driver; 1978 desc.shash.flags = 0; 1979 *(u32 *)desc.ctx = crc; 1980 1981 err = crypto_shash_update(&desc.shash, address, length); 1982 BUG_ON(err); 1983 1984 return *(u32 *)desc.ctx; 1985 } 1986 1987 #ifdef __KERNEL__ 1988 1989 /* hash info structure used by the directory hash */ 1990 struct dx_hash_info 1991 { 1992 u32 hash; 1993 u32 minor_hash; 1994 int hash_version; 1995 u32 *seed; 1996 }; 1997 1998 1999 /* 32 and 64 bit signed EOF for dx directories */ 2000 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2001 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2002 2003 2004 /* 2005 * Control parameters used by ext4_htree_next_block 2006 */ 2007 #define HASH_NB_ALWAYS 1 2008 2009 struct ext4_filename { 2010 const struct qstr *usr_fname; 2011 struct ext4_str disk_name; 2012 struct dx_hash_info hinfo; 2013 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2014 struct ext4_str crypto_buf; 2015 #endif 2016 }; 2017 2018 #define fname_name(p) ((p)->disk_name.name) 2019 #define fname_len(p) ((p)->disk_name.len) 2020 2021 /* 2022 * Describe an inode's exact location on disk and in memory 2023 */ 2024 struct ext4_iloc 2025 { 2026 struct buffer_head *bh; 2027 unsigned long offset; 2028 ext4_group_t block_group; 2029 }; 2030 2031 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2032 { 2033 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2034 } 2035 2036 /* 2037 * This structure is stuffed into the struct file's private_data field 2038 * for directories. It is where we put information so that we can do 2039 * readdir operations in hash tree order. 2040 */ 2041 struct dir_private_info { 2042 struct rb_root root; 2043 struct rb_node *curr_node; 2044 struct fname *extra_fname; 2045 loff_t last_pos; 2046 __u32 curr_hash; 2047 __u32 curr_minor_hash; 2048 __u32 next_hash; 2049 }; 2050 2051 /* calculate the first block number of the group */ 2052 static inline ext4_fsblk_t 2053 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2054 { 2055 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2056 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2057 } 2058 2059 /* 2060 * Special error return code only used by dx_probe() and its callers. 2061 */ 2062 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2063 2064 /* 2065 * Timeout and state flag for lazy initialization inode thread. 2066 */ 2067 #define EXT4_DEF_LI_WAIT_MULT 10 2068 #define EXT4_DEF_LI_MAX_START_DELAY 5 2069 #define EXT4_LAZYINIT_QUIT 0x0001 2070 #define EXT4_LAZYINIT_RUNNING 0x0002 2071 2072 /* 2073 * Lazy inode table initialization info 2074 */ 2075 struct ext4_lazy_init { 2076 unsigned long li_state; 2077 struct list_head li_request_list; 2078 struct mutex li_list_mtx; 2079 }; 2080 2081 struct ext4_li_request { 2082 struct super_block *lr_super; 2083 struct ext4_sb_info *lr_sbi; 2084 ext4_group_t lr_next_group; 2085 struct list_head lr_request; 2086 unsigned long lr_next_sched; 2087 unsigned long lr_timeout; 2088 }; 2089 2090 struct ext4_features { 2091 struct kobject f_kobj; 2092 struct completion f_kobj_unregister; 2093 }; 2094 2095 /* 2096 * This structure will be used for multiple mount protection. It will be 2097 * written into the block number saved in the s_mmp_block field in the 2098 * superblock. Programs that check MMP should assume that if 2099 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2100 * to use the filesystem, regardless of how old the timestamp is. 2101 */ 2102 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2103 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2104 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2105 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2106 2107 struct mmp_struct { 2108 __le32 mmp_magic; /* Magic number for MMP */ 2109 __le32 mmp_seq; /* Sequence no. updated periodically */ 2110 2111 /* 2112 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2113 * purposes and do not affect the correctness of the algorithm 2114 */ 2115 __le64 mmp_time; /* Time last updated */ 2116 char mmp_nodename[64]; /* Node which last updated MMP block */ 2117 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2118 2119 /* 2120 * mmp_check_interval is used to verify if the MMP block has been 2121 * updated on the block device. The value is updated based on the 2122 * maximum time to write the MMP block during an update cycle. 2123 */ 2124 __le16 mmp_check_interval; 2125 2126 __le16 mmp_pad1; 2127 __le32 mmp_pad2[226]; 2128 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2129 }; 2130 2131 /* arguments passed to the mmp thread */ 2132 struct mmpd_data { 2133 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2134 struct super_block *sb; /* super block of the fs */ 2135 }; 2136 2137 /* 2138 * Check interval multiplier 2139 * The MMP block is written every update interval and initially checked every 2140 * update interval x the multiplier (the value is then adapted based on the 2141 * write latency). The reason is that writes can be delayed under load and we 2142 * don't want readers to incorrectly assume that the filesystem is no longer 2143 * in use. 2144 */ 2145 #define EXT4_MMP_CHECK_MULT 2UL 2146 2147 /* 2148 * Minimum interval for MMP checking in seconds. 2149 */ 2150 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2151 2152 /* 2153 * Maximum interval for MMP checking in seconds. 2154 */ 2155 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2156 2157 /* 2158 * Function prototypes 2159 */ 2160 2161 /* 2162 * Ok, these declarations are also in <linux/kernel.h> but none of the 2163 * ext4 source programs needs to include it so they are duplicated here. 2164 */ 2165 # define NORET_TYPE /**/ 2166 # define ATTRIB_NORET __attribute__((noreturn)) 2167 # define NORET_AND noreturn, 2168 2169 /* bitmap.c */ 2170 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2171 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2172 struct ext4_group_desc *gdp, 2173 struct buffer_head *bh, int sz); 2174 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2175 struct ext4_group_desc *gdp, 2176 struct buffer_head *bh, int sz); 2177 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2178 struct ext4_group_desc *gdp, 2179 struct buffer_head *bh); 2180 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2181 struct ext4_group_desc *gdp, 2182 struct buffer_head *bh); 2183 2184 /* balloc.c */ 2185 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2186 ext4_fsblk_t blocknr, 2187 ext4_group_t *blockgrpp, 2188 ext4_grpblk_t *offsetp); 2189 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2190 ext4_fsblk_t block); 2191 2192 extern unsigned int ext4_block_group(struct super_block *sb, 2193 ext4_fsblk_t blocknr); 2194 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2195 ext4_fsblk_t blocknr); 2196 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2197 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2198 ext4_group_t group); 2199 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2200 ext4_fsblk_t goal, 2201 unsigned int flags, 2202 unsigned long *count, 2203 int *errp); 2204 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2205 s64 nclusters, unsigned int flags); 2206 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2207 extern void ext4_check_blocks_bitmap(struct super_block *); 2208 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2209 ext4_group_t block_group, 2210 struct buffer_head ** bh); 2211 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2212 2213 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2214 ext4_group_t block_group); 2215 extern int ext4_wait_block_bitmap(struct super_block *sb, 2216 ext4_group_t block_group, 2217 struct buffer_head *bh); 2218 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2219 ext4_group_t block_group); 2220 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2221 ext4_group_t block_group, 2222 struct ext4_group_desc *gdp); 2223 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2224 2225 /* crypto_policy.c */ 2226 int ext4_is_child_context_consistent_with_parent(struct inode *parent, 2227 struct inode *child); 2228 int ext4_inherit_context(struct inode *parent, struct inode *child); 2229 void ext4_to_hex(char *dst, char *src, size_t src_size); 2230 int ext4_process_policy(const struct ext4_encryption_policy *policy, 2231 struct inode *inode); 2232 int ext4_get_policy(struct inode *inode, 2233 struct ext4_encryption_policy *policy); 2234 2235 /* crypto.c */ 2236 extern struct kmem_cache *ext4_crypt_info_cachep; 2237 bool ext4_valid_contents_enc_mode(uint32_t mode); 2238 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size); 2239 extern struct workqueue_struct *ext4_read_workqueue; 2240 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode); 2241 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx); 2242 void ext4_restore_control_page(struct page *data_page); 2243 struct page *ext4_encrypt(struct inode *inode, 2244 struct page *plaintext_page); 2245 int ext4_decrypt(struct page *page); 2246 int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex); 2247 2248 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2249 int ext4_init_crypto(void); 2250 void ext4_exit_crypto(void); 2251 static inline int ext4_sb_has_crypto(struct super_block *sb) 2252 { 2253 return ext4_has_feature_encrypt(sb); 2254 } 2255 #else 2256 static inline int ext4_init_crypto(void) { return 0; } 2257 static inline void ext4_exit_crypto(void) { } 2258 static inline int ext4_sb_has_crypto(struct super_block *sb) 2259 { 2260 return 0; 2261 } 2262 #endif 2263 2264 /* crypto_fname.c */ 2265 bool ext4_valid_filenames_enc_mode(uint32_t mode); 2266 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize); 2267 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen); 2268 int ext4_fname_crypto_alloc_buffer(struct inode *inode, 2269 u32 ilen, struct ext4_str *crypto_str); 2270 int _ext4_fname_disk_to_usr(struct inode *inode, 2271 struct dx_hash_info *hinfo, 2272 const struct ext4_str *iname, 2273 struct ext4_str *oname); 2274 int ext4_fname_disk_to_usr(struct inode *inode, 2275 struct dx_hash_info *hinfo, 2276 const struct ext4_dir_entry_2 *de, 2277 struct ext4_str *oname); 2278 int ext4_fname_usr_to_disk(struct inode *inode, 2279 const struct qstr *iname, 2280 struct ext4_str *oname); 2281 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2282 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str); 2283 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2284 int lookup, struct ext4_filename *fname); 2285 void ext4_fname_free_filename(struct ext4_filename *fname); 2286 #else 2287 static inline 2288 int ext4_setup_fname_crypto(struct inode *inode) 2289 { 2290 return 0; 2291 } 2292 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { } 2293 static inline int ext4_fname_setup_filename(struct inode *dir, 2294 const struct qstr *iname, 2295 int lookup, struct ext4_filename *fname) 2296 { 2297 fname->usr_fname = iname; 2298 fname->disk_name.name = (unsigned char *) iname->name; 2299 fname->disk_name.len = iname->len; 2300 return 0; 2301 } 2302 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2303 #endif 2304 2305 2306 /* crypto_key.c */ 2307 void ext4_free_crypt_info(struct ext4_crypt_info *ci); 2308 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci); 2309 int _ext4_get_encryption_info(struct inode *inode); 2310 2311 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2312 int ext4_has_encryption_key(struct inode *inode); 2313 2314 static inline int ext4_get_encryption_info(struct inode *inode) 2315 { 2316 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info; 2317 2318 if (!ci || 2319 (ci->ci_keyring_key && 2320 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) | 2321 (1 << KEY_FLAG_REVOKED) | 2322 (1 << KEY_FLAG_DEAD))))) 2323 return _ext4_get_encryption_info(inode); 2324 return 0; 2325 } 2326 2327 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode) 2328 { 2329 return EXT4_I(inode)->i_crypt_info; 2330 } 2331 2332 #else 2333 static inline int ext4_has_encryption_key(struct inode *inode) 2334 { 2335 return 0; 2336 } 2337 static inline int ext4_get_encryption_info(struct inode *inode) 2338 { 2339 return 0; 2340 } 2341 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode) 2342 { 2343 return NULL; 2344 } 2345 #endif 2346 2347 2348 /* dir.c */ 2349 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2350 struct file *, 2351 struct ext4_dir_entry_2 *, 2352 struct buffer_head *, char *, int, 2353 unsigned int); 2354 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2355 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2356 (de), (bh), (buf), (size), (offset))) 2357 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2358 __u32 minor_hash, 2359 struct ext4_dir_entry_2 *dirent, 2360 struct ext4_str *ent_name); 2361 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2362 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2363 struct buffer_head *bh, 2364 void *buf, int buf_size, 2365 struct ext4_filename *fname, 2366 struct ext4_dir_entry_2 **dest_de); 2367 int ext4_insert_dentry(struct inode *dir, 2368 struct inode *inode, 2369 struct ext4_dir_entry_2 *de, 2370 int buf_size, 2371 struct ext4_filename *fname); 2372 static inline void ext4_update_dx_flag(struct inode *inode) 2373 { 2374 if (!ext4_has_feature_dir_index(inode->i_sb)) 2375 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2376 } 2377 static unsigned char ext4_filetype_table[] = { 2378 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2379 }; 2380 2381 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2382 { 2383 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2384 return DT_UNKNOWN; 2385 2386 return ext4_filetype_table[filetype]; 2387 } 2388 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2389 void *buf, int buf_size); 2390 2391 /* fsync.c */ 2392 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2393 2394 /* hash.c */ 2395 extern int ext4fs_dirhash(const char *name, int len, struct 2396 dx_hash_info *hinfo); 2397 2398 /* ialloc.c */ 2399 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2400 const struct qstr *qstr, __u32 goal, 2401 uid_t *owner, int handle_type, 2402 unsigned int line_no, int nblocks); 2403 2404 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2405 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2406 0, 0, 0) 2407 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2408 type, nblocks) \ 2409 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2410 (type), __LINE__, (nblocks)) 2411 2412 2413 extern void ext4_free_inode(handle_t *, struct inode *); 2414 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2415 extern unsigned long ext4_count_free_inodes(struct super_block *); 2416 extern unsigned long ext4_count_dirs(struct super_block *); 2417 extern void ext4_check_inodes_bitmap(struct super_block *); 2418 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2419 extern int ext4_init_inode_table(struct super_block *sb, 2420 ext4_group_t group, int barrier); 2421 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2422 2423 /* mballoc.c */ 2424 extern const struct file_operations ext4_seq_mb_groups_fops; 2425 extern long ext4_mb_stats; 2426 extern long ext4_mb_max_to_scan; 2427 extern int ext4_mb_init(struct super_block *); 2428 extern int ext4_mb_release(struct super_block *); 2429 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2430 struct ext4_allocation_request *, int *); 2431 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2432 extern void ext4_discard_preallocations(struct inode *); 2433 extern int __init ext4_init_mballoc(void); 2434 extern void ext4_exit_mballoc(void); 2435 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2436 struct buffer_head *bh, ext4_fsblk_t block, 2437 unsigned long count, int flags); 2438 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2439 ext4_group_t ngroups); 2440 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2441 ext4_group_t i, struct ext4_group_desc *desc); 2442 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2443 ext4_fsblk_t block, unsigned long count); 2444 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2445 2446 /* inode.c */ 2447 int ext4_inode_is_fast_symlink(struct inode *inode); 2448 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2449 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2450 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2451 struct buffer_head *bh_result, int create); 2452 int ext4_get_block_dax(struct inode *inode, sector_t iblock, 2453 struct buffer_head *bh_result, int create); 2454 int ext4_get_block(struct inode *inode, sector_t iblock, 2455 struct buffer_head *bh_result, int create); 2456 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2457 struct buffer_head *bh, int create); 2458 int ext4_walk_page_buffers(handle_t *handle, 2459 struct buffer_head *head, 2460 unsigned from, 2461 unsigned to, 2462 int *partial, 2463 int (*fn)(handle_t *handle, 2464 struct buffer_head *bh)); 2465 int do_journal_get_write_access(handle_t *handle, 2466 struct buffer_head *bh); 2467 #define FALL_BACK_TO_NONDELALLOC 1 2468 #define CONVERT_INLINE_DATA 2 2469 2470 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2471 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2472 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2473 extern int ext4_setattr(struct dentry *, struct iattr *); 2474 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2475 struct kstat *stat); 2476 extern void ext4_evict_inode(struct inode *); 2477 extern void ext4_clear_inode(struct inode *); 2478 extern int ext4_sync_inode(handle_t *, struct inode *); 2479 extern void ext4_dirty_inode(struct inode *, int); 2480 extern int ext4_change_inode_journal_flag(struct inode *, int); 2481 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2482 extern int ext4_inode_attach_jinode(struct inode *inode); 2483 extern int ext4_can_truncate(struct inode *inode); 2484 extern void ext4_truncate(struct inode *); 2485 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2486 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2487 extern void ext4_set_inode_flags(struct inode *); 2488 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2489 extern int ext4_alloc_da_blocks(struct inode *inode); 2490 extern void ext4_set_aops(struct inode *inode); 2491 extern int ext4_writepage_trans_blocks(struct inode *); 2492 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2493 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2494 loff_t lstart, loff_t lend); 2495 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2496 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf); 2497 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2498 extern void ext4_da_update_reserve_space(struct inode *inode, 2499 int used, int quota_claim); 2500 2501 /* indirect.c */ 2502 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2503 struct ext4_map_blocks *map, int flags); 2504 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter, 2505 loff_t offset); 2506 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2507 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2508 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2509 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2510 ext4_lblk_t start, ext4_lblk_t end); 2511 2512 /* ioctl.c */ 2513 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2514 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2515 2516 /* migrate.c */ 2517 extern int ext4_ext_migrate(struct inode *); 2518 extern int ext4_ind_migrate(struct inode *inode); 2519 2520 /* namei.c */ 2521 extern int ext4_dirent_csum_verify(struct inode *inode, 2522 struct ext4_dir_entry *dirent); 2523 extern int ext4_orphan_add(handle_t *, struct inode *); 2524 extern int ext4_orphan_del(handle_t *, struct inode *); 2525 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2526 __u32 start_minor_hash, __u32 *next_hash); 2527 extern int ext4_search_dir(struct buffer_head *bh, 2528 char *search_buf, 2529 int buf_size, 2530 struct inode *dir, 2531 struct ext4_filename *fname, 2532 const struct qstr *d_name, 2533 unsigned int offset, 2534 struct ext4_dir_entry_2 **res_dir); 2535 extern int ext4_generic_delete_entry(handle_t *handle, 2536 struct inode *dir, 2537 struct ext4_dir_entry_2 *de_del, 2538 struct buffer_head *bh, 2539 void *entry_buf, 2540 int buf_size, 2541 int csum_size); 2542 extern int ext4_empty_dir(struct inode *inode); 2543 2544 /* resize.c */ 2545 extern int ext4_group_add(struct super_block *sb, 2546 struct ext4_new_group_data *input); 2547 extern int ext4_group_extend(struct super_block *sb, 2548 struct ext4_super_block *es, 2549 ext4_fsblk_t n_blocks_count); 2550 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2551 2552 /* super.c */ 2553 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2554 extern int ext4_calculate_overhead(struct super_block *sb); 2555 extern void ext4_superblock_csum_set(struct super_block *sb); 2556 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2557 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2558 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2559 ext4_group_t ngroup); 2560 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2561 char nbuf[16]); 2562 2563 extern __printf(4, 5) 2564 void __ext4_error(struct super_block *, const char *, unsigned int, 2565 const char *, ...); 2566 extern __printf(5, 6) 2567 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2568 const char *, ...); 2569 extern __printf(5, 6) 2570 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2571 const char *, ...); 2572 extern void __ext4_std_error(struct super_block *, const char *, 2573 unsigned int, int); 2574 extern __printf(4, 5) 2575 void __ext4_abort(struct super_block *, const char *, unsigned int, 2576 const char *, ...); 2577 extern __printf(4, 5) 2578 void __ext4_warning(struct super_block *, const char *, unsigned int, 2579 const char *, ...); 2580 extern __printf(4, 5) 2581 void __ext4_warning_inode(const struct inode *inode, const char *function, 2582 unsigned int line, const char *fmt, ...); 2583 extern __printf(3, 4) 2584 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2585 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2586 const char *, unsigned int, const char *); 2587 extern __printf(7, 8) 2588 void __ext4_grp_locked_error(const char *, unsigned int, 2589 struct super_block *, ext4_group_t, 2590 unsigned long, ext4_fsblk_t, 2591 const char *, ...); 2592 2593 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2594 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2595 2596 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2597 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2598 2599 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2600 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2601 2602 #ifdef CONFIG_PRINTK 2603 2604 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2605 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2606 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2607 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2608 #define ext4_error(sb, fmt, ...) \ 2609 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2610 #define ext4_abort(sb, fmt, ...) \ 2611 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2612 #define ext4_warning(sb, fmt, ...) \ 2613 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2614 #define ext4_warning_inode(inode, fmt, ...) \ 2615 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2616 #define ext4_msg(sb, level, fmt, ...) \ 2617 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2618 #define dump_mmp_msg(sb, mmp, msg) \ 2619 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2620 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2621 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2622 fmt, ##__VA_ARGS__) 2623 2624 #else 2625 2626 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2627 do { \ 2628 no_printk(fmt, ##__VA_ARGS__); \ 2629 __ext4_error_inode(inode, "", 0, block, " "); \ 2630 } while (0) 2631 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2632 do { \ 2633 no_printk(fmt, ##__VA_ARGS__); \ 2634 __ext4_error_file(file, "", 0, block, " "); \ 2635 } while (0) 2636 #define ext4_error(sb, fmt, ...) \ 2637 do { \ 2638 no_printk(fmt, ##__VA_ARGS__); \ 2639 __ext4_error(sb, "", 0, " "); \ 2640 } while (0) 2641 #define ext4_abort(sb, fmt, ...) \ 2642 do { \ 2643 no_printk(fmt, ##__VA_ARGS__); \ 2644 __ext4_abort(sb, "", 0, " "); \ 2645 } while (0) 2646 #define ext4_warning(sb, fmt, ...) \ 2647 do { \ 2648 no_printk(fmt, ##__VA_ARGS__); \ 2649 __ext4_warning(sb, "", 0, " "); \ 2650 } while (0) 2651 #define ext4_warning_inode(inode, fmt, ...) \ 2652 do { \ 2653 no_printk(fmt, ##__VA_ARGS__); \ 2654 __ext4_warning_inode(inode, "", 0, " "); \ 2655 } while (0) 2656 #define ext4_msg(sb, level, fmt, ...) \ 2657 do { \ 2658 no_printk(fmt, ##__VA_ARGS__); \ 2659 __ext4_msg(sb, "", " "); \ 2660 } while (0) 2661 #define dump_mmp_msg(sb, mmp, msg) \ 2662 __dump_mmp_msg(sb, mmp, "", 0, "") 2663 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2664 do { \ 2665 no_printk(fmt, ##__VA_ARGS__); \ 2666 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2667 } while (0) 2668 2669 #endif 2670 2671 extern void ext4_update_dynamic_rev(struct super_block *sb); 2672 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2673 __u32 compat); 2674 extern int ext4_update_rocompat_feature(handle_t *handle, 2675 struct super_block *sb, __u32 rocompat); 2676 extern int ext4_update_incompat_feature(handle_t *handle, 2677 struct super_block *sb, __u32 incompat); 2678 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2679 struct ext4_group_desc *bg); 2680 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2681 struct ext4_group_desc *bg); 2682 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2683 struct ext4_group_desc *bg); 2684 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2685 struct ext4_group_desc *bg); 2686 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2687 struct ext4_group_desc *bg); 2688 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2689 struct ext4_group_desc *bg); 2690 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2691 struct ext4_group_desc *bg); 2692 extern void ext4_block_bitmap_set(struct super_block *sb, 2693 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2694 extern void ext4_inode_bitmap_set(struct super_block *sb, 2695 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2696 extern void ext4_inode_table_set(struct super_block *sb, 2697 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2698 extern void ext4_free_group_clusters_set(struct super_block *sb, 2699 struct ext4_group_desc *bg, 2700 __u32 count); 2701 extern void ext4_free_inodes_set(struct super_block *sb, 2702 struct ext4_group_desc *bg, __u32 count); 2703 extern void ext4_used_dirs_set(struct super_block *sb, 2704 struct ext4_group_desc *bg, __u32 count); 2705 extern void ext4_itable_unused_set(struct super_block *sb, 2706 struct ext4_group_desc *bg, __u32 count); 2707 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2708 struct ext4_group_desc *gdp); 2709 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2710 struct ext4_group_desc *gdp); 2711 extern int ext4_register_li_request(struct super_block *sb, 2712 ext4_group_t first_not_zeroed); 2713 2714 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2715 { 2716 return ext4_has_feature_gdt_csum(sb) || 2717 EXT4_SB(sb)->s_chksum_driver != NULL; 2718 } 2719 2720 static inline int ext4_has_metadata_csum(struct super_block *sb) 2721 { 2722 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2723 !EXT4_SB(sb)->s_chksum_driver); 2724 2725 return (EXT4_SB(sb)->s_chksum_driver != NULL); 2726 } 2727 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2728 { 2729 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2730 le32_to_cpu(es->s_blocks_count_lo); 2731 } 2732 2733 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2734 { 2735 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2736 le32_to_cpu(es->s_r_blocks_count_lo); 2737 } 2738 2739 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2740 { 2741 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2742 le32_to_cpu(es->s_free_blocks_count_lo); 2743 } 2744 2745 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2746 ext4_fsblk_t blk) 2747 { 2748 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2749 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2750 } 2751 2752 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2753 ext4_fsblk_t blk) 2754 { 2755 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2756 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2757 } 2758 2759 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2760 ext4_fsblk_t blk) 2761 { 2762 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2763 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2764 } 2765 2766 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2767 { 2768 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2769 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2770 le32_to_cpu(raw_inode->i_size_lo); 2771 else 2772 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2773 } 2774 2775 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2776 { 2777 raw_inode->i_size_lo = cpu_to_le32(i_size); 2778 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2779 } 2780 2781 static inline 2782 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2783 ext4_group_t group) 2784 { 2785 struct ext4_group_info ***grp_info; 2786 long indexv, indexh; 2787 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2788 grp_info = EXT4_SB(sb)->s_group_info; 2789 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2790 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2791 return grp_info[indexv][indexh]; 2792 } 2793 2794 /* 2795 * Reading s_groups_count requires using smp_rmb() afterwards. See 2796 * the locking protocol documented in the comments of ext4_group_add() 2797 * in resize.c 2798 */ 2799 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2800 { 2801 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2802 2803 smp_rmb(); 2804 return ngroups; 2805 } 2806 2807 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2808 ext4_group_t block_group) 2809 { 2810 return block_group >> sbi->s_log_groups_per_flex; 2811 } 2812 2813 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2814 { 2815 return 1 << sbi->s_log_groups_per_flex; 2816 } 2817 2818 #define ext4_std_error(sb, errno) \ 2819 do { \ 2820 if ((errno)) \ 2821 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2822 } while (0) 2823 2824 #ifdef CONFIG_SMP 2825 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2826 * counters. So we need to make sure we have free clusters more 2827 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2828 */ 2829 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2830 #else 2831 #define EXT4_FREECLUSTERS_WATERMARK 0 2832 #endif 2833 2834 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2835 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2836 { 2837 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2838 !mutex_is_locked(&inode->i_mutex)); 2839 down_write(&EXT4_I(inode)->i_data_sem); 2840 if (newsize > EXT4_I(inode)->i_disksize) 2841 EXT4_I(inode)->i_disksize = newsize; 2842 up_write(&EXT4_I(inode)->i_data_sem); 2843 } 2844 2845 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2846 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2847 { 2848 int changed = 0; 2849 2850 if (newsize > inode->i_size) { 2851 i_size_write(inode, newsize); 2852 changed = 1; 2853 } 2854 if (newsize > EXT4_I(inode)->i_disksize) { 2855 ext4_update_i_disksize(inode, newsize); 2856 changed |= 2; 2857 } 2858 return changed; 2859 } 2860 2861 struct ext4_group_info { 2862 unsigned long bb_state; 2863 struct rb_root bb_free_root; 2864 ext4_grpblk_t bb_first_free; /* first free block */ 2865 ext4_grpblk_t bb_free; /* total free blocks */ 2866 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2867 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2868 struct list_head bb_prealloc_list; 2869 #ifdef DOUBLE_CHECK 2870 void *bb_bitmap; 2871 #endif 2872 struct rw_semaphore alloc_sem; 2873 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2874 * regions, index is order. 2875 * bb_counters[3] = 5 means 2876 * 5 free 8-block regions. */ 2877 }; 2878 2879 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2880 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2881 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2882 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2883 2884 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2885 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2886 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2887 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2888 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2889 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2890 2891 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2892 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2893 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2894 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2895 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2896 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2897 2898 #define EXT4_MAX_CONTENTION 8 2899 #define EXT4_CONTENTION_THRESHOLD 2 2900 2901 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2902 ext4_group_t group) 2903 { 2904 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2905 } 2906 2907 /* 2908 * Returns true if the filesystem is busy enough that attempts to 2909 * access the block group locks has run into contention. 2910 */ 2911 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2912 { 2913 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2914 } 2915 2916 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2917 { 2918 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2919 if (spin_trylock(lock)) 2920 /* 2921 * We're able to grab the lock right away, so drop the 2922 * lock contention counter. 2923 */ 2924 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2925 else { 2926 /* 2927 * The lock is busy, so bump the contention counter, 2928 * and then wait on the spin lock. 2929 */ 2930 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2931 EXT4_MAX_CONTENTION); 2932 spin_lock(lock); 2933 } 2934 } 2935 2936 static inline void ext4_unlock_group(struct super_block *sb, 2937 ext4_group_t group) 2938 { 2939 spin_unlock(ext4_group_lock_ptr(sb, group)); 2940 } 2941 2942 /* 2943 * Block validity checking 2944 */ 2945 #define ext4_check_indirect_blockref(inode, bh) \ 2946 ext4_check_blockref(__func__, __LINE__, inode, \ 2947 (__le32 *)(bh)->b_data, \ 2948 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2949 2950 #define ext4_ind_check_inode(inode) \ 2951 ext4_check_blockref(__func__, __LINE__, inode, \ 2952 EXT4_I(inode)->i_data, \ 2953 EXT4_NDIR_BLOCKS) 2954 2955 /* 2956 * Inodes and files operations 2957 */ 2958 2959 /* dir.c */ 2960 extern const struct file_operations ext4_dir_operations; 2961 2962 /* file.c */ 2963 extern const struct inode_operations ext4_file_inode_operations; 2964 extern const struct file_operations ext4_file_operations; 2965 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2966 2967 /* inline.c */ 2968 extern int ext4_get_max_inline_size(struct inode *inode); 2969 extern int ext4_find_inline_data_nolock(struct inode *inode); 2970 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2971 unsigned int len); 2972 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2973 2974 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2975 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2976 struct inode *inode, 2977 loff_t pos, unsigned len, 2978 unsigned flags, 2979 struct page **pagep); 2980 extern int ext4_write_inline_data_end(struct inode *inode, 2981 loff_t pos, unsigned len, 2982 unsigned copied, 2983 struct page *page); 2984 extern struct buffer_head * 2985 ext4_journalled_write_inline_data(struct inode *inode, 2986 unsigned len, 2987 struct page *page); 2988 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2989 struct inode *inode, 2990 loff_t pos, unsigned len, 2991 unsigned flags, 2992 struct page **pagep, 2993 void **fsdata); 2994 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2995 unsigned len, unsigned copied, 2996 struct page *page); 2997 extern int ext4_try_add_inline_entry(handle_t *handle, 2998 struct ext4_filename *fname, 2999 struct dentry *dentry, 3000 struct inode *inode); 3001 extern int ext4_try_create_inline_dir(handle_t *handle, 3002 struct inode *parent, 3003 struct inode *inode); 3004 extern int ext4_read_inline_dir(struct file *filp, 3005 struct dir_context *ctx, 3006 int *has_inline_data); 3007 extern int htree_inlinedir_to_tree(struct file *dir_file, 3008 struct inode *dir, ext4_lblk_t block, 3009 struct dx_hash_info *hinfo, 3010 __u32 start_hash, __u32 start_minor_hash, 3011 int *has_inline_data); 3012 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3013 struct ext4_filename *fname, 3014 const struct qstr *d_name, 3015 struct ext4_dir_entry_2 **res_dir, 3016 int *has_inline_data); 3017 extern int ext4_delete_inline_entry(handle_t *handle, 3018 struct inode *dir, 3019 struct ext4_dir_entry_2 *de_del, 3020 struct buffer_head *bh, 3021 int *has_inline_data); 3022 extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 3023 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3024 struct ext4_dir_entry_2 **parent_de, 3025 int *retval); 3026 extern int ext4_inline_data_fiemap(struct inode *inode, 3027 struct fiemap_extent_info *fieinfo, 3028 int *has_inline, __u64 start, __u64 len); 3029 extern int ext4_try_to_evict_inline_data(handle_t *handle, 3030 struct inode *inode, 3031 int needed); 3032 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3033 3034 extern int ext4_convert_inline_data(struct inode *inode); 3035 3036 static inline int ext4_has_inline_data(struct inode *inode) 3037 { 3038 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3039 EXT4_I(inode)->i_inline_off; 3040 } 3041 3042 /* namei.c */ 3043 extern const struct inode_operations ext4_dir_inode_operations; 3044 extern const struct inode_operations ext4_special_inode_operations; 3045 extern struct dentry *ext4_get_parent(struct dentry *child); 3046 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3047 struct ext4_dir_entry_2 *de, 3048 int blocksize, int csum_size, 3049 unsigned int parent_ino, int dotdot_real_len); 3050 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3051 unsigned int blocksize); 3052 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3053 struct inode *inode, 3054 struct buffer_head *bh); 3055 #define S_SHIFT 12 3056 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 3057 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3058 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3059 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3060 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3061 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3062 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3063 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3064 }; 3065 3066 static inline void ext4_set_de_type(struct super_block *sb, 3067 struct ext4_dir_entry_2 *de, 3068 umode_t mode) { 3069 if (ext4_has_feature_filetype(sb)) 3070 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3071 } 3072 3073 /* readpages.c */ 3074 extern int ext4_mpage_readpages(struct address_space *mapping, 3075 struct list_head *pages, struct page *page, 3076 unsigned nr_pages); 3077 3078 /* symlink.c */ 3079 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3080 extern const struct inode_operations ext4_symlink_inode_operations; 3081 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3082 3083 /* sysfs.c */ 3084 extern int ext4_register_sysfs(struct super_block *sb); 3085 extern void ext4_unregister_sysfs(struct super_block *sb); 3086 extern int __init ext4_init_sysfs(void); 3087 extern void ext4_exit_sysfs(void); 3088 3089 /* block_validity */ 3090 extern void ext4_release_system_zone(struct super_block *sb); 3091 extern int ext4_setup_system_zone(struct super_block *sb); 3092 extern int __init ext4_init_system_zone(void); 3093 extern void ext4_exit_system_zone(void); 3094 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3095 ext4_fsblk_t start_blk, 3096 unsigned int count); 3097 extern int ext4_check_blockref(const char *, unsigned int, 3098 struct inode *, __le32 *, unsigned int); 3099 3100 /* extents.c */ 3101 struct ext4_ext_path; 3102 struct ext4_extent; 3103 3104 /* 3105 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3106 * __le32. 3107 */ 3108 #define EXT_MAX_BLOCKS 0xffffffff 3109 3110 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3111 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3112 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3113 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3114 struct ext4_map_blocks *map, int flags); 3115 extern void ext4_ext_truncate(handle_t *, struct inode *); 3116 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3117 ext4_lblk_t end); 3118 extern void ext4_ext_init(struct super_block *); 3119 extern void ext4_ext_release(struct super_block *); 3120 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3121 loff_t len); 3122 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3123 loff_t offset, ssize_t len); 3124 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3125 struct ext4_map_blocks *map, int flags); 3126 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3127 ext4_lblk_t lblocks); 3128 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3129 int num, 3130 struct ext4_ext_path *path); 3131 extern int ext4_can_extents_be_merged(struct inode *inode, 3132 struct ext4_extent *ex1, 3133 struct ext4_extent *ex2); 3134 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3135 struct ext4_ext_path **, 3136 struct ext4_extent *, int); 3137 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3138 struct ext4_ext_path **, 3139 int flags); 3140 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3141 extern int ext4_ext_check_inode(struct inode *inode); 3142 extern int ext4_find_delalloc_range(struct inode *inode, 3143 ext4_lblk_t lblk_start, 3144 ext4_lblk_t lblk_end); 3145 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3146 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3147 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3148 __u64 start, __u64 len); 3149 extern int ext4_ext_precache(struct inode *inode); 3150 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3151 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3152 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3153 struct inode *inode2, ext4_lblk_t lblk1, 3154 ext4_lblk_t lblk2, ext4_lblk_t count, 3155 int mark_unwritten,int *err); 3156 3157 /* move_extent.c */ 3158 extern void ext4_double_down_write_data_sem(struct inode *first, 3159 struct inode *second); 3160 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3161 struct inode *donor_inode); 3162 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3163 __u64 start_orig, __u64 start_donor, 3164 __u64 len, __u64 *moved_len); 3165 3166 /* page-io.c */ 3167 extern int __init ext4_init_pageio(void); 3168 extern void ext4_exit_pageio(void); 3169 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3170 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3171 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3172 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3173 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3174 struct writeback_control *wbc); 3175 extern void ext4_end_io_rsv_work(struct work_struct *work); 3176 extern void ext4_io_submit(struct ext4_io_submit *io); 3177 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3178 struct page *page, 3179 int len, 3180 struct writeback_control *wbc, 3181 bool keep_towrite); 3182 3183 /* mmp.c */ 3184 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3185 3186 /* 3187 * Add new method to test whether block and inode bitmaps are properly 3188 * initialized. With uninit_bg reading the block from disk is not enough 3189 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3190 */ 3191 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3192 3193 static inline int bitmap_uptodate(struct buffer_head *bh) 3194 { 3195 return (buffer_uptodate(bh) && 3196 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3197 } 3198 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3199 { 3200 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3201 } 3202 3203 /* 3204 * Disable DIO read nolock optimization, so new dioreaders will be forced 3205 * to grab i_mutex 3206 */ 3207 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 3208 { 3209 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3210 smp_mb(); 3211 } 3212 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 3213 { 3214 smp_mb(); 3215 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3216 } 3217 3218 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3219 3220 /* For ioend & aio unwritten conversion wait queues */ 3221 #define EXT4_WQ_HASH_SZ 37 3222 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3223 EXT4_WQ_HASH_SZ]) 3224 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 3225 EXT4_WQ_HASH_SZ]) 3226 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3227 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 3228 3229 #define EXT4_RESIZING 0 3230 extern int ext4_resize_begin(struct super_block *sb); 3231 extern void ext4_resize_end(struct super_block *sb); 3232 3233 #endif /* __KERNEL__ */ 3234 3235 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3236 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3237 3238 #endif /* _EXT4_H */ 3239