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#ifdef __KERNEL__
  33#include <linux/compat.h>
  34#endif
  35
  36/*
  37 * The fourth extended filesystem constants/structures
  38 */
  39
  40/*
  41 * Define EXT4FS_DEBUG to produce debug messages
  42 */
  43#undef EXT4FS_DEBUG
  44
  45/*
  46 * Debug code
  47 */
  48#ifdef EXT4FS_DEBUG
  49#define ext4_debug(f, a...)                                             \
  50        do {                                                            \
  51                printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
  52                        __FILE__, __LINE__, __func__);                  \
  53                printk(KERN_DEBUG f, ## a);                             \
  54        } while (0)
  55#else
  56#define ext4_debug(f, a...)     do {} while (0)
  57#endif
  58
  59#define EXT4_ERROR_INODE(inode, fmt, a...) \
  60        ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
  61
  62#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)                 \
  63        ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
  64
  65#define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
  66        ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
  67
  68/* data type for block offset of block group */
  69typedef int ext4_grpblk_t;
  70
  71/* data type for filesystem-wide blocks number */
  72typedef unsigned long long ext4_fsblk_t;
  73
  74/* data type for file logical block number */
  75typedef __u32 ext4_lblk_t;
  76
  77/* data type for block group number */
  78typedef unsigned int ext4_group_t;
  79
  80/*
  81 * Flags used in mballoc's allocation_context flags field.
  82 *
  83 * Also used to show what's going on for debugging purposes when the
  84 * flag field is exported via the traceport interface
  85 */
  86
  87/* prefer goal again. length */
  88#define EXT4_MB_HINT_MERGE              0x0001
  89/* blocks already reserved */
  90#define EXT4_MB_HINT_RESERVED           0x0002
  91/* metadata is being allocated */
  92#define EXT4_MB_HINT_METADATA           0x0004
  93/* first blocks in the file */
  94#define EXT4_MB_HINT_FIRST              0x0008
  95/* search for the best chunk */
  96#define EXT4_MB_HINT_BEST               0x0010
  97/* data is being allocated */
  98#define EXT4_MB_HINT_DATA               0x0020
  99/* don't preallocate (for tails) */
 100#define EXT4_MB_HINT_NOPREALLOC         0x0040
 101/* allocate for locality group */
 102#define EXT4_MB_HINT_GROUP_ALLOC        0x0080
 103/* allocate goal blocks or none */
 104#define EXT4_MB_HINT_GOAL_ONLY          0x0100
 105/* goal is meaningful */
 106#define EXT4_MB_HINT_TRY_GOAL           0x0200
 107/* blocks already pre-reserved by delayed allocation */
 108#define EXT4_MB_DELALLOC_RESERVED       0x0400
 109/* We are doing stream allocation */
 110#define EXT4_MB_STREAM_ALLOC            0x0800
 111/* Use reserved root blocks if needed */
 112#define EXT4_MB_USE_ROOT_BLOCKS         0x1000
 113
 114struct ext4_allocation_request {
 115        /* target inode for block we're allocating */
 116        struct inode *inode;
 117        /* how many blocks we want to allocate */
 118        unsigned int len;
 119        /* logical block in target inode */
 120        ext4_lblk_t logical;
 121        /* the closest logical allocated block to the left */
 122        ext4_lblk_t lleft;
 123        /* the closest logical allocated block to the right */
 124        ext4_lblk_t lright;
 125        /* phys. target (a hint) */
 126        ext4_fsblk_t goal;
 127        /* phys. block for the closest logical allocated block to the left */
 128        ext4_fsblk_t pleft;
 129        /* phys. block for the closest logical allocated block to the right */
 130        ext4_fsblk_t pright;
 131        /* flags. see above EXT4_MB_HINT_* */
 132        unsigned int flags;
 133};
 134
 135/*
 136 * Logical to physical block mapping, used by ext4_map_blocks()
 137 *
 138 * This structure is used to pass requests into ext4_map_blocks() as
 139 * well as to store the information returned by ext4_map_blocks().  It
 140 * takes less room on the stack than a struct buffer_head.
 141 */
 142#define EXT4_MAP_NEW            (1 << BH_New)
 143#define EXT4_MAP_MAPPED         (1 << BH_Mapped)
 144#define EXT4_MAP_UNWRITTEN      (1 << BH_Unwritten)
 145#define EXT4_MAP_BOUNDARY       (1 << BH_Boundary)
 146#define EXT4_MAP_UNINIT         (1 << BH_Uninit)
 147#define EXT4_MAP_FLAGS          (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
 148                                 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
 149                                 EXT4_MAP_UNINIT)
 150
 151struct ext4_map_blocks {
 152        ext4_fsblk_t m_pblk;
 153        ext4_lblk_t m_lblk;
 154        unsigned int m_len;
 155        unsigned int m_flags;
 156};
 157
 158/*
 159 * For delayed allocation tracking
 160 */
 161struct mpage_da_data {
 162        struct inode *inode;
 163        sector_t b_blocknr;             /* start block number of extent */
 164        size_t b_size;                  /* size of extent */
 165        unsigned long b_state;          /* state of the extent */
 166        unsigned long first_page, next_page;    /* extent of pages */
 167        struct writeback_control *wbc;
 168        int io_done;
 169        int pages_written;
 170        int retval;
 171};
 172
 173/*
 174 * Flags for ext4_io_end->flags
 175 */
 176#define EXT4_IO_END_UNWRITTEN   0x0001
 177#define EXT4_IO_END_ERROR       0x0002
 178
 179struct ext4_io_page {
 180        struct page     *p_page;
 181        atomic_t        p_count;
 182};
 183
 184#define MAX_IO_PAGES 128
 185
 186typedef struct ext4_io_end {
 187        struct list_head        list;           /* per-file finished IO list */
 188        struct inode            *inode;         /* file being written to */
 189        unsigned int            flag;           /* unwritten or not */
 190        struct page             *page;          /* page struct for buffer write */
 191        loff_t                  offset;         /* offset in the file */
 192        ssize_t                 size;           /* size of the extent */
 193        struct work_struct      work;           /* data work queue */
 194        struct kiocb            *iocb;          /* iocb struct for AIO */
 195        int                     result;         /* error value for AIO */
 196        int                     num_io_pages;
 197        struct ext4_io_page     *pages[MAX_IO_PAGES];
 198} ext4_io_end_t;
 199
 200struct ext4_io_submit {
 201        int                     io_op;
 202        struct bio              *io_bio;
 203        ext4_io_end_t           *io_end;
 204        struct ext4_io_page     *io_page;
 205        sector_t                io_next_block;
 206};
 207
 208/*
 209 * Special inodes numbers
 210 */
 211#define EXT4_BAD_INO             1      /* Bad blocks inode */
 212#define EXT4_ROOT_INO            2      /* Root inode */
 213#define EXT4_USR_QUOTA_INO       3      /* User quota inode */
 214#define EXT4_GRP_QUOTA_INO       4      /* Group quota inode */
 215#define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
 216#define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
 217#define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
 218#define EXT4_JOURNAL_INO         8      /* Journal inode */
 219
 220/* First non-reserved inode for old ext4 filesystems */
 221#define EXT4_GOOD_OLD_FIRST_INO 11
 222
 223/*
 224 * Maximal count of links to a file
 225 */
 226#define EXT4_LINK_MAX           65000
 227
 228/*
 229 * Macro-instructions used to manage several block sizes
 230 */
 231#define EXT4_MIN_BLOCK_SIZE             1024
 232#define EXT4_MAX_BLOCK_SIZE             65536
 233#define EXT4_MIN_BLOCK_LOG_SIZE         10
 234#define EXT4_MAX_BLOCK_LOG_SIZE         16
 235#ifdef __KERNEL__
 236# define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
 237#else
 238# define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
 239#endif
 240#define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
 241#ifdef __KERNEL__
 242# define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
 243#else
 244# define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
 245#endif
 246#ifdef __KERNEL__
 247#define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
 248#define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
 249#define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
 250#else
 251#define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 252                                 EXT4_GOOD_OLD_INODE_SIZE : \
 253                                 (s)->s_inode_size)
 254#define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 255                                 EXT4_GOOD_OLD_FIRST_INO : \
 256                                 (s)->s_first_ino)
 257#endif
 258#define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
 259
 260/*
 261 * Structure of a blocks group descriptor
 262 */
 263struct ext4_group_desc
 264{
 265        __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
 266        __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
 267        __le32  bg_inode_table_lo;      /* Inodes table block */
 268        __le16  bg_free_blocks_count_lo;/* Free blocks count */
 269        __le16  bg_free_inodes_count_lo;/* Free inodes count */
 270        __le16  bg_used_dirs_count_lo;  /* Directories count */
 271        __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
 272        __u32   bg_reserved[2];         /* Likely block/inode bitmap checksum */
 273        __le16  bg_itable_unused_lo;    /* Unused inodes count */
 274        __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
 275        __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
 276        __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
 277        __le32  bg_inode_table_hi;      /* Inodes table block MSB */
 278        __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
 279        __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
 280        __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
 281        __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
 282        __u32   bg_reserved2[3];
 283};
 284
 285/*
 286 * Structure of a flex block group info
 287 */
 288
 289struct flex_groups {
 290        atomic_t free_inodes;
 291        atomic_t free_blocks;
 292        atomic_t used_dirs;
 293};
 294
 295#define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
 296#define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
 297#define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
 298
 299/*
 300 * Macro-instructions used to manage group descriptors
 301 */
 302#define EXT4_MIN_DESC_SIZE              32
 303#define EXT4_MIN_DESC_SIZE_64BIT        64
 304#define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
 305#define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
 306#ifdef __KERNEL__
 307# define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
 308# define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
 309# define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
 310# define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
 311#else
 312# define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
 313# define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
 314# define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
 315#endif
 316
 317/*
 318 * Constants relative to the data blocks
 319 */
 320#define EXT4_NDIR_BLOCKS                12
 321#define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
 322#define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
 323#define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
 324#define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
 325
 326/*
 327 * Inode flags
 328 */
 329#define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
 330#define EXT4_UNRM_FL                    0x00000002 /* Undelete */
 331#define EXT4_COMPR_FL                   0x00000004 /* Compress file */
 332#define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
 333#define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
 334#define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
 335#define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
 336#define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
 337/* Reserved for compression usage... */
 338#define EXT4_DIRTY_FL                   0x00000100
 339#define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
 340#define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
 341#define EXT4_ECOMPR_FL                  0x00000800 /* Compression error */
 342/* End compression flags --- maybe not all used */
 343#define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
 344#define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
 345#define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
 346#define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
 347#define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
 348#define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
 349#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
 350#define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
 351#define EXT4_EA_INODE_FL                0x00200000 /* Inode used for large EA */
 352#define EXT4_EOFBLOCKS_FL               0x00400000 /* Blocks allocated beyond EOF */
 353#define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
 354
 355#define EXT4_FL_USER_VISIBLE            0x004BDFFF /* User visible flags */
 356#define EXT4_FL_USER_MODIFIABLE         0x004B80FF /* User modifiable flags */
 357
 358/* Flags that should be inherited by new inodes from their parent. */
 359#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
 360                           EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
 361                           EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
 362                           EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
 363                           EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
 364
 365/* Flags that are appropriate for regular files (all but dir-specific ones). */
 366#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
 367
 368/* Flags that are appropriate for non-directories/regular files. */
 369#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
 370
 371/* Mask out flags that are inappropriate for the given type of inode. */
 372static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 373{
 374        if (S_ISDIR(mode))
 375                return flags;
 376        else if (S_ISREG(mode))
 377                return flags & EXT4_REG_FLMASK;
 378        else
 379                return flags & EXT4_OTHER_FLMASK;
 380}
 381
 382/*
 383 * Inode flags used for atomic set/get
 384 */
 385enum {
 386        EXT4_INODE_SECRM        = 0,    /* Secure deletion */
 387        EXT4_INODE_UNRM         = 1,    /* Undelete */
 388        EXT4_INODE_COMPR        = 2,    /* Compress file */
 389        EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
 390        EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
 391        EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
 392        EXT4_INODE_NODUMP       = 6,    /* do not dump file */
 393        EXT4_INODE_NOATIME      = 7,    /* do not update atime */
 394/* Reserved for compression usage... */
 395        EXT4_INODE_DIRTY        = 8,
 396        EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
 397        EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
 398        EXT4_INODE_ECOMPR       = 11,   /* Compression error */
 399/* End compression flags --- maybe not all used */
 400        EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
 401        EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
 402        EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
 403        EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
 404        EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
 405        EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
 406        EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
 407        EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
 408        EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
 409        EXT4_INODE_EOFBLOCKS    = 22,   /* Blocks allocated beyond EOF */
 410        EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
 411};
 412
 413#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
 414#define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
 415        printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
 416                EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
 417
 418/*
 419 * Since it's pretty easy to mix up bit numbers and hex values, and we
 420 * can't do a compile-time test for ENUM values, we use a run-time
 421 * test to make sure that EXT4_XXX_FL is consistent with respect to
 422 * EXT4_INODE_XXX.  If all is well the printk and BUG_ON will all drop
 423 * out so it won't cost any extra space in the compiled kernel image.
 424 * But it's important that these values are the same, since we are
 425 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
 426 * must be consistent with the values of FS_XXX_FL defined in
 427 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
 428 * ext4 filesystems, and of course the values defined in e2fsprogs.
 429 *
 430 * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
 431 */
 432static inline void ext4_check_flag_values(void)
 433{
 434        CHECK_FLAG_VALUE(SECRM);
 435        CHECK_FLAG_VALUE(UNRM);
 436        CHECK_FLAG_VALUE(COMPR);
 437        CHECK_FLAG_VALUE(SYNC);
 438        CHECK_FLAG_VALUE(IMMUTABLE);
 439        CHECK_FLAG_VALUE(APPEND);
 440        CHECK_FLAG_VALUE(NODUMP);
 441        CHECK_FLAG_VALUE(NOATIME);
 442        CHECK_FLAG_VALUE(DIRTY);
 443        CHECK_FLAG_VALUE(COMPRBLK);
 444        CHECK_FLAG_VALUE(NOCOMPR);
 445        CHECK_FLAG_VALUE(ECOMPR);
 446        CHECK_FLAG_VALUE(INDEX);
 447        CHECK_FLAG_VALUE(IMAGIC);
 448        CHECK_FLAG_VALUE(JOURNAL_DATA);
 449        CHECK_FLAG_VALUE(NOTAIL);
 450        CHECK_FLAG_VALUE(DIRSYNC);
 451        CHECK_FLAG_VALUE(TOPDIR);
 452        CHECK_FLAG_VALUE(HUGE_FILE);
 453        CHECK_FLAG_VALUE(EXTENTS);
 454        CHECK_FLAG_VALUE(EA_INODE);
 455        CHECK_FLAG_VALUE(EOFBLOCKS);
 456        CHECK_FLAG_VALUE(RESERVED);
 457}
 458
 459/* Used to pass group descriptor data when online resize is done */
 460struct ext4_new_group_input {
 461        __u32 group;            /* Group number for this data */
 462        __u64 block_bitmap;     /* Absolute block number of block bitmap */
 463        __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
 464        __u64 inode_table;      /* Absolute block number of inode table start */
 465        __u32 blocks_count;     /* Total number of blocks in this group */
 466        __u16 reserved_blocks;  /* Number of reserved blocks in this group */
 467        __u16 unused;
 468};
 469
 470#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 471struct compat_ext4_new_group_input {
 472        u32 group;
 473        compat_u64 block_bitmap;
 474        compat_u64 inode_bitmap;
 475        compat_u64 inode_table;
 476        u32 blocks_count;
 477        u16 reserved_blocks;
 478        u16 unused;
 479};
 480#endif
 481
 482/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
 483struct ext4_new_group_data {
 484        __u32 group;
 485        __u64 block_bitmap;
 486        __u64 inode_bitmap;
 487        __u64 inode_table;
 488        __u32 blocks_count;
 489        __u16 reserved_blocks;
 490        __u16 unused;
 491        __u32 free_blocks_count;
 492};
 493
 494/*
 495 * Flags used by ext4_map_blocks()
 496 */
 497        /* Allocate any needed blocks and/or convert an unitialized
 498           extent to be an initialized ext4 */
 499#define EXT4_GET_BLOCKS_CREATE                  0x0001
 500        /* Request the creation of an unitialized extent */
 501#define EXT4_GET_BLOCKS_UNINIT_EXT              0x0002
 502#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT       (EXT4_GET_BLOCKS_UNINIT_EXT|\
 503                                                 EXT4_GET_BLOCKS_CREATE)
 504        /* Caller is from the delayed allocation writeout path,
 505           so set the magic i_delalloc_reserve_flag after taking the
 506           inode allocation semaphore for */
 507#define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
 508        /* caller is from the direct IO path, request to creation of an
 509        unitialized extents if not allocated, split the uninitialized
 510        extent if blocks has been preallocated already*/
 511#define EXT4_GET_BLOCKS_PRE_IO                  0x0008
 512#define EXT4_GET_BLOCKS_CONVERT                 0x0010
 513#define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
 514                                         EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
 515        /* Convert extent to initialized after IO complete */
 516#define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
 517                                         EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
 518        /* Punch out blocks of an extent */
 519#define EXT4_GET_BLOCKS_PUNCH_OUT_EXT           0x0020
 520        /* Don't normalize allocation size (used for fallocate) */
 521#define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
 522
 523/*
 524 * Flags used by ext4_free_blocks
 525 */
 526#define EXT4_FREE_BLOCKS_METADATA       0x0001
 527#define EXT4_FREE_BLOCKS_FORGET         0x0002
 528#define EXT4_FREE_BLOCKS_VALIDATED      0x0004
 529
 530/*
 531 * ioctl commands
 532 */
 533#define EXT4_IOC_GETFLAGS               FS_IOC_GETFLAGS
 534#define EXT4_IOC_SETFLAGS               FS_IOC_SETFLAGS
 535#define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
 536#define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
 537#define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
 538#define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
 539#ifdef CONFIG_JBD2_DEBUG
 540#define EXT4_IOC_WAIT_FOR_READONLY      _IOR('f', 99, long)
 541#endif
 542#define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
 543#define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
 544#define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
 545#define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
 546#define EXT4_IOC_MIGRATE                _IO('f', 9)
 547 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
 548 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
 549#define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
 550#define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
 551
 552#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 553/*
 554 * ioctl commands in 32 bit emulation
 555 */
 556#define EXT4_IOC32_GETFLAGS             FS_IOC32_GETFLAGS
 557#define EXT4_IOC32_SETFLAGS             FS_IOC32_SETFLAGS
 558#define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
 559#define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
 560#define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
 561#define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
 562#define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
 563#define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
 564#ifdef CONFIG_JBD2_DEBUG
 565#define EXT4_IOC32_WAIT_FOR_READONLY    _IOR('f', 99, int)
 566#endif
 567#define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
 568#define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
 569#endif
 570
 571/* Max physical block we can address w/o extents */
 572#define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
 573
 574/*
 575 * Structure of an inode on the disk
 576 */
 577struct ext4_inode {
 578        __le16  i_mode;         /* File mode */
 579        __le16  i_uid;          /* Low 16 bits of Owner Uid */
 580        __le32  i_size_lo;      /* Size in bytes */
 581        __le32  i_atime;        /* Access time */
 582        __le32  i_ctime;        /* Inode Change time */
 583        __le32  i_mtime;        /* Modification time */
 584        __le32  i_dtime;        /* Deletion Time */
 585        __le16  i_gid;          /* Low 16 bits of Group Id */
 586        __le16  i_links_count;  /* Links count */
 587        __le32  i_blocks_lo;    /* Blocks count */
 588        __le32  i_flags;        /* File flags */
 589        union {
 590                struct {
 591                        __le32  l_i_version;
 592                } linux1;
 593                struct {
 594                        __u32  h_i_translator;
 595                } hurd1;
 596                struct {
 597                        __u32  m_i_reserved1;
 598                } masix1;
 599        } osd1;                         /* OS dependent 1 */
 600        __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
 601        __le32  i_generation;   /* File version (for NFS) */
 602        __le32  i_file_acl_lo;  /* File ACL */
 603        __le32  i_size_high;
 604        __le32  i_obso_faddr;   /* Obsoleted fragment address */
 605        union {
 606                struct {
 607                        __le16  l_i_blocks_high; /* were l_i_reserved1 */
 608                        __le16  l_i_file_acl_high;
 609                        __le16  l_i_uid_high;   /* these 2 fields */
 610                        __le16  l_i_gid_high;   /* were reserved2[0] */
 611                        __u32   l_i_reserved2;
 612                } linux2;
 613                struct {
 614                        __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 615                        __u16   h_i_mode_high;
 616                        __u16   h_i_uid_high;
 617                        __u16   h_i_gid_high;
 618                        __u32   h_i_author;
 619                } hurd2;
 620                struct {
 621                        __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 622                        __le16  m_i_file_acl_high;
 623                        __u32   m_i_reserved2[2];
 624                } masix2;
 625        } osd2;                         /* OS dependent 2 */
 626        __le16  i_extra_isize;
 627        __le16  i_pad1;
 628        __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
 629        __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
 630        __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
 631        __le32  i_crtime;       /* File Creation time */
 632        __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
 633        __le32  i_version_hi;   /* high 32 bits for 64-bit version */
 634};
 635
 636struct move_extent {
 637        __u32 reserved;         /* should be zero */
 638        __u32 donor_fd;         /* donor file descriptor */
 639        __u64 orig_start;       /* logical start offset in block for orig */
 640        __u64 donor_start;      /* logical start offset in block for donor */
 641        __u64 len;              /* block length to be moved */
 642        __u64 moved_len;        /* moved block length */
 643};
 644
 645#define EXT4_EPOCH_BITS 2
 646#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
 647#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
 648
 649/*
 650 * Extended fields will fit into an inode if the filesystem was formatted
 651 * with large inodes (-I 256 or larger) and there are not currently any EAs
 652 * consuming all of the available space. For new inodes we always reserve
 653 * enough space for the kernel's known extended fields, but for inodes
 654 * created with an old kernel this might not have been the case. None of
 655 * the extended inode fields is critical for correct filesystem operation.
 656 * This macro checks if a certain field fits in the inode. Note that
 657 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
 658 */
 659#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
 660        ((offsetof(typeof(*ext4_inode), field) +        \
 661          sizeof((ext4_inode)->field))                  \
 662        <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
 663            (einode)->i_extra_isize))                   \
 664
 665static inline __le32 ext4_encode_extra_time(struct timespec *time)
 666{
 667       return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
 668                           (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
 669                          ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
 670}
 671
 672static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
 673{
 674       if (sizeof(time->tv_sec) > 4)
 675               time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
 676                               << 32;
 677       time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 678}
 679
 680#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                          \
 681do {                                                                           \
 682        (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);               \
 683        if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 684                (raw_inode)->xtime ## _extra =                                 \
 685                                ext4_encode_extra_time(&(inode)->xtime);       \
 686} while (0)
 687
 688#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
 689do {                                                                           \
 690        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 691                (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
 692        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 693                (raw_inode)->xtime ## _extra =                                 \
 694                                ext4_encode_extra_time(&(einode)->xtime);      \
 695} while (0)
 696
 697#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                          \
 698do {                                                                           \
 699        (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
 700        if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 701                ext4_decode_extra_time(&(inode)->xtime,                        \
 702                                       raw_inode->xtime ## _extra);            \
 703        else                                                                   \
 704                (inode)->xtime.tv_nsec = 0;                                    \
 705} while (0)
 706
 707#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
 708do {                                                                           \
 709        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 710                (einode)->xtime.tv_sec =                                       \
 711                        (signed)le32_to_cpu((raw_inode)->xtime);               \
 712        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 713                ext4_decode_extra_time(&(einode)->xtime,                       \
 714                                       raw_inode->xtime ## _extra);            \
 715        else                                                                   \
 716                (einode)->xtime.tv_nsec = 0;                                   \
 717} while (0)
 718
 719#define i_disk_version osd1.linux1.l_i_version
 720
 721#if defined(__KERNEL__) || defined(__linux__)
 722#define i_reserved1     osd1.linux1.l_i_reserved1
 723#define i_file_acl_high osd2.linux2.l_i_file_acl_high
 724#define i_blocks_high   osd2.linux2.l_i_blocks_high
 725#define i_uid_low       i_uid
 726#define i_gid_low       i_gid
 727#define i_uid_high      osd2.linux2.l_i_uid_high
 728#define i_gid_high      osd2.linux2.l_i_gid_high
 729#define i_reserved2     osd2.linux2.l_i_reserved2
 730
 731#elif defined(__GNU__)
 732
 733#define i_translator    osd1.hurd1.h_i_translator
 734#define i_uid_high      osd2.hurd2.h_i_uid_high
 735#define i_gid_high      osd2.hurd2.h_i_gid_high
 736#define i_author        osd2.hurd2.h_i_author
 737
 738#elif defined(__masix__)
 739
 740#define i_reserved1     osd1.masix1.m_i_reserved1
 741#define i_file_acl_high osd2.masix2.m_i_file_acl_high
 742#define i_reserved2     osd2.masix2.m_i_reserved2
 743
 744#endif /* defined(__KERNEL__) || defined(__linux__) */
 745
 746/*
 747 * storage for cached extent
 748 * If ec_len == 0, then the cache is invalid.
 749 * If ec_start == 0, then the cache represents a gap (null mapping)
 750 */
 751struct ext4_ext_cache {
 752        ext4_fsblk_t    ec_start;
 753        ext4_lblk_t     ec_block;
 754        __u32           ec_len; /* must be 32bit to return holes */
 755};
 756
 757/*
 758 * fourth extended file system inode data in memory
 759 */
 760struct ext4_inode_info {
 761        __le32  i_data[15];     /* unconverted */
 762        __u32   i_dtime;
 763        ext4_fsblk_t    i_file_acl;
 764
 765        /*
 766         * i_block_group is the number of the block group which contains
 767         * this file's inode.  Constant across the lifetime of the inode,
 768         * it is ued for making block allocation decisions - we try to
 769         * place a file's data blocks near its inode block, and new inodes
 770         * near to their parent directory's inode.
 771         */
 772        ext4_group_t    i_block_group;
 773        ext4_lblk_t     i_dir_start_lookup;
 774#if (BITS_PER_LONG < 64)
 775        unsigned long   i_state_flags;          /* Dynamic state flags */
 776#endif
 777        unsigned long   i_flags;
 778
 779#ifdef CONFIG_EXT4_FS_XATTR
 780        /*
 781         * Extended attributes can be read independently of the main file
 782         * data. Taking i_mutex even when reading would cause contention
 783         * between readers of EAs and writers of regular file data, so
 784         * instead we synchronize on xattr_sem when reading or changing
 785         * EAs.
 786         */
 787        struct rw_semaphore xattr_sem;
 788#endif
 789
 790        struct list_head i_orphan;      /* unlinked but open inodes */
 791
 792        /*
 793         * i_disksize keeps track of what the inode size is ON DISK, not
 794         * in memory.  During truncate, i_size is set to the new size by
 795         * the VFS prior to calling ext4_truncate(), but the filesystem won't
 796         * set i_disksize to 0 until the truncate is actually under way.
 797         *
 798         * The intent is that i_disksize always represents the blocks which
 799         * are used by this file.  This allows recovery to restart truncate
 800         * on orphans if we crash during truncate.  We actually write i_disksize
 801         * into the on-disk inode when writing inodes out, instead of i_size.
 802         *
 803         * The only time when i_disksize and i_size may be different is when
 804         * a truncate is in progress.  The only things which change i_disksize
 805         * are ext4_get_block (growth) and ext4_truncate (shrinkth).
 806         */
 807        loff_t  i_disksize;
 808
 809        /*
 810         * i_data_sem is for serialising ext4_truncate() against
 811         * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
 812         * data tree are chopped off during truncate. We can't do that in
 813         * ext4 because whenever we perform intermediate commits during
 814         * truncate, the inode and all the metadata blocks *must* be in a
 815         * consistent state which allows truncation of the orphans to restart
 816         * during recovery.  Hence we must fix the get_block-vs-truncate race
 817         * by other means, so we have i_data_sem.
 818         */
 819        struct rw_semaphore i_data_sem;
 820        struct inode vfs_inode;
 821        struct jbd2_inode *jinode;
 822
 823        struct ext4_ext_cache i_cached_extent;
 824        /*
 825         * File creation time. Its function is same as that of
 826         * struct timespec i_{a,c,m}time in the generic inode.
 827         */
 828        struct timespec i_crtime;
 829
 830        /* mballoc */
 831        struct list_head i_prealloc_list;
 832        spinlock_t i_prealloc_lock;
 833
 834        /* ialloc */
 835        ext4_group_t    i_last_alloc_group;
 836
 837        /* allocation reservation info for delalloc */
 838        unsigned int i_reserved_data_blocks;
 839        unsigned int i_reserved_meta_blocks;
 840        unsigned int i_allocated_meta_blocks;
 841        ext4_lblk_t i_da_metadata_calc_last_lblock;
 842        int i_da_metadata_calc_len;
 843
 844        /* on-disk additional length */
 845        __u16 i_extra_isize;
 846
 847#ifdef CONFIG_QUOTA
 848        /* quota space reservation, managed internally by quota code */
 849        qsize_t i_reserved_quota;
 850#endif
 851
 852        /* completed IOs that might need unwritten extents handling */
 853        struct list_head i_completed_io_list;
 854        spinlock_t i_completed_io_lock;
 855        atomic_t i_ioend_count; /* Number of outstanding io_end structs */
 856        /* current io_end structure for async DIO write*/
 857        ext4_io_end_t *cur_aio_dio;
 858        atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
 859
 860        spinlock_t i_block_reservation_lock;
 861
 862        /*
 863         * Transactions that contain inode's metadata needed to complete
 864         * fsync and fdatasync, respectively.
 865         */
 866        tid_t i_sync_tid;
 867        tid_t i_datasync_tid;
 868};
 869
 870/*
 871 * File system states
 872 */
 873#define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
 874#define EXT4_ERROR_FS                   0x0002  /* Errors detected */
 875#define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
 876
 877/*
 878 * Misc. filesystem flags
 879 */
 880#define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
 881#define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
 882#define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
 883
 884/*
 885 * Mount flags
 886 */
 887#define EXT4_MOUNT_OLDALLOC             0x00002  /* Don't use the new Orlov allocator */
 888#define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
 889#define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
 890#define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
 891#define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
 892#define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
 893#define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
 894#define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
 895#define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
 896#define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
 897#define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
 898#define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
 899#define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
 900#define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
 901#define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
 902#define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
 903#define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
 904#define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
 905#define EXT4_MOUNT_QUOTA                0x80000 /* Some quota option set */
 906#define EXT4_MOUNT_USRQUOTA             0x100000 /* "old" user quota */
 907#define EXT4_MOUNT_GRPQUOTA             0x200000 /* "old" group quota */
 908#define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
 909#define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
 910#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
 911#define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
 912#define EXT4_MOUNT_MBLK_IO_SUBMIT       0x4000000 /* multi-block io submits */
 913#define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
 914#define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
 915#define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
 916#define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
 917#define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
 918
 919#define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
 920                                                ~EXT4_MOUNT_##opt
 921#define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
 922                                                EXT4_MOUNT_##opt
 923#define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
 924                                         EXT4_MOUNT_##opt)
 925
 926#define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
 927                                                ~EXT4_MOUNT2_##opt
 928#define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
 929                                                EXT4_MOUNT2_##opt
 930#define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
 931                                         EXT4_MOUNT2_##opt)
 932
 933#define ext4_set_bit                    __test_and_set_bit_le
 934#define ext4_set_bit_atomic             ext2_set_bit_atomic
 935#define ext4_clear_bit                  __test_and_clear_bit_le
 936#define ext4_clear_bit_atomic           ext2_clear_bit_atomic
 937#define ext4_test_bit                   test_bit_le
 938#define ext4_find_first_zero_bit        find_first_zero_bit_le
 939#define ext4_find_next_zero_bit         find_next_zero_bit_le
 940#define ext4_find_next_bit              find_next_bit_le
 941
 942/*
 943 * Maximal mount counts between two filesystem checks
 944 */
 945#define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
 946#define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
 947
 948/*
 949 * Behaviour when detecting errors
 950 */
 951#define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
 952#define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
 953#define EXT4_ERRORS_PANIC               3       /* Panic */
 954#define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
 955
 956/*
 957 * Structure of the super block
 958 */
 959struct ext4_super_block {
 960/*00*/  __le32  s_inodes_count;         /* Inodes count */
 961        __le32  s_blocks_count_lo;      /* Blocks count */
 962        __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
 963        __le32  s_free_blocks_count_lo; /* Free blocks count */
 964/*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
 965        __le32  s_first_data_block;     /* First Data Block */
 966        __le32  s_log_block_size;       /* Block size */
 967        __le32  s_obso_log_frag_size;   /* Obsoleted fragment size */
 968/*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
 969        __le32  s_obso_frags_per_group; /* Obsoleted fragments per group */
 970        __le32  s_inodes_per_group;     /* # Inodes per group */
 971        __le32  s_mtime;                /* Mount time */
 972/*30*/  __le32  s_wtime;                /* Write time */
 973        __le16  s_mnt_count;            /* Mount count */
 974        __le16  s_max_mnt_count;        /* Maximal mount count */
 975        __le16  s_magic;                /* Magic signature */
 976        __le16  s_state;                /* File system state */
 977        __le16  s_errors;               /* Behaviour when detecting errors */
 978        __le16  s_minor_rev_level;      /* minor revision level */
 979/*40*/  __le32  s_lastcheck;            /* time of last check */
 980        __le32  s_checkinterval;        /* max. time between checks */
 981        __le32  s_creator_os;           /* OS */
 982        __le32  s_rev_level;            /* Revision level */
 983/*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
 984        __le16  s_def_resgid;           /* Default gid for reserved blocks */
 985        /*
 986         * These fields are for EXT4_DYNAMIC_REV superblocks only.
 987         *
 988         * Note: the difference between the compatible feature set and
 989         * the incompatible feature set is that if there is a bit set
 990         * in the incompatible feature set that the kernel doesn't
 991         * know about, it should refuse to mount the filesystem.
 992         *
 993         * e2fsck's requirements are more strict; if it doesn't know
 994         * about a feature in either the compatible or incompatible
 995         * feature set, it must abort and not try to meddle with
 996         * things it doesn't understand...
 997         */
 998        __le32  s_first_ino;            /* First non-reserved inode */
 999        __le16  s_inode_size;           /* size of inode structure */
1000        __le16  s_block_group_nr;       /* block group # of this superblock */
1001        __le32  s_feature_compat;       /* compatible feature set */
1002/*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1003        __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1004/*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1005/*78*/  char    s_volume_name[16];      /* volume name */
1006/*88*/  char    s_last_mounted[64];     /* directory where last mounted */
1007/*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1008        /*
1009         * Performance hints.  Directory preallocation should only
1010         * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1011         */
1012        __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1013        __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1014        __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1015        /*
1016         * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1017         */
1018/*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1019/*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1020        __le32  s_journal_dev;          /* device number of journal file */
1021        __le32  s_last_orphan;          /* start of list of inodes to delete */
1022        __le32  s_hash_seed[4];         /* HTREE hash seed */
1023        __u8    s_def_hash_version;     /* Default hash version to use */
1024        __u8    s_jnl_backup_type;
1025        __le16  s_desc_size;            /* size of group descriptor */
1026/*100*/ __le32  s_default_mount_opts;
1027        __le32  s_first_meta_bg;        /* First metablock block group */
1028        __le32  s_mkfs_time;            /* When the filesystem was created */
1029        __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1030        /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1031/*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1032        __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1033        __le32  s_free_blocks_count_hi; /* Free blocks count */
1034        __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1035        __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1036        __le32  s_flags;                /* Miscellaneous flags */
1037        __le16  s_raid_stride;          /* RAID stride */
1038        __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1039        __le64  s_mmp_block;            /* Block for multi-mount protection */
1040        __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1041        __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1042        __u8    s_reserved_char_pad;
1043        __le16  s_reserved_pad;
1044        __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1045        __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1046        __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1047        __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1048                                              snapshot's future use */
1049        __le32  s_snapshot_list;        /* inode number of the head of the
1050                                           on-disk snapshot list */
1051#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1052        __le32  s_error_count;          /* number of fs errors */
1053        __le32  s_first_error_time;     /* first time an error happened */
1054        __le32  s_first_error_ino;      /* inode involved in first error */
1055        __le64  s_first_error_block;    /* block involved of first error */
1056        __u8    s_first_error_func[32]; /* function where the error happened */
1057        __le32  s_first_error_line;     /* line number where error happened */
1058        __le32  s_last_error_time;      /* most recent time of an error */
1059        __le32  s_last_error_ino;       /* inode involved in last error */
1060        __le32  s_last_error_line;      /* line number where error happened */
1061        __le64  s_last_error_block;     /* block involved of last error */
1062        __u8    s_last_error_func[32];  /* function where the error happened */
1063#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1064        __u8    s_mount_opts[64];
1065        __le32  s_reserved[112];        /* Padding to the end of the block */
1066};
1067
1068#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1069
1070#ifdef __KERNEL__
1071
1072/*
1073 * run-time mount flags
1074 */
1075#define EXT4_MF_MNTDIR_SAMPLED  0x0001
1076#define EXT4_MF_FS_ABORTED      0x0002  /* Fatal error detected */
1077
1078/*
1079 * fourth extended-fs super-block data in memory
1080 */
1081struct ext4_sb_info {
1082        unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1083        unsigned long s_inodes_per_block;/* Number of inodes per block */
1084        unsigned long s_blocks_per_group;/* Number of blocks in a group */
1085        unsigned long s_inodes_per_group;/* Number of inodes in a group */
1086        unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1087        unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1088        unsigned long s_desc_per_block; /* Number of group descriptors per block */
1089        ext4_group_t s_groups_count;    /* Number of groups in the fs */
1090        ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1091        unsigned long s_overhead_last;  /* Last calculated overhead */
1092        unsigned long s_blocks_last;    /* Last seen block count */
1093        loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1094        struct buffer_head * s_sbh;     /* Buffer containing the super block */
1095        struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1096        struct buffer_head **s_group_desc;
1097        unsigned int s_mount_opt;
1098        unsigned int s_mount_opt2;
1099        unsigned int s_mount_flags;
1100        ext4_fsblk_t s_sb_block;
1101        uid_t s_resuid;
1102        gid_t s_resgid;
1103        unsigned short s_mount_state;
1104        unsigned short s_pad;
1105        int s_addr_per_block_bits;
1106        int s_desc_per_block_bits;
1107        int s_inode_size;
1108        int s_first_ino;
1109        unsigned int s_inode_readahead_blks;
1110        unsigned int s_inode_goal;
1111        spinlock_t s_next_gen_lock;
1112        u32 s_next_generation;
1113        u32 s_hash_seed[4];
1114        int s_def_hash_version;
1115        int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1116        struct percpu_counter s_freeblocks_counter;
1117        struct percpu_counter s_freeinodes_counter;
1118        struct percpu_counter s_dirs_counter;
1119        struct percpu_counter s_dirtyblocks_counter;
1120        struct blockgroup_lock *s_blockgroup_lock;
1121        struct proc_dir_entry *s_proc;
1122        struct kobject s_kobj;
1123        struct completion s_kobj_unregister;
1124
1125        /* Journaling */
1126        struct journal_s *s_journal;
1127        struct list_head s_orphan;
1128        struct mutex s_orphan_lock;
1129        struct mutex s_resize_lock;
1130        unsigned long s_commit_interval;
1131        u32 s_max_batch_time;
1132        u32 s_min_batch_time;
1133        struct block_device *journal_bdev;
1134#ifdef CONFIG_JBD2_DEBUG
1135        struct timer_list turn_ro_timer;        /* For turning read-only (crash simulation) */
1136        wait_queue_head_t ro_wait_queue;        /* For people waiting for the fs to go read-only */
1137#endif
1138#ifdef CONFIG_QUOTA
1139        char *s_qf_names[MAXQUOTAS];            /* Names of quota files with journalled quota */
1140        int s_jquota_fmt;                       /* Format of quota to use */
1141#endif
1142        unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1143        struct rb_root system_blks;
1144
1145#ifdef EXTENTS_STATS
1146        /* ext4 extents stats */
1147        unsigned long s_ext_min;
1148        unsigned long s_ext_max;
1149        unsigned long s_depth_max;
1150        spinlock_t s_ext_stats_lock;
1151        unsigned long s_ext_blocks;
1152        unsigned long s_ext_extents;
1153#endif
1154        /* ext4 extent cache stats */
1155        unsigned long extent_cache_hits;
1156        unsigned long extent_cache_misses;
1157
1158        /* for buddy allocator */
1159        struct ext4_group_info ***s_group_info;
1160        struct inode *s_buddy_cache;
1161        spinlock_t s_md_lock;
1162        unsigned short *s_mb_offsets;
1163        unsigned int *s_mb_maxs;
1164
1165        /* tunables */
1166        unsigned long s_stripe;
1167        unsigned int s_mb_stream_request;
1168        unsigned int s_mb_max_to_scan;
1169        unsigned int s_mb_min_to_scan;
1170        unsigned int s_mb_stats;
1171        unsigned int s_mb_order2_reqs;
1172        unsigned int s_mb_group_prealloc;
1173        unsigned int s_max_writeback_mb_bump;
1174        /* where last allocation was done - for stream allocation */
1175        unsigned long s_mb_last_group;
1176        unsigned long s_mb_last_start;
1177
1178        /* stats for buddy allocator */
1179        atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1180        atomic_t s_bal_success; /* we found long enough chunks */
1181        atomic_t s_bal_allocated;       /* in blocks */
1182        atomic_t s_bal_ex_scanned;      /* total extents scanned */
1183        atomic_t s_bal_goals;   /* goal hits */
1184        atomic_t s_bal_breaks;  /* too long searches */
1185        atomic_t s_bal_2orders; /* 2^order hits */
1186        spinlock_t s_bal_lock;
1187        unsigned long s_mb_buddies_generated;
1188        unsigned long long s_mb_generation_time;
1189        atomic_t s_mb_lost_chunks;
1190        atomic_t s_mb_preallocated;
1191        atomic_t s_mb_discarded;
1192        atomic_t s_lock_busy;
1193
1194        /* locality groups */
1195        struct ext4_locality_group __percpu *s_locality_groups;
1196
1197        /* for write statistics */
1198        unsigned long s_sectors_written_start;
1199        u64 s_kbytes_written;
1200
1201        unsigned int s_log_groups_per_flex;
1202        struct flex_groups *s_flex_groups;
1203
1204        /* workqueue for dio unwritten */
1205        struct workqueue_struct *dio_unwritten_wq;
1206
1207        /* timer for periodic error stats printing */
1208        struct timer_list s_err_report;
1209
1210        /* Lazy inode table initialization info */
1211        struct ext4_li_request *s_li_request;
1212        /* Wait multiplier for lazy initialization thread */
1213        unsigned int s_li_wait_mult;
1214
1215        /* Kernel thread for multiple mount protection */
1216        struct task_struct *s_mmp_tsk;
1217};
1218
1219static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1220{
1221        return sb->s_fs_info;
1222}
1223static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1224{
1225        return container_of(inode, struct ext4_inode_info, vfs_inode);
1226}
1227
1228static inline struct timespec ext4_current_time(struct inode *inode)
1229{
1230        return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1231                current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1232}
1233
1234static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1235{
1236        return ino == EXT4_ROOT_INO ||
1237                ino == EXT4_JOURNAL_INO ||
1238                ino == EXT4_RESIZE_INO ||
1239                (ino >= EXT4_FIRST_INO(sb) &&
1240                 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1241}
1242
1243/*
1244 * Inode dynamic state flags
1245 */
1246enum {
1247        EXT4_STATE_JDATA,               /* journaled data exists */
1248        EXT4_STATE_NEW,                 /* inode is newly created */
1249        EXT4_STATE_XATTR,               /* has in-inode xattrs */
1250        EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1251        EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1252        EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1253        EXT4_STATE_DIO_UNWRITTEN,       /* need convert on dio done*/
1254        EXT4_STATE_NEWENTRY,            /* File just added to dir */
1255        EXT4_STATE_DELALLOC_RESERVED,   /* blks already reserved for delalloc */
1256};
1257
1258#define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1259static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1260{                                                                       \
1261        return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1262}                                                                       \
1263static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1264{                                                                       \
1265        set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1266}                                                                       \
1267static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1268{                                                                       \
1269        clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1270}
1271
1272EXT4_INODE_BIT_FNS(flag, flags, 0)
1273#if (BITS_PER_LONG < 64)
1274EXT4_INODE_BIT_FNS(state, state_flags, 0)
1275
1276static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1277{
1278        (ei)->i_state_flags = 0;
1279}
1280#else
1281EXT4_INODE_BIT_FNS(state, flags, 32)
1282
1283static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1284{
1285        /* We depend on the fact that callers will set i_flags */
1286}
1287#endif
1288#else
1289/* Assume that user mode programs are passing in an ext4fs superblock, not
1290 * a kernel struct super_block.  This will allow us to call the feature-test
1291 * macros from user land. */
1292#define EXT4_SB(sb)     (sb)
1293#endif
1294
1295#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1296
1297/*
1298 * Codes for operating systems
1299 */
1300#define EXT4_OS_LINUX           0
1301#define EXT4_OS_HURD            1
1302#define EXT4_OS_MASIX           2
1303#define EXT4_OS_FREEBSD         3
1304#define EXT4_OS_LITES           4
1305
1306/*
1307 * Revision levels
1308 */
1309#define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1310#define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1311
1312#define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1313#define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1314
1315#define EXT4_GOOD_OLD_INODE_SIZE 128
1316
1317/*
1318 * Feature set definitions
1319 */
1320
1321#define EXT4_HAS_COMPAT_FEATURE(sb,mask)                        \
1322        ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1323#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)                     \
1324        ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1325#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)                      \
1326        ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1327#define EXT4_SET_COMPAT_FEATURE(sb,mask)                        \
1328        EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1329#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)                     \
1330        EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1331#define EXT4_SET_INCOMPAT_FEATURE(sb,mask)                      \
1332        EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1333#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)                      \
1334        EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1335#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)                   \
1336        EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1337#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)                    \
1338        EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1339
1340#define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1341#define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1342#define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1343#define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1344#define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1345#define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1346
1347#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1348#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1349#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1350#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1351#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1352#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1353#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1354#define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1355
1356#define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1357#define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1358#define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1359#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1360#define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1361#define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1362#define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1363#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1364#define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1365#define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1366#define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1367
1368#define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1369#define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1370                                         EXT4_FEATURE_INCOMPAT_META_BG)
1371#define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1372                                         EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1373                                         EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1374
1375#define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1376#define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1377                                         EXT4_FEATURE_INCOMPAT_RECOVER| \
1378                                         EXT4_FEATURE_INCOMPAT_META_BG)
1379#define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1380                                         EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1381                                         EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1382
1383#define EXT4_FEATURE_COMPAT_SUPP        EXT2_FEATURE_COMPAT_EXT_ATTR
1384#define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1385                                         EXT4_FEATURE_INCOMPAT_RECOVER| \
1386                                         EXT4_FEATURE_INCOMPAT_META_BG| \
1387                                         EXT4_FEATURE_INCOMPAT_EXTENTS| \
1388                                         EXT4_FEATURE_INCOMPAT_64BIT| \
1389                                         EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1390                                         EXT4_FEATURE_INCOMPAT_MMP)
1391#define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1392                                         EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1393                                         EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1394                                         EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1395                                         EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1396                                         EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1397                                         EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1398
1399/*
1400 * Default values for user and/or group using reserved blocks
1401 */
1402#define EXT4_DEF_RESUID         0
1403#define EXT4_DEF_RESGID         0
1404
1405#define EXT4_DEF_INODE_READAHEAD_BLKS   32
1406
1407/*
1408 * Default mount options
1409 */
1410#define EXT4_DEFM_DEBUG         0x0001
1411#define EXT4_DEFM_BSDGROUPS     0x0002
1412#define EXT4_DEFM_XATTR_USER    0x0004
1413#define EXT4_DEFM_ACL           0x0008
1414#define EXT4_DEFM_UID16         0x0010
1415#define EXT4_DEFM_JMODE         0x0060
1416#define EXT4_DEFM_JMODE_DATA    0x0020
1417#define EXT4_DEFM_JMODE_ORDERED 0x0040
1418#define EXT4_DEFM_JMODE_WBACK   0x0060
1419#define EXT4_DEFM_NOBARRIER     0x0100
1420#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1421#define EXT4_DEFM_DISCARD       0x0400
1422#define EXT4_DEFM_NODELALLOC    0x0800
1423
1424/*
1425 * Default journal batch times
1426 */
1427#define EXT4_DEF_MIN_BATCH_TIME 0
1428#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1429
1430/*
1431 * Minimum number of groups in a flexgroup before we separate out
1432 * directories into the first block group of a flexgroup
1433 */
1434#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1435
1436/*
1437 * Structure of a directory entry
1438 */
1439#define EXT4_NAME_LEN 255
1440
1441struct ext4_dir_entry {
1442        __le32  inode;                  /* Inode number */
1443        __le16  rec_len;                /* Directory entry length */
1444        __le16  name_len;               /* Name length */
1445        char    name[EXT4_NAME_LEN];    /* File name */
1446};
1447
1448/*
1449 * The new version of the directory entry.  Since EXT4 structures are
1450 * stored in intel byte order, and the name_len field could never be
1451 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1452 * file_type field.
1453 */
1454struct ext4_dir_entry_2 {
1455        __le32  inode;                  /* Inode number */
1456        __le16  rec_len;                /* Directory entry length */
1457        __u8    name_len;               /* Name length */
1458        __u8    file_type;
1459        char    name[EXT4_NAME_LEN];    /* File name */
1460};
1461
1462/*
1463 * Ext4 directory file types.  Only the low 3 bits are used.  The
1464 * other bits are reserved for now.
1465 */
1466#define EXT4_FT_UNKNOWN         0
1467#define EXT4_FT_REG_FILE        1
1468#define EXT4_FT_DIR             2
1469#define EXT4_FT_CHRDEV          3
1470#define EXT4_FT_BLKDEV          4
1471#define EXT4_FT_FIFO            5
1472#define EXT4_FT_SOCK            6
1473#define EXT4_FT_SYMLINK         7
1474
1475#define EXT4_FT_MAX             8
1476
1477/*
1478 * EXT4_DIR_PAD defines the directory entries boundaries
1479 *
1480 * NOTE: It must be a multiple of 4
1481 */
1482#define EXT4_DIR_PAD                    4
1483#define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
1484#define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
1485                                         ~EXT4_DIR_ROUND)
1486#define EXT4_MAX_REC_LEN                ((1<<16)-1)
1487
1488/*
1489 * If we ever get support for fs block sizes > page_size, we'll need
1490 * to remove the #if statements in the next two functions...
1491 */
1492static inline unsigned int
1493ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1494{
1495        unsigned len = le16_to_cpu(dlen);
1496
1497#if (PAGE_CACHE_SIZE >= 65536)
1498        if (len == EXT4_MAX_REC_LEN || len == 0)
1499                return blocksize;
1500        return (len & 65532) | ((len & 3) << 16);
1501#else
1502        return len;
1503#endif
1504}
1505
1506static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1507{
1508        if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1509                BUG();
1510#if (PAGE_CACHE_SIZE >= 65536)
1511        if (len < 65536)
1512                return cpu_to_le16(len);
1513        if (len == blocksize) {
1514                if (blocksize == 65536)
1515                        return cpu_to_le16(EXT4_MAX_REC_LEN);
1516                else
1517                        return cpu_to_le16(0);
1518        }
1519        return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1520#else
1521        return cpu_to_le16(len);
1522#endif
1523}
1524
1525/*
1526 * Hash Tree Directory indexing
1527 * (c) Daniel Phillips, 2001
1528 */
1529
1530#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1531                                      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1532                    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1533#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1534#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1535
1536/* Legal values for the dx_root hash_version field: */
1537
1538#define DX_HASH_LEGACY          0
1539#define DX_HASH_HALF_MD4        1
1540#define DX_HASH_TEA             2
1541#define DX_HASH_LEGACY_UNSIGNED 3
1542#define DX_HASH_HALF_MD4_UNSIGNED       4
1543#define DX_HASH_TEA_UNSIGNED            5
1544
1545#ifdef __KERNEL__
1546
1547/* hash info structure used by the directory hash */
1548struct dx_hash_info
1549{
1550        u32             hash;
1551        u32             minor_hash;
1552        int             hash_version;
1553        u32             *seed;
1554};
1555
1556#define EXT4_HTREE_EOF  0x7fffffff
1557
1558/*
1559 * Control parameters used by ext4_htree_next_block
1560 */
1561#define HASH_NB_ALWAYS          1
1562
1563
1564/*
1565 * Describe an inode's exact location on disk and in memory
1566 */
1567struct ext4_iloc
1568{
1569        struct buffer_head *bh;
1570        unsigned long offset;
1571        ext4_group_t block_group;
1572};
1573
1574static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1575{
1576        return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1577}
1578
1579/*
1580 * This structure is stuffed into the struct file's private_data field
1581 * for directories.  It is where we put information so that we can do
1582 * readdir operations in hash tree order.
1583 */
1584struct dir_private_info {
1585        struct rb_root  root;
1586        struct rb_node  *curr_node;
1587        struct fname    *extra_fname;
1588        loff_t          last_pos;
1589        __u32           curr_hash;
1590        __u32           curr_minor_hash;
1591        __u32           next_hash;
1592};
1593
1594/* calculate the first block number of the group */
1595static inline ext4_fsblk_t
1596ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1597{
1598        return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1599                le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1600}
1601
1602/*
1603 * Special error return code only used by dx_probe() and its callers.
1604 */
1605#define ERR_BAD_DX_DIR  -75000
1606
1607void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1608                        ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1609
1610/*
1611 * Timeout and state flag for lazy initialization inode thread.
1612 */
1613#define EXT4_DEF_LI_WAIT_MULT                   10
1614#define EXT4_DEF_LI_MAX_START_DELAY             5
1615#define EXT4_LAZYINIT_QUIT                      0x0001
1616#define EXT4_LAZYINIT_RUNNING                   0x0002
1617
1618/*
1619 * Lazy inode table initialization info
1620 */
1621struct ext4_lazy_init {
1622        unsigned long           li_state;
1623        struct list_head        li_request_list;
1624        struct mutex            li_list_mtx;
1625};
1626
1627struct ext4_li_request {
1628        struct super_block      *lr_super;
1629        struct ext4_sb_info     *lr_sbi;
1630        ext4_group_t            lr_next_group;
1631        struct list_head        lr_request;
1632        unsigned long           lr_next_sched;
1633        unsigned long           lr_timeout;
1634};
1635
1636struct ext4_features {
1637        struct kobject f_kobj;
1638        struct completion f_kobj_unregister;
1639};
1640
1641/*
1642 * This structure will be used for multiple mount protection. It will be
1643 * written into the block number saved in the s_mmp_block field in the
1644 * superblock. Programs that check MMP should assume that if
1645 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1646 * to use the filesystem, regardless of how old the timestamp is.
1647 */
1648#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1649#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1650#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1651#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1652
1653struct mmp_struct {
1654        __le32  mmp_magic;              /* Magic number for MMP */
1655        __le32  mmp_seq;                /* Sequence no. updated periodically */
1656
1657        /*
1658         * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1659         * purposes and do not affect the correctness of the algorithm
1660         */
1661        __le64  mmp_time;               /* Time last updated */
1662        char    mmp_nodename[64];       /* Node which last updated MMP block */
1663        char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
1664
1665        /*
1666         * mmp_check_interval is used to verify if the MMP block has been
1667         * updated on the block device. The value is updated based on the
1668         * maximum time to write the MMP block during an update cycle.
1669         */
1670        __le16  mmp_check_interval;
1671
1672        __le16  mmp_pad1;
1673        __le32  mmp_pad2[227];
1674};
1675
1676/* arguments passed to the mmp thread */
1677struct mmpd_data {
1678        struct buffer_head *bh; /* bh from initial read_mmp_block() */
1679        struct super_block *sb;  /* super block of the fs */
1680};
1681
1682/*
1683 * Check interval multiplier
1684 * The MMP block is written every update interval and initially checked every
1685 * update interval x the multiplier (the value is then adapted based on the
1686 * write latency). The reason is that writes can be delayed under load and we
1687 * don't want readers to incorrectly assume that the filesystem is no longer
1688 * in use.
1689 */
1690#define EXT4_MMP_CHECK_MULT             2UL
1691
1692/*
1693 * Minimum interval for MMP checking in seconds.
1694 */
1695#define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
1696
1697/*
1698 * Maximum interval for MMP checking in seconds.
1699 */
1700#define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
1701
1702/*
1703 * Function prototypes
1704 */
1705
1706/*
1707 * Ok, these declarations are also in <linux/kernel.h> but none of the
1708 * ext4 source programs needs to include it so they are duplicated here.
1709 */
1710# define NORET_TYPE     /**/
1711# define ATTRIB_NORET   __attribute__((noreturn))
1712# define NORET_AND      noreturn,
1713
1714/* bitmap.c */
1715extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1716
1717/* balloc.c */
1718extern unsigned int ext4_block_group(struct super_block *sb,
1719                        ext4_fsblk_t blocknr);
1720extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1721                        ext4_fsblk_t blocknr);
1722extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1723extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1724                        ext4_group_t group);
1725extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1726                                         ext4_fsblk_t goal,
1727                                         unsigned int flags,
1728                                         unsigned long *count,
1729                                         int *errp);
1730extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi,
1731                                  s64 nblocks, unsigned int flags);
1732extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1733extern void ext4_check_blocks_bitmap(struct super_block *);
1734extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1735                                                    ext4_group_t block_group,
1736                                                    struct buffer_head ** bh);
1737extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1738struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1739                                      ext4_group_t block_group);
1740extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1741                                       struct buffer_head *bh,
1742                                       ext4_group_t group,
1743                                       struct ext4_group_desc *desc);
1744#define ext4_free_blocks_after_init(sb, group, desc)                    \
1745                ext4_init_block_bitmap(sb, NULL, group, desc)
1746
1747/* dir.c */
1748extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1749                                  struct file *,
1750                                  struct ext4_dir_entry_2 *,
1751                                  struct buffer_head *, unsigned int);
1752#define ext4_check_dir_entry(dir, filp, de, bh, offset)                 \
1753        unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1754                                        (de), (bh), (offset)))
1755extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1756                                    __u32 minor_hash,
1757                                    struct ext4_dir_entry_2 *dirent);
1758extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1759
1760/* fsync.c */
1761extern int ext4_sync_file(struct file *, int);
1762extern int ext4_flush_completed_IO(struct inode *);
1763
1764/* hash.c */
1765extern int ext4fs_dirhash(const char *name, int len, struct
1766                          dx_hash_info *hinfo);
1767
1768/* ialloc.c */
1769extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1770                                    const struct qstr *qstr, __u32 goal);
1771extern void ext4_free_inode(handle_t *, struct inode *);
1772extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1773extern unsigned long ext4_count_free_inodes(struct super_block *);
1774extern unsigned long ext4_count_dirs(struct super_block *);
1775extern void ext4_check_inodes_bitmap(struct super_block *);
1776extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1777extern int ext4_init_inode_table(struct super_block *sb,
1778                                 ext4_group_t group, int barrier);
1779
1780/* mballoc.c */
1781extern long ext4_mb_stats;
1782extern long ext4_mb_max_to_scan;
1783extern int ext4_mb_init(struct super_block *, int);
1784extern int ext4_mb_release(struct super_block *);
1785extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1786                                struct ext4_allocation_request *, int *);
1787extern int ext4_mb_reserve_blocks(struct super_block *, int);
1788extern void ext4_discard_preallocations(struct inode *);
1789extern int __init ext4_init_mballoc(void);
1790extern void ext4_exit_mballoc(void);
1791extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1792                             struct buffer_head *bh, ext4_fsblk_t block,
1793                             unsigned long count, int flags);
1794extern int ext4_mb_add_groupinfo(struct super_block *sb,
1795                ext4_group_t i, struct ext4_group_desc *desc);
1796extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1797                                ext4_fsblk_t block, unsigned long count);
1798extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
1799
1800/* inode.c */
1801struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1802                                                ext4_lblk_t, int, int *);
1803struct buffer_head *ext4_bread(handle_t *, struct inode *,
1804                                                ext4_lblk_t, int, int *);
1805int ext4_get_block(struct inode *inode, sector_t iblock,
1806                                struct buffer_head *bh_result, int create);
1807
1808extern struct inode *ext4_iget(struct super_block *, unsigned long);
1809extern int  ext4_write_inode(struct inode *, struct writeback_control *);
1810extern int  ext4_setattr(struct dentry *, struct iattr *);
1811extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1812                                struct kstat *stat);
1813extern void ext4_evict_inode(struct inode *);
1814extern void ext4_clear_inode(struct inode *);
1815extern int  ext4_sync_inode(handle_t *, struct inode *);
1816extern void ext4_dirty_inode(struct inode *, int);
1817extern int ext4_change_inode_journal_flag(struct inode *, int);
1818extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1819extern int ext4_can_truncate(struct inode *inode);
1820extern void ext4_truncate(struct inode *);
1821extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1822extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1823extern void ext4_set_inode_flags(struct inode *);
1824extern void ext4_get_inode_flags(struct ext4_inode_info *);
1825extern int ext4_alloc_da_blocks(struct inode *inode);
1826extern void ext4_set_aops(struct inode *inode);
1827extern int ext4_writepage_trans_blocks(struct inode *);
1828extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1829extern int ext4_block_truncate_page(handle_t *handle,
1830                struct address_space *mapping, loff_t from);
1831extern int ext4_block_zero_page_range(handle_t *handle,
1832                struct address_space *mapping, loff_t from, loff_t length);
1833extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1834extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1835extern void ext4_da_update_reserve_space(struct inode *inode,
1836                                        int used, int quota_claim);
1837/* ioctl.c */
1838extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1839extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1840
1841/* migrate.c */
1842extern int ext4_ext_migrate(struct inode *);
1843
1844/* namei.c */
1845extern int ext4_orphan_add(handle_t *, struct inode *);
1846extern int ext4_orphan_del(handle_t *, struct inode *);
1847extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1848                                __u32 start_minor_hash, __u32 *next_hash);
1849
1850/* resize.c */
1851extern int ext4_group_add(struct super_block *sb,
1852                                struct ext4_new_group_data *input);
1853extern int ext4_group_extend(struct super_block *sb,
1854                                struct ext4_super_block *es,
1855                                ext4_fsblk_t n_blocks_count);
1856
1857/* super.c */
1858extern void __ext4_error(struct super_block *, const char *, unsigned int,
1859                         const char *, ...)
1860        __attribute__ ((format (printf, 4, 5)));
1861#define ext4_error(sb, message...)      __ext4_error(sb, __func__,      \
1862                                                     __LINE__, ## message)
1863extern void ext4_error_inode(struct inode *, const char *, unsigned int,
1864                             ext4_fsblk_t, const char *, ...)
1865        __attribute__ ((format (printf, 5, 6)));
1866extern void ext4_error_file(struct file *, const char *, unsigned int,
1867                            ext4_fsblk_t, const char *, ...)
1868        __attribute__ ((format (printf, 5, 6)));
1869extern void __ext4_std_error(struct super_block *, const char *,
1870                             unsigned int, int);
1871extern void __ext4_abort(struct super_block *, const char *, unsigned int,
1872                       const char *, ...)
1873        __attribute__ ((format (printf, 4, 5)));
1874#define ext4_abort(sb, message...)      __ext4_abort(sb, __func__, \
1875                                                       __LINE__, ## message)
1876extern void __ext4_warning(struct super_block *, const char *, unsigned int,
1877                          const char *, ...)
1878        __attribute__ ((format (printf, 4, 5)));
1879#define ext4_warning(sb, message...)    __ext4_warning(sb, __func__, \
1880                                                       __LINE__, ## message)
1881extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1882        __attribute__ ((format (printf, 3, 4)));
1883extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
1884                           const char *, unsigned int, const char *);
1885#define dump_mmp_msg(sb, mmp, msg)      __dump_mmp_msg(sb, mmp, __func__, \
1886                                                       __LINE__, msg)
1887extern void __ext4_grp_locked_error(const char *, unsigned int, \
1888                                    struct super_block *, ext4_group_t, \
1889                                    unsigned long, ext4_fsblk_t, \
1890                                    const char *, ...)
1891        __attribute__ ((format (printf, 7, 8)));
1892#define ext4_grp_locked_error(sb, grp, message...) \
1893        __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1894extern void ext4_update_dynamic_rev(struct super_block *sb);
1895extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1896                                        __u32 compat);
1897extern int ext4_update_rocompat_feature(handle_t *handle,
1898                                        struct super_block *sb, __u32 rocompat);
1899extern int ext4_update_incompat_feature(handle_t *handle,
1900                                        struct super_block *sb, __u32 incompat);
1901extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1902                                      struct ext4_group_desc *bg);
1903extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1904                                      struct ext4_group_desc *bg);
1905extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1906                                     struct ext4_group_desc *bg);
1907extern __u32 ext4_free_blks_count(struct super_block *sb,
1908                                struct ext4_group_desc *bg);
1909extern __u32 ext4_free_inodes_count(struct super_block *sb,
1910                                 struct ext4_group_desc *bg);
1911extern __u32 ext4_used_dirs_count(struct super_block *sb,
1912                                struct ext4_group_desc *bg);
1913extern __u32 ext4_itable_unused_count(struct super_block *sb,
1914                                   struct ext4_group_desc *bg);
1915extern void ext4_block_bitmap_set(struct super_block *sb,
1916                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1917extern void ext4_inode_bitmap_set(struct super_block *sb,
1918                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1919extern void ext4_inode_table_set(struct super_block *sb,
1920                                 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1921extern void ext4_free_blks_set(struct super_block *sb,
1922                               struct ext4_group_desc *bg, __u32 count);
1923extern void ext4_free_inodes_set(struct super_block *sb,
1924                                struct ext4_group_desc *bg, __u32 count);
1925extern void ext4_used_dirs_set(struct super_block *sb,
1926                                struct ext4_group_desc *bg, __u32 count);
1927extern void ext4_itable_unused_set(struct super_block *sb,
1928                                   struct ext4_group_desc *bg, __u32 count);
1929extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1930                                   struct ext4_group_desc *gdp);
1931extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1932                                       struct ext4_group_desc *gdp);
1933
1934static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1935{
1936        return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1937                le32_to_cpu(es->s_blocks_count_lo);
1938}
1939
1940static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1941{
1942        return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1943                le32_to_cpu(es->s_r_blocks_count_lo);
1944}
1945
1946static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1947{
1948        return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1949                le32_to_cpu(es->s_free_blocks_count_lo);
1950}
1951
1952static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1953                                         ext4_fsblk_t blk)
1954{
1955        es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1956        es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1957}
1958
1959static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1960                                              ext4_fsblk_t blk)
1961{
1962        es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1963        es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1964}
1965
1966static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1967                                           ext4_fsblk_t blk)
1968{
1969        es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1970        es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1971}
1972
1973static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1974{
1975        if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1976                return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1977                        le32_to_cpu(raw_inode->i_size_lo);
1978        else
1979                return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1980}
1981
1982static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1983{
1984        raw_inode->i_size_lo = cpu_to_le32(i_size);
1985        raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1986}
1987
1988static inline
1989struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1990                                            ext4_group_t group)
1991{
1992         struct ext4_group_info ***grp_info;
1993         long indexv, indexh;
1994         grp_info = EXT4_SB(sb)->s_group_info;
1995         indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1996         indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1997         return grp_info[indexv][indexh];
1998}
1999
2000/*
2001 * Reading s_groups_count requires using smp_rmb() afterwards.  See
2002 * the locking protocol documented in the comments of ext4_group_add()
2003 * in resize.c
2004 */
2005static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2006{
2007        ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
2008
2009        smp_rmb();
2010        return ngroups;
2011}
2012
2013static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2014                                             ext4_group_t block_group)
2015{
2016        return block_group >> sbi->s_log_groups_per_flex;
2017}
2018
2019static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2020{
2021        return 1 << sbi->s_log_groups_per_flex;
2022}
2023
2024#define ext4_std_error(sb, errno)                               \
2025do {                                                            \
2026        if ((errno))                                            \
2027                __ext4_std_error((sb), __func__, __LINE__, (errno));    \
2028} while (0)
2029
2030#ifdef CONFIG_SMP
2031/* Each CPU can accumulate percpu_counter_batch blocks in their local
2032 * counters. So we need to make sure we have free blocks more
2033 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2034 */
2035#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2036#else
2037#define EXT4_FREEBLOCKS_WATERMARK 0
2038#endif
2039
2040static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2041{
2042        /*
2043         * XXX: replace with spinlock if seen contended -bzzz
2044         */
2045        down_write(&EXT4_I(inode)->i_data_sem);
2046        if (newsize > EXT4_I(inode)->i_disksize)
2047                EXT4_I(inode)->i_disksize = newsize;
2048        up_write(&EXT4_I(inode)->i_data_sem);
2049        return ;
2050}
2051
2052struct ext4_group_info {
2053        unsigned long   bb_state;
2054        struct rb_root  bb_free_root;
2055        ext4_grpblk_t   bb_first_free;  /* first free block */
2056        ext4_grpblk_t   bb_free;        /* total free blocks */
2057        ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
2058        ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
2059        struct          list_head bb_prealloc_list;
2060#ifdef DOUBLE_CHECK
2061        void            *bb_bitmap;
2062#endif
2063        struct rw_semaphore alloc_sem;
2064        ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
2065                                         * regions, index is order.
2066                                         * bb_counters[3] = 5 means
2067                                         * 5 free 8-block regions. */
2068};
2069
2070#define EXT4_GROUP_INFO_NEED_INIT_BIT   0
2071
2072#define EXT4_MB_GRP_NEED_INIT(grp)      \
2073        (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2074
2075#define EXT4_MAX_CONTENTION             8
2076#define EXT4_CONTENTION_THRESHOLD       2
2077
2078static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2079                                              ext4_group_t group)
2080{
2081        return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2082}
2083
2084/*
2085 * Returns true if the filesystem is busy enough that attempts to
2086 * access the block group locks has run into contention.
2087 */
2088static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2089{
2090        return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2091}
2092
2093static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2094{
2095        spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2096        if (spin_trylock(lock))
2097                /*
2098                 * We're able to grab the lock right away, so drop the
2099                 * lock contention counter.
2100                 */
2101                atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2102        else {
2103                /*
2104                 * The lock is busy, so bump the contention counter,
2105                 * and then wait on the spin lock.
2106                 */
2107                atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2108                                  EXT4_MAX_CONTENTION);
2109                spin_lock(lock);
2110        }
2111}
2112
2113static inline void ext4_unlock_group(struct super_block *sb,
2114                                        ext4_group_t group)
2115{
2116        spin_unlock(ext4_group_lock_ptr(sb, group));
2117}
2118
2119static inline void ext4_mark_super_dirty(struct super_block *sb)
2120{
2121        if (EXT4_SB(sb)->s_journal == NULL)
2122                sb->s_dirt =1;
2123}
2124
2125/*
2126 * Inodes and files operations
2127 */
2128
2129/* dir.c */
2130extern const struct file_operations ext4_dir_operations;
2131
2132/* file.c */
2133extern const struct inode_operations ext4_file_inode_operations;
2134extern const struct file_operations ext4_file_operations;
2135extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2136
2137/* namei.c */
2138extern const struct inode_operations ext4_dir_inode_operations;
2139extern const struct inode_operations ext4_special_inode_operations;
2140extern struct dentry *ext4_get_parent(struct dentry *child);
2141
2142/* symlink.c */
2143extern const struct inode_operations ext4_symlink_inode_operations;
2144extern const struct inode_operations ext4_fast_symlink_inode_operations;
2145
2146/* block_validity */
2147extern void ext4_release_system_zone(struct super_block *sb);
2148extern int ext4_setup_system_zone(struct super_block *sb);
2149extern int __init ext4_init_system_zone(void);
2150extern void ext4_exit_system_zone(void);
2151extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2152                                 ext4_fsblk_t start_blk,
2153                                 unsigned int count);
2154
2155/* extents.c */
2156extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2157extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2158extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2159                                       int chunk);
2160extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2161                               struct ext4_map_blocks *map, int flags);
2162extern void ext4_ext_truncate(struct inode *);
2163extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2164                                loff_t length);
2165extern void ext4_ext_init(struct super_block *);
2166extern void ext4_ext_release(struct super_block *);
2167extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2168                          loff_t len);
2169extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2170                          ssize_t len);
2171extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2172                           struct ext4_map_blocks *map, int flags);
2173extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2174                        __u64 start, __u64 len);
2175/* move_extent.c */
2176extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2177                             __u64 start_orig, __u64 start_donor,
2178                             __u64 len, __u64 *moved_len);
2179
2180/* page-io.c */
2181extern int __init ext4_init_pageio(void);
2182extern void ext4_exit_pageio(void);
2183extern void ext4_ioend_wait(struct inode *);
2184extern void ext4_free_io_end(ext4_io_end_t *io);
2185extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2186extern int ext4_end_io_nolock(ext4_io_end_t *io);
2187extern void ext4_io_submit(struct ext4_io_submit *io);
2188extern int ext4_bio_write_page(struct ext4_io_submit *io,
2189                               struct page *page,
2190                               int len,
2191                               struct writeback_control *wbc);
2192
2193/* mmp.c */
2194extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2195
2196/* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2197enum ext4_state_bits {
2198        BH_Uninit       /* blocks are allocated but uninitialized on disk */
2199          = BH_JBDPrivateStart,
2200};
2201
2202BUFFER_FNS(Uninit, uninit)
2203TAS_BUFFER_FNS(Uninit, uninit)
2204
2205/*
2206 * Add new method to test wether block and inode bitmaps are properly
2207 * initialized. With uninit_bg reading the block from disk is not enough
2208 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2209 */
2210#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2211
2212static inline int bitmap_uptodate(struct buffer_head *bh)
2213{
2214        return (buffer_uptodate(bh) &&
2215                        test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2216}
2217static inline void set_bitmap_uptodate(struct buffer_head *bh)
2218{
2219        set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2220}
2221
2222#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2223
2224/* For ioend & aio unwritten conversion wait queues */
2225#define EXT4_WQ_HASH_SZ         37
2226#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
2227                                            EXT4_WQ_HASH_SZ])
2228#define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
2229                                             EXT4_WQ_HASH_SZ])
2230extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2231extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2232
2233#endif  /* __KERNEL__ */
2234
2235#endif  /* _EXT4_H */
2236