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