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 /* Convert written extents to unwritten */ 559 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 560 561 /* 562 * The bit position of these flags must not overlap with any of the 563 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 564 * read_extent_tree_block(), ext4_split_extent_at(), 565 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 566 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 567 * caching the extents when reading from the extent tree while a 568 * truncate or punch hole operation is in progress. 569 */ 570 #define EXT4_EX_NOCACHE 0x40000000 571 #define EXT4_EX_FORCE_CACHE 0x20000000 572 573 /* 574 * Flags used by ext4_free_blocks 575 */ 576 #define EXT4_FREE_BLOCKS_METADATA 0x0001 577 #define EXT4_FREE_BLOCKS_FORGET 0x0002 578 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 579 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 580 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 581 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 582 583 /* Encryption algorithms */ 584 #define EXT4_ENCRYPTION_MODE_INVALID 0 585 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1 586 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2 587 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3 588 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4 589 590 #include "ext4_crypto.h" 591 592 /* 593 * ioctl commands 594 */ 595 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 596 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 597 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 598 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 599 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 600 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 601 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 602 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 603 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 604 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 605 #define EXT4_IOC_MIGRATE _IO('f', 9) 606 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 607 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 608 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 609 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 610 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 611 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 612 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 613 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy) 614 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16]) 615 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy) 616 617 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 618 /* 619 * ioctl commands in 32 bit emulation 620 */ 621 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 622 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 623 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 624 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 625 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 626 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 627 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 628 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 629 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 630 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 631 #endif 632 633 /* Max physical block we can address w/o extents */ 634 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 635 636 /* 637 * Structure of an inode on the disk 638 */ 639 struct ext4_inode { 640 __le16 i_mode; /* File mode */ 641 __le16 i_uid; /* Low 16 bits of Owner Uid */ 642 __le32 i_size_lo; /* Size in bytes */ 643 __le32 i_atime; /* Access time */ 644 __le32 i_ctime; /* Inode Change time */ 645 __le32 i_mtime; /* Modification time */ 646 __le32 i_dtime; /* Deletion Time */ 647 __le16 i_gid; /* Low 16 bits of Group Id */ 648 __le16 i_links_count; /* Links count */ 649 __le32 i_blocks_lo; /* Blocks count */ 650 __le32 i_flags; /* File flags */ 651 union { 652 struct { 653 __le32 l_i_version; 654 } linux1; 655 struct { 656 __u32 h_i_translator; 657 } hurd1; 658 struct { 659 __u32 m_i_reserved1; 660 } masix1; 661 } osd1; /* OS dependent 1 */ 662 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 663 __le32 i_generation; /* File version (for NFS) */ 664 __le32 i_file_acl_lo; /* File ACL */ 665 __le32 i_size_high; 666 __le32 i_obso_faddr; /* Obsoleted fragment address */ 667 union { 668 struct { 669 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 670 __le16 l_i_file_acl_high; 671 __le16 l_i_uid_high; /* these 2 fields */ 672 __le16 l_i_gid_high; /* were reserved2[0] */ 673 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 674 __le16 l_i_reserved; 675 } linux2; 676 struct { 677 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 678 __u16 h_i_mode_high; 679 __u16 h_i_uid_high; 680 __u16 h_i_gid_high; 681 __u32 h_i_author; 682 } hurd2; 683 struct { 684 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 685 __le16 m_i_file_acl_high; 686 __u32 m_i_reserved2[2]; 687 } masix2; 688 } osd2; /* OS dependent 2 */ 689 __le16 i_extra_isize; 690 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 691 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 692 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 693 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 694 __le32 i_crtime; /* File Creation time */ 695 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 696 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 697 __le32 i_projid; /* Project ID */ 698 }; 699 700 struct move_extent { 701 __u32 reserved; /* should be zero */ 702 __u32 donor_fd; /* donor file descriptor */ 703 __u64 orig_start; /* logical start offset in block for orig */ 704 __u64 donor_start; /* logical start offset in block for donor */ 705 __u64 len; /* block length to be moved */ 706 __u64 moved_len; /* moved block length */ 707 }; 708 709 #define EXT4_EPOCH_BITS 2 710 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 711 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 712 713 /* 714 * Extended fields will fit into an inode if the filesystem was formatted 715 * with large inodes (-I 256 or larger) and there are not currently any EAs 716 * consuming all of the available space. For new inodes we always reserve 717 * enough space for the kernel's known extended fields, but for inodes 718 * created with an old kernel this might not have been the case. None of 719 * the extended inode fields is critical for correct filesystem operation. 720 * This macro checks if a certain field fits in the inode. Note that 721 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 722 */ 723 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 724 ((offsetof(typeof(*ext4_inode), field) + \ 725 sizeof((ext4_inode)->field)) \ 726 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 727 (einode)->i_extra_isize)) \ 728 729 /* 730 * We use an encoding that preserves the times for extra epoch "00": 731 * 732 * extra msb of adjust for signed 733 * epoch 32-bit 32-bit tv_sec to 734 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 735 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 736 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 737 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 738 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 739 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 740 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 741 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 742 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 743 * 744 * Note that previous versions of the kernel on 64-bit systems would 745 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 746 * 1970. e2fsck will correct this, assuming that it is run on the 747 * affected filesystem before 2242. 748 */ 749 750 static inline __le32 ext4_encode_extra_time(struct timespec *time) 751 { 752 u32 extra = sizeof(time->tv_sec) > 4 ? 753 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0; 754 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 755 } 756 757 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 758 { 759 if (unlikely(sizeof(time->tv_sec) > 4 && 760 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) { 761 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0) 762 /* Handle legacy encoding of pre-1970 dates with epoch 763 * bits 1,1. We assume that by kernel version 4.20, 764 * everyone will have run fsck over the affected 765 * filesystems to correct the problem. (This 766 * backwards compatibility may be removed before this 767 * time, at the discretion of the ext4 developers.) 768 */ 769 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 770 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 771 extra_bits = 0; 772 time->tv_sec += extra_bits << 32; 773 #else 774 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 775 #endif 776 } 777 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 778 } 779 780 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 781 do { \ 782 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 783 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 784 (raw_inode)->xtime ## _extra = \ 785 ext4_encode_extra_time(&(inode)->xtime); \ 786 } while (0) 787 788 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 789 do { \ 790 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 791 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 792 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 793 (raw_inode)->xtime ## _extra = \ 794 ext4_encode_extra_time(&(einode)->xtime); \ 795 } while (0) 796 797 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 798 do { \ 799 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 800 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 801 ext4_decode_extra_time(&(inode)->xtime, \ 802 raw_inode->xtime ## _extra); \ 803 else \ 804 (inode)->xtime.tv_nsec = 0; \ 805 } while (0) 806 807 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 808 do { \ 809 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 810 (einode)->xtime.tv_sec = \ 811 (signed)le32_to_cpu((raw_inode)->xtime); \ 812 else \ 813 (einode)->xtime.tv_sec = 0; \ 814 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 815 ext4_decode_extra_time(&(einode)->xtime, \ 816 raw_inode->xtime ## _extra); \ 817 else \ 818 (einode)->xtime.tv_nsec = 0; \ 819 } while (0) 820 821 #define i_disk_version osd1.linux1.l_i_version 822 823 #if defined(__KERNEL__) || defined(__linux__) 824 #define i_reserved1 osd1.linux1.l_i_reserved1 825 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 826 #define i_blocks_high osd2.linux2.l_i_blocks_high 827 #define i_uid_low i_uid 828 #define i_gid_low i_gid 829 #define i_uid_high osd2.linux2.l_i_uid_high 830 #define i_gid_high osd2.linux2.l_i_gid_high 831 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 832 833 #elif defined(__GNU__) 834 835 #define i_translator osd1.hurd1.h_i_translator 836 #define i_uid_high osd2.hurd2.h_i_uid_high 837 #define i_gid_high osd2.hurd2.h_i_gid_high 838 #define i_author osd2.hurd2.h_i_author 839 840 #elif defined(__masix__) 841 842 #define i_reserved1 osd1.masix1.m_i_reserved1 843 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 844 #define i_reserved2 osd2.masix2.m_i_reserved2 845 846 #endif /* defined(__KERNEL__) || defined(__linux__) */ 847 848 #include "extents_status.h" 849 850 /* 851 * fourth extended file system inode data in memory 852 */ 853 struct ext4_inode_info { 854 __le32 i_data[15]; /* unconverted */ 855 __u32 i_dtime; 856 ext4_fsblk_t i_file_acl; 857 858 /* 859 * i_block_group is the number of the block group which contains 860 * this file's inode. Constant across the lifetime of the inode, 861 * it is ued for making block allocation decisions - we try to 862 * place a file's data blocks near its inode block, and new inodes 863 * near to their parent directory's inode. 864 */ 865 ext4_group_t i_block_group; 866 ext4_lblk_t i_dir_start_lookup; 867 #if (BITS_PER_LONG < 64) 868 unsigned long i_state_flags; /* Dynamic state flags */ 869 #endif 870 unsigned long i_flags; 871 872 /* 873 * Extended attributes can be read independently of the main file 874 * data. Taking i_mutex even when reading would cause contention 875 * between readers of EAs and writers of regular file data, so 876 * instead we synchronize on xattr_sem when reading or changing 877 * EAs. 878 */ 879 struct rw_semaphore xattr_sem; 880 881 struct list_head i_orphan; /* unlinked but open inodes */ 882 883 /* 884 * i_disksize keeps track of what the inode size is ON DISK, not 885 * in memory. During truncate, i_size is set to the new size by 886 * the VFS prior to calling ext4_truncate(), but the filesystem won't 887 * set i_disksize to 0 until the truncate is actually under way. 888 * 889 * The intent is that i_disksize always represents the blocks which 890 * are used by this file. This allows recovery to restart truncate 891 * on orphans if we crash during truncate. We actually write i_disksize 892 * into the on-disk inode when writing inodes out, instead of i_size. 893 * 894 * The only time when i_disksize and i_size may be different is when 895 * a truncate is in progress. The only things which change i_disksize 896 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 897 */ 898 loff_t i_disksize; 899 900 /* 901 * i_data_sem is for serialising ext4_truncate() against 902 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 903 * data tree are chopped off during truncate. We can't do that in 904 * ext4 because whenever we perform intermediate commits during 905 * truncate, the inode and all the metadata blocks *must* be in a 906 * consistent state which allows truncation of the orphans to restart 907 * during recovery. Hence we must fix the get_block-vs-truncate race 908 * by other means, so we have i_data_sem. 909 */ 910 struct rw_semaphore i_data_sem; 911 /* 912 * i_mmap_sem is for serializing page faults with truncate / punch hole 913 * operations. We have to make sure that new page cannot be faulted in 914 * a section of the inode that is being punched. We cannot easily use 915 * i_data_sem for this since we need protection for the whole punch 916 * operation and i_data_sem ranks below transaction start so we have 917 * to occasionally drop it. 918 */ 919 struct rw_semaphore i_mmap_sem; 920 struct inode vfs_inode; 921 struct jbd2_inode *jinode; 922 923 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 924 925 /* 926 * File creation time. Its function is same as that of 927 * struct timespec i_{a,c,m}time in the generic inode. 928 */ 929 struct timespec i_crtime; 930 931 /* mballoc */ 932 struct list_head i_prealloc_list; 933 spinlock_t i_prealloc_lock; 934 935 /* extents status tree */ 936 struct ext4_es_tree i_es_tree; 937 rwlock_t i_es_lock; 938 struct list_head i_es_list; 939 unsigned int i_es_all_nr; /* protected by i_es_lock */ 940 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 941 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 942 extents to shrink. Protected by 943 i_es_lock */ 944 945 /* ialloc */ 946 ext4_group_t i_last_alloc_group; 947 948 /* allocation reservation info for delalloc */ 949 /* In case of bigalloc, these refer to clusters rather than blocks */ 950 unsigned int i_reserved_data_blocks; 951 unsigned int i_reserved_meta_blocks; 952 unsigned int i_allocated_meta_blocks; 953 ext4_lblk_t i_da_metadata_calc_last_lblock; 954 int i_da_metadata_calc_len; 955 956 /* on-disk additional length */ 957 __u16 i_extra_isize; 958 959 /* Indicate the inline data space. */ 960 u16 i_inline_off; 961 u16 i_inline_size; 962 963 #ifdef CONFIG_QUOTA 964 /* quota space reservation, managed internally by quota code */ 965 qsize_t i_reserved_quota; 966 #endif 967 968 /* Lock protecting lists below */ 969 spinlock_t i_completed_io_lock; 970 /* 971 * Completed IOs that need unwritten extents handling and have 972 * transaction reserved 973 */ 974 struct list_head i_rsv_conversion_list; 975 /* 976 * Completed IOs that need unwritten extents handling and don't have 977 * transaction reserved 978 */ 979 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 980 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 981 struct work_struct i_rsv_conversion_work; 982 983 spinlock_t i_block_reservation_lock; 984 985 /* 986 * Transactions that contain inode's metadata needed to complete 987 * fsync and fdatasync, respectively. 988 */ 989 tid_t i_sync_tid; 990 tid_t i_datasync_tid; 991 992 #ifdef CONFIG_QUOTA 993 struct dquot *i_dquot[MAXQUOTAS]; 994 #endif 995 996 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 997 __u32 i_csum_seed; 998 999 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1000 /* Encryption params */ 1001 struct ext4_crypt_info *i_crypt_info; 1002 #endif 1003 }; 1004 1005 /* 1006 * File system states 1007 */ 1008 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1009 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1010 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1011 1012 /* 1013 * Misc. filesystem flags 1014 */ 1015 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1016 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1017 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1018 1019 /* 1020 * Mount flags set via mount options or defaults 1021 */ 1022 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1023 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1024 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1025 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1026 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1027 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1028 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1029 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1030 #ifdef CONFIG_FS_DAX 1031 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1032 #else 1033 #define EXT4_MOUNT_DAX 0 1034 #endif 1035 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1036 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1037 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1038 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1039 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1040 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1041 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1042 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1043 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1044 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1045 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 1046 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 1047 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 1048 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1049 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1050 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1051 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1052 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1053 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1054 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1055 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1056 1057 /* 1058 * Mount flags set either automatically (could not be set by mount option) 1059 * based on per file system feature or property or in special cases such as 1060 * distinguishing between explicit mount option definition and default. 1061 */ 1062 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1063 specified delalloc */ 1064 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1065 size of blocksize * 8 1066 blocks */ 1067 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1068 file systems */ 1069 1070 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1071 specified journal checksum */ 1072 1073 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1074 ~EXT4_MOUNT_##opt 1075 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1076 EXT4_MOUNT_##opt 1077 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1078 EXT4_MOUNT_##opt) 1079 1080 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1081 ~EXT4_MOUNT2_##opt 1082 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1083 EXT4_MOUNT2_##opt 1084 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1085 EXT4_MOUNT2_##opt) 1086 1087 #define ext4_test_and_set_bit __test_and_set_bit_le 1088 #define ext4_set_bit __set_bit_le 1089 #define ext4_set_bit_atomic ext2_set_bit_atomic 1090 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1091 #define ext4_clear_bit __clear_bit_le 1092 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1093 #define ext4_test_bit test_bit_le 1094 #define ext4_find_next_zero_bit find_next_zero_bit_le 1095 #define ext4_find_next_bit find_next_bit_le 1096 1097 extern void ext4_set_bits(void *bm, int cur, int len); 1098 1099 /* 1100 * Maximal mount counts between two filesystem checks 1101 */ 1102 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1103 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1104 1105 /* 1106 * Behaviour when detecting errors 1107 */ 1108 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1109 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1110 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1111 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1112 1113 /* Metadata checksum algorithm codes */ 1114 #define EXT4_CRC32C_CHKSUM 1 1115 1116 /* 1117 * Structure of the super block 1118 */ 1119 struct ext4_super_block { 1120 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1121 __le32 s_blocks_count_lo; /* Blocks count */ 1122 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1123 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1124 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1125 __le32 s_first_data_block; /* First Data Block */ 1126 __le32 s_log_block_size; /* Block size */ 1127 __le32 s_log_cluster_size; /* Allocation cluster size */ 1128 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1129 __le32 s_clusters_per_group; /* # Clusters per group */ 1130 __le32 s_inodes_per_group; /* # Inodes per group */ 1131 __le32 s_mtime; /* Mount time */ 1132 /*30*/ __le32 s_wtime; /* Write time */ 1133 __le16 s_mnt_count; /* Mount count */ 1134 __le16 s_max_mnt_count; /* Maximal mount count */ 1135 __le16 s_magic; /* Magic signature */ 1136 __le16 s_state; /* File system state */ 1137 __le16 s_errors; /* Behaviour when detecting errors */ 1138 __le16 s_minor_rev_level; /* minor revision level */ 1139 /*40*/ __le32 s_lastcheck; /* time of last check */ 1140 __le32 s_checkinterval; /* max. time between checks */ 1141 __le32 s_creator_os; /* OS */ 1142 __le32 s_rev_level; /* Revision level */ 1143 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1144 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1145 /* 1146 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1147 * 1148 * Note: the difference between the compatible feature set and 1149 * the incompatible feature set is that if there is a bit set 1150 * in the incompatible feature set that the kernel doesn't 1151 * know about, it should refuse to mount the filesystem. 1152 * 1153 * e2fsck's requirements are more strict; if it doesn't know 1154 * about a feature in either the compatible or incompatible 1155 * feature set, it must abort and not try to meddle with 1156 * things it doesn't understand... 1157 */ 1158 __le32 s_first_ino; /* First non-reserved inode */ 1159 __le16 s_inode_size; /* size of inode structure */ 1160 __le16 s_block_group_nr; /* block group # of this superblock */ 1161 __le32 s_feature_compat; /* compatible feature set */ 1162 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1163 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1164 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1165 /*78*/ char s_volume_name[16]; /* volume name */ 1166 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1167 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1168 /* 1169 * Performance hints. Directory preallocation should only 1170 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1171 */ 1172 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1173 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1174 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1175 /* 1176 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1177 */ 1178 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1179 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1180 __le32 s_journal_dev; /* device number of journal file */ 1181 __le32 s_last_orphan; /* start of list of inodes to delete */ 1182 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1183 __u8 s_def_hash_version; /* Default hash version to use */ 1184 __u8 s_jnl_backup_type; 1185 __le16 s_desc_size; /* size of group descriptor */ 1186 /*100*/ __le32 s_default_mount_opts; 1187 __le32 s_first_meta_bg; /* First metablock block group */ 1188 __le32 s_mkfs_time; /* When the filesystem was created */ 1189 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1190 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1191 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1192 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1193 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1194 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1195 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1196 __le32 s_flags; /* Miscellaneous flags */ 1197 __le16 s_raid_stride; /* RAID stride */ 1198 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1199 __le64 s_mmp_block; /* Block for multi-mount protection */ 1200 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1201 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1202 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1203 __u8 s_encryption_level; /* versioning level for encryption */ 1204 __u8 s_reserved_pad; /* Padding to next 32bits */ 1205 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1206 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1207 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1208 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1209 snapshot's future use */ 1210 __le32 s_snapshot_list; /* inode number of the head of the 1211 on-disk snapshot list */ 1212 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1213 __le32 s_error_count; /* number of fs errors */ 1214 __le32 s_first_error_time; /* first time an error happened */ 1215 __le32 s_first_error_ino; /* inode involved in first error */ 1216 __le64 s_first_error_block; /* block involved of first error */ 1217 __u8 s_first_error_func[32]; /* function where the error happened */ 1218 __le32 s_first_error_line; /* line number where error happened */ 1219 __le32 s_last_error_time; /* most recent time of an error */ 1220 __le32 s_last_error_ino; /* inode involved in last error */ 1221 __le32 s_last_error_line; /* line number where error happened */ 1222 __le64 s_last_error_block; /* block involved of last error */ 1223 __u8 s_last_error_func[32]; /* function where the error happened */ 1224 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1225 __u8 s_mount_opts[64]; 1226 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1227 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1228 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1229 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1230 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1231 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1232 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1233 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1234 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1235 __le32 s_reserved[98]; /* Padding to the end of the block */ 1236 __le32 s_checksum; /* crc32c(superblock) */ 1237 }; 1238 1239 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1240 1241 #ifdef __KERNEL__ 1242 1243 /* 1244 * run-time mount flags 1245 */ 1246 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1247 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1248 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1249 1250 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1251 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1252 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1253 #else 1254 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1255 #endif 1256 1257 /* Number of quota types we support */ 1258 #define EXT4_MAXQUOTAS 2 1259 1260 /* 1261 * fourth extended-fs super-block data in memory 1262 */ 1263 struct ext4_sb_info { 1264 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1265 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1266 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1267 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1268 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1269 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1270 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1271 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1272 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1273 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1274 unsigned long s_overhead; /* # of fs overhead clusters */ 1275 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1276 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1277 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1278 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1279 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1280 struct buffer_head **s_group_desc; 1281 unsigned int s_mount_opt; 1282 unsigned int s_mount_opt2; 1283 unsigned int s_mount_flags; 1284 unsigned int s_def_mount_opt; 1285 ext4_fsblk_t s_sb_block; 1286 atomic64_t s_resv_clusters; 1287 kuid_t s_resuid; 1288 kgid_t s_resgid; 1289 unsigned short s_mount_state; 1290 unsigned short s_pad; 1291 int s_addr_per_block_bits; 1292 int s_desc_per_block_bits; 1293 int s_inode_size; 1294 int s_first_ino; 1295 unsigned int s_inode_readahead_blks; 1296 unsigned int s_inode_goal; 1297 spinlock_t s_next_gen_lock; 1298 u32 s_next_generation; 1299 u32 s_hash_seed[4]; 1300 int s_def_hash_version; 1301 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1302 struct percpu_counter s_freeclusters_counter; 1303 struct percpu_counter s_freeinodes_counter; 1304 struct percpu_counter s_dirs_counter; 1305 struct percpu_counter s_dirtyclusters_counter; 1306 struct blockgroup_lock *s_blockgroup_lock; 1307 struct proc_dir_entry *s_proc; 1308 struct kobject s_kobj; 1309 struct completion s_kobj_unregister; 1310 struct super_block *s_sb; 1311 1312 /* Journaling */ 1313 struct journal_s *s_journal; 1314 struct list_head s_orphan; 1315 struct mutex s_orphan_lock; 1316 unsigned long s_resize_flags; /* Flags indicating if there 1317 is a resizer */ 1318 unsigned long s_commit_interval; 1319 u32 s_max_batch_time; 1320 u32 s_min_batch_time; 1321 struct block_device *journal_bdev; 1322 #ifdef CONFIG_QUOTA 1323 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1324 int s_jquota_fmt; /* Format of quota to use */ 1325 #endif 1326 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1327 struct rb_root system_blks; 1328 1329 #ifdef EXTENTS_STATS 1330 /* ext4 extents stats */ 1331 unsigned long s_ext_min; 1332 unsigned long s_ext_max; 1333 unsigned long s_depth_max; 1334 spinlock_t s_ext_stats_lock; 1335 unsigned long s_ext_blocks; 1336 unsigned long s_ext_extents; 1337 #endif 1338 1339 /* for buddy allocator */ 1340 struct ext4_group_info ***s_group_info; 1341 struct inode *s_buddy_cache; 1342 spinlock_t s_md_lock; 1343 unsigned short *s_mb_offsets; 1344 unsigned int *s_mb_maxs; 1345 unsigned int s_group_info_size; 1346 1347 /* tunables */ 1348 unsigned long s_stripe; 1349 unsigned int s_mb_stream_request; 1350 unsigned int s_mb_max_to_scan; 1351 unsigned int s_mb_min_to_scan; 1352 unsigned int s_mb_stats; 1353 unsigned int s_mb_order2_reqs; 1354 unsigned int s_mb_group_prealloc; 1355 unsigned int s_max_dir_size_kb; 1356 /* where last allocation was done - for stream allocation */ 1357 unsigned long s_mb_last_group; 1358 unsigned long s_mb_last_start; 1359 1360 /* stats for buddy allocator */ 1361 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1362 atomic_t s_bal_success; /* we found long enough chunks */ 1363 atomic_t s_bal_allocated; /* in blocks */ 1364 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1365 atomic_t s_bal_goals; /* goal hits */ 1366 atomic_t s_bal_breaks; /* too long searches */ 1367 atomic_t s_bal_2orders; /* 2^order hits */ 1368 spinlock_t s_bal_lock; 1369 unsigned long s_mb_buddies_generated; 1370 unsigned long long s_mb_generation_time; 1371 atomic_t s_mb_lost_chunks; 1372 atomic_t s_mb_preallocated; 1373 atomic_t s_mb_discarded; 1374 atomic_t s_lock_busy; 1375 1376 /* locality groups */ 1377 struct ext4_locality_group __percpu *s_locality_groups; 1378 1379 /* for write statistics */ 1380 unsigned long s_sectors_written_start; 1381 u64 s_kbytes_written; 1382 1383 /* the size of zero-out chunk */ 1384 unsigned int s_extent_max_zeroout_kb; 1385 1386 unsigned int s_log_groups_per_flex; 1387 struct flex_groups *s_flex_groups; 1388 ext4_group_t s_flex_groups_allocated; 1389 1390 /* workqueue for reserved extent conversions (buffered io) */ 1391 struct workqueue_struct *rsv_conversion_wq; 1392 1393 /* timer for periodic error stats printing */ 1394 struct timer_list s_err_report; 1395 1396 /* Lazy inode table initialization info */ 1397 struct ext4_li_request *s_li_request; 1398 /* Wait multiplier for lazy initialization thread */ 1399 unsigned int s_li_wait_mult; 1400 1401 /* Kernel thread for multiple mount protection */ 1402 struct task_struct *s_mmp_tsk; 1403 1404 /* record the last minlen when FITRIM is called. */ 1405 atomic_t s_last_trim_minblks; 1406 1407 /* Reference to checksum algorithm driver via cryptoapi */ 1408 struct crypto_shash *s_chksum_driver; 1409 1410 /* Precomputed FS UUID checksum for seeding other checksums */ 1411 __u32 s_csum_seed; 1412 1413 /* Reclaim extents from extent status tree */ 1414 struct shrinker s_es_shrinker; 1415 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1416 long s_es_nr_inode; 1417 struct ext4_es_stats s_es_stats; 1418 struct mb_cache *s_mb_cache; 1419 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1420 1421 /* Ratelimit ext4 messages. */ 1422 struct ratelimit_state s_err_ratelimit_state; 1423 struct ratelimit_state s_warning_ratelimit_state; 1424 struct ratelimit_state s_msg_ratelimit_state; 1425 }; 1426 1427 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1428 { 1429 return sb->s_fs_info; 1430 } 1431 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1432 { 1433 return container_of(inode, struct ext4_inode_info, vfs_inode); 1434 } 1435 1436 static inline struct timespec ext4_current_time(struct inode *inode) 1437 { 1438 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1439 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1440 } 1441 1442 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1443 { 1444 return ino == EXT4_ROOT_INO || 1445 ino == EXT4_USR_QUOTA_INO || 1446 ino == EXT4_GRP_QUOTA_INO || 1447 ino == EXT4_BOOT_LOADER_INO || 1448 ino == EXT4_JOURNAL_INO || 1449 ino == EXT4_RESIZE_INO || 1450 (ino >= EXT4_FIRST_INO(sb) && 1451 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1452 } 1453 1454 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1455 struct ext4_io_end *io_end) 1456 { 1457 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1458 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1459 atomic_inc(&EXT4_I(inode)->i_unwritten); 1460 } 1461 } 1462 1463 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1464 { 1465 return inode->i_private; 1466 } 1467 1468 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1469 { 1470 inode->i_private = io; 1471 } 1472 1473 /* 1474 * Inode dynamic state flags 1475 */ 1476 enum { 1477 EXT4_STATE_JDATA, /* journaled data exists */ 1478 EXT4_STATE_NEW, /* inode is newly created */ 1479 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1480 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1481 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1482 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1483 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1484 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1485 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1486 nolocking */ 1487 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1488 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */ 1489 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1490 }; 1491 1492 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1493 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1494 { \ 1495 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1496 } \ 1497 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1498 { \ 1499 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1500 } \ 1501 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1502 { \ 1503 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1504 } 1505 1506 /* Add these declarations here only so that these functions can be 1507 * found by name. Otherwise, they are very hard to locate. */ 1508 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1509 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1510 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1511 EXT4_INODE_BIT_FNS(flag, flags, 0) 1512 1513 /* Add these declarations here only so that these functions can be 1514 * found by name. Otherwise, they are very hard to locate. */ 1515 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1516 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1517 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1518 #if (BITS_PER_LONG < 64) 1519 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1520 1521 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1522 { 1523 (ei)->i_state_flags = 0; 1524 } 1525 #else 1526 EXT4_INODE_BIT_FNS(state, flags, 32) 1527 1528 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1529 { 1530 /* We depend on the fact that callers will set i_flags */ 1531 } 1532 #endif 1533 #else 1534 /* Assume that user mode programs are passing in an ext4fs superblock, not 1535 * a kernel struct super_block. This will allow us to call the feature-test 1536 * macros from user land. */ 1537 #define EXT4_SB(sb) (sb) 1538 #endif 1539 1540 /* 1541 * Returns true if the inode is inode is encrypted 1542 */ 1543 static inline int ext4_encrypted_inode(struct inode *inode) 1544 { 1545 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1546 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 1547 #else 1548 return 0; 1549 #endif 1550 } 1551 1552 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1553 1554 /* 1555 * Codes for operating systems 1556 */ 1557 #define EXT4_OS_LINUX 0 1558 #define EXT4_OS_HURD 1 1559 #define EXT4_OS_MASIX 2 1560 #define EXT4_OS_FREEBSD 3 1561 #define EXT4_OS_LITES 4 1562 1563 /* 1564 * Revision levels 1565 */ 1566 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1567 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1568 1569 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1570 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1571 1572 #define EXT4_GOOD_OLD_INODE_SIZE 128 1573 1574 /* 1575 * Feature set definitions 1576 */ 1577 1578 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */ 1579 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1580 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1581 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1582 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1583 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1584 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1585 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1586 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1587 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1588 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1589 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1590 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1591 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1592 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1593 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1594 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1595 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1596 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1597 1598 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1599 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1600 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1601 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1602 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1603 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1604 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1605 1606 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1607 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1608 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1609 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1610 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1611 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1612 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1613 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1614 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1615 /* 1616 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1617 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1618 * all other data structures' checksums. However, the METADATA_CSUM and 1619 * GDT_CSUM bits are mutually exclusive. 1620 */ 1621 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1622 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1623 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1624 1625 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1626 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1627 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1628 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1629 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1630 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1631 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1632 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1633 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1634 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1635 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1636 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1637 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1638 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1639 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1640 1641 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1642 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1643 { \ 1644 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1645 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1646 } \ 1647 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1648 { \ 1649 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1650 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1651 } \ 1652 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1653 { \ 1654 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1655 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1656 } 1657 1658 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1659 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1660 { \ 1661 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1662 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1663 } \ 1664 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1665 { \ 1666 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1667 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1668 } \ 1669 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1670 { \ 1671 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1672 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1673 } 1674 1675 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1676 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1677 { \ 1678 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1679 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1680 } \ 1681 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1682 { \ 1683 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1684 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1685 } \ 1686 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1687 { \ 1688 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1689 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1690 } 1691 1692 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1693 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1694 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1695 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1696 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1697 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1698 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1699 1700 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1701 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1702 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1703 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1704 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1705 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1706 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1707 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1708 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1709 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1710 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1711 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1712 1713 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1714 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1715 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1716 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1717 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1718 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1719 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1720 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1721 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1722 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1723 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1724 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1725 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1726 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1727 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1728 1729 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1730 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1731 EXT4_FEATURE_INCOMPAT_META_BG) 1732 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1733 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1734 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1735 1736 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1737 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1738 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1739 EXT4_FEATURE_INCOMPAT_META_BG) 1740 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1741 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1742 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1743 1744 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1745 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1746 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1747 EXT4_FEATURE_INCOMPAT_META_BG| \ 1748 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1749 EXT4_FEATURE_INCOMPAT_64BIT| \ 1750 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1751 EXT4_FEATURE_INCOMPAT_MMP | \ 1752 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1753 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1754 EXT4_FEATURE_INCOMPAT_CSUM_SEED) 1755 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1756 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1757 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1758 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1759 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1760 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1761 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1762 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1763 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1764 EXT4_FEATURE_RO_COMPAT_QUOTA) 1765 1766 #define EXTN_FEATURE_FUNCS(ver) \ 1767 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1768 { \ 1769 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1770 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1771 } \ 1772 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1773 { \ 1774 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1775 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1776 } \ 1777 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1778 { \ 1779 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1780 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1781 } 1782 1783 EXTN_FEATURE_FUNCS(2) 1784 EXTN_FEATURE_FUNCS(3) 1785 EXTN_FEATURE_FUNCS(4) 1786 1787 static inline bool ext4_has_compat_features(struct super_block *sb) 1788 { 1789 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1790 } 1791 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1792 { 1793 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1794 } 1795 static inline bool ext4_has_incompat_features(struct super_block *sb) 1796 { 1797 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1798 } 1799 1800 /* 1801 * Default values for user and/or group using reserved blocks 1802 */ 1803 #define EXT4_DEF_RESUID 0 1804 #define EXT4_DEF_RESGID 0 1805 1806 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1807 1808 /* 1809 * Default mount options 1810 */ 1811 #define EXT4_DEFM_DEBUG 0x0001 1812 #define EXT4_DEFM_BSDGROUPS 0x0002 1813 #define EXT4_DEFM_XATTR_USER 0x0004 1814 #define EXT4_DEFM_ACL 0x0008 1815 #define EXT4_DEFM_UID16 0x0010 1816 #define EXT4_DEFM_JMODE 0x0060 1817 #define EXT4_DEFM_JMODE_DATA 0x0020 1818 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1819 #define EXT4_DEFM_JMODE_WBACK 0x0060 1820 #define EXT4_DEFM_NOBARRIER 0x0100 1821 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1822 #define EXT4_DEFM_DISCARD 0x0400 1823 #define EXT4_DEFM_NODELALLOC 0x0800 1824 1825 /* 1826 * Default journal batch times 1827 */ 1828 #define EXT4_DEF_MIN_BATCH_TIME 0 1829 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1830 1831 /* 1832 * Minimum number of groups in a flexgroup before we separate out 1833 * directories into the first block group of a flexgroup 1834 */ 1835 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1836 1837 /* 1838 * Structure of a directory entry 1839 */ 1840 #define EXT4_NAME_LEN 255 1841 1842 struct ext4_dir_entry { 1843 __le32 inode; /* Inode number */ 1844 __le16 rec_len; /* Directory entry length */ 1845 __le16 name_len; /* Name length */ 1846 char name[EXT4_NAME_LEN]; /* File name */ 1847 }; 1848 1849 /* 1850 * The new version of the directory entry. Since EXT4 structures are 1851 * stored in intel byte order, and the name_len field could never be 1852 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1853 * file_type field. 1854 */ 1855 struct ext4_dir_entry_2 { 1856 __le32 inode; /* Inode number */ 1857 __le16 rec_len; /* Directory entry length */ 1858 __u8 name_len; /* Name length */ 1859 __u8 file_type; 1860 char name[EXT4_NAME_LEN]; /* File name */ 1861 }; 1862 1863 /* 1864 * This is a bogus directory entry at the end of each leaf block that 1865 * records checksums. 1866 */ 1867 struct ext4_dir_entry_tail { 1868 __le32 det_reserved_zero1; /* Pretend to be unused */ 1869 __le16 det_rec_len; /* 12 */ 1870 __u8 det_reserved_zero2; /* Zero name length */ 1871 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1872 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1873 }; 1874 1875 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1876 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1877 ((blocksize) - \ 1878 sizeof(struct ext4_dir_entry_tail)))) 1879 1880 /* 1881 * Ext4 directory file types. Only the low 3 bits are used. The 1882 * other bits are reserved for now. 1883 */ 1884 #define EXT4_FT_UNKNOWN 0 1885 #define EXT4_FT_REG_FILE 1 1886 #define EXT4_FT_DIR 2 1887 #define EXT4_FT_CHRDEV 3 1888 #define EXT4_FT_BLKDEV 4 1889 #define EXT4_FT_FIFO 5 1890 #define EXT4_FT_SOCK 6 1891 #define EXT4_FT_SYMLINK 7 1892 1893 #define EXT4_FT_MAX 8 1894 1895 #define EXT4_FT_DIR_CSUM 0xDE 1896 1897 /* 1898 * EXT4_DIR_PAD defines the directory entries boundaries 1899 * 1900 * NOTE: It must be a multiple of 4 1901 */ 1902 #define EXT4_DIR_PAD 4 1903 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1904 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1905 ~EXT4_DIR_ROUND) 1906 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1907 1908 /* 1909 * If we ever get support for fs block sizes > page_size, we'll need 1910 * to remove the #if statements in the next two functions... 1911 */ 1912 static inline unsigned int 1913 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1914 { 1915 unsigned len = le16_to_cpu(dlen); 1916 1917 #if (PAGE_CACHE_SIZE >= 65536) 1918 if (len == EXT4_MAX_REC_LEN || len == 0) 1919 return blocksize; 1920 return (len & 65532) | ((len & 3) << 16); 1921 #else 1922 return len; 1923 #endif 1924 } 1925 1926 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1927 { 1928 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1929 BUG(); 1930 #if (PAGE_CACHE_SIZE >= 65536) 1931 if (len < 65536) 1932 return cpu_to_le16(len); 1933 if (len == blocksize) { 1934 if (blocksize == 65536) 1935 return cpu_to_le16(EXT4_MAX_REC_LEN); 1936 else 1937 return cpu_to_le16(0); 1938 } 1939 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1940 #else 1941 return cpu_to_le16(len); 1942 #endif 1943 } 1944 1945 /* 1946 * Hash Tree Directory indexing 1947 * (c) Daniel Phillips, 2001 1948 */ 1949 1950 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1951 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1952 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1953 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1954 1955 /* Legal values for the dx_root hash_version field: */ 1956 1957 #define DX_HASH_LEGACY 0 1958 #define DX_HASH_HALF_MD4 1 1959 #define DX_HASH_TEA 2 1960 #define DX_HASH_LEGACY_UNSIGNED 3 1961 #define DX_HASH_HALF_MD4_UNSIGNED 4 1962 #define DX_HASH_TEA_UNSIGNED 5 1963 1964 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1965 const void *address, unsigned int length) 1966 { 1967 struct { 1968 struct shash_desc shash; 1969 char ctx[4]; 1970 } desc; 1971 int err; 1972 1973 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1974 1975 desc.shash.tfm = sbi->s_chksum_driver; 1976 desc.shash.flags = 0; 1977 *(u32 *)desc.ctx = crc; 1978 1979 err = crypto_shash_update(&desc.shash, address, length); 1980 BUG_ON(err); 1981 1982 return *(u32 *)desc.ctx; 1983 } 1984 1985 #ifdef __KERNEL__ 1986 1987 /* hash info structure used by the directory hash */ 1988 struct dx_hash_info 1989 { 1990 u32 hash; 1991 u32 minor_hash; 1992 int hash_version; 1993 u32 *seed; 1994 }; 1995 1996 1997 /* 32 and 64 bit signed EOF for dx directories */ 1998 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1999 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2000 2001 2002 /* 2003 * Control parameters used by ext4_htree_next_block 2004 */ 2005 #define HASH_NB_ALWAYS 1 2006 2007 struct ext4_filename { 2008 const struct qstr *usr_fname; 2009 struct ext4_str disk_name; 2010 struct dx_hash_info hinfo; 2011 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2012 struct ext4_str crypto_buf; 2013 #endif 2014 }; 2015 2016 #define fname_name(p) ((p)->disk_name.name) 2017 #define fname_len(p) ((p)->disk_name.len) 2018 2019 /* 2020 * Describe an inode's exact location on disk and in memory 2021 */ 2022 struct ext4_iloc 2023 { 2024 struct buffer_head *bh; 2025 unsigned long offset; 2026 ext4_group_t block_group; 2027 }; 2028 2029 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2030 { 2031 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2032 } 2033 2034 /* 2035 * This structure is stuffed into the struct file's private_data field 2036 * for directories. It is where we put information so that we can do 2037 * readdir operations in hash tree order. 2038 */ 2039 struct dir_private_info { 2040 struct rb_root root; 2041 struct rb_node *curr_node; 2042 struct fname *extra_fname; 2043 loff_t last_pos; 2044 __u32 curr_hash; 2045 __u32 curr_minor_hash; 2046 __u32 next_hash; 2047 }; 2048 2049 /* calculate the first block number of the group */ 2050 static inline ext4_fsblk_t 2051 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2052 { 2053 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2054 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2055 } 2056 2057 /* 2058 * Special error return code only used by dx_probe() and its callers. 2059 */ 2060 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2061 2062 /* 2063 * Timeout and state flag for lazy initialization inode thread. 2064 */ 2065 #define EXT4_DEF_LI_WAIT_MULT 10 2066 #define EXT4_DEF_LI_MAX_START_DELAY 5 2067 #define EXT4_LAZYINIT_QUIT 0x0001 2068 #define EXT4_LAZYINIT_RUNNING 0x0002 2069 2070 /* 2071 * Lazy inode table initialization info 2072 */ 2073 struct ext4_lazy_init { 2074 unsigned long li_state; 2075 struct list_head li_request_list; 2076 struct mutex li_list_mtx; 2077 }; 2078 2079 struct ext4_li_request { 2080 struct super_block *lr_super; 2081 struct ext4_sb_info *lr_sbi; 2082 ext4_group_t lr_next_group; 2083 struct list_head lr_request; 2084 unsigned long lr_next_sched; 2085 unsigned long lr_timeout; 2086 }; 2087 2088 struct ext4_features { 2089 struct kobject f_kobj; 2090 struct completion f_kobj_unregister; 2091 }; 2092 2093 /* 2094 * This structure will be used for multiple mount protection. It will be 2095 * written into the block number saved in the s_mmp_block field in the 2096 * superblock. Programs that check MMP should assume that if 2097 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2098 * to use the filesystem, regardless of how old the timestamp is. 2099 */ 2100 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2101 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2102 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2103 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2104 2105 struct mmp_struct { 2106 __le32 mmp_magic; /* Magic number for MMP */ 2107 __le32 mmp_seq; /* Sequence no. updated periodically */ 2108 2109 /* 2110 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2111 * purposes and do not affect the correctness of the algorithm 2112 */ 2113 __le64 mmp_time; /* Time last updated */ 2114 char mmp_nodename[64]; /* Node which last updated MMP block */ 2115 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2116 2117 /* 2118 * mmp_check_interval is used to verify if the MMP block has been 2119 * updated on the block device. The value is updated based on the 2120 * maximum time to write the MMP block during an update cycle. 2121 */ 2122 __le16 mmp_check_interval; 2123 2124 __le16 mmp_pad1; 2125 __le32 mmp_pad2[226]; 2126 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2127 }; 2128 2129 /* arguments passed to the mmp thread */ 2130 struct mmpd_data { 2131 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2132 struct super_block *sb; /* super block of the fs */ 2133 }; 2134 2135 /* 2136 * Check interval multiplier 2137 * The MMP block is written every update interval and initially checked every 2138 * update interval x the multiplier (the value is then adapted based on the 2139 * write latency). The reason is that writes can be delayed under load and we 2140 * don't want readers to incorrectly assume that the filesystem is no longer 2141 * in use. 2142 */ 2143 #define EXT4_MMP_CHECK_MULT 2UL 2144 2145 /* 2146 * Minimum interval for MMP checking in seconds. 2147 */ 2148 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2149 2150 /* 2151 * Maximum interval for MMP checking in seconds. 2152 */ 2153 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2154 2155 /* 2156 * Function prototypes 2157 */ 2158 2159 /* 2160 * Ok, these declarations are also in <linux/kernel.h> but none of the 2161 * ext4 source programs needs to include it so they are duplicated here. 2162 */ 2163 # define NORET_TYPE /**/ 2164 # define ATTRIB_NORET __attribute__((noreturn)) 2165 # define NORET_AND noreturn, 2166 2167 /* bitmap.c */ 2168 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2169 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2170 struct ext4_group_desc *gdp, 2171 struct buffer_head *bh, int sz); 2172 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2173 struct ext4_group_desc *gdp, 2174 struct buffer_head *bh, int sz); 2175 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2176 struct ext4_group_desc *gdp, 2177 struct buffer_head *bh); 2178 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2179 struct ext4_group_desc *gdp, 2180 struct buffer_head *bh); 2181 2182 /* balloc.c */ 2183 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2184 ext4_fsblk_t blocknr, 2185 ext4_group_t *blockgrpp, 2186 ext4_grpblk_t *offsetp); 2187 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2188 ext4_fsblk_t block); 2189 2190 extern unsigned int ext4_block_group(struct super_block *sb, 2191 ext4_fsblk_t blocknr); 2192 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2193 ext4_fsblk_t blocknr); 2194 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2195 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2196 ext4_group_t group); 2197 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2198 ext4_fsblk_t goal, 2199 unsigned int flags, 2200 unsigned long *count, 2201 int *errp); 2202 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2203 s64 nclusters, unsigned int flags); 2204 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2205 extern void ext4_check_blocks_bitmap(struct super_block *); 2206 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2207 ext4_group_t block_group, 2208 struct buffer_head ** bh); 2209 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2210 2211 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2212 ext4_group_t block_group); 2213 extern int ext4_wait_block_bitmap(struct super_block *sb, 2214 ext4_group_t block_group, 2215 struct buffer_head *bh); 2216 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2217 ext4_group_t block_group); 2218 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2219 ext4_group_t block_group, 2220 struct ext4_group_desc *gdp); 2221 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2222 2223 /* crypto_policy.c */ 2224 int ext4_is_child_context_consistent_with_parent(struct inode *parent, 2225 struct inode *child); 2226 int ext4_inherit_context(struct inode *parent, struct inode *child); 2227 void ext4_to_hex(char *dst, char *src, size_t src_size); 2228 int ext4_process_policy(const struct ext4_encryption_policy *policy, 2229 struct inode *inode); 2230 int ext4_get_policy(struct inode *inode, 2231 struct ext4_encryption_policy *policy); 2232 2233 /* crypto.c */ 2234 extern struct kmem_cache *ext4_crypt_info_cachep; 2235 bool ext4_valid_contents_enc_mode(uint32_t mode); 2236 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size); 2237 extern struct workqueue_struct *ext4_read_workqueue; 2238 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode); 2239 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx); 2240 void ext4_restore_control_page(struct page *data_page); 2241 struct page *ext4_encrypt(struct inode *inode, 2242 struct page *plaintext_page); 2243 int ext4_decrypt(struct page *page); 2244 int ext4_encrypted_zeroout(struct inode *inode, ext4_lblk_t lblk, 2245 ext4_fsblk_t pblk, ext4_lblk_t len); 2246 2247 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2248 int ext4_init_crypto(void); 2249 void ext4_exit_crypto(void); 2250 static inline int ext4_sb_has_crypto(struct super_block *sb) 2251 { 2252 return ext4_has_feature_encrypt(sb); 2253 } 2254 #else 2255 static inline int ext4_init_crypto(void) { return 0; } 2256 static inline void ext4_exit_crypto(void) { } 2257 static inline int ext4_sb_has_crypto(struct super_block *sb) 2258 { 2259 return 0; 2260 } 2261 #endif 2262 2263 /* crypto_fname.c */ 2264 bool ext4_valid_filenames_enc_mode(uint32_t mode); 2265 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize); 2266 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen); 2267 int ext4_fname_crypto_alloc_buffer(struct inode *inode, 2268 u32 ilen, struct ext4_str *crypto_str); 2269 int _ext4_fname_disk_to_usr(struct inode *inode, 2270 struct dx_hash_info *hinfo, 2271 const struct ext4_str *iname, 2272 struct ext4_str *oname); 2273 int ext4_fname_disk_to_usr(struct inode *inode, 2274 struct dx_hash_info *hinfo, 2275 const struct ext4_dir_entry_2 *de, 2276 struct ext4_str *oname); 2277 int ext4_fname_usr_to_disk(struct inode *inode, 2278 const struct qstr *iname, 2279 struct ext4_str *oname); 2280 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2281 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str); 2282 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2283 int lookup, struct ext4_filename *fname); 2284 void ext4_fname_free_filename(struct ext4_filename *fname); 2285 #else 2286 static inline 2287 int ext4_setup_fname_crypto(struct inode *inode) 2288 { 2289 return 0; 2290 } 2291 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { } 2292 static inline int ext4_fname_setup_filename(struct inode *dir, 2293 const struct qstr *iname, 2294 int lookup, struct ext4_filename *fname) 2295 { 2296 fname->usr_fname = iname; 2297 fname->disk_name.name = (unsigned char *) iname->name; 2298 fname->disk_name.len = iname->len; 2299 return 0; 2300 } 2301 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2302 #endif 2303 2304 2305 /* crypto_key.c */ 2306 void ext4_free_crypt_info(struct ext4_crypt_info *ci); 2307 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci); 2308 int _ext4_get_encryption_info(struct inode *inode); 2309 2310 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2311 int ext4_has_encryption_key(struct inode *inode); 2312 2313 static inline int ext4_get_encryption_info(struct inode *inode) 2314 { 2315 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info; 2316 2317 if (!ci || 2318 (ci->ci_keyring_key && 2319 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) | 2320 (1 << KEY_FLAG_REVOKED) | 2321 (1 << KEY_FLAG_DEAD))))) 2322 return _ext4_get_encryption_info(inode); 2323 return 0; 2324 } 2325 2326 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode) 2327 { 2328 return EXT4_I(inode)->i_crypt_info; 2329 } 2330 2331 #else 2332 static inline int ext4_has_encryption_key(struct inode *inode) 2333 { 2334 return 0; 2335 } 2336 static inline int ext4_get_encryption_info(struct inode *inode) 2337 { 2338 return 0; 2339 } 2340 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode) 2341 { 2342 return NULL; 2343 } 2344 #endif 2345 2346 2347 /* dir.c */ 2348 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2349 struct file *, 2350 struct ext4_dir_entry_2 *, 2351 struct buffer_head *, char *, int, 2352 unsigned int); 2353 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2354 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2355 (de), (bh), (buf), (size), (offset))) 2356 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2357 __u32 minor_hash, 2358 struct ext4_dir_entry_2 *dirent, 2359 struct ext4_str *ent_name); 2360 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2361 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2362 struct buffer_head *bh, 2363 void *buf, int buf_size, 2364 struct ext4_filename *fname, 2365 struct ext4_dir_entry_2 **dest_de); 2366 int ext4_insert_dentry(struct inode *dir, 2367 struct inode *inode, 2368 struct ext4_dir_entry_2 *de, 2369 int buf_size, 2370 struct ext4_filename *fname); 2371 static inline void ext4_update_dx_flag(struct inode *inode) 2372 { 2373 if (!ext4_has_feature_dir_index(inode->i_sb)) 2374 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2375 } 2376 static unsigned char ext4_filetype_table[] = { 2377 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2378 }; 2379 2380 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2381 { 2382 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2383 return DT_UNKNOWN; 2384 2385 return ext4_filetype_table[filetype]; 2386 } 2387 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2388 void *buf, int buf_size); 2389 2390 /* fsync.c */ 2391 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2392 2393 /* hash.c */ 2394 extern int ext4fs_dirhash(const char *name, int len, struct 2395 dx_hash_info *hinfo); 2396 2397 /* ialloc.c */ 2398 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2399 const struct qstr *qstr, __u32 goal, 2400 uid_t *owner, int handle_type, 2401 unsigned int line_no, int nblocks); 2402 2403 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2404 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2405 0, 0, 0) 2406 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2407 type, nblocks) \ 2408 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2409 (type), __LINE__, (nblocks)) 2410 2411 2412 extern void ext4_free_inode(handle_t *, struct inode *); 2413 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2414 extern unsigned long ext4_count_free_inodes(struct super_block *); 2415 extern unsigned long ext4_count_dirs(struct super_block *); 2416 extern void ext4_check_inodes_bitmap(struct super_block *); 2417 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2418 extern int ext4_init_inode_table(struct super_block *sb, 2419 ext4_group_t group, int barrier); 2420 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2421 2422 /* mballoc.c */ 2423 extern const struct file_operations ext4_seq_mb_groups_fops; 2424 extern long ext4_mb_stats; 2425 extern long ext4_mb_max_to_scan; 2426 extern int ext4_mb_init(struct super_block *); 2427 extern int ext4_mb_release(struct super_block *); 2428 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2429 struct ext4_allocation_request *, int *); 2430 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2431 extern void ext4_discard_preallocations(struct inode *); 2432 extern int __init ext4_init_mballoc(void); 2433 extern void ext4_exit_mballoc(void); 2434 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2435 struct buffer_head *bh, ext4_fsblk_t block, 2436 unsigned long count, int flags); 2437 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2438 ext4_group_t ngroups); 2439 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2440 ext4_group_t i, struct ext4_group_desc *desc); 2441 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2442 ext4_fsblk_t block, unsigned long count); 2443 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2444 2445 /* inode.c */ 2446 int ext4_inode_is_fast_symlink(struct inode *inode); 2447 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2448 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2449 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2450 struct buffer_head *bh_result, int create); 2451 int ext4_get_block_dax(struct inode *inode, sector_t iblock, 2452 struct buffer_head *bh_result, int create); 2453 int ext4_get_block(struct inode *inode, sector_t iblock, 2454 struct buffer_head *bh_result, int create); 2455 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2456 struct buffer_head *bh, int create); 2457 int ext4_walk_page_buffers(handle_t *handle, 2458 struct buffer_head *head, 2459 unsigned from, 2460 unsigned to, 2461 int *partial, 2462 int (*fn)(handle_t *handle, 2463 struct buffer_head *bh)); 2464 int do_journal_get_write_access(handle_t *handle, 2465 struct buffer_head *bh); 2466 #define FALL_BACK_TO_NONDELALLOC 1 2467 #define CONVERT_INLINE_DATA 2 2468 2469 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2470 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2471 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2472 extern int ext4_setattr(struct dentry *, struct iattr *); 2473 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2474 struct kstat *stat); 2475 extern void ext4_evict_inode(struct inode *); 2476 extern void ext4_clear_inode(struct inode *); 2477 extern int ext4_sync_inode(handle_t *, struct inode *); 2478 extern void ext4_dirty_inode(struct inode *, int); 2479 extern int ext4_change_inode_journal_flag(struct inode *, int); 2480 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2481 extern int ext4_inode_attach_jinode(struct inode *inode); 2482 extern int ext4_can_truncate(struct inode *inode); 2483 extern void ext4_truncate(struct inode *); 2484 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2485 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2486 extern void ext4_set_inode_flags(struct inode *); 2487 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2488 extern int ext4_alloc_da_blocks(struct inode *inode); 2489 extern void ext4_set_aops(struct inode *inode); 2490 extern int ext4_writepage_trans_blocks(struct inode *); 2491 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2492 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2493 loff_t lstart, loff_t lend); 2494 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2495 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf); 2496 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2497 extern void ext4_da_update_reserve_space(struct inode *inode, 2498 int used, int quota_claim); 2499 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2500 ext4_fsblk_t pblk, ext4_lblk_t len); 2501 2502 /* indirect.c */ 2503 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2504 struct ext4_map_blocks *map, int flags); 2505 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter, 2506 loff_t offset); 2507 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2508 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2509 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2510 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2511 ext4_lblk_t start, ext4_lblk_t end); 2512 2513 /* ioctl.c */ 2514 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2515 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2516 2517 /* migrate.c */ 2518 extern int ext4_ext_migrate(struct inode *); 2519 extern int ext4_ind_migrate(struct inode *inode); 2520 2521 /* namei.c */ 2522 extern int ext4_dirent_csum_verify(struct inode *inode, 2523 struct ext4_dir_entry *dirent); 2524 extern int ext4_orphan_add(handle_t *, struct inode *); 2525 extern int ext4_orphan_del(handle_t *, struct inode *); 2526 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2527 __u32 start_minor_hash, __u32 *next_hash); 2528 extern int ext4_search_dir(struct buffer_head *bh, 2529 char *search_buf, 2530 int buf_size, 2531 struct inode *dir, 2532 struct ext4_filename *fname, 2533 const struct qstr *d_name, 2534 unsigned int offset, 2535 struct ext4_dir_entry_2 **res_dir); 2536 extern int ext4_generic_delete_entry(handle_t *handle, 2537 struct inode *dir, 2538 struct ext4_dir_entry_2 *de_del, 2539 struct buffer_head *bh, 2540 void *entry_buf, 2541 int buf_size, 2542 int csum_size); 2543 extern int ext4_empty_dir(struct inode *inode); 2544 2545 /* resize.c */ 2546 extern int ext4_group_add(struct super_block *sb, 2547 struct ext4_new_group_data *input); 2548 extern int ext4_group_extend(struct super_block *sb, 2549 struct ext4_super_block *es, 2550 ext4_fsblk_t n_blocks_count); 2551 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2552 2553 /* super.c */ 2554 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2555 extern int ext4_calculate_overhead(struct super_block *sb); 2556 extern void ext4_superblock_csum_set(struct super_block *sb); 2557 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2558 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2559 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2560 ext4_group_t ngroup); 2561 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2562 char nbuf[16]); 2563 2564 extern __printf(4, 5) 2565 void __ext4_error(struct super_block *, const char *, unsigned int, 2566 const char *, ...); 2567 extern __printf(5, 6) 2568 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2569 const char *, ...); 2570 extern __printf(5, 6) 2571 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2572 const char *, ...); 2573 extern void __ext4_std_error(struct super_block *, const char *, 2574 unsigned int, int); 2575 extern __printf(4, 5) 2576 void __ext4_abort(struct super_block *, const char *, unsigned int, 2577 const char *, ...); 2578 extern __printf(4, 5) 2579 void __ext4_warning(struct super_block *, const char *, unsigned int, 2580 const char *, ...); 2581 extern __printf(4, 5) 2582 void __ext4_warning_inode(const struct inode *inode, const char *function, 2583 unsigned int line, const char *fmt, ...); 2584 extern __printf(3, 4) 2585 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2586 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2587 const char *, unsigned int, const char *); 2588 extern __printf(7, 8) 2589 void __ext4_grp_locked_error(const char *, unsigned int, 2590 struct super_block *, ext4_group_t, 2591 unsigned long, ext4_fsblk_t, 2592 const char *, ...); 2593 2594 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2595 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2596 2597 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2598 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2599 2600 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2601 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2602 2603 #ifdef CONFIG_PRINTK 2604 2605 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2606 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2607 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2608 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2609 #define ext4_error(sb, fmt, ...) \ 2610 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2611 #define ext4_abort(sb, fmt, ...) \ 2612 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2613 #define ext4_warning(sb, fmt, ...) \ 2614 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2615 #define ext4_warning_inode(inode, fmt, ...) \ 2616 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2617 #define ext4_msg(sb, level, fmt, ...) \ 2618 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2619 #define dump_mmp_msg(sb, mmp, msg) \ 2620 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2621 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2622 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2623 fmt, ##__VA_ARGS__) 2624 2625 #else 2626 2627 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2628 do { \ 2629 no_printk(fmt, ##__VA_ARGS__); \ 2630 __ext4_error_inode(inode, "", 0, block, " "); \ 2631 } while (0) 2632 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2633 do { \ 2634 no_printk(fmt, ##__VA_ARGS__); \ 2635 __ext4_error_file(file, "", 0, block, " "); \ 2636 } while (0) 2637 #define ext4_error(sb, fmt, ...) \ 2638 do { \ 2639 no_printk(fmt, ##__VA_ARGS__); \ 2640 __ext4_error(sb, "", 0, " "); \ 2641 } while (0) 2642 #define ext4_abort(sb, fmt, ...) \ 2643 do { \ 2644 no_printk(fmt, ##__VA_ARGS__); \ 2645 __ext4_abort(sb, "", 0, " "); \ 2646 } while (0) 2647 #define ext4_warning(sb, fmt, ...) \ 2648 do { \ 2649 no_printk(fmt, ##__VA_ARGS__); \ 2650 __ext4_warning(sb, "", 0, " "); \ 2651 } while (0) 2652 #define ext4_warning_inode(inode, fmt, ...) \ 2653 do { \ 2654 no_printk(fmt, ##__VA_ARGS__); \ 2655 __ext4_warning_inode(inode, "", 0, " "); \ 2656 } while (0) 2657 #define ext4_msg(sb, level, fmt, ...) \ 2658 do { \ 2659 no_printk(fmt, ##__VA_ARGS__); \ 2660 __ext4_msg(sb, "", " "); \ 2661 } while (0) 2662 #define dump_mmp_msg(sb, mmp, msg) \ 2663 __dump_mmp_msg(sb, mmp, "", 0, "") 2664 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2665 do { \ 2666 no_printk(fmt, ##__VA_ARGS__); \ 2667 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2668 } while (0) 2669 2670 #endif 2671 2672 extern void ext4_update_dynamic_rev(struct super_block *sb); 2673 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2674 __u32 compat); 2675 extern int ext4_update_rocompat_feature(handle_t *handle, 2676 struct super_block *sb, __u32 rocompat); 2677 extern int ext4_update_incompat_feature(handle_t *handle, 2678 struct super_block *sb, __u32 incompat); 2679 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2680 struct ext4_group_desc *bg); 2681 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2682 struct ext4_group_desc *bg); 2683 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2684 struct ext4_group_desc *bg); 2685 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2686 struct ext4_group_desc *bg); 2687 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2688 struct ext4_group_desc *bg); 2689 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2690 struct ext4_group_desc *bg); 2691 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2692 struct ext4_group_desc *bg); 2693 extern void ext4_block_bitmap_set(struct super_block *sb, 2694 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2695 extern void ext4_inode_bitmap_set(struct super_block *sb, 2696 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2697 extern void ext4_inode_table_set(struct super_block *sb, 2698 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2699 extern void ext4_free_group_clusters_set(struct super_block *sb, 2700 struct ext4_group_desc *bg, 2701 __u32 count); 2702 extern void ext4_free_inodes_set(struct super_block *sb, 2703 struct ext4_group_desc *bg, __u32 count); 2704 extern void ext4_used_dirs_set(struct super_block *sb, 2705 struct ext4_group_desc *bg, __u32 count); 2706 extern void ext4_itable_unused_set(struct super_block *sb, 2707 struct ext4_group_desc *bg, __u32 count); 2708 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2709 struct ext4_group_desc *gdp); 2710 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2711 struct ext4_group_desc *gdp); 2712 extern int ext4_register_li_request(struct super_block *sb, 2713 ext4_group_t first_not_zeroed); 2714 2715 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2716 { 2717 return ext4_has_feature_gdt_csum(sb) || 2718 EXT4_SB(sb)->s_chksum_driver != NULL; 2719 } 2720 2721 static inline int ext4_has_metadata_csum(struct super_block *sb) 2722 { 2723 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2724 !EXT4_SB(sb)->s_chksum_driver); 2725 2726 return (EXT4_SB(sb)->s_chksum_driver != NULL); 2727 } 2728 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2729 { 2730 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2731 le32_to_cpu(es->s_blocks_count_lo); 2732 } 2733 2734 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2735 { 2736 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2737 le32_to_cpu(es->s_r_blocks_count_lo); 2738 } 2739 2740 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2741 { 2742 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2743 le32_to_cpu(es->s_free_blocks_count_lo); 2744 } 2745 2746 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2747 ext4_fsblk_t blk) 2748 { 2749 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2750 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2751 } 2752 2753 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2754 ext4_fsblk_t blk) 2755 { 2756 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2757 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2758 } 2759 2760 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2761 ext4_fsblk_t blk) 2762 { 2763 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2764 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2765 } 2766 2767 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2768 { 2769 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2770 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2771 le32_to_cpu(raw_inode->i_size_lo); 2772 else 2773 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2774 } 2775 2776 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2777 { 2778 raw_inode->i_size_lo = cpu_to_le32(i_size); 2779 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2780 } 2781 2782 static inline 2783 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2784 ext4_group_t group) 2785 { 2786 struct ext4_group_info ***grp_info; 2787 long indexv, indexh; 2788 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2789 grp_info = EXT4_SB(sb)->s_group_info; 2790 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2791 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2792 return grp_info[indexv][indexh]; 2793 } 2794 2795 /* 2796 * Reading s_groups_count requires using smp_rmb() afterwards. See 2797 * the locking protocol documented in the comments of ext4_group_add() 2798 * in resize.c 2799 */ 2800 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2801 { 2802 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2803 2804 smp_rmb(); 2805 return ngroups; 2806 } 2807 2808 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2809 ext4_group_t block_group) 2810 { 2811 return block_group >> sbi->s_log_groups_per_flex; 2812 } 2813 2814 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2815 { 2816 return 1 << sbi->s_log_groups_per_flex; 2817 } 2818 2819 #define ext4_std_error(sb, errno) \ 2820 do { \ 2821 if ((errno)) \ 2822 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2823 } while (0) 2824 2825 #ifdef CONFIG_SMP 2826 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2827 * counters. So we need to make sure we have free clusters more 2828 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2829 */ 2830 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2831 #else 2832 #define EXT4_FREECLUSTERS_WATERMARK 0 2833 #endif 2834 2835 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2836 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2837 { 2838 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2839 !mutex_is_locked(&inode->i_mutex)); 2840 down_write(&EXT4_I(inode)->i_data_sem); 2841 if (newsize > EXT4_I(inode)->i_disksize) 2842 EXT4_I(inode)->i_disksize = newsize; 2843 up_write(&EXT4_I(inode)->i_data_sem); 2844 } 2845 2846 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2847 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2848 { 2849 int changed = 0; 2850 2851 if (newsize > inode->i_size) { 2852 i_size_write(inode, newsize); 2853 changed = 1; 2854 } 2855 if (newsize > EXT4_I(inode)->i_disksize) { 2856 ext4_update_i_disksize(inode, newsize); 2857 changed |= 2; 2858 } 2859 return changed; 2860 } 2861 2862 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2863 loff_t len); 2864 2865 struct ext4_group_info { 2866 unsigned long bb_state; 2867 struct rb_root bb_free_root; 2868 ext4_grpblk_t bb_first_free; /* first free block */ 2869 ext4_grpblk_t bb_free; /* total free blocks */ 2870 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2871 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2872 struct list_head bb_prealloc_list; 2873 #ifdef DOUBLE_CHECK 2874 void *bb_bitmap; 2875 #endif 2876 struct rw_semaphore alloc_sem; 2877 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2878 * regions, index is order. 2879 * bb_counters[3] = 5 means 2880 * 5 free 8-block regions. */ 2881 }; 2882 2883 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2884 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2885 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2886 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2887 2888 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2889 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2890 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2891 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2892 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2893 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2894 2895 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2896 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2897 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2898 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2899 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2900 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2901 2902 #define EXT4_MAX_CONTENTION 8 2903 #define EXT4_CONTENTION_THRESHOLD 2 2904 2905 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2906 ext4_group_t group) 2907 { 2908 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2909 } 2910 2911 /* 2912 * Returns true if the filesystem is busy enough that attempts to 2913 * access the block group locks has run into contention. 2914 */ 2915 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2916 { 2917 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2918 } 2919 2920 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2921 { 2922 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2923 if (spin_trylock(lock)) 2924 /* 2925 * We're able to grab the lock right away, so drop the 2926 * lock contention counter. 2927 */ 2928 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2929 else { 2930 /* 2931 * The lock is busy, so bump the contention counter, 2932 * and then wait on the spin lock. 2933 */ 2934 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2935 EXT4_MAX_CONTENTION); 2936 spin_lock(lock); 2937 } 2938 } 2939 2940 static inline void ext4_unlock_group(struct super_block *sb, 2941 ext4_group_t group) 2942 { 2943 spin_unlock(ext4_group_lock_ptr(sb, group)); 2944 } 2945 2946 /* 2947 * Block validity checking 2948 */ 2949 #define ext4_check_indirect_blockref(inode, bh) \ 2950 ext4_check_blockref(__func__, __LINE__, inode, \ 2951 (__le32 *)(bh)->b_data, \ 2952 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2953 2954 #define ext4_ind_check_inode(inode) \ 2955 ext4_check_blockref(__func__, __LINE__, inode, \ 2956 EXT4_I(inode)->i_data, \ 2957 EXT4_NDIR_BLOCKS) 2958 2959 /* 2960 * Inodes and files operations 2961 */ 2962 2963 /* dir.c */ 2964 extern const struct file_operations ext4_dir_operations; 2965 2966 /* file.c */ 2967 extern const struct inode_operations ext4_file_inode_operations; 2968 extern const struct file_operations ext4_file_operations; 2969 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2970 2971 /* inline.c */ 2972 extern int ext4_get_max_inline_size(struct inode *inode); 2973 extern int ext4_find_inline_data_nolock(struct inode *inode); 2974 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2975 unsigned int len); 2976 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2977 2978 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2979 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2980 struct inode *inode, 2981 loff_t pos, unsigned len, 2982 unsigned flags, 2983 struct page **pagep); 2984 extern int ext4_write_inline_data_end(struct inode *inode, 2985 loff_t pos, unsigned len, 2986 unsigned copied, 2987 struct page *page); 2988 extern struct buffer_head * 2989 ext4_journalled_write_inline_data(struct inode *inode, 2990 unsigned len, 2991 struct page *page); 2992 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2993 struct inode *inode, 2994 loff_t pos, unsigned len, 2995 unsigned flags, 2996 struct page **pagep, 2997 void **fsdata); 2998 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2999 unsigned len, unsigned copied, 3000 struct page *page); 3001 extern int ext4_try_add_inline_entry(handle_t *handle, 3002 struct ext4_filename *fname, 3003 struct dentry *dentry, 3004 struct inode *inode); 3005 extern int ext4_try_create_inline_dir(handle_t *handle, 3006 struct inode *parent, 3007 struct inode *inode); 3008 extern int ext4_read_inline_dir(struct file *filp, 3009 struct dir_context *ctx, 3010 int *has_inline_data); 3011 extern int htree_inlinedir_to_tree(struct file *dir_file, 3012 struct inode *dir, ext4_lblk_t block, 3013 struct dx_hash_info *hinfo, 3014 __u32 start_hash, __u32 start_minor_hash, 3015 int *has_inline_data); 3016 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3017 struct ext4_filename *fname, 3018 const struct qstr *d_name, 3019 struct ext4_dir_entry_2 **res_dir, 3020 int *has_inline_data); 3021 extern int ext4_delete_inline_entry(handle_t *handle, 3022 struct inode *dir, 3023 struct ext4_dir_entry_2 *de_del, 3024 struct buffer_head *bh, 3025 int *has_inline_data); 3026 extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 3027 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3028 struct ext4_dir_entry_2 **parent_de, 3029 int *retval); 3030 extern int ext4_inline_data_fiemap(struct inode *inode, 3031 struct fiemap_extent_info *fieinfo, 3032 int *has_inline, __u64 start, __u64 len); 3033 extern int ext4_try_to_evict_inline_data(handle_t *handle, 3034 struct inode *inode, 3035 int needed); 3036 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3037 3038 extern int ext4_convert_inline_data(struct inode *inode); 3039 3040 static inline int ext4_has_inline_data(struct inode *inode) 3041 { 3042 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3043 EXT4_I(inode)->i_inline_off; 3044 } 3045 3046 /* namei.c */ 3047 extern const struct inode_operations ext4_dir_inode_operations; 3048 extern const struct inode_operations ext4_special_inode_operations; 3049 extern struct dentry *ext4_get_parent(struct dentry *child); 3050 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3051 struct ext4_dir_entry_2 *de, 3052 int blocksize, int csum_size, 3053 unsigned int parent_ino, int dotdot_real_len); 3054 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3055 unsigned int blocksize); 3056 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3057 struct inode *inode, 3058 struct buffer_head *bh); 3059 #define S_SHIFT 12 3060 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 3061 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3062 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3063 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3064 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3065 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3066 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3067 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3068 }; 3069 3070 static inline void ext4_set_de_type(struct super_block *sb, 3071 struct ext4_dir_entry_2 *de, 3072 umode_t mode) { 3073 if (ext4_has_feature_filetype(sb)) 3074 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3075 } 3076 3077 /* readpages.c */ 3078 extern int ext4_mpage_readpages(struct address_space *mapping, 3079 struct list_head *pages, struct page *page, 3080 unsigned nr_pages); 3081 3082 /* symlink.c */ 3083 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3084 extern const struct inode_operations ext4_symlink_inode_operations; 3085 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3086 3087 /* sysfs.c */ 3088 extern int ext4_register_sysfs(struct super_block *sb); 3089 extern void ext4_unregister_sysfs(struct super_block *sb); 3090 extern int __init ext4_init_sysfs(void); 3091 extern void ext4_exit_sysfs(void); 3092 3093 /* block_validity */ 3094 extern void ext4_release_system_zone(struct super_block *sb); 3095 extern int ext4_setup_system_zone(struct super_block *sb); 3096 extern int __init ext4_init_system_zone(void); 3097 extern void ext4_exit_system_zone(void); 3098 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3099 ext4_fsblk_t start_blk, 3100 unsigned int count); 3101 extern int ext4_check_blockref(const char *, unsigned int, 3102 struct inode *, __le32 *, unsigned int); 3103 3104 /* extents.c */ 3105 struct ext4_ext_path; 3106 struct ext4_extent; 3107 3108 /* 3109 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3110 * __le32. 3111 */ 3112 #define EXT_MAX_BLOCKS 0xffffffff 3113 3114 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3115 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3116 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3117 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3118 struct ext4_map_blocks *map, int flags); 3119 extern void ext4_ext_truncate(handle_t *, struct inode *); 3120 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3121 ext4_lblk_t end); 3122 extern void ext4_ext_init(struct super_block *); 3123 extern void ext4_ext_release(struct super_block *); 3124 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3125 loff_t len); 3126 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3127 loff_t offset, ssize_t len); 3128 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3129 struct ext4_map_blocks *map, int flags); 3130 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3131 ext4_lblk_t lblocks); 3132 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3133 int num, 3134 struct ext4_ext_path *path); 3135 extern int ext4_can_extents_be_merged(struct inode *inode, 3136 struct ext4_extent *ex1, 3137 struct ext4_extent *ex2); 3138 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3139 struct ext4_ext_path **, 3140 struct ext4_extent *, int); 3141 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3142 struct ext4_ext_path **, 3143 int flags); 3144 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3145 extern int ext4_ext_check_inode(struct inode *inode); 3146 extern int ext4_find_delalloc_range(struct inode *inode, 3147 ext4_lblk_t lblk_start, 3148 ext4_lblk_t lblk_end); 3149 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3150 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3151 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3152 __u64 start, __u64 len); 3153 extern int ext4_ext_precache(struct inode *inode); 3154 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3155 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3156 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3157 struct inode *inode2, ext4_lblk_t lblk1, 3158 ext4_lblk_t lblk2, ext4_lblk_t count, 3159 int mark_unwritten,int *err); 3160 3161 /* move_extent.c */ 3162 extern void ext4_double_down_write_data_sem(struct inode *first, 3163 struct inode *second); 3164 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3165 struct inode *donor_inode); 3166 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3167 __u64 start_orig, __u64 start_donor, 3168 __u64 len, __u64 *moved_len); 3169 3170 /* page-io.c */ 3171 extern int __init ext4_init_pageio(void); 3172 extern void ext4_exit_pageio(void); 3173 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3174 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3175 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3176 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3177 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3178 struct writeback_control *wbc); 3179 extern void ext4_end_io_rsv_work(struct work_struct *work); 3180 extern void ext4_io_submit(struct ext4_io_submit *io); 3181 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3182 struct page *page, 3183 int len, 3184 struct writeback_control *wbc, 3185 bool keep_towrite); 3186 3187 /* mmp.c */ 3188 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3189 3190 /* 3191 * Add new method to test whether block and inode bitmaps are properly 3192 * initialized. With uninit_bg reading the block from disk is not enough 3193 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3194 */ 3195 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3196 3197 static inline int bitmap_uptodate(struct buffer_head *bh) 3198 { 3199 return (buffer_uptodate(bh) && 3200 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3201 } 3202 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3203 { 3204 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3205 } 3206 3207 /* 3208 * Disable DIO read nolock optimization, so new dioreaders will be forced 3209 * to grab i_mutex 3210 */ 3211 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 3212 { 3213 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3214 smp_mb(); 3215 } 3216 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 3217 { 3218 smp_mb(); 3219 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3220 } 3221 3222 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3223 3224 /* For ioend & aio unwritten conversion wait queues */ 3225 #define EXT4_WQ_HASH_SZ 37 3226 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3227 EXT4_WQ_HASH_SZ]) 3228 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 3229 EXT4_WQ_HASH_SZ]) 3230 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3231 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 3232 3233 #define EXT4_RESIZING 0 3234 extern int ext4_resize_begin(struct super_block *sb); 3235 extern void ext4_resize_end(struct super_block *sb); 3236 3237 #endif /* __KERNEL__ */ 3238 3239 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3240 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3241 3242 #endif /* _EXT4_H */ 3243