linux/fs/ext4/ext4.h
<<
>>
Prefs
   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#include <crypto/hash.h>
  33#ifdef __KERNEL__
  34#include <linux/compat.h>
  35#endif
  36
  37/*
  38 * The fourth extended filesystem constants/structures
  39 */
  40
  41/*
  42 * Define EXT4FS_DEBUG to produce debug messages
  43 */
  44#undef EXT4FS_DEBUG
  45
  46/*
  47 * Debug code
  48 */
  49#ifdef EXT4FS_DEBUG
  50#define ext4_debug(f, a...)                                             \
  51        do {                                                            \
  52                printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
  53                        __FILE__, __LINE__, __func__);                  \
  54                printk(KERN_DEBUG f, ## a);                             \
  55        } while (0)
  56#else
  57#define ext4_debug(fmt, ...)    no_printk(fmt, ##__VA_ARGS__)
  58#endif
  59
  60/*
  61 * Turn on EXT_DEBUG to get lots of info about extents operations.
  62 */
  63#define EXT_DEBUG__
  64#ifdef EXT_DEBUG
  65#define ext_debug(fmt, ...)     printk(fmt, ##__VA_ARGS__)
  66#else
  67#define ext_debug(fmt, ...)     no_printk(fmt, ##__VA_ARGS__)
  68#endif
  69
  70#define EXT4_ERROR_INODE(inode, fmt, a...) \
  71        ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
  72
  73#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)                 \
  74        ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
  75
  76#define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
  77        ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
  78
  79/* data type for block offset of block group */
  80typedef int ext4_grpblk_t;
  81
  82/* data type for filesystem-wide blocks number */
  83typedef unsigned long long ext4_fsblk_t;
  84
  85/* data type for file logical block number */
  86typedef __u32 ext4_lblk_t;
  87
  88/* data type for block group number */
  89typedef unsigned int ext4_group_t;
  90
  91/*
  92 * Flags used in mballoc's allocation_context flags field.
  93 *
  94 * Also used to show what's going on for debugging purposes when the
  95 * flag field is exported via the traceport interface
  96 */
  97
  98/* prefer goal again. length */
  99#define EXT4_MB_HINT_MERGE              0x0001
 100/* blocks already reserved */
 101#define EXT4_MB_HINT_RESERVED           0x0002
 102/* metadata is being allocated */
 103#define EXT4_MB_HINT_METADATA           0x0004
 104/* first blocks in the file */
 105#define EXT4_MB_HINT_FIRST              0x0008
 106/* search for the best chunk */
 107#define EXT4_MB_HINT_BEST               0x0010
 108/* data is being allocated */
 109#define EXT4_MB_HINT_DATA               0x0020
 110/* don't preallocate (for tails) */
 111#define EXT4_MB_HINT_NOPREALLOC         0x0040
 112/* allocate for locality group */
 113#define EXT4_MB_HINT_GROUP_ALLOC        0x0080
 114/* allocate goal blocks or none */
 115#define EXT4_MB_HINT_GOAL_ONLY          0x0100
 116/* goal is meaningful */
 117#define EXT4_MB_HINT_TRY_GOAL           0x0200
 118/* blocks already pre-reserved by delayed allocation */
 119#define EXT4_MB_DELALLOC_RESERVED       0x0400
 120/* We are doing stream allocation */
 121#define EXT4_MB_STREAM_ALLOC            0x0800
 122/* Use reserved root blocks if needed */
 123#define EXT4_MB_USE_ROOT_BLOCKS         0x1000
 124
 125struct ext4_allocation_request {
 126        /* target inode for block we're allocating */
 127        struct inode *inode;
 128        /* how many blocks we want to allocate */
 129        unsigned int len;
 130        /* logical block in target inode */
 131        ext4_lblk_t logical;
 132        /* the closest logical allocated block to the left */
 133        ext4_lblk_t lleft;
 134        /* the closest logical allocated block to the right */
 135        ext4_lblk_t lright;
 136        /* phys. target (a hint) */
 137        ext4_fsblk_t goal;
 138        /* phys. block for the closest logical allocated block to the left */
 139        ext4_fsblk_t pleft;
 140        /* phys. block for the closest logical allocated block to the right */
 141        ext4_fsblk_t pright;
 142        /* flags. see above EXT4_MB_HINT_* */
 143        unsigned int flags;
 144};
 145
 146/*
 147 * Logical to physical block mapping, used by ext4_map_blocks()
 148 *
 149 * This structure is used to pass requests into ext4_map_blocks() as
 150 * well as to store the information returned by ext4_map_blocks().  It
 151 * takes less room on the stack than a struct buffer_head.
 152 */
 153#define EXT4_MAP_NEW            (1 << BH_New)
 154#define EXT4_MAP_MAPPED         (1 << BH_Mapped)
 155#define EXT4_MAP_UNWRITTEN      (1 << BH_Unwritten)
 156#define EXT4_MAP_BOUNDARY       (1 << BH_Boundary)
 157#define EXT4_MAP_UNINIT         (1 << BH_Uninit)
 158/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
 159 * ext4_map_blocks wants to know whether or not the underlying cluster has
 160 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
 161 * the requested mapping was from previously mapped (or delayed allocated)
 162 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
 163 * should never appear on buffer_head's state flags.
 164 */
 165#define EXT4_MAP_FROM_CLUSTER   (1 << BH_AllocFromCluster)
 166#define EXT4_MAP_FLAGS          (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
 167                                 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
 168                                 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
 169
 170struct ext4_map_blocks {
 171        ext4_fsblk_t m_pblk;
 172        ext4_lblk_t m_lblk;
 173        unsigned int m_len;
 174        unsigned int m_flags;
 175};
 176
 177/*
 178 * For delayed allocation tracking
 179 */
 180struct mpage_da_data {
 181        struct inode *inode;
 182        sector_t b_blocknr;             /* start block number of extent */
 183        size_t b_size;                  /* size of extent */
 184        unsigned long b_state;          /* state of the extent */
 185        unsigned long first_page, next_page;    /* extent of pages */
 186        struct writeback_control *wbc;
 187        int io_done;
 188        int pages_written;
 189        int retval;
 190};
 191
 192/*
 193 * Flags for ext4_io_end->flags
 194 */
 195#define EXT4_IO_END_UNWRITTEN   0x0001
 196#define EXT4_IO_END_ERROR       0x0002
 197#define EXT4_IO_END_QUEUED      0x0004
 198#define EXT4_IO_END_DIRECT      0x0008
 199
 200struct ext4_io_page {
 201        struct page     *p_page;
 202        atomic_t        p_count;
 203};
 204
 205#define MAX_IO_PAGES 128
 206
 207/*
 208 * For converting uninitialized extents on a work queue.
 209 *
 210 * 'page' is only used from the writepage() path; 'pages' is only used for
 211 * buffered writes; they are used to keep page references until conversion
 212 * takes place.  For AIO/DIO, neither field is filled in.
 213 */
 214typedef struct ext4_io_end {
 215        struct list_head        list;           /* per-file finished IO list */
 216        struct inode            *inode;         /* file being written to */
 217        unsigned int            flag;           /* unwritten or not */
 218        struct page             *page;          /* for writepage() path */
 219        loff_t                  offset;         /* offset in the file */
 220        ssize_t                 size;           /* size of the extent */
 221        struct work_struct      work;           /* data work queue */
 222        struct kiocb            *iocb;          /* iocb struct for AIO */
 223        int                     result;         /* error value for AIO */
 224        int                     num_io_pages;   /* for writepages() */
 225        struct ext4_io_page     *pages[MAX_IO_PAGES]; /* for writepages() */
 226} ext4_io_end_t;
 227
 228struct ext4_io_submit {
 229        int                     io_op;
 230        struct bio              *io_bio;
 231        ext4_io_end_t           *io_end;
 232        struct ext4_io_page     *io_page;
 233        sector_t                io_next_block;
 234};
 235
 236/*
 237 * Special inodes numbers
 238 */
 239#define EXT4_BAD_INO             1      /* Bad blocks inode */
 240#define EXT4_ROOT_INO            2      /* Root inode */
 241#define EXT4_USR_QUOTA_INO       3      /* User quota inode */
 242#define EXT4_GRP_QUOTA_INO       4      /* Group quota inode */
 243#define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
 244#define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
 245#define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
 246#define EXT4_JOURNAL_INO         8      /* Journal inode */
 247
 248/* First non-reserved inode for old ext4 filesystems */
 249#define EXT4_GOOD_OLD_FIRST_INO 11
 250
 251/*
 252 * Maximal count of links to a file
 253 */
 254#define EXT4_LINK_MAX           65000
 255
 256/*
 257 * Macro-instructions used to manage several block sizes
 258 */
 259#define EXT4_MIN_BLOCK_SIZE             1024
 260#define EXT4_MAX_BLOCK_SIZE             65536
 261#define EXT4_MIN_BLOCK_LOG_SIZE         10
 262#define EXT4_MAX_BLOCK_LOG_SIZE         16
 263#ifdef __KERNEL__
 264# define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
 265#else
 266# define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
 267#endif
 268#define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
 269#define EXT4_CLUSTER_SIZE(s)            (EXT4_BLOCK_SIZE(s) << \
 270                                         EXT4_SB(s)->s_cluster_bits)
 271#ifdef __KERNEL__
 272# define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
 273# define EXT4_CLUSTER_BITS(s)           (EXT4_SB(s)->s_cluster_bits)
 274#else
 275# define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
 276#endif
 277#ifdef __KERNEL__
 278#define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
 279#define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
 280#define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
 281#else
 282#define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 283                                 EXT4_GOOD_OLD_INODE_SIZE : \
 284                                 (s)->s_inode_size)
 285#define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 286                                 EXT4_GOOD_OLD_FIRST_INO : \
 287                                 (s)->s_first_ino)
 288#endif
 289#define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
 290
 291/* Translate a block number to a cluster number */
 292#define EXT4_B2C(sbi, blk)      ((blk) >> (sbi)->s_cluster_bits)
 293/* Translate a cluster number to a block number */
 294#define EXT4_C2B(sbi, cluster)  ((cluster) << (sbi)->s_cluster_bits)
 295/* Translate # of blks to # of clusters */
 296#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
 297                                 (sbi)->s_cluster_bits)
 298
 299/*
 300 * Structure of a blocks group descriptor
 301 */
 302struct ext4_group_desc
 303{
 304        __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
 305        __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
 306        __le32  bg_inode_table_lo;      /* Inodes table block */
 307        __le16  bg_free_blocks_count_lo;/* Free blocks count */
 308        __le16  bg_free_inodes_count_lo;/* Free inodes count */
 309        __le16  bg_used_dirs_count_lo;  /* Directories count */
 310        __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
 311        __le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
 312        __le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
 313        __le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
 314        __le16  bg_itable_unused_lo;    /* Unused inodes count */
 315        __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
 316        __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
 317        __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
 318        __le32  bg_inode_table_hi;      /* Inodes table block MSB */
 319        __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
 320        __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
 321        __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
 322        __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
 323        __le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
 324        __le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
 325        __le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
 326        __u32   bg_reserved;
 327};
 328
 329#define EXT4_BG_INODE_BITMAP_CSUM_HI_END        \
 330        (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
 331         sizeof(__le16))
 332#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END        \
 333        (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
 334         sizeof(__le16))
 335
 336/*
 337 * Structure of a flex block group info
 338 */
 339
 340struct flex_groups {
 341        atomic_t free_inodes;
 342        atomic_t free_clusters;
 343        atomic_t used_dirs;
 344};
 345
 346#define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
 347#define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
 348#define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
 349
 350/*
 351 * Macro-instructions used to manage group descriptors
 352 */
 353#define EXT4_MIN_DESC_SIZE              32
 354#define EXT4_MIN_DESC_SIZE_64BIT        64
 355#define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
 356#define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
 357#ifdef __KERNEL__
 358# define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
 359# define EXT4_CLUSTERS_PER_GROUP(s)     (EXT4_SB(s)->s_clusters_per_group)
 360# define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
 361# define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
 362# define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
 363#else
 364# define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
 365# define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
 366# define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
 367#endif
 368
 369/*
 370 * Constants relative to the data blocks
 371 */
 372#define EXT4_NDIR_BLOCKS                12
 373#define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
 374#define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
 375#define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
 376#define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
 377
 378/*
 379 * Inode flags
 380 */
 381#define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
 382#define EXT4_UNRM_FL                    0x00000002 /* Undelete */
 383#define EXT4_COMPR_FL                   0x00000004 /* Compress file */
 384#define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
 385#define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
 386#define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
 387#define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
 388#define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
 389/* Reserved for compression usage... */
 390#define EXT4_DIRTY_FL                   0x00000100
 391#define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
 392#define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
 393#define EXT4_ECOMPR_FL                  0x00000800 /* Compression error */
 394/* End compression flags --- maybe not all used */
 395#define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
 396#define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
 397#define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
 398#define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
 399#define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
 400#define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
 401#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
 402#define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
 403#define EXT4_EA_INODE_FL                0x00200000 /* Inode used for large EA */
 404#define EXT4_EOFBLOCKS_FL               0x00400000 /* Blocks allocated beyond EOF */
 405#define EXT4_INLINE_DATA_FL             0x10000000 /* Inode has inline data. */
 406#define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
 407
 408#define EXT4_FL_USER_VISIBLE            0x004BDFFF /* User visible flags */
 409#define EXT4_FL_USER_MODIFIABLE         0x004B80FF /* User modifiable flags */
 410
 411/* Flags that should be inherited by new inodes from their parent. */
 412#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
 413                           EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
 414                           EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
 415                           EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
 416
 417/* Flags that are appropriate for regular files (all but dir-specific ones). */
 418#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
 419
 420/* Flags that are appropriate for non-directories/regular files. */
 421#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
 422
 423/* Mask out flags that are inappropriate for the given type of inode. */
 424static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 425{
 426        if (S_ISDIR(mode))
 427                return flags;
 428        else if (S_ISREG(mode))
 429                return flags & EXT4_REG_FLMASK;
 430        else
 431                return flags & EXT4_OTHER_FLMASK;
 432}
 433
 434/*
 435 * Inode flags used for atomic set/get
 436 */
 437enum {
 438        EXT4_INODE_SECRM        = 0,    /* Secure deletion */
 439        EXT4_INODE_UNRM         = 1,    /* Undelete */
 440        EXT4_INODE_COMPR        = 2,    /* Compress file */
 441        EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
 442        EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
 443        EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
 444        EXT4_INODE_NODUMP       = 6,    /* do not dump file */
 445        EXT4_INODE_NOATIME      = 7,    /* do not update atime */
 446/* Reserved for compression usage... */
 447        EXT4_INODE_DIRTY        = 8,
 448        EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
 449        EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
 450        EXT4_INODE_ECOMPR       = 11,   /* Compression error */
 451/* End compression flags --- maybe not all used */
 452        EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
 453        EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
 454        EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
 455        EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
 456        EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
 457        EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
 458        EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
 459        EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
 460        EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
 461        EXT4_INODE_EOFBLOCKS    = 22,   /* Blocks allocated beyond EOF */
 462        EXT4_INODE_INLINE_DATA  = 28,   /* Data in inode. */
 463        EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
 464};
 465
 466/*
 467 * Since it's pretty easy to mix up bit numbers and hex values, we use a
 468 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
 469 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
 470 * any extra space in the compiled kernel image, otherwise, the build will fail.
 471 * It's important that these values are the same, since we are using
 472 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
 473 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
 474 * values found in ext2, ext3 and ext4 filesystems, and of course the values
 475 * defined in e2fsprogs.
 476 *
 477 * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
 478 */
 479#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
 480#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
 481
 482static inline void ext4_check_flag_values(void)
 483{
 484        CHECK_FLAG_VALUE(SECRM);
 485        CHECK_FLAG_VALUE(UNRM);
 486        CHECK_FLAG_VALUE(COMPR);
 487        CHECK_FLAG_VALUE(SYNC);
 488        CHECK_FLAG_VALUE(IMMUTABLE);
 489        CHECK_FLAG_VALUE(APPEND);
 490        CHECK_FLAG_VALUE(NODUMP);
 491        CHECK_FLAG_VALUE(NOATIME);
 492        CHECK_FLAG_VALUE(DIRTY);
 493        CHECK_FLAG_VALUE(COMPRBLK);
 494        CHECK_FLAG_VALUE(NOCOMPR);
 495        CHECK_FLAG_VALUE(ECOMPR);
 496        CHECK_FLAG_VALUE(INDEX);
 497        CHECK_FLAG_VALUE(IMAGIC);
 498        CHECK_FLAG_VALUE(JOURNAL_DATA);
 499        CHECK_FLAG_VALUE(NOTAIL);
 500        CHECK_FLAG_VALUE(DIRSYNC);
 501        CHECK_FLAG_VALUE(TOPDIR);
 502        CHECK_FLAG_VALUE(HUGE_FILE);
 503        CHECK_FLAG_VALUE(EXTENTS);
 504        CHECK_FLAG_VALUE(EA_INODE);
 505        CHECK_FLAG_VALUE(EOFBLOCKS);
 506        CHECK_FLAG_VALUE(INLINE_DATA);
 507        CHECK_FLAG_VALUE(RESERVED);
 508}
 509
 510/* Used to pass group descriptor data when online resize is done */
 511struct ext4_new_group_input {
 512        __u32 group;            /* Group number for this data */
 513        __u64 block_bitmap;     /* Absolute block number of block bitmap */
 514        __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
 515        __u64 inode_table;      /* Absolute block number of inode table start */
 516        __u32 blocks_count;     /* Total number of blocks in this group */
 517        __u16 reserved_blocks;  /* Number of reserved blocks in this group */
 518        __u16 unused;
 519};
 520
 521#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 522struct compat_ext4_new_group_input {
 523        u32 group;
 524        compat_u64 block_bitmap;
 525        compat_u64 inode_bitmap;
 526        compat_u64 inode_table;
 527        u32 blocks_count;
 528        u16 reserved_blocks;
 529        u16 unused;
 530};
 531#endif
 532
 533/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
 534struct ext4_new_group_data {
 535        __u32 group;
 536        __u64 block_bitmap;
 537        __u64 inode_bitmap;
 538        __u64 inode_table;
 539        __u32 blocks_count;
 540        __u16 reserved_blocks;
 541        __u16 unused;
 542        __u32 free_blocks_count;
 543};
 544
 545/* Indexes used to index group tables in ext4_new_group_data */
 546enum {
 547        BLOCK_BITMAP = 0,       /* block bitmap */
 548        INODE_BITMAP,           /* inode bitmap */
 549        INODE_TABLE,            /* inode tables */
 550        GROUP_TABLE_COUNT,
 551};
 552
 553/*
 554 * Flags used by ext4_map_blocks()
 555 */
 556        /* Allocate any needed blocks and/or convert an unitialized
 557           extent to be an initialized ext4 */
 558#define EXT4_GET_BLOCKS_CREATE                  0x0001
 559        /* Request the creation of an unitialized extent */
 560#define EXT4_GET_BLOCKS_UNINIT_EXT              0x0002
 561#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT       (EXT4_GET_BLOCKS_UNINIT_EXT|\
 562                                                 EXT4_GET_BLOCKS_CREATE)
 563        /* Caller is from the delayed allocation writeout path,
 564           so set the magic i_delalloc_reserve_flag after taking the
 565           inode allocation semaphore for */
 566#define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
 567        /* caller is from the direct IO path, request to creation of an
 568        unitialized extents if not allocated, split the uninitialized
 569        extent if blocks has been preallocated already*/
 570#define EXT4_GET_BLOCKS_PRE_IO                  0x0008
 571#define EXT4_GET_BLOCKS_CONVERT                 0x0010
 572#define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
 573                                         EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
 574        /* Convert extent to initialized after IO complete */
 575#define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
 576                                         EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
 577        /* Punch out blocks of an extent */
 578#define EXT4_GET_BLOCKS_PUNCH_OUT_EXT           0x0020
 579        /* Don't normalize allocation size (used for fallocate) */
 580#define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
 581        /* Request will not result in inode size update (user for fallocate) */
 582#define EXT4_GET_BLOCKS_KEEP_SIZE               0x0080
 583        /* Do not take i_data_sem locking in ext4_map_blocks */
 584#define EXT4_GET_BLOCKS_NO_LOCK                 0x0100
 585
 586/*
 587 * Flags used by ext4_free_blocks
 588 */
 589#define EXT4_FREE_BLOCKS_METADATA       0x0001
 590#define EXT4_FREE_BLOCKS_FORGET         0x0002
 591#define EXT4_FREE_BLOCKS_VALIDATED      0x0004
 592#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
 593#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER   0x0010
 594#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER    0x0020
 595
 596/*
 597 * Flags used by ext4_discard_partial_page_buffers
 598 */
 599#define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED   0x0001
 600
 601/*
 602 * ioctl commands
 603 */
 604#define EXT4_IOC_GETFLAGS               FS_IOC_GETFLAGS
 605#define EXT4_IOC_SETFLAGS               FS_IOC_SETFLAGS
 606#define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
 607#define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
 608#define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
 609#define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
 610#define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
 611#define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
 612#define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
 613#define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
 614#define EXT4_IOC_MIGRATE                _IO('f', 9)
 615 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
 616 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
 617#define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
 618#define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
 619#define EXT4_IOC_RESIZE_FS              _IOW('f', 16, __u64)
 620
 621#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 622/*
 623 * ioctl commands in 32 bit emulation
 624 */
 625#define EXT4_IOC32_GETFLAGS             FS_IOC32_GETFLAGS
 626#define EXT4_IOC32_SETFLAGS             FS_IOC32_SETFLAGS
 627#define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
 628#define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
 629#define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
 630#define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
 631#define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
 632#define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
 633#define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
 634#define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
 635#endif
 636
 637/* Max physical block we can address w/o extents */
 638#define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
 639
 640/*
 641 * Structure of an inode on the disk
 642 */
 643struct ext4_inode {
 644        __le16  i_mode;         /* File mode */
 645        __le16  i_uid;          /* Low 16 bits of Owner Uid */
 646        __le32  i_size_lo;      /* Size in bytes */
 647        __le32  i_atime;        /* Access time */
 648        __le32  i_ctime;        /* Inode Change time */
 649        __le32  i_mtime;        /* Modification time */
 650        __le32  i_dtime;        /* Deletion Time */
 651        __le16  i_gid;          /* Low 16 bits of Group Id */
 652        __le16  i_links_count;  /* Links count */
 653        __le32  i_blocks_lo;    /* Blocks count */
 654        __le32  i_flags;        /* File flags */
 655        union {
 656                struct {
 657                        __le32  l_i_version;
 658                } linux1;
 659                struct {
 660                        __u32  h_i_translator;
 661                } hurd1;
 662                struct {
 663                        __u32  m_i_reserved1;
 664                } masix1;
 665        } osd1;                         /* OS dependent 1 */
 666        __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
 667        __le32  i_generation;   /* File version (for NFS) */
 668        __le32  i_file_acl_lo;  /* File ACL */
 669        __le32  i_size_high;
 670        __le32  i_obso_faddr;   /* Obsoleted fragment address */
 671        union {
 672                struct {
 673                        __le16  l_i_blocks_high; /* were l_i_reserved1 */
 674                        __le16  l_i_file_acl_high;
 675                        __le16  l_i_uid_high;   /* these 2 fields */
 676                        __le16  l_i_gid_high;   /* were reserved2[0] */
 677                        __le16  l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
 678                        __le16  l_i_reserved;
 679                } linux2;
 680                struct {
 681                        __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 682                        __u16   h_i_mode_high;
 683                        __u16   h_i_uid_high;
 684                        __u16   h_i_gid_high;
 685                        __u32   h_i_author;
 686                } hurd2;
 687                struct {
 688                        __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 689                        __le16  m_i_file_acl_high;
 690                        __u32   m_i_reserved2[2];
 691                } masix2;
 692        } osd2;                         /* OS dependent 2 */
 693        __le16  i_extra_isize;
 694        __le16  i_checksum_hi;  /* crc32c(uuid+inum+inode) BE */
 695        __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
 696        __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
 697        __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
 698        __le32  i_crtime;       /* File Creation time */
 699        __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
 700        __le32  i_version_hi;   /* high 32 bits for 64-bit version */
 701};
 702
 703struct move_extent {
 704        __u32 reserved;         /* should be zero */
 705        __u32 donor_fd;         /* donor file descriptor */
 706        __u64 orig_start;       /* logical start offset in block for orig */
 707        __u64 donor_start;      /* logical start offset in block for donor */
 708        __u64 len;              /* block length to be moved */
 709        __u64 moved_len;        /* moved block length */
 710};
 711
 712#define EXT4_EPOCH_BITS 2
 713#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
 714#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
 715
 716/*
 717 * Extended fields will fit into an inode if the filesystem was formatted
 718 * with large inodes (-I 256 or larger) and there are not currently any EAs
 719 * consuming all of the available space. For new inodes we always reserve
 720 * enough space for the kernel's known extended fields, but for inodes
 721 * created with an old kernel this might not have been the case. None of
 722 * the extended inode fields is critical for correct filesystem operation.
 723 * This macro checks if a certain field fits in the inode. Note that
 724 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
 725 */
 726#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
 727        ((offsetof(typeof(*ext4_inode), field) +        \
 728          sizeof((ext4_inode)->field))                  \
 729        <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
 730            (einode)->i_extra_isize))                   \
 731
 732static inline __le32 ext4_encode_extra_time(struct timespec *time)
 733{
 734       return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
 735                           (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
 736                          ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
 737}
 738
 739static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
 740{
 741       if (sizeof(time->tv_sec) > 4)
 742               time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
 743                               << 32;
 744       time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 745}
 746
 747#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                          \
 748do {                                                                           \
 749        (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);               \
 750        if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 751                (raw_inode)->xtime ## _extra =                                 \
 752                                ext4_encode_extra_time(&(inode)->xtime);       \
 753} while (0)
 754
 755#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
 756do {                                                                           \
 757        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 758                (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
 759        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 760                (raw_inode)->xtime ## _extra =                                 \
 761                                ext4_encode_extra_time(&(einode)->xtime);      \
 762} while (0)
 763
 764#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                          \
 765do {                                                                           \
 766        (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
 767        if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 768                ext4_decode_extra_time(&(inode)->xtime,                        \
 769                                       raw_inode->xtime ## _extra);            \
 770        else                                                                   \
 771                (inode)->xtime.tv_nsec = 0;                                    \
 772} while (0)
 773
 774#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
 775do {                                                                           \
 776        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 777                (einode)->xtime.tv_sec =                                       \
 778                        (signed)le32_to_cpu((raw_inode)->xtime);               \
 779        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 780                ext4_decode_extra_time(&(einode)->xtime,                       \
 781                                       raw_inode->xtime ## _extra);            \
 782        else                                                                   \
 783                (einode)->xtime.tv_nsec = 0;                                   \
 784} while (0)
 785
 786#define i_disk_version osd1.linux1.l_i_version
 787
 788#if defined(__KERNEL__) || defined(__linux__)
 789#define i_reserved1     osd1.linux1.l_i_reserved1
 790#define i_file_acl_high osd2.linux2.l_i_file_acl_high
 791#define i_blocks_high   osd2.linux2.l_i_blocks_high
 792#define i_uid_low       i_uid
 793#define i_gid_low       i_gid
 794#define i_uid_high      osd2.linux2.l_i_uid_high
 795#define i_gid_high      osd2.linux2.l_i_gid_high
 796#define i_checksum_lo   osd2.linux2.l_i_checksum_lo
 797
 798#elif defined(__GNU__)
 799
 800#define i_translator    osd1.hurd1.h_i_translator
 801#define i_uid_high      osd2.hurd2.h_i_uid_high
 802#define i_gid_high      osd2.hurd2.h_i_gid_high
 803#define i_author        osd2.hurd2.h_i_author
 804
 805#elif defined(__masix__)
 806
 807#define i_reserved1     osd1.masix1.m_i_reserved1
 808#define i_file_acl_high osd2.masix2.m_i_file_acl_high
 809#define i_reserved2     osd2.masix2.m_i_reserved2
 810
 811#endif /* defined(__KERNEL__) || defined(__linux__) */
 812
 813/*
 814 * storage for cached extent
 815 * If ec_len == 0, then the cache is invalid.
 816 * If ec_start == 0, then the cache represents a gap (null mapping)
 817 */
 818struct ext4_ext_cache {
 819        ext4_fsblk_t    ec_start;
 820        ext4_lblk_t     ec_block;
 821        __u32           ec_len; /* must be 32bit to return holes */
 822};
 823
 824#include "extents_status.h"
 825
 826/*
 827 * fourth extended file system inode data in memory
 828 */
 829struct ext4_inode_info {
 830        __le32  i_data[15];     /* unconverted */
 831        __u32   i_dtime;
 832        ext4_fsblk_t    i_file_acl;
 833
 834        /*
 835         * i_block_group is the number of the block group which contains
 836         * this file's inode.  Constant across the lifetime of the inode,
 837         * it is ued for making block allocation decisions - we try to
 838         * place a file's data blocks near its inode block, and new inodes
 839         * near to their parent directory's inode.
 840         */
 841        ext4_group_t    i_block_group;
 842        ext4_lblk_t     i_dir_start_lookup;
 843#if (BITS_PER_LONG < 64)
 844        unsigned long   i_state_flags;          /* Dynamic state flags */
 845#endif
 846        unsigned long   i_flags;
 847
 848        /*
 849         * Extended attributes can be read independently of the main file
 850         * data. Taking i_mutex even when reading would cause contention
 851         * between readers of EAs and writers of regular file data, so
 852         * instead we synchronize on xattr_sem when reading or changing
 853         * EAs.
 854         */
 855        struct rw_semaphore xattr_sem;
 856
 857        struct list_head i_orphan;      /* unlinked but open inodes */
 858
 859        /*
 860         * i_disksize keeps track of what the inode size is ON DISK, not
 861         * in memory.  During truncate, i_size is set to the new size by
 862         * the VFS prior to calling ext4_truncate(), but the filesystem won't
 863         * set i_disksize to 0 until the truncate is actually under way.
 864         *
 865         * The intent is that i_disksize always represents the blocks which
 866         * are used by this file.  This allows recovery to restart truncate
 867         * on orphans if we crash during truncate.  We actually write i_disksize
 868         * into the on-disk inode when writing inodes out, instead of i_size.
 869         *
 870         * The only time when i_disksize and i_size may be different is when
 871         * a truncate is in progress.  The only things which change i_disksize
 872         * are ext4_get_block (growth) and ext4_truncate (shrinkth).
 873         */
 874        loff_t  i_disksize;
 875
 876        /*
 877         * i_data_sem is for serialising ext4_truncate() against
 878         * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
 879         * data tree are chopped off during truncate. We can't do that in
 880         * ext4 because whenever we perform intermediate commits during
 881         * truncate, the inode and all the metadata blocks *must* be in a
 882         * consistent state which allows truncation of the orphans to restart
 883         * during recovery.  Hence we must fix the get_block-vs-truncate race
 884         * by other means, so we have i_data_sem.
 885         */
 886        struct rw_semaphore i_data_sem;
 887        struct inode vfs_inode;
 888        struct jbd2_inode *jinode;
 889
 890        struct ext4_ext_cache i_cached_extent;
 891        /*
 892         * File creation time. Its function is same as that of
 893         * struct timespec i_{a,c,m}time in the generic inode.
 894         */
 895        struct timespec i_crtime;
 896
 897        /* mballoc */
 898        struct list_head i_prealloc_list;
 899        spinlock_t i_prealloc_lock;
 900
 901        /* extents status tree */
 902        struct ext4_es_tree i_es_tree;
 903        rwlock_t i_es_lock;
 904
 905        /* ialloc */
 906        ext4_group_t    i_last_alloc_group;
 907
 908        /* allocation reservation info for delalloc */
 909        /* In case of bigalloc, these refer to clusters rather than blocks */
 910        unsigned int i_reserved_data_blocks;
 911        unsigned int i_reserved_meta_blocks;
 912        unsigned int i_allocated_meta_blocks;
 913        ext4_lblk_t i_da_metadata_calc_last_lblock;
 914        int i_da_metadata_calc_len;
 915
 916        /* on-disk additional length */
 917        __u16 i_extra_isize;
 918
 919        /* Indicate the inline data space. */
 920        u16 i_inline_off;
 921        u16 i_inline_size;
 922
 923#ifdef CONFIG_QUOTA
 924        /* quota space reservation, managed internally by quota code */
 925        qsize_t i_reserved_quota;
 926#endif
 927
 928        /* completed IOs that might need unwritten extents handling */
 929        struct list_head i_completed_io_list;
 930        spinlock_t i_completed_io_lock;
 931        atomic_t i_ioend_count; /* Number of outstanding io_end structs */
 932        atomic_t i_unwritten; /* Nr. of inflight conversions pending */
 933
 934        spinlock_t i_block_reservation_lock;
 935
 936        /*
 937         * Transactions that contain inode's metadata needed to complete
 938         * fsync and fdatasync, respectively.
 939         */
 940        tid_t i_sync_tid;
 941        tid_t i_datasync_tid;
 942
 943        /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
 944        __u32 i_csum_seed;
 945};
 946
 947/*
 948 * File system states
 949 */
 950#define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
 951#define EXT4_ERROR_FS                   0x0002  /* Errors detected */
 952#define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
 953
 954/*
 955 * Misc. filesystem flags
 956 */
 957#define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
 958#define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
 959#define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
 960
 961/*
 962 * Mount flags
 963 */
 964#define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
 965#define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
 966#define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
 967#define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
 968#define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
 969#define EXT4_MOUNT_ERRORS_MASK          0x00070
 970#define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
 971#define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
 972#define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
 973#define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
 974#define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
 975#define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
 976#define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
 977#define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
 978#define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
 979#define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
 980#define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
 981#define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
 982#define EXT4_MOUNT_QUOTA                0x80000 /* Some quota option set */
 983#define EXT4_MOUNT_USRQUOTA             0x100000 /* "old" user quota */
 984#define EXT4_MOUNT_GRPQUOTA             0x200000 /* "old" group quota */
 985#define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
 986#define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
 987#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
 988#define EXT4_MOUNT_MBLK_IO_SUBMIT       0x4000000 /* multi-block io submits */
 989#define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
 990#define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
 991#define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
 992#define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
 993#define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
 994
 995#define EXT4_MOUNT2_EXPLICIT_DELALLOC   0x00000001 /* User explicitly
 996                                                      specified delalloc */
 997
 998#define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
 999                                                ~EXT4_MOUNT_##opt
1000#define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
1001                                                EXT4_MOUNT_##opt
1002#define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
1003                                         EXT4_MOUNT_##opt)
1004
1005#define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
1006                                                ~EXT4_MOUNT2_##opt
1007#define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
1008                                                EXT4_MOUNT2_##opt
1009#define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
1010                                         EXT4_MOUNT2_##opt)
1011
1012#define ext4_test_and_set_bit           __test_and_set_bit_le
1013#define ext4_set_bit                    __set_bit_le
1014#define ext4_set_bit_atomic             ext2_set_bit_atomic
1015#define ext4_test_and_clear_bit         __test_and_clear_bit_le
1016#define ext4_clear_bit                  __clear_bit_le
1017#define ext4_clear_bit_atomic           ext2_clear_bit_atomic
1018#define ext4_test_bit                   test_bit_le
1019#define ext4_find_next_zero_bit         find_next_zero_bit_le
1020#define ext4_find_next_bit              find_next_bit_le
1021
1022extern void ext4_set_bits(void *bm, int cur, int len);
1023
1024/*
1025 * Maximal mount counts between two filesystem checks
1026 */
1027#define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
1028#define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
1029
1030/*
1031 * Behaviour when detecting errors
1032 */
1033#define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
1034#define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
1035#define EXT4_ERRORS_PANIC               3       /* Panic */
1036#define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
1037
1038/* Metadata checksum algorithm codes */
1039#define EXT4_CRC32C_CHKSUM              1
1040
1041/*
1042 * Structure of the super block
1043 */
1044struct ext4_super_block {
1045/*00*/  __le32  s_inodes_count;         /* Inodes count */
1046        __le32  s_blocks_count_lo;      /* Blocks count */
1047        __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
1048        __le32  s_free_blocks_count_lo; /* Free blocks count */
1049/*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
1050        __le32  s_first_data_block;     /* First Data Block */
1051        __le32  s_log_block_size;       /* Block size */
1052        __le32  s_log_cluster_size;     /* Allocation cluster size */
1053/*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
1054        __le32  s_clusters_per_group;   /* # Clusters per group */
1055        __le32  s_inodes_per_group;     /* # Inodes per group */
1056        __le32  s_mtime;                /* Mount time */
1057/*30*/  __le32  s_wtime;                /* Write time */
1058        __le16  s_mnt_count;            /* Mount count */
1059        __le16  s_max_mnt_count;        /* Maximal mount count */
1060        __le16  s_magic;                /* Magic signature */
1061        __le16  s_state;                /* File system state */
1062        __le16  s_errors;               /* Behaviour when detecting errors */
1063        __le16  s_minor_rev_level;      /* minor revision level */
1064/*40*/  __le32  s_lastcheck;            /* time of last check */
1065        __le32  s_checkinterval;        /* max. time between checks */
1066        __le32  s_creator_os;           /* OS */
1067        __le32  s_rev_level;            /* Revision level */
1068/*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
1069        __le16  s_def_resgid;           /* Default gid for reserved blocks */
1070        /*
1071         * These fields are for EXT4_DYNAMIC_REV superblocks only.
1072         *
1073         * Note: the difference between the compatible feature set and
1074         * the incompatible feature set is that if there is a bit set
1075         * in the incompatible feature set that the kernel doesn't
1076         * know about, it should refuse to mount the filesystem.
1077         *
1078         * e2fsck's requirements are more strict; if it doesn't know
1079         * about a feature in either the compatible or incompatible
1080         * feature set, it must abort and not try to meddle with
1081         * things it doesn't understand...
1082         */
1083        __le32  s_first_ino;            /* First non-reserved inode */
1084        __le16  s_inode_size;           /* size of inode structure */
1085        __le16  s_block_group_nr;       /* block group # of this superblock */
1086        __le32  s_feature_compat;       /* compatible feature set */
1087/*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1088        __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1089/*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1090/*78*/  char    s_volume_name[16];      /* volume name */
1091/*88*/  char    s_last_mounted[64];     /* directory where last mounted */
1092/*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1093        /*
1094         * Performance hints.  Directory preallocation should only
1095         * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1096         */
1097        __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1098        __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1099        __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1100        /*
1101         * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1102         */
1103/*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1104/*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1105        __le32  s_journal_dev;          /* device number of journal file */
1106        __le32  s_last_orphan;          /* start of list of inodes to delete */
1107        __le32  s_hash_seed[4];         /* HTREE hash seed */
1108        __u8    s_def_hash_version;     /* Default hash version to use */
1109        __u8    s_jnl_backup_type;
1110        __le16  s_desc_size;            /* size of group descriptor */
1111/*100*/ __le32  s_default_mount_opts;
1112        __le32  s_first_meta_bg;        /* First metablock block group */
1113        __le32  s_mkfs_time;            /* When the filesystem was created */
1114        __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1115        /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1116/*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1117        __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1118        __le32  s_free_blocks_count_hi; /* Free blocks count */
1119        __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1120        __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1121        __le32  s_flags;                /* Miscellaneous flags */
1122        __le16  s_raid_stride;          /* RAID stride */
1123        __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1124        __le64  s_mmp_block;            /* Block for multi-mount protection */
1125        __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1126        __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1127        __u8    s_checksum_type;        /* metadata checksum algorithm used */
1128        __le16  s_reserved_pad;
1129        __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1130        __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1131        __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1132        __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1133                                              snapshot's future use */
1134        __le32  s_snapshot_list;        /* inode number of the head of the
1135                                           on-disk snapshot list */
1136#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1137        __le32  s_error_count;          /* number of fs errors */
1138        __le32  s_first_error_time;     /* first time an error happened */
1139        __le32  s_first_error_ino;      /* inode involved in first error */
1140        __le64  s_first_error_block;    /* block involved of first error */
1141        __u8    s_first_error_func[32]; /* function where the error happened */
1142        __le32  s_first_error_line;     /* line number where error happened */
1143        __le32  s_last_error_time;      /* most recent time of an error */
1144        __le32  s_last_error_ino;       /* inode involved in last error */
1145        __le32  s_last_error_line;      /* line number where error happened */
1146        __le64  s_last_error_block;     /* block involved of last error */
1147        __u8    s_last_error_func[32];  /* function where the error happened */
1148#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1149        __u8    s_mount_opts[64];
1150        __le32  s_usr_quota_inum;       /* inode for tracking user quota */
1151        __le32  s_grp_quota_inum;       /* inode for tracking group quota */
1152        __le32  s_overhead_clusters;    /* overhead blocks/clusters in fs */
1153        __le32  s_reserved[108];        /* Padding to the end of the block */
1154        __le32  s_checksum;             /* crc32c(superblock) */
1155};
1156
1157#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1158
1159#ifdef __KERNEL__
1160
1161/*
1162 * run-time mount flags
1163 */
1164#define EXT4_MF_MNTDIR_SAMPLED  0x0001
1165#define EXT4_MF_FS_ABORTED      0x0002  /* Fatal error detected */
1166
1167/*
1168 * fourth extended-fs super-block data in memory
1169 */
1170struct ext4_sb_info {
1171        unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1172        unsigned long s_inodes_per_block;/* Number of inodes per block */
1173        unsigned long s_blocks_per_group;/* Number of blocks in a group */
1174        unsigned long s_clusters_per_group; /* Number of clusters in a group */
1175        unsigned long s_inodes_per_group;/* Number of inodes in a group */
1176        unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1177        unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1178        unsigned long s_desc_per_block; /* Number of group descriptors per block */
1179        ext4_group_t s_groups_count;    /* Number of groups in the fs */
1180        ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1181        unsigned long s_overhead;  /* # of fs overhead clusters */
1182        unsigned int s_cluster_ratio;   /* Number of blocks per cluster */
1183        unsigned int s_cluster_bits;    /* log2 of s_cluster_ratio */
1184        loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1185        struct buffer_head * s_sbh;     /* Buffer containing the super block */
1186        struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1187        struct buffer_head **s_group_desc;
1188        unsigned int s_mount_opt;
1189        unsigned int s_mount_opt2;
1190        unsigned int s_mount_flags;
1191        unsigned int s_def_mount_opt;
1192        ext4_fsblk_t s_sb_block;
1193        kuid_t s_resuid;
1194        kgid_t s_resgid;
1195        unsigned short s_mount_state;
1196        unsigned short s_pad;
1197        int s_addr_per_block_bits;
1198        int s_desc_per_block_bits;
1199        int s_inode_size;
1200        int s_first_ino;
1201        unsigned int s_inode_readahead_blks;
1202        unsigned int s_inode_goal;
1203        spinlock_t s_next_gen_lock;
1204        u32 s_next_generation;
1205        u32 s_hash_seed[4];
1206        int s_def_hash_version;
1207        int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1208        struct percpu_counter s_freeclusters_counter;
1209        struct percpu_counter s_freeinodes_counter;
1210        struct percpu_counter s_dirs_counter;
1211        struct percpu_counter s_dirtyclusters_counter;
1212        struct blockgroup_lock *s_blockgroup_lock;
1213        struct proc_dir_entry *s_proc;
1214        struct kobject s_kobj;
1215        struct completion s_kobj_unregister;
1216        struct super_block *s_sb;
1217
1218        /* Journaling */
1219        struct journal_s *s_journal;
1220        struct list_head s_orphan;
1221        struct mutex s_orphan_lock;
1222        unsigned long s_resize_flags;           /* Flags indicating if there
1223                                                   is a resizer */
1224        unsigned long s_commit_interval;
1225        u32 s_max_batch_time;
1226        u32 s_min_batch_time;
1227        struct block_device *journal_bdev;
1228#ifdef CONFIG_QUOTA
1229        char *s_qf_names[MAXQUOTAS];            /* Names of quota files with journalled quota */
1230        int s_jquota_fmt;                       /* Format of quota to use */
1231#endif
1232        unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1233        struct rb_root system_blks;
1234
1235#ifdef EXTENTS_STATS
1236        /* ext4 extents stats */
1237        unsigned long s_ext_min;
1238        unsigned long s_ext_max;
1239        unsigned long s_depth_max;
1240        spinlock_t s_ext_stats_lock;
1241        unsigned long s_ext_blocks;
1242        unsigned long s_ext_extents;
1243#endif
1244
1245        /* for buddy allocator */
1246        struct ext4_group_info ***s_group_info;
1247        struct inode *s_buddy_cache;
1248        spinlock_t s_md_lock;
1249        unsigned short *s_mb_offsets;
1250        unsigned int *s_mb_maxs;
1251        unsigned int s_group_info_size;
1252
1253        /* tunables */
1254        unsigned long s_stripe;
1255        unsigned int s_mb_stream_request;
1256        unsigned int s_mb_max_to_scan;
1257        unsigned int s_mb_min_to_scan;
1258        unsigned int s_mb_stats;
1259        unsigned int s_mb_order2_reqs;
1260        unsigned int s_mb_group_prealloc;
1261        unsigned int s_max_writeback_mb_bump;
1262        unsigned int s_max_dir_size_kb;
1263        /* where last allocation was done - for stream allocation */
1264        unsigned long s_mb_last_group;
1265        unsigned long s_mb_last_start;
1266
1267        /* stats for buddy allocator */
1268        atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1269        atomic_t s_bal_success; /* we found long enough chunks */
1270        atomic_t s_bal_allocated;       /* in blocks */
1271        atomic_t s_bal_ex_scanned;      /* total extents scanned */
1272        atomic_t s_bal_goals;   /* goal hits */
1273        atomic_t s_bal_breaks;  /* too long searches */
1274        atomic_t s_bal_2orders; /* 2^order hits */
1275        spinlock_t s_bal_lock;
1276        unsigned long s_mb_buddies_generated;
1277        unsigned long long s_mb_generation_time;
1278        atomic_t s_mb_lost_chunks;
1279        atomic_t s_mb_preallocated;
1280        atomic_t s_mb_discarded;
1281        atomic_t s_lock_busy;
1282
1283        /* locality groups */
1284        struct ext4_locality_group __percpu *s_locality_groups;
1285
1286        /* for write statistics */
1287        unsigned long s_sectors_written_start;
1288        u64 s_kbytes_written;
1289
1290        /* the size of zero-out chunk */
1291        unsigned int s_extent_max_zeroout_kb;
1292
1293        unsigned int s_log_groups_per_flex;
1294        struct flex_groups *s_flex_groups;
1295        ext4_group_t s_flex_groups_allocated;
1296
1297        /* workqueue for dio unwritten */
1298        struct workqueue_struct *dio_unwritten_wq;
1299
1300        /* timer for periodic error stats printing */
1301        struct timer_list s_err_report;
1302
1303        /* Lazy inode table initialization info */
1304        struct ext4_li_request *s_li_request;
1305        /* Wait multiplier for lazy initialization thread */
1306        unsigned int s_li_wait_mult;
1307
1308        /* Kernel thread for multiple mount protection */
1309        struct task_struct *s_mmp_tsk;
1310
1311        /* record the last minlen when FITRIM is called. */
1312        atomic_t s_last_trim_minblks;
1313
1314        /* Reference to checksum algorithm driver via cryptoapi */
1315        struct crypto_shash *s_chksum_driver;
1316
1317        /* Precomputed FS UUID checksum for seeding other checksums */
1318        __u32 s_csum_seed;
1319};
1320
1321static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1322{
1323        return sb->s_fs_info;
1324}
1325static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1326{
1327        return container_of(inode, struct ext4_inode_info, vfs_inode);
1328}
1329
1330static inline struct timespec ext4_current_time(struct inode *inode)
1331{
1332        return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1333                current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1334}
1335
1336static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1337{
1338        return ino == EXT4_ROOT_INO ||
1339                ino == EXT4_USR_QUOTA_INO ||
1340                ino == EXT4_GRP_QUOTA_INO ||
1341                ino == EXT4_JOURNAL_INO ||
1342                ino == EXT4_RESIZE_INO ||
1343                (ino >= EXT4_FIRST_INO(sb) &&
1344                 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1345}
1346
1347static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1348                                              struct ext4_io_end *io_end)
1349{
1350        if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1351                io_end->flag |= EXT4_IO_END_UNWRITTEN;
1352                atomic_inc(&EXT4_I(inode)->i_unwritten);
1353        }
1354}
1355
1356static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1357{
1358        return inode->i_private;
1359}
1360
1361static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1362{
1363        inode->i_private = io;
1364}
1365
1366/*
1367 * Inode dynamic state flags
1368 */
1369enum {
1370        EXT4_STATE_JDATA,               /* journaled data exists */
1371        EXT4_STATE_NEW,                 /* inode is newly created */
1372        EXT4_STATE_XATTR,               /* has in-inode xattrs */
1373        EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1374        EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1375        EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1376        EXT4_STATE_DIO_UNWRITTEN,       /* need convert on dio done*/
1377        EXT4_STATE_NEWENTRY,            /* File just added to dir */
1378        EXT4_STATE_DELALLOC_RESERVED,   /* blks already reserved for delalloc */
1379        EXT4_STATE_DIOREAD_LOCK,        /* Disable support for dio read
1380                                           nolocking */
1381        EXT4_STATE_MAY_INLINE_DATA,     /* may have in-inode data */
1382};
1383
1384#define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1385static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1386{                                                                       \
1387        return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1388}                                                                       \
1389static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1390{                                                                       \
1391        set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1392}                                                                       \
1393static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1394{                                                                       \
1395        clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1396}
1397
1398EXT4_INODE_BIT_FNS(flag, flags, 0)
1399#if (BITS_PER_LONG < 64)
1400EXT4_INODE_BIT_FNS(state, state_flags, 0)
1401
1402static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1403{
1404        (ei)->i_state_flags = 0;
1405}
1406#else
1407EXT4_INODE_BIT_FNS(state, flags, 32)
1408
1409static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1410{
1411        /* We depend on the fact that callers will set i_flags */
1412}
1413#endif
1414#else
1415/* Assume that user mode programs are passing in an ext4fs superblock, not
1416 * a kernel struct super_block.  This will allow us to call the feature-test
1417 * macros from user land. */
1418#define EXT4_SB(sb)     (sb)
1419#endif
1420
1421#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1422
1423/*
1424 * Codes for operating systems
1425 */
1426#define EXT4_OS_LINUX           0
1427#define EXT4_OS_HURD            1
1428#define EXT4_OS_MASIX           2
1429#define EXT4_OS_FREEBSD         3
1430#define EXT4_OS_LITES           4
1431
1432/*
1433 * Revision levels
1434 */
1435#define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1436#define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1437
1438#define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1439#define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1440
1441#define EXT4_GOOD_OLD_INODE_SIZE 128
1442
1443/*
1444 * Feature set definitions
1445 */
1446
1447#define EXT4_HAS_COMPAT_FEATURE(sb,mask)                        \
1448        ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1449#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)                     \
1450        ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1451#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)                      \
1452        ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1453#define EXT4_SET_COMPAT_FEATURE(sb,mask)                        \
1454        EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1455#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)                     \
1456        EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1457#define EXT4_SET_INCOMPAT_FEATURE(sb,mask)                      \
1458        EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1459#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)                      \
1460        EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1461#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)                   \
1462        EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1463#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)                    \
1464        EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1465
1466#define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1467#define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1468#define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1469#define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1470#define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1471#define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1472
1473#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1474#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1475#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1476#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1477#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1478#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1479#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1480#define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1481#define EXT4_FEATURE_RO_COMPAT_BIGALLOC         0x0200
1482/*
1483 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1484 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1485 * all other data structures' checksums.  However, the METADATA_CSUM and
1486 * GDT_CSUM bits are mutually exclusive.
1487 */
1488#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM    0x0400
1489
1490#define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1491#define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1492#define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1493#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1494#define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1495#define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1496#define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1497#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1498#define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1499#define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1500#define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1501#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM  0x2000 /* use crc32c for bg */
1502#define EXT4_FEATURE_INCOMPAT_LARGEDIR          0x4000 /* >2GB or 3-lvl htree */
1503#define EXT4_FEATURE_INCOMPAT_INLINE_DATA       0x8000 /* data in inode */
1504
1505#define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1506#define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1507                                         EXT4_FEATURE_INCOMPAT_META_BG)
1508#define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1509                                         EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1510                                         EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1511
1512#define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1513#define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1514                                         EXT4_FEATURE_INCOMPAT_RECOVER| \
1515                                         EXT4_FEATURE_INCOMPAT_META_BG)
1516#define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1517                                         EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1518                                         EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1519
1520#define EXT4_FEATURE_COMPAT_SUPP        EXT2_FEATURE_COMPAT_EXT_ATTR
1521#define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1522                                         EXT4_FEATURE_INCOMPAT_RECOVER| \
1523                                         EXT4_FEATURE_INCOMPAT_META_BG| \
1524                                         EXT4_FEATURE_INCOMPAT_EXTENTS| \
1525                                         EXT4_FEATURE_INCOMPAT_64BIT| \
1526                                         EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1527                                         EXT4_FEATURE_INCOMPAT_MMP |    \
1528                                         EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1529#define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1530                                         EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1531                                         EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1532                                         EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1533                                         EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1534                                         EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1535                                         EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1536                                         EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1537                                         EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1538                                         EXT4_FEATURE_RO_COMPAT_QUOTA)
1539
1540/*
1541 * Default values for user and/or group using reserved blocks
1542 */
1543#define EXT4_DEF_RESUID         0
1544#define EXT4_DEF_RESGID         0
1545
1546#define EXT4_DEF_INODE_READAHEAD_BLKS   32
1547
1548/*
1549 * Default mount options
1550 */
1551#define EXT4_DEFM_DEBUG         0x0001
1552#define EXT4_DEFM_BSDGROUPS     0x0002
1553#define EXT4_DEFM_XATTR_USER    0x0004
1554#define EXT4_DEFM_ACL           0x0008
1555#define EXT4_DEFM_UID16         0x0010
1556#define EXT4_DEFM_JMODE         0x0060
1557#define EXT4_DEFM_JMODE_DATA    0x0020
1558#define EXT4_DEFM_JMODE_ORDERED 0x0040
1559#define EXT4_DEFM_JMODE_WBACK   0x0060
1560#define EXT4_DEFM_NOBARRIER     0x0100
1561#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1562#define EXT4_DEFM_DISCARD       0x0400
1563#define EXT4_DEFM_NODELALLOC    0x0800
1564
1565/*
1566 * Default journal batch times
1567 */
1568#define EXT4_DEF_MIN_BATCH_TIME 0
1569#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1570
1571/*
1572 * Minimum number of groups in a flexgroup before we separate out
1573 * directories into the first block group of a flexgroup
1574 */
1575#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1576
1577/*
1578 * Structure of a directory entry
1579 */
1580#define EXT4_NAME_LEN 255
1581
1582struct ext4_dir_entry {
1583        __le32  inode;                  /* Inode number */
1584        __le16  rec_len;                /* Directory entry length */
1585        __le16  name_len;               /* Name length */
1586        char    name[EXT4_NAME_LEN];    /* File name */
1587};
1588
1589/*
1590 * The new version of the directory entry.  Since EXT4 structures are
1591 * stored in intel byte order, and the name_len field could never be
1592 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1593 * file_type field.
1594 */
1595struct ext4_dir_entry_2 {
1596        __le32  inode;                  /* Inode number */
1597        __le16  rec_len;                /* Directory entry length */
1598        __u8    name_len;               /* Name length */
1599        __u8    file_type;
1600        char    name[EXT4_NAME_LEN];    /* File name */
1601};
1602
1603/*
1604 * This is a bogus directory entry at the end of each leaf block that
1605 * records checksums.
1606 */
1607struct ext4_dir_entry_tail {
1608        __le32  det_reserved_zero1;     /* Pretend to be unused */
1609        __le16  det_rec_len;            /* 12 */
1610        __u8    det_reserved_zero2;     /* Zero name length */
1611        __u8    det_reserved_ft;        /* 0xDE, fake file type */
1612        __le32  det_checksum;           /* crc32c(uuid+inum+dirblock) */
1613};
1614
1615#define EXT4_DIRENT_TAIL(block, blocksize) \
1616        ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1617                                        ((blocksize) - \
1618                                         sizeof(struct ext4_dir_entry_tail))))
1619
1620/*
1621 * Ext4 directory file types.  Only the low 3 bits are used.  The
1622 * other bits are reserved for now.
1623 */
1624#define EXT4_FT_UNKNOWN         0
1625#define EXT4_FT_REG_FILE        1
1626#define EXT4_FT_DIR             2
1627#define EXT4_FT_CHRDEV          3
1628#define EXT4_FT_BLKDEV          4
1629#define EXT4_FT_FIFO            5
1630#define EXT4_FT_SOCK            6
1631#define EXT4_FT_SYMLINK         7
1632
1633#define EXT4_FT_MAX             8
1634
1635#define EXT4_FT_DIR_CSUM        0xDE
1636
1637/*
1638 * EXT4_DIR_PAD defines the directory entries boundaries
1639 *
1640 * NOTE: It must be a multiple of 4
1641 */
1642#define EXT4_DIR_PAD                    4
1643#define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
1644#define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
1645                                         ~EXT4_DIR_ROUND)
1646#define EXT4_MAX_REC_LEN                ((1<<16)-1)
1647
1648/*
1649 * If we ever get support for fs block sizes > page_size, we'll need
1650 * to remove the #if statements in the next two functions...
1651 */
1652static inline unsigned int
1653ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1654{
1655        unsigned len = le16_to_cpu(dlen);
1656
1657#if (PAGE_CACHE_SIZE >= 65536)
1658        if (len == EXT4_MAX_REC_LEN || len == 0)
1659                return blocksize;
1660        return (len & 65532) | ((len & 3) << 16);
1661#else
1662        return len;
1663#endif
1664}
1665
1666static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1667{
1668        if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1669                BUG();
1670#if (PAGE_CACHE_SIZE >= 65536)
1671        if (len < 65536)
1672                return cpu_to_le16(len);
1673        if (len == blocksize) {
1674                if (blocksize == 65536)
1675                        return cpu_to_le16(EXT4_MAX_REC_LEN);
1676                else
1677                        return cpu_to_le16(0);
1678        }
1679        return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1680#else
1681        return cpu_to_le16(len);
1682#endif
1683}
1684
1685/*
1686 * Hash Tree Directory indexing
1687 * (c) Daniel Phillips, 2001
1688 */
1689
1690#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1691                                      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1692                    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1693#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1694#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1695
1696/* Legal values for the dx_root hash_version field: */
1697
1698#define DX_HASH_LEGACY          0
1699#define DX_HASH_HALF_MD4        1
1700#define DX_HASH_TEA             2
1701#define DX_HASH_LEGACY_UNSIGNED 3
1702#define DX_HASH_HALF_MD4_UNSIGNED       4
1703#define DX_HASH_TEA_UNSIGNED            5
1704
1705static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1706                              const void *address, unsigned int length)
1707{
1708        struct {
1709                struct shash_desc shash;
1710                char ctx[4];
1711        } desc;
1712        int err;
1713
1714        BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1715
1716        desc.shash.tfm = sbi->s_chksum_driver;
1717        desc.shash.flags = 0;
1718        *(u32 *)desc.ctx = crc;
1719
1720        err = crypto_shash_update(&desc.shash, address, length);
1721        BUG_ON(err);
1722
1723        return *(u32 *)desc.ctx;
1724}
1725
1726#ifdef __KERNEL__
1727
1728/* hash info structure used by the directory hash */
1729struct dx_hash_info
1730{
1731        u32             hash;
1732        u32             minor_hash;
1733        int             hash_version;
1734        u32             *seed;
1735};
1736
1737
1738/* 32 and 64 bit signed EOF for dx directories */
1739#define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1740#define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1741
1742
1743/*
1744 * Control parameters used by ext4_htree_next_block
1745 */
1746#define HASH_NB_ALWAYS          1
1747
1748
1749/*
1750 * Describe an inode's exact location on disk and in memory
1751 */
1752struct ext4_iloc
1753{
1754        struct buffer_head *bh;
1755        unsigned long offset;
1756        ext4_group_t block_group;
1757};
1758
1759static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1760{
1761        return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1762}
1763
1764/*
1765 * This structure is stuffed into the struct file's private_data field
1766 * for directories.  It is where we put information so that we can do
1767 * readdir operations in hash tree order.
1768 */
1769struct dir_private_info {
1770        struct rb_root  root;
1771        struct rb_node  *curr_node;
1772        struct fname    *extra_fname;
1773        loff_t          last_pos;
1774        __u32           curr_hash;
1775        __u32           curr_minor_hash;
1776        __u32           next_hash;
1777};
1778
1779/* calculate the first block number of the group */
1780static inline ext4_fsblk_t
1781ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1782{
1783        return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1784                le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1785}
1786
1787/*
1788 * Special error return code only used by dx_probe() and its callers.
1789 */
1790#define ERR_BAD_DX_DIR  -75000
1791
1792void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1793                        ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1794
1795/*
1796 * Timeout and state flag for lazy initialization inode thread.
1797 */
1798#define EXT4_DEF_LI_WAIT_MULT                   10
1799#define EXT4_DEF_LI_MAX_START_DELAY             5
1800#define EXT4_LAZYINIT_QUIT                      0x0001
1801#define EXT4_LAZYINIT_RUNNING                   0x0002
1802
1803/*
1804 * Lazy inode table initialization info
1805 */
1806struct ext4_lazy_init {
1807        unsigned long           li_state;
1808        struct list_head        li_request_list;
1809        struct mutex            li_list_mtx;
1810};
1811
1812struct ext4_li_request {
1813        struct super_block      *lr_super;
1814        struct ext4_sb_info     *lr_sbi;
1815        ext4_group_t            lr_next_group;
1816        struct list_head        lr_request;
1817        unsigned long           lr_next_sched;
1818        unsigned long           lr_timeout;
1819};
1820
1821struct ext4_features {
1822        struct kobject f_kobj;
1823        struct completion f_kobj_unregister;
1824};
1825
1826/*
1827 * This structure will be used for multiple mount protection. It will be
1828 * written into the block number saved in the s_mmp_block field in the
1829 * superblock. Programs that check MMP should assume that if
1830 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1831 * to use the filesystem, regardless of how old the timestamp is.
1832 */
1833#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1834#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1835#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1836#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1837
1838struct mmp_struct {
1839        __le32  mmp_magic;              /* Magic number for MMP */
1840        __le32  mmp_seq;                /* Sequence no. updated periodically */
1841
1842        /*
1843         * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1844         * purposes and do not affect the correctness of the algorithm
1845         */
1846        __le64  mmp_time;               /* Time last updated */
1847        char    mmp_nodename[64];       /* Node which last updated MMP block */
1848        char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
1849
1850        /*
1851         * mmp_check_interval is used to verify if the MMP block has been
1852         * updated on the block device. The value is updated based on the
1853         * maximum time to write the MMP block during an update cycle.
1854         */
1855        __le16  mmp_check_interval;
1856
1857        __le16  mmp_pad1;
1858        __le32  mmp_pad2[226];
1859        __le32  mmp_checksum;           /* crc32c(uuid+mmp_block) */
1860};
1861
1862/* arguments passed to the mmp thread */
1863struct mmpd_data {
1864        struct buffer_head *bh; /* bh from initial read_mmp_block() */
1865        struct super_block *sb;  /* super block of the fs */
1866};
1867
1868/*
1869 * Check interval multiplier
1870 * The MMP block is written every update interval and initially checked every
1871 * update interval x the multiplier (the value is then adapted based on the
1872 * write latency). The reason is that writes can be delayed under load and we
1873 * don't want readers to incorrectly assume that the filesystem is no longer
1874 * in use.
1875 */
1876#define EXT4_MMP_CHECK_MULT             2UL
1877
1878/*
1879 * Minimum interval for MMP checking in seconds.
1880 */
1881#define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
1882
1883/*
1884 * Maximum interval for MMP checking in seconds.
1885 */
1886#define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
1887
1888/*
1889 * Function prototypes
1890 */
1891
1892/*
1893 * Ok, these declarations are also in <linux/kernel.h> but none of the
1894 * ext4 source programs needs to include it so they are duplicated here.
1895 */
1896# define NORET_TYPE     /**/
1897# define ATTRIB_NORET   __attribute__((noreturn))
1898# define NORET_AND      noreturn,
1899
1900/* bitmap.c */
1901extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1902void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1903                                struct ext4_group_desc *gdp,
1904                                struct buffer_head *bh, int sz);
1905int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1906                                  struct ext4_group_desc *gdp,
1907                                  struct buffer_head *bh, int sz);
1908void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1909                                struct ext4_group_desc *gdp,
1910                                struct buffer_head *bh);
1911int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1912                                  struct ext4_group_desc *gdp,
1913                                  struct buffer_head *bh);
1914
1915/* balloc.c */
1916extern void ext4_validate_block_bitmap(struct super_block *sb,
1917                                       struct ext4_group_desc *desc,
1918                                       unsigned int block_group,
1919                                       struct buffer_head *bh);
1920extern unsigned int ext4_block_group(struct super_block *sb,
1921                        ext4_fsblk_t blocknr);
1922extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1923                        ext4_fsblk_t blocknr);
1924extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1925extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1926                        ext4_group_t group);
1927extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1928                                         ext4_fsblk_t goal,
1929                                         unsigned int flags,
1930                                         unsigned long *count,
1931                                         int *errp);
1932extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1933                                    s64 nclusters, unsigned int flags);
1934extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1935extern void ext4_check_blocks_bitmap(struct super_block *);
1936extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1937                                                    ext4_group_t block_group,
1938                                                    struct buffer_head ** bh);
1939extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1940
1941extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1942                                                ext4_group_t block_group);
1943extern int ext4_wait_block_bitmap(struct super_block *sb,
1944                                  ext4_group_t block_group,
1945                                  struct buffer_head *bh);
1946extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1947                                                  ext4_group_t block_group);
1948extern void ext4_init_block_bitmap(struct super_block *sb,
1949                                   struct buffer_head *bh,
1950                                   ext4_group_t group,
1951                                   struct ext4_group_desc *desc);
1952extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1953                                              ext4_group_t block_group,
1954                                              struct ext4_group_desc *gdp);
1955extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1956                                           ext4_group_t block_group,
1957                                           struct ext4_group_desc *gdp);
1958ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1959
1960/* dir.c */
1961extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1962                                  struct file *,
1963                                  struct ext4_dir_entry_2 *,
1964                                  struct buffer_head *, char *, int,
1965                                  unsigned int);
1966#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)      \
1967        unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1968                                        (de), (bh), (buf), (size), (offset)))
1969extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1970                                    __u32 minor_hash,
1971                                    struct ext4_dir_entry_2 *dirent);
1972extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1973extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1974                             struct buffer_head *bh,
1975                             void *buf, int buf_size,
1976                             const char *name, int namelen,
1977                             struct ext4_dir_entry_2 **dest_de);
1978void ext4_insert_dentry(struct inode *inode,
1979                        struct ext4_dir_entry_2 *de,
1980                        int buf_size,
1981                        const char *name, int namelen);
1982static inline void ext4_update_dx_flag(struct inode *inode)
1983{
1984        if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1985                                     EXT4_FEATURE_COMPAT_DIR_INDEX))
1986                ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1987}
1988static unsigned char ext4_filetype_table[] = {
1989        DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
1990};
1991
1992static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
1993{
1994        if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
1995            (filetype >= EXT4_FT_MAX))
1996                return DT_UNKNOWN;
1997
1998        return ext4_filetype_table[filetype];
1999}
2000
2001/* fsync.c */
2002extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2003extern int ext4_flush_unwritten_io(struct inode *);
2004
2005/* hash.c */
2006extern int ext4fs_dirhash(const char *name, int len, struct
2007                          dx_hash_info *hinfo);
2008
2009/* ialloc.c */
2010extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t,
2011                                    const struct qstr *qstr, __u32 goal,
2012                                    uid_t *owner);
2013extern void ext4_free_inode(handle_t *, struct inode *);
2014extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2015extern unsigned long ext4_count_free_inodes(struct super_block *);
2016extern unsigned long ext4_count_dirs(struct super_block *);
2017extern void ext4_check_inodes_bitmap(struct super_block *);
2018extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2019extern int ext4_init_inode_table(struct super_block *sb,
2020                                 ext4_group_t group, int barrier);
2021extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2022
2023/* mballoc.c */
2024extern long ext4_mb_stats;
2025extern long ext4_mb_max_to_scan;
2026extern int ext4_mb_init(struct super_block *);
2027extern int ext4_mb_release(struct super_block *);
2028extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2029                                struct ext4_allocation_request *, int *);
2030extern int ext4_mb_reserve_blocks(struct super_block *, int);
2031extern void ext4_discard_preallocations(struct inode *);
2032extern int __init ext4_init_mballoc(void);
2033extern void ext4_exit_mballoc(void);
2034extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2035                             struct buffer_head *bh, ext4_fsblk_t block,
2036                             unsigned long count, int flags);
2037extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2038                                   ext4_group_t ngroups);
2039extern int ext4_mb_add_groupinfo(struct super_block *sb,
2040                ext4_group_t i, struct ext4_group_desc *desc);
2041extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2042                                ext4_fsblk_t block, unsigned long count);
2043extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2044
2045/* inode.c */
2046struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2047                                                ext4_lblk_t, int, int *);
2048struct buffer_head *ext4_bread(handle_t *, struct inode *,
2049                                                ext4_lblk_t, int, int *);
2050int ext4_get_block_write(struct inode *inode, sector_t iblock,
2051                         struct buffer_head *bh_result, int create);
2052int ext4_get_block(struct inode *inode, sector_t iblock,
2053                                struct buffer_head *bh_result, int create);
2054int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2055                           struct buffer_head *bh, int create);
2056int ext4_walk_page_buffers(handle_t *handle,
2057                           struct buffer_head *head,
2058                           unsigned from,
2059                           unsigned to,
2060                           int *partial,
2061                           int (*fn)(handle_t *handle,
2062                                     struct buffer_head *bh));
2063int do_journal_get_write_access(handle_t *handle,
2064                                struct buffer_head *bh);
2065#define FALL_BACK_TO_NONDELALLOC 1
2066#define CONVERT_INLINE_DATA      2
2067
2068extern struct inode *ext4_iget(struct super_block *, unsigned long);
2069extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2070extern int  ext4_setattr(struct dentry *, struct iattr *);
2071extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2072                                struct kstat *stat);
2073extern void ext4_evict_inode(struct inode *);
2074extern void ext4_clear_inode(struct inode *);
2075extern int  ext4_sync_inode(handle_t *, struct inode *);
2076extern void ext4_dirty_inode(struct inode *, int);
2077extern int ext4_change_inode_journal_flag(struct inode *, int);
2078extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2079extern int ext4_can_truncate(struct inode *inode);
2080extern void ext4_truncate(struct inode *);
2081extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2082extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2083extern void ext4_set_inode_flags(struct inode *);
2084extern void ext4_get_inode_flags(struct ext4_inode_info *);
2085extern int ext4_alloc_da_blocks(struct inode *inode);
2086extern void ext4_set_aops(struct inode *inode);
2087extern int ext4_writepage_trans_blocks(struct inode *);
2088extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2089extern int ext4_discard_partial_page_buffers(handle_t *handle,
2090                struct address_space *mapping, loff_t from,
2091                loff_t length, int flags);
2092extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2093extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2094extern void ext4_da_update_reserve_space(struct inode *inode,
2095                                        int used, int quota_claim);
2096
2097/* indirect.c */
2098extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2099                                struct ext4_map_blocks *map, int flags);
2100extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2101                                const struct iovec *iov, loff_t offset,
2102                                unsigned long nr_segs);
2103extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2104extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2105extern void ext4_ind_truncate(struct inode *inode);
2106
2107/* ioctl.c */
2108extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2109extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2110
2111/* migrate.c */
2112extern int ext4_ext_migrate(struct inode *);
2113
2114/* namei.c */
2115extern int ext4_dirent_csum_verify(struct inode *inode,
2116                                   struct ext4_dir_entry *dirent);
2117extern int ext4_orphan_add(handle_t *, struct inode *);
2118extern int ext4_orphan_del(handle_t *, struct inode *);
2119extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2120                                __u32 start_minor_hash, __u32 *next_hash);
2121extern int search_dir(struct buffer_head *bh,
2122                      char *search_buf,
2123                      int buf_size,
2124                      struct inode *dir,
2125                      const struct qstr *d_name,
2126                      unsigned int offset,
2127                      struct ext4_dir_entry_2 **res_dir);
2128extern int ext4_generic_delete_entry(handle_t *handle,
2129                                     struct inode *dir,
2130                                     struct ext4_dir_entry_2 *de_del,
2131                                     struct buffer_head *bh,
2132                                     void *entry_buf,
2133                                     int buf_size,
2134                                     int csum_size);
2135
2136/* resize.c */
2137extern int ext4_group_add(struct super_block *sb,
2138                                struct ext4_new_group_data *input);
2139extern int ext4_group_extend(struct super_block *sb,
2140                                struct ext4_super_block *es,
2141                                ext4_fsblk_t n_blocks_count);
2142extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2143
2144/* super.c */
2145extern int ext4_calculate_overhead(struct super_block *sb);
2146extern int ext4_superblock_csum_verify(struct super_block *sb,
2147                                       struct ext4_super_block *es);
2148extern void ext4_superblock_csum_set(struct super_block *sb);
2149extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2150extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2151extern void ext4_kvfree(void *ptr);
2152extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2153                                    ext4_group_t ngroup);
2154extern __printf(4, 5)
2155void __ext4_error(struct super_block *, const char *, unsigned int,
2156                  const char *, ...);
2157#define ext4_error(sb, message...)      __ext4_error(sb, __func__,      \
2158                                                     __LINE__, ## message)
2159extern __printf(5, 6)
2160void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2161                      const char *, ...);
2162extern __printf(5, 6)
2163void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2164                     const char *, ...);
2165extern void __ext4_std_error(struct super_block *, const char *,
2166                             unsigned int, int);
2167extern __printf(4, 5)
2168void __ext4_abort(struct super_block *, const char *, unsigned int,
2169                  const char *, ...);
2170#define ext4_abort(sb, message...)      __ext4_abort(sb, __func__, \
2171                                                       __LINE__, ## message)
2172extern __printf(4, 5)
2173void __ext4_warning(struct super_block *, const char *, unsigned int,
2174                    const char *, ...);
2175#define ext4_warning(sb, message...)    __ext4_warning(sb, __func__, \
2176                                                       __LINE__, ## message)
2177extern __printf(3, 4)
2178void ext4_msg(struct super_block *, const char *, const char *, ...);
2179extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2180                           const char *, unsigned int, const char *);
2181#define dump_mmp_msg(sb, mmp, msg)      __dump_mmp_msg(sb, mmp, __func__, \
2182                                                       __LINE__, msg)
2183extern __printf(7, 8)
2184void __ext4_grp_locked_error(const char *, unsigned int,
2185                             struct super_block *, ext4_group_t,
2186                             unsigned long, ext4_fsblk_t,
2187                             const char *, ...);
2188#define ext4_grp_locked_error(sb, grp, message...) \
2189        __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2190extern void ext4_update_dynamic_rev(struct super_block *sb);
2191extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2192                                        __u32 compat);
2193extern int ext4_update_rocompat_feature(handle_t *handle,
2194                                        struct super_block *sb, __u32 rocompat);
2195extern int ext4_update_incompat_feature(handle_t *handle,
2196                                        struct super_block *sb, __u32 incompat);
2197extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2198                                      struct ext4_group_desc *bg);
2199extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2200                                      struct ext4_group_desc *bg);
2201extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2202                                     struct ext4_group_desc *bg);
2203extern __u32 ext4_free_group_clusters(struct super_block *sb,
2204                                      struct ext4_group_desc *bg);
2205extern __u32 ext4_free_inodes_count(struct super_block *sb,
2206                                 struct ext4_group_desc *bg);
2207extern __u32 ext4_used_dirs_count(struct super_block *sb,
2208                                struct ext4_group_desc *bg);
2209extern __u32 ext4_itable_unused_count(struct super_block *sb,
2210                                   struct ext4_group_desc *bg);
2211extern void ext4_block_bitmap_set(struct super_block *sb,
2212                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2213extern void ext4_inode_bitmap_set(struct super_block *sb,
2214                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2215extern void ext4_inode_table_set(struct super_block *sb,
2216                                 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2217extern void ext4_free_group_clusters_set(struct super_block *sb,
2218                                         struct ext4_group_desc *bg,
2219                                         __u32 count);
2220extern void ext4_free_inodes_set(struct super_block *sb,
2221                                struct ext4_group_desc *bg, __u32 count);
2222extern void ext4_used_dirs_set(struct super_block *sb,
2223                                struct ext4_group_desc *bg, __u32 count);
2224extern void ext4_itable_unused_set(struct super_block *sb,
2225                                   struct ext4_group_desc *bg, __u32 count);
2226extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2227                                       struct ext4_group_desc *gdp);
2228extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2229                                     struct ext4_group_desc *gdp);
2230
2231static inline int ext4_has_group_desc_csum(struct super_block *sb)
2232{
2233        return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2234                                          EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2235                                          EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2236}
2237
2238static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2239{
2240        return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2241                le32_to_cpu(es->s_blocks_count_lo);
2242}
2243
2244static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2245{
2246        return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2247                le32_to_cpu(es->s_r_blocks_count_lo);
2248}
2249
2250static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2251{
2252        return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2253                le32_to_cpu(es->s_free_blocks_count_lo);
2254}
2255
2256static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2257                                         ext4_fsblk_t blk)
2258{
2259        es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2260        es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2261}
2262
2263static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2264                                              ext4_fsblk_t blk)
2265{
2266        es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2267        es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2268}
2269
2270static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2271                                           ext4_fsblk_t blk)
2272{
2273        es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2274        es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2275}
2276
2277static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2278{
2279        if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2280                return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2281                        le32_to_cpu(raw_inode->i_size_lo);
2282        else
2283                return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2284}
2285
2286static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2287{
2288        raw_inode->i_size_lo = cpu_to_le32(i_size);
2289        raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2290}
2291
2292static inline
2293struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2294                                            ext4_group_t group)
2295{
2296         struct ext4_group_info ***grp_info;
2297         long indexv, indexh;
2298         grp_info = EXT4_SB(sb)->s_group_info;
2299         indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2300         indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2301         return grp_info[indexv][indexh];
2302}
2303
2304/*
2305 * Reading s_groups_count requires using smp_rmb() afterwards.  See
2306 * the locking protocol documented in the comments of ext4_group_add()
2307 * in resize.c
2308 */
2309static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2310{
2311        ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
2312
2313        smp_rmb();
2314        return ngroups;
2315}
2316
2317static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2318                                             ext4_group_t block_group)
2319{
2320        return block_group >> sbi->s_log_groups_per_flex;
2321}
2322
2323static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2324{
2325        return 1 << sbi->s_log_groups_per_flex;
2326}
2327
2328#define ext4_std_error(sb, errno)                               \
2329do {                                                            \
2330        if ((errno))                                            \
2331                __ext4_std_error((sb), __func__, __LINE__, (errno));    \
2332} while (0)
2333
2334#ifdef CONFIG_SMP
2335/* Each CPU can accumulate percpu_counter_batch clusters in their local
2336 * counters. So we need to make sure we have free clusters more
2337 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2338 */
2339#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2340#else
2341#define EXT4_FREECLUSTERS_WATERMARK 0
2342#endif
2343
2344static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2345{
2346        /*
2347         * XXX: replace with spinlock if seen contended -bzzz
2348         */
2349        down_write(&EXT4_I(inode)->i_data_sem);
2350        if (newsize > EXT4_I(inode)->i_disksize)
2351                EXT4_I(inode)->i_disksize = newsize;
2352        up_write(&EXT4_I(inode)->i_data_sem);
2353        return ;
2354}
2355
2356struct ext4_group_info {
2357        unsigned long   bb_state;
2358        struct rb_root  bb_free_root;
2359        ext4_grpblk_t   bb_first_free;  /* first free block */
2360        ext4_grpblk_t   bb_free;        /* total free blocks */
2361        ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
2362        ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
2363        struct          list_head bb_prealloc_list;
2364#ifdef DOUBLE_CHECK
2365        void            *bb_bitmap;
2366#endif
2367        struct rw_semaphore alloc_sem;
2368        ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
2369                                         * regions, index is order.
2370                                         * bb_counters[3] = 5 means
2371                                         * 5 free 8-block regions. */
2372};
2373
2374#define EXT4_GROUP_INFO_NEED_INIT_BIT           0
2375#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT         1
2376
2377#define EXT4_MB_GRP_NEED_INIT(grp)      \
2378        (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2379
2380#define EXT4_MB_GRP_WAS_TRIMMED(grp)    \
2381        (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2382#define EXT4_MB_GRP_SET_TRIMMED(grp)    \
2383        (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2384#define EXT4_MB_GRP_CLEAR_TRIMMED(grp)  \
2385        (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2386
2387#define EXT4_MAX_CONTENTION             8
2388#define EXT4_CONTENTION_THRESHOLD       2
2389
2390static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2391                                              ext4_group_t group)
2392{
2393        return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2394}
2395
2396/*
2397 * Returns true if the filesystem is busy enough that attempts to
2398 * access the block group locks has run into contention.
2399 */
2400static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2401{
2402        return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2403}
2404
2405static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2406{
2407        spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2408        if (spin_trylock(lock))
2409                /*
2410                 * We're able to grab the lock right away, so drop the
2411                 * lock contention counter.
2412                 */
2413                atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2414        else {
2415                /*
2416                 * The lock is busy, so bump the contention counter,
2417                 * and then wait on the spin lock.
2418                 */
2419                atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2420                                  EXT4_MAX_CONTENTION);
2421                spin_lock(lock);
2422        }
2423}
2424
2425static inline void ext4_unlock_group(struct super_block *sb,
2426                                        ext4_group_t group)
2427{
2428        spin_unlock(ext4_group_lock_ptr(sb, group));
2429}
2430
2431/*
2432 * Block validity checking
2433 */
2434#define ext4_check_indirect_blockref(inode, bh)                         \
2435        ext4_check_blockref(__func__, __LINE__, inode,                  \
2436                            (__le32 *)(bh)->b_data,                     \
2437                            EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2438
2439#define ext4_ind_check_inode(inode)                                     \
2440        ext4_check_blockref(__func__, __LINE__, inode,                  \
2441                            EXT4_I(inode)->i_data,                      \
2442                            EXT4_NDIR_BLOCKS)
2443
2444/*
2445 * Inodes and files operations
2446 */
2447
2448/* dir.c */
2449extern const struct file_operations ext4_dir_operations;
2450
2451/* file.c */
2452extern const struct inode_operations ext4_file_inode_operations;
2453extern const struct file_operations ext4_file_operations;
2454extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2455extern void ext4_unwritten_wait(struct inode *inode);
2456
2457/* namei.c */
2458extern const struct inode_operations ext4_dir_inode_operations;
2459extern const struct inode_operations ext4_special_inode_operations;
2460extern struct dentry *ext4_get_parent(struct dentry *child);
2461extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2462                                 struct ext4_dir_entry_2 *de,
2463                                 int blocksize, int csum_size,
2464                                 unsigned int parent_ino, int dotdot_real_len);
2465extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2466                                   unsigned int blocksize);
2467extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2468                                         struct inode *inode,
2469                                         struct buffer_head *bh);
2470
2471/* symlink.c */
2472extern const struct inode_operations ext4_symlink_inode_operations;
2473extern const struct inode_operations ext4_fast_symlink_inode_operations;
2474
2475/* block_validity */
2476extern void ext4_release_system_zone(struct super_block *sb);
2477extern int ext4_setup_system_zone(struct super_block *sb);
2478extern int __init ext4_init_system_zone(void);
2479extern void ext4_exit_system_zone(void);
2480extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2481                                 ext4_fsblk_t start_blk,
2482                                 unsigned int count);
2483extern int ext4_check_blockref(const char *, unsigned int,
2484                               struct inode *, __le32 *, unsigned int);
2485
2486/* extents.c */
2487struct ext4_ext_path;
2488struct ext4_extent;
2489
2490extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2491extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2492extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2493                                       int chunk);
2494extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2495                               struct ext4_map_blocks *map, int flags);
2496extern void ext4_ext_truncate(struct inode *);
2497extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2498                                loff_t length);
2499extern void ext4_ext_init(struct super_block *);
2500extern void ext4_ext_release(struct super_block *);
2501extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2502                          loff_t len);
2503extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2504                          ssize_t len);
2505extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2506                           struct ext4_map_blocks *map, int flags);
2507extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2508                                         ext4_lblk_t lblocks);
2509extern int ext4_extent_tree_init(handle_t *, struct inode *);
2510extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2511                                                   int num,
2512                                                   struct ext4_ext_path *path);
2513extern int ext4_can_extents_be_merged(struct inode *inode,
2514                                      struct ext4_extent *ex1,
2515                                      struct ext4_extent *ex2);
2516extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2517                                  struct ext4_ext_path *,
2518                                  struct ext4_extent *, int);
2519extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2520                                                  struct ext4_ext_path *);
2521extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2522extern int ext4_ext_check_inode(struct inode *inode);
2523extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2524extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2525                        __u64 start, __u64 len);
2526
2527
2528/* move_extent.c */
2529extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2530                             __u64 start_orig, __u64 start_donor,
2531                             __u64 len, __u64 *moved_len);
2532
2533/* page-io.c */
2534extern int __init ext4_init_pageio(void);
2535extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2536extern void ext4_exit_pageio(void);
2537extern void ext4_ioend_wait(struct inode *);
2538extern void ext4_free_io_end(ext4_io_end_t *io);
2539extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2540extern void ext4_io_submit(struct ext4_io_submit *io);
2541extern int ext4_bio_write_page(struct ext4_io_submit *io,
2542                               struct page *page,
2543                               int len,
2544                               struct writeback_control *wbc);
2545
2546/* mmp.c */
2547extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2548extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2549extern int ext4_mmp_csum_verify(struct super_block *sb,
2550                                struct mmp_struct *mmp);
2551
2552/* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2553enum ext4_state_bits {
2554        BH_Uninit       /* blocks are allocated but uninitialized on disk */
2555          = BH_JBDPrivateStart,
2556        BH_AllocFromCluster,    /* allocated blocks were part of already
2557                                 * allocated cluster. Note that this flag will
2558                                 * never, ever appear in a buffer_head's state
2559                                 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2560                                 * this is used. */
2561};
2562
2563BUFFER_FNS(Uninit, uninit)
2564TAS_BUFFER_FNS(Uninit, uninit)
2565
2566/*
2567 * Add new method to test whether block and inode bitmaps are properly
2568 * initialized. With uninit_bg reading the block from disk is not enough
2569 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2570 */
2571#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2572
2573static inline int bitmap_uptodate(struct buffer_head *bh)
2574{
2575        return (buffer_uptodate(bh) &&
2576                        test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2577}
2578static inline void set_bitmap_uptodate(struct buffer_head *bh)
2579{
2580        set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2581}
2582
2583/*
2584 * Disable DIO read nolock optimization, so new dioreaders will be forced
2585 * to grab i_mutex
2586 */
2587static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2588{
2589        ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2590        smp_mb();
2591}
2592static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2593{
2594        smp_mb();
2595        ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2596}
2597
2598#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2599
2600/* For ioend & aio unwritten conversion wait queues */
2601#define EXT4_WQ_HASH_SZ         37
2602#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
2603                                            EXT4_WQ_HASH_SZ])
2604#define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
2605                                             EXT4_WQ_HASH_SZ])
2606extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2607extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2608
2609#define EXT4_RESIZING   0
2610extern int ext4_resize_begin(struct super_block *sb);
2611extern void ext4_resize_end(struct super_block *sb);
2612
2613#endif  /* __KERNEL__ */
2614
2615#endif  /* _EXT4_H */
2616