1 /* 2 * ext4.h 3 * 4 * Copyright (C) 1992, 1993, 1994, 1995 5 * Remy Card (card@masi.ibp.fr) 6 * Laboratoire MASI - Institut Blaise Pascal 7 * Universite Pierre et Marie Curie (Paris VI) 8 * 9 * from 10 * 11 * linux/include/linux/minix_fs.h 12 * 13 * Copyright (C) 1991, 1992 Linus Torvalds 14 */ 15 16 #ifndef _EXT4_H 17 #define _EXT4_H 18 19 #include <linux/types.h> 20 #include <linux/blkdev.h> 21 #include <linux/magic.h> 22 #include <linux/jbd2.h> 23 #include <linux/quota.h> 24 #include <linux/rwsem.h> 25 #include <linux/rbtree.h> 26 #include <linux/seqlock.h> 27 #include <linux/mutex.h> 28 #include <linux/timer.h> 29 #include <linux/wait.h> 30 #include <linux/blockgroup_lock.h> 31 #include <linux/percpu_counter.h> 32 33 /* 34 * The fourth extended filesystem constants/structures 35 */ 36 37 /* 38 * Define EXT4FS_DEBUG to produce debug messages 39 */ 40 #undef EXT4FS_DEBUG 41 42 /* 43 * Debug code 44 */ 45 #ifdef EXT4FS_DEBUG 46 #define ext4_debug(f, a...) \ 47 do { \ 48 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 49 __FILE__, __LINE__, __func__); \ 50 printk(KERN_DEBUG f, ## a); \ 51 } while (0) 52 #else 53 #define ext4_debug(f, a...) do {} while (0) 54 #endif 55 56 /* data type for block offset of block group */ 57 typedef int ext4_grpblk_t; 58 59 /* data type for filesystem-wide blocks number */ 60 typedef unsigned long long ext4_fsblk_t; 61 62 /* data type for file logical block number */ 63 typedef __u32 ext4_lblk_t; 64 65 /* data type for block group number */ 66 typedef unsigned int ext4_group_t; 67 68 /* 69 * Flags used in mballoc's allocation_context flags field. 70 * 71 * Also used to show what's going on for debugging purposes when the 72 * flag field is exported via the traceport interface 73 */ 74 75 /* prefer goal again. length */ 76 #define EXT4_MB_HINT_MERGE 0x0001 77 /* blocks already reserved */ 78 #define EXT4_MB_HINT_RESERVED 0x0002 79 /* metadata is being allocated */ 80 #define EXT4_MB_HINT_METADATA 0x0004 81 /* first blocks in the file */ 82 #define EXT4_MB_HINT_FIRST 0x0008 83 /* search for the best chunk */ 84 #define EXT4_MB_HINT_BEST 0x0010 85 /* data is being allocated */ 86 #define EXT4_MB_HINT_DATA 0x0020 87 /* don't preallocate (for tails) */ 88 #define EXT4_MB_HINT_NOPREALLOC 0x0040 89 /* allocate for locality group */ 90 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 91 /* allocate goal blocks or none */ 92 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 93 /* goal is meaningful */ 94 #define EXT4_MB_HINT_TRY_GOAL 0x0200 95 /* blocks already pre-reserved by delayed allocation */ 96 #define EXT4_MB_DELALLOC_RESERVED 0x0400 97 /* We are doing stream allocation */ 98 #define EXT4_MB_STREAM_ALLOC 0x0800 99 100 101 struct ext4_allocation_request { 102 /* target inode for block we're allocating */ 103 struct inode *inode; 104 /* how many blocks we want to allocate */ 105 unsigned int len; 106 /* logical block in target inode */ 107 ext4_lblk_t logical; 108 /* the closest logical allocated block to the left */ 109 ext4_lblk_t lleft; 110 /* the closest logical allocated block to the right */ 111 ext4_lblk_t lright; 112 /* phys. target (a hint) */ 113 ext4_fsblk_t goal; 114 /* phys. block for the closest logical allocated block to the left */ 115 ext4_fsblk_t pleft; 116 /* phys. block for the closest logical allocated block to the right */ 117 ext4_fsblk_t pright; 118 /* flags. see above EXT4_MB_HINT_* */ 119 unsigned int flags; 120 }; 121 122 /* 123 * For delayed allocation tracking 124 */ 125 struct mpage_da_data { 126 struct inode *inode; 127 sector_t b_blocknr; /* start block number of extent */ 128 size_t b_size; /* size of extent */ 129 unsigned long b_state; /* state of the extent */ 130 unsigned long first_page, next_page; /* extent of pages */ 131 struct writeback_control *wbc; 132 int io_done; 133 int pages_written; 134 int retval; 135 }; 136 #define DIO_AIO_UNWRITTEN 0x1 137 typedef struct ext4_io_end { 138 struct list_head list; /* per-file finished AIO list */ 139 struct inode *inode; /* file being written to */ 140 unsigned int flag; /* unwritten or not */ 141 int error; /* I/O error code */ 142 loff_t offset; /* offset in the file */ 143 ssize_t size; /* size of the extent */ 144 struct work_struct work; /* data work queue */ 145 } ext4_io_end_t; 146 147 /* 148 * Special inodes numbers 149 */ 150 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 151 #define EXT4_ROOT_INO 2 /* Root inode */ 152 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 153 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 154 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 155 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 156 157 /* First non-reserved inode for old ext4 filesystems */ 158 #define EXT4_GOOD_OLD_FIRST_INO 11 159 160 /* 161 * Maximal count of links to a file 162 */ 163 #define EXT4_LINK_MAX 65000 164 165 /* 166 * Macro-instructions used to manage several block sizes 167 */ 168 #define EXT4_MIN_BLOCK_SIZE 1024 169 #define EXT4_MAX_BLOCK_SIZE 65536 170 #define EXT4_MIN_BLOCK_LOG_SIZE 10 171 #ifdef __KERNEL__ 172 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 173 #else 174 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 175 #endif 176 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 177 #ifdef __KERNEL__ 178 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 179 #else 180 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 181 #endif 182 #ifdef __KERNEL__ 183 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 184 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 185 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 186 #else 187 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 188 EXT4_GOOD_OLD_INODE_SIZE : \ 189 (s)->s_inode_size) 190 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 191 EXT4_GOOD_OLD_FIRST_INO : \ 192 (s)->s_first_ino) 193 #endif 194 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 195 196 /* 197 * Structure of a blocks group descriptor 198 */ 199 struct ext4_group_desc 200 { 201 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 202 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 203 __le32 bg_inode_table_lo; /* Inodes table block */ 204 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 205 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 206 __le16 bg_used_dirs_count_lo; /* Directories count */ 207 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 208 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */ 209 __le16 bg_itable_unused_lo; /* Unused inodes count */ 210 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 211 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 212 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 213 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 214 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 215 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 216 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 217 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 218 __u32 bg_reserved2[3]; 219 }; 220 221 /* 222 * Structure of a flex block group info 223 */ 224 225 struct flex_groups { 226 atomic_t free_inodes; 227 atomic_t free_blocks; 228 atomic_t used_dirs; 229 }; 230 231 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 232 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 233 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 234 235 /* 236 * Macro-instructions used to manage group descriptors 237 */ 238 #define EXT4_MIN_DESC_SIZE 32 239 #define EXT4_MIN_DESC_SIZE_64BIT 64 240 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 241 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 242 #ifdef __KERNEL__ 243 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 244 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 245 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 246 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 247 #else 248 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 249 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 250 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 251 #endif 252 253 /* 254 * Constants relative to the data blocks 255 */ 256 #define EXT4_NDIR_BLOCKS 12 257 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 258 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 259 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 260 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 261 262 /* 263 * Inode flags 264 */ 265 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 266 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 267 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 268 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 269 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 270 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 271 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 272 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 273 /* Reserved for compression usage... */ 274 #define EXT4_DIRTY_FL 0x00000100 275 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 276 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 277 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 278 /* End compression flags --- maybe not all used */ 279 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 280 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 281 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 282 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 283 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 284 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 285 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 286 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 287 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 288 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 289 290 #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */ 291 #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */ 292 293 /* Flags that should be inherited by new inodes from their parent. */ 294 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 295 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\ 296 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 297 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 298 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 299 300 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 301 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 302 303 /* Flags that are appropriate for non-directories/regular files. */ 304 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 305 306 /* Mask out flags that are inappropriate for the given type of inode. */ 307 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 308 { 309 if (S_ISDIR(mode)) 310 return flags; 311 else if (S_ISREG(mode)) 312 return flags & EXT4_REG_FLMASK; 313 else 314 return flags & EXT4_OTHER_FLMASK; 315 } 316 317 /* Used to pass group descriptor data when online resize is done */ 318 struct ext4_new_group_input { 319 __u32 group; /* Group number for this data */ 320 __u64 block_bitmap; /* Absolute block number of block bitmap */ 321 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 322 __u64 inode_table; /* Absolute block number of inode table start */ 323 __u32 blocks_count; /* Total number of blocks in this group */ 324 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 325 __u16 unused; 326 }; 327 328 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 329 struct ext4_new_group_data { 330 __u32 group; 331 __u64 block_bitmap; 332 __u64 inode_bitmap; 333 __u64 inode_table; 334 __u32 blocks_count; 335 __u16 reserved_blocks; 336 __u16 unused; 337 __u32 free_blocks_count; 338 }; 339 340 /* 341 * Flags used by ext4_get_blocks() 342 */ 343 /* Allocate any needed blocks and/or convert an unitialized 344 extent to be an initialized ext4 */ 345 #define EXT4_GET_BLOCKS_CREATE 0x0001 346 /* Request the creation of an unitialized extent */ 347 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 348 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 349 EXT4_GET_BLOCKS_CREATE) 350 /* Caller is from the delayed allocation writeout path, 351 so set the magic i_delalloc_reserve_flag after taking the 352 inode allocation semaphore for */ 353 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 354 /* caller is from the direct IO path, request to creation of an 355 unitialized extents if not allocated, split the uninitialized 356 extent if blocks has been preallocated already*/ 357 #define EXT4_GET_BLOCKS_DIO 0x0008 358 #define EXT4_GET_BLOCKS_CONVERT 0x0010 359 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\ 360 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 361 /* Convert extent to initialized after direct IO complete */ 362 #define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 363 EXT4_GET_BLOCKS_DIO_CREATE_EXT) 364 365 /* 366 * Flags used by ext4_free_blocks 367 */ 368 #define EXT4_FREE_BLOCKS_METADATA 0x0001 369 #define EXT4_FREE_BLOCKS_FORGET 0x0002 370 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 371 372 /* 373 * ioctl commands 374 */ 375 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 376 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 377 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 378 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 379 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 380 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 381 #ifdef CONFIG_JBD2_DEBUG 382 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long) 383 #endif 384 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 385 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 386 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 387 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 388 #define EXT4_IOC_MIGRATE _IO('f', 9) 389 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 390 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 391 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 392 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 393 394 /* 395 * ioctl commands in 32 bit emulation 396 */ 397 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 398 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 399 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 400 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 401 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 402 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 403 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 404 #ifdef CONFIG_JBD2_DEBUG 405 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int) 406 #endif 407 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 408 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 409 410 411 /* 412 * Mount options 413 */ 414 struct ext4_mount_options { 415 unsigned long s_mount_opt; 416 uid_t s_resuid; 417 gid_t s_resgid; 418 unsigned long s_commit_interval; 419 u32 s_min_batch_time, s_max_batch_time; 420 #ifdef CONFIG_QUOTA 421 int s_jquota_fmt; 422 char *s_qf_names[MAXQUOTAS]; 423 #endif 424 }; 425 426 /* Max physical block we can addres w/o extents */ 427 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 428 429 /* 430 * Structure of an inode on the disk 431 */ 432 struct ext4_inode { 433 __le16 i_mode; /* File mode */ 434 __le16 i_uid; /* Low 16 bits of Owner Uid */ 435 __le32 i_size_lo; /* Size in bytes */ 436 __le32 i_atime; /* Access time */ 437 __le32 i_ctime; /* Inode Change time */ 438 __le32 i_mtime; /* Modification time */ 439 __le32 i_dtime; /* Deletion Time */ 440 __le16 i_gid; /* Low 16 bits of Group Id */ 441 __le16 i_links_count; /* Links count */ 442 __le32 i_blocks_lo; /* Blocks count */ 443 __le32 i_flags; /* File flags */ 444 union { 445 struct { 446 __le32 l_i_version; 447 } linux1; 448 struct { 449 __u32 h_i_translator; 450 } hurd1; 451 struct { 452 __u32 m_i_reserved1; 453 } masix1; 454 } osd1; /* OS dependent 1 */ 455 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 456 __le32 i_generation; /* File version (for NFS) */ 457 __le32 i_file_acl_lo; /* File ACL */ 458 __le32 i_size_high; 459 __le32 i_obso_faddr; /* Obsoleted fragment address */ 460 union { 461 struct { 462 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 463 __le16 l_i_file_acl_high; 464 __le16 l_i_uid_high; /* these 2 fields */ 465 __le16 l_i_gid_high; /* were reserved2[0] */ 466 __u32 l_i_reserved2; 467 } linux2; 468 struct { 469 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 470 __u16 h_i_mode_high; 471 __u16 h_i_uid_high; 472 __u16 h_i_gid_high; 473 __u32 h_i_author; 474 } hurd2; 475 struct { 476 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 477 __le16 m_i_file_acl_high; 478 __u32 m_i_reserved2[2]; 479 } masix2; 480 } osd2; /* OS dependent 2 */ 481 __le16 i_extra_isize; 482 __le16 i_pad1; 483 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 484 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 485 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 486 __le32 i_crtime; /* File Creation time */ 487 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 488 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 489 }; 490 491 struct move_extent { 492 __u32 reserved; /* should be zero */ 493 __u32 donor_fd; /* donor file descriptor */ 494 __u64 orig_start; /* logical start offset in block for orig */ 495 __u64 donor_start; /* logical start offset in block for donor */ 496 __u64 len; /* block length to be moved */ 497 __u64 moved_len; /* moved block length */ 498 }; 499 500 #define EXT4_EPOCH_BITS 2 501 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 502 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 503 504 /* 505 * Extended fields will fit into an inode if the filesystem was formatted 506 * with large inodes (-I 256 or larger) and there are not currently any EAs 507 * consuming all of the available space. For new inodes we always reserve 508 * enough space for the kernel's known extended fields, but for inodes 509 * created with an old kernel this might not have been the case. None of 510 * the extended inode fields is critical for correct filesystem operation. 511 * This macro checks if a certain field fits in the inode. Note that 512 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 513 */ 514 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 515 ((offsetof(typeof(*ext4_inode), field) + \ 516 sizeof((ext4_inode)->field)) \ 517 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 518 (einode)->i_extra_isize)) \ 519 520 static inline __le32 ext4_encode_extra_time(struct timespec *time) 521 { 522 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 523 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 524 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 525 } 526 527 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 528 { 529 if (sizeof(time->tv_sec) > 4) 530 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 531 << 32; 532 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 533 } 534 535 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 536 do { \ 537 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 538 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 539 (raw_inode)->xtime ## _extra = \ 540 ext4_encode_extra_time(&(inode)->xtime); \ 541 } while (0) 542 543 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 544 do { \ 545 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 546 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 547 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 548 (raw_inode)->xtime ## _extra = \ 549 ext4_encode_extra_time(&(einode)->xtime); \ 550 } while (0) 551 552 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 553 do { \ 554 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 555 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 556 ext4_decode_extra_time(&(inode)->xtime, \ 557 raw_inode->xtime ## _extra); \ 558 } while (0) 559 560 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 561 do { \ 562 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 563 (einode)->xtime.tv_sec = \ 564 (signed)le32_to_cpu((raw_inode)->xtime); \ 565 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 566 ext4_decode_extra_time(&(einode)->xtime, \ 567 raw_inode->xtime ## _extra); \ 568 } while (0) 569 570 #define i_disk_version osd1.linux1.l_i_version 571 572 #if defined(__KERNEL__) || defined(__linux__) 573 #define i_reserved1 osd1.linux1.l_i_reserved1 574 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 575 #define i_blocks_high osd2.linux2.l_i_blocks_high 576 #define i_uid_low i_uid 577 #define i_gid_low i_gid 578 #define i_uid_high osd2.linux2.l_i_uid_high 579 #define i_gid_high osd2.linux2.l_i_gid_high 580 #define i_reserved2 osd2.linux2.l_i_reserved2 581 582 #elif defined(__GNU__) 583 584 #define i_translator osd1.hurd1.h_i_translator 585 #define i_uid_high osd2.hurd2.h_i_uid_high 586 #define i_gid_high osd2.hurd2.h_i_gid_high 587 #define i_author osd2.hurd2.h_i_author 588 589 #elif defined(__masix__) 590 591 #define i_reserved1 osd1.masix1.m_i_reserved1 592 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 593 #define i_reserved2 osd2.masix2.m_i_reserved2 594 595 #endif /* defined(__KERNEL__) || defined(__linux__) */ 596 597 /* 598 * storage for cached extent 599 */ 600 struct ext4_ext_cache { 601 ext4_fsblk_t ec_start; 602 ext4_lblk_t ec_block; 603 __u32 ec_len; /* must be 32bit to return holes */ 604 __u32 ec_type; 605 }; 606 607 /* 608 * fourth extended file system inode data in memory 609 */ 610 struct ext4_inode_info { 611 __le32 i_data[15]; /* unconverted */ 612 __u32 i_flags; 613 ext4_fsblk_t i_file_acl; 614 __u32 i_dtime; 615 616 /* 617 * i_block_group is the number of the block group which contains 618 * this file's inode. Constant across the lifetime of the inode, 619 * it is ued for making block allocation decisions - we try to 620 * place a file's data blocks near its inode block, and new inodes 621 * near to their parent directory's inode. 622 */ 623 ext4_group_t i_block_group; 624 unsigned long i_state_flags; /* Dynamic state flags */ 625 626 ext4_lblk_t i_dir_start_lookup; 627 #ifdef CONFIG_EXT4_FS_XATTR 628 /* 629 * Extended attributes can be read independently of the main file 630 * data. Taking i_mutex even when reading would cause contention 631 * between readers of EAs and writers of regular file data, so 632 * instead we synchronize on xattr_sem when reading or changing 633 * EAs. 634 */ 635 struct rw_semaphore xattr_sem; 636 #endif 637 638 struct list_head i_orphan; /* unlinked but open inodes */ 639 640 /* 641 * i_disksize keeps track of what the inode size is ON DISK, not 642 * in memory. During truncate, i_size is set to the new size by 643 * the VFS prior to calling ext4_truncate(), but the filesystem won't 644 * set i_disksize to 0 until the truncate is actually under way. 645 * 646 * The intent is that i_disksize always represents the blocks which 647 * are used by this file. This allows recovery to restart truncate 648 * on orphans if we crash during truncate. We actually write i_disksize 649 * into the on-disk inode when writing inodes out, instead of i_size. 650 * 651 * The only time when i_disksize and i_size may be different is when 652 * a truncate is in progress. The only things which change i_disksize 653 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 654 */ 655 loff_t i_disksize; 656 657 /* 658 * i_data_sem is for serialising ext4_truncate() against 659 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 660 * data tree are chopped off during truncate. We can't do that in 661 * ext4 because whenever we perform intermediate commits during 662 * truncate, the inode and all the metadata blocks *must* be in a 663 * consistent state which allows truncation of the orphans to restart 664 * during recovery. Hence we must fix the get_block-vs-truncate race 665 * by other means, so we have i_data_sem. 666 */ 667 struct rw_semaphore i_data_sem; 668 struct inode vfs_inode; 669 struct jbd2_inode jinode; 670 671 struct ext4_ext_cache i_cached_extent; 672 /* 673 * File creation time. Its function is same as that of 674 * struct timespec i_{a,c,m}time in the generic inode. 675 */ 676 struct timespec i_crtime; 677 678 /* mballoc */ 679 struct list_head i_prealloc_list; 680 spinlock_t i_prealloc_lock; 681 682 /* ialloc */ 683 ext4_group_t i_last_alloc_group; 684 685 /* allocation reservation info for delalloc */ 686 unsigned int i_reserved_data_blocks; 687 unsigned int i_reserved_meta_blocks; 688 unsigned int i_allocated_meta_blocks; 689 unsigned short i_delalloc_reserved_flag; 690 sector_t i_da_metadata_calc_last_lblock; 691 int i_da_metadata_calc_len; 692 693 /* on-disk additional length */ 694 __u16 i_extra_isize; 695 696 spinlock_t i_block_reservation_lock; 697 #ifdef CONFIG_QUOTA 698 /* quota space reservation, managed internally by quota code */ 699 qsize_t i_reserved_quota; 700 #endif 701 702 /* completed async DIOs that might need unwritten extents handling */ 703 struct list_head i_aio_dio_complete_list; 704 /* current io_end structure for async DIO write*/ 705 ext4_io_end_t *cur_aio_dio; 706 707 /* 708 * Transactions that contain inode's metadata needed to complete 709 * fsync and fdatasync, respectively. 710 */ 711 tid_t i_sync_tid; 712 tid_t i_datasync_tid; 713 }; 714 715 /* 716 * File system states 717 */ 718 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 719 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 720 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 721 722 /* 723 * Misc. filesystem flags 724 */ 725 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 726 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 727 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 728 729 /* 730 * Mount flags 731 */ 732 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */ 733 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 734 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 735 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 736 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 737 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 738 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 739 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 740 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 741 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 742 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 743 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 744 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 745 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 746 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 747 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 748 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 749 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 750 #define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */ 751 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 752 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 753 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 754 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 755 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 756 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */ 757 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 758 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 759 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 760 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 761 762 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt 763 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt 764 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 765 EXT4_MOUNT_##opt) 766 767 #define ext4_set_bit ext2_set_bit 768 #define ext4_set_bit_atomic ext2_set_bit_atomic 769 #define ext4_clear_bit ext2_clear_bit 770 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 771 #define ext4_test_bit ext2_test_bit 772 #define ext4_find_first_zero_bit ext2_find_first_zero_bit 773 #define ext4_find_next_zero_bit ext2_find_next_zero_bit 774 #define ext4_find_next_bit ext2_find_next_bit 775 776 /* 777 * Maximal mount counts between two filesystem checks 778 */ 779 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 780 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 781 782 /* 783 * Behaviour when detecting errors 784 */ 785 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 786 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 787 #define EXT4_ERRORS_PANIC 3 /* Panic */ 788 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 789 790 /* 791 * Structure of the super block 792 */ 793 struct ext4_super_block { 794 /*00*/ __le32 s_inodes_count; /* Inodes count */ 795 __le32 s_blocks_count_lo; /* Blocks count */ 796 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 797 __le32 s_free_blocks_count_lo; /* Free blocks count */ 798 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 799 __le32 s_first_data_block; /* First Data Block */ 800 __le32 s_log_block_size; /* Block size */ 801 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */ 802 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 803 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */ 804 __le32 s_inodes_per_group; /* # Inodes per group */ 805 __le32 s_mtime; /* Mount time */ 806 /*30*/ __le32 s_wtime; /* Write time */ 807 __le16 s_mnt_count; /* Mount count */ 808 __le16 s_max_mnt_count; /* Maximal mount count */ 809 __le16 s_magic; /* Magic signature */ 810 __le16 s_state; /* File system state */ 811 __le16 s_errors; /* Behaviour when detecting errors */ 812 __le16 s_minor_rev_level; /* minor revision level */ 813 /*40*/ __le32 s_lastcheck; /* time of last check */ 814 __le32 s_checkinterval; /* max. time between checks */ 815 __le32 s_creator_os; /* OS */ 816 __le32 s_rev_level; /* Revision level */ 817 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 818 __le16 s_def_resgid; /* Default gid for reserved blocks */ 819 /* 820 * These fields are for EXT4_DYNAMIC_REV superblocks only. 821 * 822 * Note: the difference between the compatible feature set and 823 * the incompatible feature set is that if there is a bit set 824 * in the incompatible feature set that the kernel doesn't 825 * know about, it should refuse to mount the filesystem. 826 * 827 * e2fsck's requirements are more strict; if it doesn't know 828 * about a feature in either the compatible or incompatible 829 * feature set, it must abort and not try to meddle with 830 * things it doesn't understand... 831 */ 832 __le32 s_first_ino; /* First non-reserved inode */ 833 __le16 s_inode_size; /* size of inode structure */ 834 __le16 s_block_group_nr; /* block group # of this superblock */ 835 __le32 s_feature_compat; /* compatible feature set */ 836 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 837 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 838 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 839 /*78*/ char s_volume_name[16]; /* volume name */ 840 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 841 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 842 /* 843 * Performance hints. Directory preallocation should only 844 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 845 */ 846 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 847 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 848 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 849 /* 850 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 851 */ 852 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 853 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 854 __le32 s_journal_dev; /* device number of journal file */ 855 __le32 s_last_orphan; /* start of list of inodes to delete */ 856 __le32 s_hash_seed[4]; /* HTREE hash seed */ 857 __u8 s_def_hash_version; /* Default hash version to use */ 858 __u8 s_reserved_char_pad; 859 __le16 s_desc_size; /* size of group descriptor */ 860 /*100*/ __le32 s_default_mount_opts; 861 __le32 s_first_meta_bg; /* First metablock block group */ 862 __le32 s_mkfs_time; /* When the filesystem was created */ 863 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 864 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 865 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 866 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 867 __le32 s_free_blocks_count_hi; /* Free blocks count */ 868 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 869 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 870 __le32 s_flags; /* Miscellaneous flags */ 871 __le16 s_raid_stride; /* RAID stride */ 872 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */ 873 __le64 s_mmp_block; /* Block for multi-mount protection */ 874 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 875 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 876 __u8 s_reserved_char_pad2; 877 __le16 s_reserved_pad; 878 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 879 __u32 s_reserved[160]; /* Padding to the end of the block */ 880 }; 881 882 #ifdef __KERNEL__ 883 884 /* 885 * run-time mount flags 886 */ 887 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 888 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 889 890 /* 891 * fourth extended-fs super-block data in memory 892 */ 893 struct ext4_sb_info { 894 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 895 unsigned long s_inodes_per_block;/* Number of inodes per block */ 896 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 897 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 898 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 899 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 900 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 901 ext4_group_t s_groups_count; /* Number of groups in the fs */ 902 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 903 unsigned long s_overhead_last; /* Last calculated overhead */ 904 unsigned long s_blocks_last; /* Last seen block count */ 905 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 906 struct buffer_head * s_sbh; /* Buffer containing the super block */ 907 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 908 struct buffer_head **s_group_desc; 909 unsigned int s_mount_opt; 910 unsigned int s_mount_flags; 911 ext4_fsblk_t s_sb_block; 912 uid_t s_resuid; 913 gid_t s_resgid; 914 unsigned short s_mount_state; 915 unsigned short s_pad; 916 int s_addr_per_block_bits; 917 int s_desc_per_block_bits; 918 int s_inode_size; 919 int s_first_ino; 920 unsigned int s_inode_readahead_blks; 921 unsigned int s_inode_goal; 922 spinlock_t s_next_gen_lock; 923 u32 s_next_generation; 924 u32 s_hash_seed[4]; 925 int s_def_hash_version; 926 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 927 struct percpu_counter s_freeblocks_counter; 928 struct percpu_counter s_freeinodes_counter; 929 struct percpu_counter s_dirs_counter; 930 struct percpu_counter s_dirtyblocks_counter; 931 struct blockgroup_lock *s_blockgroup_lock; 932 struct proc_dir_entry *s_proc; 933 struct kobject s_kobj; 934 struct completion s_kobj_unregister; 935 936 /* Journaling */ 937 struct inode *s_journal_inode; 938 struct journal_s *s_journal; 939 struct list_head s_orphan; 940 struct mutex s_orphan_lock; 941 struct mutex s_resize_lock; 942 unsigned long s_commit_interval; 943 u32 s_max_batch_time; 944 u32 s_min_batch_time; 945 struct block_device *journal_bdev; 946 #ifdef CONFIG_JBD2_DEBUG 947 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */ 948 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */ 949 #endif 950 #ifdef CONFIG_QUOTA 951 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 952 int s_jquota_fmt; /* Format of quota to use */ 953 #endif 954 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 955 struct rb_root system_blks; 956 957 #ifdef EXTENTS_STATS 958 /* ext4 extents stats */ 959 unsigned long s_ext_min; 960 unsigned long s_ext_max; 961 unsigned long s_depth_max; 962 spinlock_t s_ext_stats_lock; 963 unsigned long s_ext_blocks; 964 unsigned long s_ext_extents; 965 #endif 966 967 /* for buddy allocator */ 968 struct ext4_group_info ***s_group_info; 969 struct inode *s_buddy_cache; 970 long s_blocks_reserved; 971 spinlock_t s_reserve_lock; 972 spinlock_t s_md_lock; 973 tid_t s_last_transaction; 974 unsigned short *s_mb_offsets; 975 unsigned int *s_mb_maxs; 976 977 /* tunables */ 978 unsigned long s_stripe; 979 unsigned int s_mb_stream_request; 980 unsigned int s_mb_max_to_scan; 981 unsigned int s_mb_min_to_scan; 982 unsigned int s_mb_stats; 983 unsigned int s_mb_order2_reqs; 984 unsigned int s_mb_group_prealloc; 985 unsigned int s_max_writeback_mb_bump; 986 /* where last allocation was done - for stream allocation */ 987 unsigned long s_mb_last_group; 988 unsigned long s_mb_last_start; 989 990 /* stats for buddy allocator */ 991 spinlock_t s_mb_pa_lock; 992 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 993 atomic_t s_bal_success; /* we found long enough chunks */ 994 atomic_t s_bal_allocated; /* in blocks */ 995 atomic_t s_bal_ex_scanned; /* total extents scanned */ 996 atomic_t s_bal_goals; /* goal hits */ 997 atomic_t s_bal_breaks; /* too long searches */ 998 atomic_t s_bal_2orders; /* 2^order hits */ 999 spinlock_t s_bal_lock; 1000 unsigned long s_mb_buddies_generated; 1001 unsigned long long s_mb_generation_time; 1002 atomic_t s_mb_lost_chunks; 1003 atomic_t s_mb_preallocated; 1004 atomic_t s_mb_discarded; 1005 atomic_t s_lock_busy; 1006 1007 /* locality groups */ 1008 struct ext4_locality_group *s_locality_groups; 1009 1010 /* for write statistics */ 1011 unsigned long s_sectors_written_start; 1012 u64 s_kbytes_written; 1013 1014 unsigned int s_log_groups_per_flex; 1015 struct flex_groups *s_flex_groups; 1016 1017 /* workqueue for dio unwritten */ 1018 struct workqueue_struct *dio_unwritten_wq; 1019 }; 1020 1021 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1022 { 1023 return sb->s_fs_info; 1024 } 1025 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1026 { 1027 return container_of(inode, struct ext4_inode_info, vfs_inode); 1028 } 1029 1030 static inline struct timespec ext4_current_time(struct inode *inode) 1031 { 1032 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1033 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1034 } 1035 1036 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1037 { 1038 return ino == EXT4_ROOT_INO || 1039 ino == EXT4_JOURNAL_INO || 1040 ino == EXT4_RESIZE_INO || 1041 (ino >= EXT4_FIRST_INO(sb) && 1042 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1043 } 1044 1045 /* 1046 * Inode dynamic state flags 1047 */ 1048 enum { 1049 EXT4_STATE_JDATA, /* journaled data exists */ 1050 EXT4_STATE_NEW, /* inode is newly created */ 1051 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1052 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1053 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1054 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1055 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1056 }; 1057 1058 static inline int ext4_test_inode_state(struct inode *inode, int bit) 1059 { 1060 return test_bit(bit, &EXT4_I(inode)->i_state_flags); 1061 } 1062 1063 static inline void ext4_set_inode_state(struct inode *inode, int bit) 1064 { 1065 set_bit(bit, &EXT4_I(inode)->i_state_flags); 1066 } 1067 1068 static inline void ext4_clear_inode_state(struct inode *inode, int bit) 1069 { 1070 clear_bit(bit, &EXT4_I(inode)->i_state_flags); 1071 } 1072 #else 1073 /* Assume that user mode programs are passing in an ext4fs superblock, not 1074 * a kernel struct super_block. This will allow us to call the feature-test 1075 * macros from user land. */ 1076 #define EXT4_SB(sb) (sb) 1077 #endif 1078 1079 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1080 1081 /* 1082 * Codes for operating systems 1083 */ 1084 #define EXT4_OS_LINUX 0 1085 #define EXT4_OS_HURD 1 1086 #define EXT4_OS_MASIX 2 1087 #define EXT4_OS_FREEBSD 3 1088 #define EXT4_OS_LITES 4 1089 1090 /* 1091 * Revision levels 1092 */ 1093 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1094 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1095 1096 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1097 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1098 1099 #define EXT4_GOOD_OLD_INODE_SIZE 128 1100 1101 /* 1102 * Feature set definitions 1103 */ 1104 1105 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1106 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1107 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1108 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1109 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1110 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1111 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1112 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1113 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1114 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1115 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1116 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1117 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1118 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1119 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1120 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1121 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1122 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1123 1124 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1125 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1126 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1127 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1128 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1129 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1130 1131 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1132 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1133 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1134 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1135 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1136 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1137 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1138 1139 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1140 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1141 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1142 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1143 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1144 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1145 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1146 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1147 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1148 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1149 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1150 1151 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1152 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1153 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1154 EXT4_FEATURE_INCOMPAT_META_BG| \ 1155 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1156 EXT4_FEATURE_INCOMPAT_64BIT| \ 1157 EXT4_FEATURE_INCOMPAT_FLEX_BG) 1158 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1159 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1160 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1161 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1162 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1163 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1164 EXT4_FEATURE_RO_COMPAT_HUGE_FILE) 1165 1166 /* 1167 * Default values for user and/or group using reserved blocks 1168 */ 1169 #define EXT4_DEF_RESUID 0 1170 #define EXT4_DEF_RESGID 0 1171 1172 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1173 1174 /* 1175 * Default mount options 1176 */ 1177 #define EXT4_DEFM_DEBUG 0x0001 1178 #define EXT4_DEFM_BSDGROUPS 0x0002 1179 #define EXT4_DEFM_XATTR_USER 0x0004 1180 #define EXT4_DEFM_ACL 0x0008 1181 #define EXT4_DEFM_UID16 0x0010 1182 #define EXT4_DEFM_JMODE 0x0060 1183 #define EXT4_DEFM_JMODE_DATA 0x0020 1184 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1185 #define EXT4_DEFM_JMODE_WBACK 0x0060 1186 1187 /* 1188 * Default journal batch times 1189 */ 1190 #define EXT4_DEF_MIN_BATCH_TIME 0 1191 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1192 1193 /* 1194 * Minimum number of groups in a flexgroup before we separate out 1195 * directories into the first block group of a flexgroup 1196 */ 1197 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1198 1199 /* 1200 * Structure of a directory entry 1201 */ 1202 #define EXT4_NAME_LEN 255 1203 1204 struct ext4_dir_entry { 1205 __le32 inode; /* Inode number */ 1206 __le16 rec_len; /* Directory entry length */ 1207 __le16 name_len; /* Name length */ 1208 char name[EXT4_NAME_LEN]; /* File name */ 1209 }; 1210 1211 /* 1212 * The new version of the directory entry. Since EXT4 structures are 1213 * stored in intel byte order, and the name_len field could never be 1214 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1215 * file_type field. 1216 */ 1217 struct ext4_dir_entry_2 { 1218 __le32 inode; /* Inode number */ 1219 __le16 rec_len; /* Directory entry length */ 1220 __u8 name_len; /* Name length */ 1221 __u8 file_type; 1222 char name[EXT4_NAME_LEN]; /* File name */ 1223 }; 1224 1225 /* 1226 * Ext4 directory file types. Only the low 3 bits are used. The 1227 * other bits are reserved for now. 1228 */ 1229 #define EXT4_FT_UNKNOWN 0 1230 #define EXT4_FT_REG_FILE 1 1231 #define EXT4_FT_DIR 2 1232 #define EXT4_FT_CHRDEV 3 1233 #define EXT4_FT_BLKDEV 4 1234 #define EXT4_FT_FIFO 5 1235 #define EXT4_FT_SOCK 6 1236 #define EXT4_FT_SYMLINK 7 1237 1238 #define EXT4_FT_MAX 8 1239 1240 /* 1241 * EXT4_DIR_PAD defines the directory entries boundaries 1242 * 1243 * NOTE: It must be a multiple of 4 1244 */ 1245 #define EXT4_DIR_PAD 4 1246 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1247 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1248 ~EXT4_DIR_ROUND) 1249 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1250 1251 /* 1252 * Hash Tree Directory indexing 1253 * (c) Daniel Phillips, 2001 1254 */ 1255 1256 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1257 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1258 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL)) 1259 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1260 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1261 1262 /* Legal values for the dx_root hash_version field: */ 1263 1264 #define DX_HASH_LEGACY 0 1265 #define DX_HASH_HALF_MD4 1 1266 #define DX_HASH_TEA 2 1267 #define DX_HASH_LEGACY_UNSIGNED 3 1268 #define DX_HASH_HALF_MD4_UNSIGNED 4 1269 #define DX_HASH_TEA_UNSIGNED 5 1270 1271 #ifdef __KERNEL__ 1272 1273 /* hash info structure used by the directory hash */ 1274 struct dx_hash_info 1275 { 1276 u32 hash; 1277 u32 minor_hash; 1278 int hash_version; 1279 u32 *seed; 1280 }; 1281 1282 #define EXT4_HTREE_EOF 0x7fffffff 1283 1284 /* 1285 * Control parameters used by ext4_htree_next_block 1286 */ 1287 #define HASH_NB_ALWAYS 1 1288 1289 1290 /* 1291 * Describe an inode's exact location on disk and in memory 1292 */ 1293 struct ext4_iloc 1294 { 1295 struct buffer_head *bh; 1296 unsigned long offset; 1297 ext4_group_t block_group; 1298 }; 1299 1300 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1301 { 1302 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1303 } 1304 1305 /* 1306 * This structure is stuffed into the struct file's private_data field 1307 * for directories. It is where we put information so that we can do 1308 * readdir operations in hash tree order. 1309 */ 1310 struct dir_private_info { 1311 struct rb_root root; 1312 struct rb_node *curr_node; 1313 struct fname *extra_fname; 1314 loff_t last_pos; 1315 __u32 curr_hash; 1316 __u32 curr_minor_hash; 1317 __u32 next_hash; 1318 }; 1319 1320 /* calculate the first block number of the group */ 1321 static inline ext4_fsblk_t 1322 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1323 { 1324 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1325 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1326 } 1327 1328 /* 1329 * Special error return code only used by dx_probe() and its callers. 1330 */ 1331 #define ERR_BAD_DX_DIR -75000 1332 1333 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr, 1334 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp); 1335 1336 extern struct proc_dir_entry *ext4_proc_root; 1337 1338 /* 1339 * Function prototypes 1340 */ 1341 1342 /* 1343 * Ok, these declarations are also in <linux/kernel.h> but none of the 1344 * ext4 source programs needs to include it so they are duplicated here. 1345 */ 1346 # define NORET_TYPE /**/ 1347 # define ATTRIB_NORET __attribute__((noreturn)) 1348 # define NORET_AND noreturn, 1349 1350 /* bitmap.c */ 1351 extern unsigned int ext4_count_free(struct buffer_head *, unsigned); 1352 1353 /* balloc.c */ 1354 extern unsigned int ext4_block_group(struct super_block *sb, 1355 ext4_fsblk_t blocknr); 1356 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1357 ext4_fsblk_t blocknr); 1358 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1359 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1360 ext4_group_t group); 1361 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1362 ext4_fsblk_t goal, unsigned long *count, int *errp); 1363 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1364 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1365 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb, 1366 ext4_fsblk_t block, unsigned long count); 1367 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *); 1368 extern void ext4_check_blocks_bitmap(struct super_block *); 1369 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1370 ext4_group_t block_group, 1371 struct buffer_head ** bh); 1372 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1373 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1374 ext4_group_t block_group); 1375 extern unsigned ext4_init_block_bitmap(struct super_block *sb, 1376 struct buffer_head *bh, 1377 ext4_group_t group, 1378 struct ext4_group_desc *desc); 1379 #define ext4_free_blocks_after_init(sb, group, desc) \ 1380 ext4_init_block_bitmap(sb, NULL, group, desc) 1381 1382 /* dir.c */ 1383 extern int ext4_check_dir_entry(const char *, struct inode *, 1384 struct ext4_dir_entry_2 *, 1385 struct buffer_head *, unsigned int); 1386 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1387 __u32 minor_hash, 1388 struct ext4_dir_entry_2 *dirent); 1389 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1390 1391 /* fsync.c */ 1392 extern int ext4_sync_file(struct file *, struct dentry *, int); 1393 1394 /* hash.c */ 1395 extern int ext4fs_dirhash(const char *name, int len, struct 1396 dx_hash_info *hinfo); 1397 1398 /* ialloc.c */ 1399 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int, 1400 const struct qstr *qstr, __u32 goal); 1401 extern void ext4_free_inode(handle_t *, struct inode *); 1402 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 1403 extern unsigned long ext4_count_free_inodes(struct super_block *); 1404 extern unsigned long ext4_count_dirs(struct super_block *); 1405 extern void ext4_check_inodes_bitmap(struct super_block *); 1406 extern unsigned ext4_init_inode_bitmap(struct super_block *sb, 1407 struct buffer_head *bh, 1408 ext4_group_t group, 1409 struct ext4_group_desc *desc); 1410 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 1411 1412 /* mballoc.c */ 1413 extern long ext4_mb_stats; 1414 extern long ext4_mb_max_to_scan; 1415 extern int ext4_mb_init(struct super_block *, int); 1416 extern int ext4_mb_release(struct super_block *); 1417 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 1418 struct ext4_allocation_request *, int *); 1419 extern int ext4_mb_reserve_blocks(struct super_block *, int); 1420 extern void ext4_discard_preallocations(struct inode *); 1421 extern int __init init_ext4_mballoc(void); 1422 extern void exit_ext4_mballoc(void); 1423 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 1424 struct buffer_head *bh, ext4_fsblk_t block, 1425 unsigned long count, int flags); 1426 extern int ext4_mb_add_groupinfo(struct super_block *sb, 1427 ext4_group_t i, struct ext4_group_desc *desc); 1428 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t); 1429 extern void ext4_mb_put_buddy_cache_lock(struct super_block *, 1430 ext4_group_t, int); 1431 /* inode.c */ 1432 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 1433 ext4_lblk_t, int, int *); 1434 struct buffer_head *ext4_bread(handle_t *, struct inode *, 1435 ext4_lblk_t, int, int *); 1436 int ext4_get_block(struct inode *inode, sector_t iblock, 1437 struct buffer_head *bh_result, int create); 1438 1439 extern struct inode *ext4_iget(struct super_block *, unsigned long); 1440 extern int ext4_write_inode(struct inode *, int); 1441 extern int ext4_setattr(struct dentry *, struct iattr *); 1442 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 1443 struct kstat *stat); 1444 extern void ext4_delete_inode(struct inode *); 1445 extern int ext4_sync_inode(handle_t *, struct inode *); 1446 extern void ext4_dirty_inode(struct inode *); 1447 extern int ext4_change_inode_journal_flag(struct inode *, int); 1448 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 1449 extern int ext4_can_truncate(struct inode *inode); 1450 extern void ext4_truncate(struct inode *); 1451 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 1452 extern void ext4_set_inode_flags(struct inode *); 1453 extern void ext4_get_inode_flags(struct ext4_inode_info *); 1454 extern int ext4_alloc_da_blocks(struct inode *inode); 1455 extern void ext4_set_aops(struct inode *inode); 1456 extern int ext4_writepage_trans_blocks(struct inode *); 1457 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks); 1458 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 1459 extern int ext4_block_truncate_page(handle_t *handle, 1460 struct address_space *mapping, loff_t from); 1461 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 1462 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 1463 extern int flush_aio_dio_completed_IO(struct inode *inode); 1464 extern void ext4_da_update_reserve_space(struct inode *inode, 1465 int used, int quota_claim); 1466 /* ioctl.c */ 1467 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 1468 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 1469 1470 /* migrate.c */ 1471 extern int ext4_ext_migrate(struct inode *); 1472 1473 /* namei.c */ 1474 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize); 1475 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize); 1476 extern int ext4_orphan_add(handle_t *, struct inode *); 1477 extern int ext4_orphan_del(handle_t *, struct inode *); 1478 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 1479 __u32 start_minor_hash, __u32 *next_hash); 1480 1481 /* resize.c */ 1482 extern int ext4_group_add(struct super_block *sb, 1483 struct ext4_new_group_data *input); 1484 extern int ext4_group_extend(struct super_block *sb, 1485 struct ext4_super_block *es, 1486 ext4_fsblk_t n_blocks_count); 1487 1488 /* super.c */ 1489 extern void ext4_error(struct super_block *, const char *, const char *, ...) 1490 __attribute__ ((format (printf, 3, 4))); 1491 extern void __ext4_std_error(struct super_block *, const char *, int); 1492 extern void ext4_abort(struct super_block *, const char *, const char *, ...) 1493 __attribute__ ((format (printf, 3, 4))); 1494 extern void ext4_warning(struct super_block *, const char *, const char *, ...) 1495 __attribute__ ((format (printf, 3, 4))); 1496 extern void ext4_msg(struct super_block *, const char *, const char *, ...) 1497 __attribute__ ((format (printf, 3, 4))); 1498 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t, 1499 const char *, const char *, ...) 1500 __attribute__ ((format (printf, 4, 5))); 1501 extern void ext4_update_dynamic_rev(struct super_block *sb); 1502 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 1503 __u32 compat); 1504 extern int ext4_update_rocompat_feature(handle_t *handle, 1505 struct super_block *sb, __u32 rocompat); 1506 extern int ext4_update_incompat_feature(handle_t *handle, 1507 struct super_block *sb, __u32 incompat); 1508 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 1509 struct ext4_group_desc *bg); 1510 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 1511 struct ext4_group_desc *bg); 1512 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 1513 struct ext4_group_desc *bg); 1514 extern __u32 ext4_free_blks_count(struct super_block *sb, 1515 struct ext4_group_desc *bg); 1516 extern __u32 ext4_free_inodes_count(struct super_block *sb, 1517 struct ext4_group_desc *bg); 1518 extern __u32 ext4_used_dirs_count(struct super_block *sb, 1519 struct ext4_group_desc *bg); 1520 extern __u32 ext4_itable_unused_count(struct super_block *sb, 1521 struct ext4_group_desc *bg); 1522 extern void ext4_block_bitmap_set(struct super_block *sb, 1523 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1524 extern void ext4_inode_bitmap_set(struct super_block *sb, 1525 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1526 extern void ext4_inode_table_set(struct super_block *sb, 1527 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1528 extern void ext4_free_blks_set(struct super_block *sb, 1529 struct ext4_group_desc *bg, __u32 count); 1530 extern void ext4_free_inodes_set(struct super_block *sb, 1531 struct ext4_group_desc *bg, __u32 count); 1532 extern void ext4_used_dirs_set(struct super_block *sb, 1533 struct ext4_group_desc *bg, __u32 count); 1534 extern void ext4_itable_unused_set(struct super_block *sb, 1535 struct ext4_group_desc *bg, __u32 count); 1536 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group, 1537 struct ext4_group_desc *gdp); 1538 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group, 1539 struct ext4_group_desc *gdp); 1540 1541 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 1542 { 1543 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 1544 le32_to_cpu(es->s_blocks_count_lo); 1545 } 1546 1547 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 1548 { 1549 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 1550 le32_to_cpu(es->s_r_blocks_count_lo); 1551 } 1552 1553 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 1554 { 1555 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 1556 le32_to_cpu(es->s_free_blocks_count_lo); 1557 } 1558 1559 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 1560 ext4_fsblk_t blk) 1561 { 1562 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 1563 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 1564 } 1565 1566 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 1567 ext4_fsblk_t blk) 1568 { 1569 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 1570 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 1571 } 1572 1573 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 1574 ext4_fsblk_t blk) 1575 { 1576 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 1577 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 1578 } 1579 1580 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 1581 { 1582 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 1583 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 1584 le32_to_cpu(raw_inode->i_size_lo); 1585 else 1586 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 1587 } 1588 1589 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 1590 { 1591 raw_inode->i_size_lo = cpu_to_le32(i_size); 1592 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 1593 } 1594 1595 static inline 1596 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 1597 ext4_group_t group) 1598 { 1599 struct ext4_group_info ***grp_info; 1600 long indexv, indexh; 1601 grp_info = EXT4_SB(sb)->s_group_info; 1602 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 1603 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 1604 return grp_info[indexv][indexh]; 1605 } 1606 1607 /* 1608 * Reading s_groups_count requires using smp_rmb() afterwards. See 1609 * the locking protocol documented in the comments of ext4_group_add() 1610 * in resize.c 1611 */ 1612 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 1613 { 1614 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 1615 1616 smp_rmb(); 1617 return ngroups; 1618 } 1619 1620 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 1621 ext4_group_t block_group) 1622 { 1623 return block_group >> sbi->s_log_groups_per_flex; 1624 } 1625 1626 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 1627 { 1628 return 1 << sbi->s_log_groups_per_flex; 1629 } 1630 1631 #define ext4_std_error(sb, errno) \ 1632 do { \ 1633 if ((errno)) \ 1634 __ext4_std_error((sb), __func__, (errno)); \ 1635 } while (0) 1636 1637 #ifdef CONFIG_SMP 1638 /* Each CPU can accumulate percpu_counter_batch blocks in their local 1639 * counters. So we need to make sure we have free blocks more 1640 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 1641 */ 1642 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 1643 #else 1644 #define EXT4_FREEBLOCKS_WATERMARK 0 1645 #endif 1646 1647 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 1648 { 1649 /* 1650 * XXX: replace with spinlock if seen contended -bzzz 1651 */ 1652 down_write(&EXT4_I(inode)->i_data_sem); 1653 if (newsize > EXT4_I(inode)->i_disksize) 1654 EXT4_I(inode)->i_disksize = newsize; 1655 up_write(&EXT4_I(inode)->i_data_sem); 1656 return ; 1657 } 1658 1659 struct ext4_group_info { 1660 unsigned long bb_state; 1661 struct rb_root bb_free_root; 1662 ext4_grpblk_t bb_first_free; /* first free block */ 1663 ext4_grpblk_t bb_free; /* total free blocks */ 1664 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 1665 struct list_head bb_prealloc_list; 1666 #ifdef DOUBLE_CHECK 1667 void *bb_bitmap; 1668 #endif 1669 struct rw_semaphore alloc_sem; 1670 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 1671 * regions, index is order. 1672 * bb_counters[3] = 5 means 1673 * 5 free 8-block regions. */ 1674 }; 1675 1676 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 1677 1678 #define EXT4_MB_GRP_NEED_INIT(grp) \ 1679 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 1680 1681 #define EXT4_MAX_CONTENTION 8 1682 #define EXT4_CONTENTION_THRESHOLD 2 1683 1684 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 1685 ext4_group_t group) 1686 { 1687 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 1688 } 1689 1690 /* 1691 * Returns true if the filesystem is busy enough that attempts to 1692 * access the block group locks has run into contention. 1693 */ 1694 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 1695 { 1696 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 1697 } 1698 1699 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 1700 { 1701 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 1702 if (spin_trylock(lock)) 1703 /* 1704 * We're able to grab the lock right away, so drop the 1705 * lock contention counter. 1706 */ 1707 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 1708 else { 1709 /* 1710 * The lock is busy, so bump the contention counter, 1711 * and then wait on the spin lock. 1712 */ 1713 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 1714 EXT4_MAX_CONTENTION); 1715 spin_lock(lock); 1716 } 1717 } 1718 1719 static inline void ext4_unlock_group(struct super_block *sb, 1720 ext4_group_t group) 1721 { 1722 spin_unlock(ext4_group_lock_ptr(sb, group)); 1723 } 1724 1725 /* 1726 * Inodes and files operations 1727 */ 1728 1729 /* dir.c */ 1730 extern const struct file_operations ext4_dir_operations; 1731 1732 /* file.c */ 1733 extern const struct inode_operations ext4_file_inode_operations; 1734 extern const struct file_operations ext4_file_operations; 1735 1736 /* namei.c */ 1737 extern const struct inode_operations ext4_dir_inode_operations; 1738 extern const struct inode_operations ext4_special_inode_operations; 1739 extern struct dentry *ext4_get_parent(struct dentry *child); 1740 1741 /* symlink.c */ 1742 extern const struct inode_operations ext4_symlink_inode_operations; 1743 extern const struct inode_operations ext4_fast_symlink_inode_operations; 1744 1745 /* block_validity */ 1746 extern void ext4_release_system_zone(struct super_block *sb); 1747 extern int ext4_setup_system_zone(struct super_block *sb); 1748 extern int __init init_ext4_system_zone(void); 1749 extern void exit_ext4_system_zone(void); 1750 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 1751 ext4_fsblk_t start_blk, 1752 unsigned int count); 1753 1754 /* extents.c */ 1755 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 1756 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 1757 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 1758 int chunk); 1759 extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode, 1760 ext4_lblk_t iblock, unsigned int max_blocks, 1761 struct buffer_head *bh_result, int flags); 1762 extern void ext4_ext_truncate(struct inode *); 1763 extern void ext4_ext_init(struct super_block *); 1764 extern void ext4_ext_release(struct super_block *); 1765 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset, 1766 loff_t len); 1767 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 1768 ssize_t len); 1769 extern int ext4_get_blocks(handle_t *handle, struct inode *inode, 1770 sector_t block, unsigned int max_blocks, 1771 struct buffer_head *bh, int flags); 1772 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 1773 __u64 start, __u64 len); 1774 /* move_extent.c */ 1775 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 1776 __u64 start_orig, __u64 start_donor, 1777 __u64 len, __u64 *moved_len); 1778 1779 1780 /* 1781 * Add new method to test wether block and inode bitmaps are properly 1782 * initialized. With uninit_bg reading the block from disk is not enough 1783 * to mark the bitmap uptodate. We need to also zero-out the bitmap 1784 */ 1785 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 1786 1787 static inline int bitmap_uptodate(struct buffer_head *bh) 1788 { 1789 return (buffer_uptodate(bh) && 1790 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 1791 } 1792 static inline void set_bitmap_uptodate(struct buffer_head *bh) 1793 { 1794 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 1795 } 1796 1797 #endif /* __KERNEL__ */ 1798 1799 #endif /* _EXT4_H */ 1800