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
  33/*
  34 * The fourth extended filesystem constants/structures
  35 */
  36
  37/*
  38 * Define EXT4FS_DEBUG to produce debug messages
  39 */
  40#undef EXT4FS_DEBUG
  41
  42/*
  43 * Debug code
  44 */
  45#ifdef EXT4FS_DEBUG
  46#define ext4_debug(f, a...)                                             \
  47        do {                                                            \
  48                printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
  49                        __FILE__, __LINE__, __func__);                  \
  50                printk(KERN_DEBUG f, ## a);                             \
  51        } while (0)
  52#else
  53#define ext4_debug(f, a...)     do {} while (0)
  54#endif
  55
  56/* data type for block offset of block group */
  57typedef int ext4_grpblk_t;
  58
  59/* data type for filesystem-wide blocks number */
  60typedef unsigned long long ext4_fsblk_t;
  61
  62/* data type for file logical block number */
  63typedef __u32 ext4_lblk_t;
  64
  65/* data type for block group number */
  66typedef unsigned int ext4_group_t;
  67
  68/*
  69 * Flags used in mballoc's allocation_context flags field.  
  70 *
  71 * Also used to show what's going on for debugging purposes when the
  72 * flag field is exported via the traceport interface
  73 */
  74
  75/* prefer goal again. length */
  76#define EXT4_MB_HINT_MERGE              0x0001
  77/* blocks already reserved */
  78#define EXT4_MB_HINT_RESERVED           0x0002
  79/* metadata is being allocated */
  80#define EXT4_MB_HINT_METADATA           0x0004
  81/* first blocks in the file */
  82#define EXT4_MB_HINT_FIRST              0x0008
  83/* search for the best chunk */
  84#define EXT4_MB_HINT_BEST               0x0010
  85/* data is being allocated */
  86#define EXT4_MB_HINT_DATA               0x0020
  87/* don't preallocate (for tails) */
  88#define EXT4_MB_HINT_NOPREALLOC         0x0040
  89/* allocate for locality group */
  90#define EXT4_MB_HINT_GROUP_ALLOC        0x0080
  91/* allocate goal blocks or none */
  92#define EXT4_MB_HINT_GOAL_ONLY          0x0100
  93/* goal is meaningful */
  94#define EXT4_MB_HINT_TRY_GOAL           0x0200
  95/* blocks already pre-reserved by delayed allocation */
  96#define EXT4_MB_DELALLOC_RESERVED       0x0400
  97/* We are doing stream allocation */
  98#define EXT4_MB_STREAM_ALLOC            0x0800
  99
 100
 101struct ext4_allocation_request {
 102        /* target inode for block we're allocating */
 103        struct inode *inode;
 104        /* how many blocks we want to allocate */
 105        unsigned int len;
 106        /* logical block in target inode */
 107        ext4_lblk_t logical;
 108        /* the closest logical allocated block to the left */
 109        ext4_lblk_t lleft;
 110        /* the closest logical allocated block to the right */
 111        ext4_lblk_t lright;
 112        /* phys. target (a hint) */
 113        ext4_fsblk_t goal;
 114        /* phys. block for the closest logical allocated block to the left */
 115        ext4_fsblk_t pleft;
 116        /* phys. block for the closest logical allocated block to the right */
 117        ext4_fsblk_t pright;
 118        /* flags. see above EXT4_MB_HINT_* */
 119        unsigned int flags;
 120};
 121
 122/*
 123 * For delayed allocation tracking
 124 */
 125struct mpage_da_data {
 126        struct inode *inode;
 127        sector_t b_blocknr;             /* start block number of extent */
 128        size_t b_size;                  /* size of extent */
 129        unsigned long b_state;          /* state of the extent */
 130        unsigned long first_page, next_page;    /* extent of pages */
 131        struct writeback_control *wbc;
 132        int io_done;
 133        int pages_written;
 134        int retval;
 135};
 136#define DIO_AIO_UNWRITTEN       0x1
 137typedef struct ext4_io_end {
 138        struct list_head        list;           /* per-file finished AIO list */
 139        struct inode            *inode;         /* file being written to */
 140        unsigned int            flag;           /* unwritten or not */
 141        int                     error;          /* I/O error code */
 142        ext4_lblk_t             offset;         /* offset in the file */
 143        size_t                  size;           /* size of the extent */
 144        struct work_struct      work;           /* data work queue */
 145} ext4_io_end_t;
 146
 147/*
 148 * Special inodes numbers
 149 */
 150#define EXT4_BAD_INO             1      /* Bad blocks inode */
 151#define EXT4_ROOT_INO            2      /* Root inode */
 152#define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
 153#define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
 154#define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
 155#define EXT4_JOURNAL_INO         8      /* Journal inode */
 156
 157/* First non-reserved inode for old ext4 filesystems */
 158#define EXT4_GOOD_OLD_FIRST_INO 11
 159
 160/*
 161 * Maximal count of links to a file
 162 */
 163#define EXT4_LINK_MAX           65000
 164
 165/*
 166 * Macro-instructions used to manage several block sizes
 167 */
 168#define EXT4_MIN_BLOCK_SIZE             1024
 169#define EXT4_MAX_BLOCK_SIZE             65536
 170#define EXT4_MIN_BLOCK_LOG_SIZE         10
 171#ifdef __KERNEL__
 172# define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
 173#else
 174# define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
 175#endif
 176#define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
 177#ifdef __KERNEL__
 178# define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
 179#else
 180# define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
 181#endif
 182#ifdef __KERNEL__
 183#define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
 184#define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
 185#define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
 186#else
 187#define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 188                                 EXT4_GOOD_OLD_INODE_SIZE : \
 189                                 (s)->s_inode_size)
 190#define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 191                                 EXT4_GOOD_OLD_FIRST_INO : \
 192                                 (s)->s_first_ino)
 193#endif
 194#define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
 195
 196/*
 197 * Structure of a blocks group descriptor
 198 */
 199struct ext4_group_desc
 200{
 201        __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
 202        __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
 203        __le32  bg_inode_table_lo;      /* Inodes table block */
 204        __le16  bg_free_blocks_count_lo;/* Free blocks count */
 205        __le16  bg_free_inodes_count_lo;/* Free inodes count */
 206        __le16  bg_used_dirs_count_lo;  /* Directories count */
 207        __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
 208        __u32   bg_reserved[2];         /* Likely block/inode bitmap checksum */
 209        __le16  bg_itable_unused_lo;    /* Unused inodes count */
 210        __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
 211        __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
 212        __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
 213        __le32  bg_inode_table_hi;      /* Inodes table block MSB */
 214        __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
 215        __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
 216        __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
 217        __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
 218        __u32   bg_reserved2[3];
 219};
 220
 221/*
 222 * Structure of a flex block group info
 223 */
 224
 225struct flex_groups {
 226        atomic_t free_inodes;
 227        atomic_t free_blocks;
 228        atomic_t used_dirs;
 229};
 230
 231#define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
 232#define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
 233#define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
 234
 235/*
 236 * Macro-instructions used to manage group descriptors
 237 */
 238#define EXT4_MIN_DESC_SIZE              32
 239#define EXT4_MIN_DESC_SIZE_64BIT        64
 240#define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
 241#define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
 242#ifdef __KERNEL__
 243# define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
 244# define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
 245# define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
 246# define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
 247#else
 248# define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
 249# define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
 250# define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
 251#endif
 252
 253/*
 254 * Constants relative to the data blocks
 255 */
 256#define EXT4_NDIR_BLOCKS                12
 257#define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
 258#define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
 259#define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
 260#define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
 261
 262/*
 263 * Inode flags
 264 */
 265#define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
 266#define EXT4_UNRM_FL                    0x00000002 /* Undelete */
 267#define EXT4_COMPR_FL                   0x00000004 /* Compress file */
 268#define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
 269#define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
 270#define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
 271#define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
 272#define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
 273/* Reserved for compression usage... */
 274#define EXT4_DIRTY_FL                   0x00000100
 275#define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
 276#define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
 277#define EXT4_ECOMPR_FL                  0x00000800 /* Compression error */
 278/* End compression flags --- maybe not all used */
 279#define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
 280#define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
 281#define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
 282#define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
 283#define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
 284#define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
 285#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
 286#define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
 287#define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
 288
 289#define EXT4_FL_USER_VISIBLE            0x000BDFFF /* User visible flags */
 290#define EXT4_FL_USER_MODIFIABLE         0x000B80FF /* User modifiable flags */
 291
 292/* Flags that should be inherited by new inodes from their parent. */
 293#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
 294                           EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
 295                           EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
 296                           EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
 297                           EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
 298
 299/* Flags that are appropriate for regular files (all but dir-specific ones). */
 300#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
 301
 302/* Flags that are appropriate for non-directories/regular files. */
 303#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
 304
 305/* Mask out flags that are inappropriate for the given type of inode. */
 306static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 307{
 308        if (S_ISDIR(mode))
 309                return flags;
 310        else if (S_ISREG(mode))
 311                return flags & EXT4_REG_FLMASK;
 312        else
 313                return flags & EXT4_OTHER_FLMASK;
 314}
 315
 316/*
 317 * Inode dynamic state flags
 318 */
 319#define EXT4_STATE_JDATA                0x00000001 /* journaled data exists */
 320#define EXT4_STATE_NEW                  0x00000002 /* inode is newly created */
 321#define EXT4_STATE_XATTR                0x00000004 /* has in-inode xattrs */
 322#define EXT4_STATE_NO_EXPAND            0x00000008 /* No space for expansion */
 323#define EXT4_STATE_DA_ALLOC_CLOSE       0x00000010 /* Alloc DA blks on close */
 324#define EXT4_STATE_EXT_MIGRATE          0x00000020 /* Inode is migrating */
 325#define EXT4_STATE_DIO_UNWRITTEN        0x00000040 /* need convert on dio done*/
 326
 327/* Used to pass group descriptor data when online resize is done */
 328struct ext4_new_group_input {
 329        __u32 group;            /* Group number for this data */
 330        __u64 block_bitmap;     /* Absolute block number of block bitmap */
 331        __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
 332        __u64 inode_table;      /* Absolute block number of inode table start */
 333        __u32 blocks_count;     /* Total number of blocks in this group */
 334        __u16 reserved_blocks;  /* Number of reserved blocks in this group */
 335        __u16 unused;
 336};
 337
 338/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
 339struct ext4_new_group_data {
 340        __u32 group;
 341        __u64 block_bitmap;
 342        __u64 inode_bitmap;
 343        __u64 inode_table;
 344        __u32 blocks_count;
 345        __u16 reserved_blocks;
 346        __u16 unused;
 347        __u32 free_blocks_count;
 348};
 349
 350/*
 351 * Flags used by ext4_get_blocks()
 352 */
 353        /* Allocate any needed blocks and/or convert an unitialized
 354           extent to be an initialized ext4 */
 355#define EXT4_GET_BLOCKS_CREATE                  0x0001
 356        /* Request the creation of an unitialized extent */
 357#define EXT4_GET_BLOCKS_UNINIT_EXT              0x0002
 358#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT       (EXT4_GET_BLOCKS_UNINIT_EXT|\
 359                                                 EXT4_GET_BLOCKS_CREATE)
 360        /* Caller is from the delayed allocation writeout path,
 361           so set the magic i_delalloc_reserve_flag after taking the 
 362           inode allocation semaphore for */
 363#define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
 364        /* Call ext4_da_update_reserve_space() after successfully 
 365           allocating the blocks */
 366#define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE    0x0008
 367        /* caller is from the direct IO path, request to creation of an
 368        unitialized extents if not allocated, split the uninitialized
 369        extent if blocks has been preallocated already*/
 370#define EXT4_GET_BLOCKS_DIO                     0x0010
 371#define EXT4_GET_BLOCKS_CONVERT                 0x0020
 372#define EXT4_GET_BLOCKS_DIO_CREATE_EXT          (EXT4_GET_BLOCKS_DIO|\
 373                                         EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
 374        /* Convert extent to initialized after direct IO complete */
 375#define EXT4_GET_BLOCKS_DIO_CONVERT_EXT         (EXT4_GET_BLOCKS_CONVERT|\
 376                                         EXT4_GET_BLOCKS_DIO_CREATE_EXT)
 377
 378/*
 379 * ioctl commands
 380 */
 381#define EXT4_IOC_GETFLAGS               FS_IOC_GETFLAGS
 382#define EXT4_IOC_SETFLAGS               FS_IOC_SETFLAGS
 383#define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
 384#define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
 385#define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
 386#define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
 387#ifdef CONFIG_JBD2_DEBUG
 388#define EXT4_IOC_WAIT_FOR_READONLY      _IOR('f', 99, long)
 389#endif
 390#define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
 391#define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
 392#define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
 393#define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
 394#define EXT4_IOC_MIGRATE                _IO('f', 9)
 395 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
 396 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
 397#define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
 398#define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
 399
 400/*
 401 * ioctl commands in 32 bit emulation
 402 */
 403#define EXT4_IOC32_GETFLAGS             FS_IOC32_GETFLAGS
 404#define EXT4_IOC32_SETFLAGS             FS_IOC32_SETFLAGS
 405#define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
 406#define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
 407#define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
 408#define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
 409#define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
 410#ifdef CONFIG_JBD2_DEBUG
 411#define EXT4_IOC32_WAIT_FOR_READONLY    _IOR('f', 99, int)
 412#endif
 413#define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
 414#define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
 415
 416
 417/*
 418 *  Mount options
 419 */
 420struct ext4_mount_options {
 421        unsigned long s_mount_opt;
 422        uid_t s_resuid;
 423        gid_t s_resgid;
 424        unsigned long s_commit_interval;
 425        u32 s_min_batch_time, s_max_batch_time;
 426#ifdef CONFIG_QUOTA
 427        int s_jquota_fmt;
 428        char *s_qf_names[MAXQUOTAS];
 429#endif
 430};
 431
 432/* Max physical block we can addres w/o extents */
 433#define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
 434
 435/*
 436 * Structure of an inode on the disk
 437 */
 438struct ext4_inode {
 439        __le16  i_mode;         /* File mode */
 440        __le16  i_uid;          /* Low 16 bits of Owner Uid */
 441        __le32  i_size_lo;      /* Size in bytes */
 442        __le32  i_atime;        /* Access time */
 443        __le32  i_ctime;        /* Inode Change time */
 444        __le32  i_mtime;        /* Modification time */
 445        __le32  i_dtime;        /* Deletion Time */
 446        __le16  i_gid;          /* Low 16 bits of Group Id */
 447        __le16  i_links_count;  /* Links count */
 448        __le32  i_blocks_lo;    /* Blocks count */
 449        __le32  i_flags;        /* File flags */
 450        union {
 451                struct {
 452                        __le32  l_i_version;
 453                } linux1;
 454                struct {
 455                        __u32  h_i_translator;
 456                } hurd1;
 457                struct {
 458                        __u32  m_i_reserved1;
 459                } masix1;
 460        } osd1;                         /* OS dependent 1 */
 461        __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
 462        __le32  i_generation;   /* File version (for NFS) */
 463        __le32  i_file_acl_lo;  /* File ACL */
 464        __le32  i_size_high;
 465        __le32  i_obso_faddr;   /* Obsoleted fragment address */
 466        union {
 467                struct {
 468                        __le16  l_i_blocks_high; /* were l_i_reserved1 */
 469                        __le16  l_i_file_acl_high;
 470                        __le16  l_i_uid_high;   /* these 2 fields */
 471                        __le16  l_i_gid_high;   /* were reserved2[0] */
 472                        __u32   l_i_reserved2;
 473                } linux2;
 474                struct {
 475                        __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 476                        __u16   h_i_mode_high;
 477                        __u16   h_i_uid_high;
 478                        __u16   h_i_gid_high;
 479                        __u32   h_i_author;
 480                } hurd2;
 481                struct {
 482                        __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 483                        __le16  m_i_file_acl_high;
 484                        __u32   m_i_reserved2[2];
 485                } masix2;
 486        } osd2;                         /* OS dependent 2 */
 487        __le16  i_extra_isize;
 488        __le16  i_pad1;
 489        __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
 490        __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
 491        __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
 492        __le32  i_crtime;       /* File Creation time */
 493        __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
 494        __le32  i_version_hi;   /* high 32 bits for 64-bit version */
 495};
 496
 497struct move_extent {
 498        __u32 reserved;         /* should be zero */
 499        __u32 donor_fd;         /* donor file descriptor */
 500        __u64 orig_start;       /* logical start offset in block for orig */
 501        __u64 donor_start;      /* logical start offset in block for donor */
 502        __u64 len;              /* block length to be moved */
 503        __u64 moved_len;        /* moved block length */
 504};
 505
 506#define EXT4_EPOCH_BITS 2
 507#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
 508#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
 509
 510/*
 511 * Extended fields will fit into an inode if the filesystem was formatted
 512 * with large inodes (-I 256 or larger) and there are not currently any EAs
 513 * consuming all of the available space. For new inodes we always reserve
 514 * enough space for the kernel's known extended fields, but for inodes
 515 * created with an old kernel this might not have been the case. None of
 516 * the extended inode fields is critical for correct filesystem operation.
 517 * This macro checks if a certain field fits in the inode. Note that
 518 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
 519 */
 520#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
 521        ((offsetof(typeof(*ext4_inode), field) +        \
 522          sizeof((ext4_inode)->field))                  \
 523        <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
 524            (einode)->i_extra_isize))                   \
 525
 526static inline __le32 ext4_encode_extra_time(struct timespec *time)
 527{
 528       return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
 529                           (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
 530                          ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
 531}
 532
 533static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
 534{
 535       if (sizeof(time->tv_sec) > 4)
 536               time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
 537                               << 32;
 538       time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 539}
 540
 541#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                          \
 542do {                                                                           \
 543        (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);               \
 544        if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 545                (raw_inode)->xtime ## _extra =                                 \
 546                                ext4_encode_extra_time(&(inode)->xtime);       \
 547} while (0)
 548
 549#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
 550do {                                                                           \
 551        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 552                (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
 553        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 554                (raw_inode)->xtime ## _extra =                                 \
 555                                ext4_encode_extra_time(&(einode)->xtime);      \
 556} while (0)
 557
 558#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                          \
 559do {                                                                           \
 560        (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
 561        if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 562                ext4_decode_extra_time(&(inode)->xtime,                        \
 563                                       raw_inode->xtime ## _extra);            \
 564} while (0)
 565
 566#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
 567do {                                                                           \
 568        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 569                (einode)->xtime.tv_sec =                                       \
 570                        (signed)le32_to_cpu((raw_inode)->xtime);               \
 571        if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 572                ext4_decode_extra_time(&(einode)->xtime,                       \
 573                                       raw_inode->xtime ## _extra);            \
 574} while (0)
 575
 576#define i_disk_version osd1.linux1.l_i_version
 577
 578#if defined(__KERNEL__) || defined(__linux__)
 579#define i_reserved1     osd1.linux1.l_i_reserved1
 580#define i_file_acl_high osd2.linux2.l_i_file_acl_high
 581#define i_blocks_high   osd2.linux2.l_i_blocks_high
 582#define i_uid_low       i_uid
 583#define i_gid_low       i_gid
 584#define i_uid_high      osd2.linux2.l_i_uid_high
 585#define i_gid_high      osd2.linux2.l_i_gid_high
 586#define i_reserved2     osd2.linux2.l_i_reserved2
 587
 588#elif defined(__GNU__)
 589
 590#define i_translator    osd1.hurd1.h_i_translator
 591#define i_uid_high      osd2.hurd2.h_i_uid_high
 592#define i_gid_high      osd2.hurd2.h_i_gid_high
 593#define i_author        osd2.hurd2.h_i_author
 594
 595#elif defined(__masix__)
 596
 597#define i_reserved1     osd1.masix1.m_i_reserved1
 598#define i_file_acl_high osd2.masix2.m_i_file_acl_high
 599#define i_reserved2     osd2.masix2.m_i_reserved2
 600
 601#endif /* defined(__KERNEL__) || defined(__linux__) */
 602
 603/*
 604 * storage for cached extent
 605 */
 606struct ext4_ext_cache {
 607        ext4_fsblk_t    ec_start;
 608        ext4_lblk_t     ec_block;
 609        __u32           ec_len; /* must be 32bit to return holes */
 610        __u32           ec_type;
 611};
 612
 613/*
 614 * fourth extended file system inode data in memory
 615 */
 616struct ext4_inode_info {
 617        __le32  i_data[15];     /* unconverted */
 618        __u32   i_flags;
 619        ext4_fsblk_t    i_file_acl;
 620        __u32   i_dtime;
 621
 622        /*
 623         * i_block_group is the number of the block group which contains
 624         * this file's inode.  Constant across the lifetime of the inode,
 625         * it is ued for making block allocation decisions - we try to
 626         * place a file's data blocks near its inode block, and new inodes
 627         * near to their parent directory's inode.
 628         */
 629        ext4_group_t    i_block_group;
 630        __u32   i_state;                /* Dynamic state flags for ext4 */
 631
 632        ext4_lblk_t             i_dir_start_lookup;
 633#ifdef CONFIG_EXT4_FS_XATTR
 634        /*
 635         * Extended attributes can be read independently of the main file
 636         * data. Taking i_mutex even when reading would cause contention
 637         * between readers of EAs and writers of regular file data, so
 638         * instead we synchronize on xattr_sem when reading or changing
 639         * EAs.
 640         */
 641        struct rw_semaphore xattr_sem;
 642#endif
 643
 644        struct list_head i_orphan;      /* unlinked but open inodes */
 645
 646        /*
 647         * i_disksize keeps track of what the inode size is ON DISK, not
 648         * in memory.  During truncate, i_size is set to the new size by
 649         * the VFS prior to calling ext4_truncate(), but the filesystem won't
 650         * set i_disksize to 0 until the truncate is actually under way.
 651         *
 652         * The intent is that i_disksize always represents the blocks which
 653         * are used by this file.  This allows recovery to restart truncate
 654         * on orphans if we crash during truncate.  We actually write i_disksize
 655         * into the on-disk inode when writing inodes out, instead of i_size.
 656         *
 657         * The only time when i_disksize and i_size may be different is when
 658         * a truncate is in progress.  The only things which change i_disksize
 659         * are ext4_get_block (growth) and ext4_truncate (shrinkth).
 660         */
 661        loff_t  i_disksize;
 662
 663        /*
 664         * i_data_sem is for serialising ext4_truncate() against
 665         * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
 666         * data tree are chopped off during truncate. We can't do that in
 667         * ext4 because whenever we perform intermediate commits during
 668         * truncate, the inode and all the metadata blocks *must* be in a
 669         * consistent state which allows truncation of the orphans to restart
 670         * during recovery.  Hence we must fix the get_block-vs-truncate race
 671         * by other means, so we have i_data_sem.
 672         */
 673        struct rw_semaphore i_data_sem;
 674        struct inode vfs_inode;
 675        struct jbd2_inode jinode;
 676
 677        struct ext4_ext_cache i_cached_extent;
 678        /*
 679         * File creation time. Its function is same as that of
 680         * struct timespec i_{a,c,m}time in the generic inode.
 681         */
 682        struct timespec i_crtime;
 683
 684        /* mballoc */
 685        struct list_head i_prealloc_list;
 686        spinlock_t i_prealloc_lock;
 687
 688        /* ialloc */
 689        ext4_group_t    i_last_alloc_group;
 690
 691        /* allocation reservation info for delalloc */
 692        unsigned int i_reserved_data_blocks;
 693        unsigned int i_reserved_meta_blocks;
 694        unsigned int i_allocated_meta_blocks;
 695        unsigned short i_delalloc_reserved_flag;
 696
 697        /* on-disk additional length */
 698        __u16 i_extra_isize;
 699
 700        spinlock_t i_block_reservation_lock;
 701
 702        /* completed async DIOs that might need unwritten extents handling */
 703        struct list_head i_aio_dio_complete_list;
 704        /* current io_end structure for async DIO write*/
 705        ext4_io_end_t *cur_aio_dio;
 706};
 707
 708/*
 709 * File system states
 710 */
 711#define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
 712#define EXT4_ERROR_FS                   0x0002  /* Errors detected */
 713#define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
 714
 715/*
 716 * Misc. filesystem flags
 717 */
 718#define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
 719#define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
 720#define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
 721
 722/*
 723 * Mount flags
 724 */
 725#define EXT4_MOUNT_OLDALLOC             0x00002  /* Don't use the new Orlov allocator */
 726#define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
 727#define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
 728#define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
 729#define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
 730#define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
 731#define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
 732#define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
 733#define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
 734#define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
 735#define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
 736#define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
 737#define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
 738#define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
 739#define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
 740#define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
 741#define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
 742#define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
 743#define EXT4_MOUNT_NOBH                 0x40000 /* No bufferheads */
 744#define EXT4_MOUNT_QUOTA                0x80000 /* Some quota option set */
 745#define EXT4_MOUNT_USRQUOTA             0x100000 /* "old" user quota */
 746#define EXT4_MOUNT_GRPQUOTA             0x200000 /* "old" group quota */
 747#define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
 748#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
 749#define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
 750#define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
 751#define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
 752#define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
 753
 754#define clear_opt(o, opt)               o &= ~EXT4_MOUNT_##opt
 755#define set_opt(o, opt)                 o |= EXT4_MOUNT_##opt
 756#define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
 757                                         EXT4_MOUNT_##opt)
 758
 759#define ext4_set_bit                    ext2_set_bit
 760#define ext4_set_bit_atomic             ext2_set_bit_atomic
 761#define ext4_clear_bit                  ext2_clear_bit
 762#define ext4_clear_bit_atomic           ext2_clear_bit_atomic
 763#define ext4_test_bit                   ext2_test_bit
 764#define ext4_find_first_zero_bit        ext2_find_first_zero_bit
 765#define ext4_find_next_zero_bit         ext2_find_next_zero_bit
 766#define ext4_find_next_bit              ext2_find_next_bit
 767
 768/*
 769 * Maximal mount counts between two filesystem checks
 770 */
 771#define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
 772#define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
 773
 774/*
 775 * Behaviour when detecting errors
 776 */
 777#define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
 778#define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
 779#define EXT4_ERRORS_PANIC               3       /* Panic */
 780#define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
 781
 782/*
 783 * Structure of the super block
 784 */
 785struct ext4_super_block {
 786/*00*/  __le32  s_inodes_count;         /* Inodes count */
 787        __le32  s_blocks_count_lo;      /* Blocks count */
 788        __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
 789        __le32  s_free_blocks_count_lo; /* Free blocks count */
 790/*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
 791        __le32  s_first_data_block;     /* First Data Block */
 792        __le32  s_log_block_size;       /* Block size */
 793        __le32  s_obso_log_frag_size;   /* Obsoleted fragment size */
 794/*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
 795        __le32  s_obso_frags_per_group; /* Obsoleted fragments per group */
 796        __le32  s_inodes_per_group;     /* # Inodes per group */
 797        __le32  s_mtime;                /* Mount time */
 798/*30*/  __le32  s_wtime;                /* Write time */
 799        __le16  s_mnt_count;            /* Mount count */
 800        __le16  s_max_mnt_count;        /* Maximal mount count */
 801        __le16  s_magic;                /* Magic signature */
 802        __le16  s_state;                /* File system state */
 803        __le16  s_errors;               /* Behaviour when detecting errors */
 804        __le16  s_minor_rev_level;      /* minor revision level */
 805/*40*/  __le32  s_lastcheck;            /* time of last check */
 806        __le32  s_checkinterval;        /* max. time between checks */
 807        __le32  s_creator_os;           /* OS */
 808        __le32  s_rev_level;            /* Revision level */
 809/*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
 810        __le16  s_def_resgid;           /* Default gid for reserved blocks */
 811        /*
 812         * These fields are for EXT4_DYNAMIC_REV superblocks only.
 813         *
 814         * Note: the difference between the compatible feature set and
 815         * the incompatible feature set is that if there is a bit set
 816         * in the incompatible feature set that the kernel doesn't
 817         * know about, it should refuse to mount the filesystem.
 818         *
 819         * e2fsck's requirements are more strict; if it doesn't know
 820         * about a feature in either the compatible or incompatible
 821         * feature set, it must abort and not try to meddle with
 822         * things it doesn't understand...
 823         */
 824        __le32  s_first_ino;            /* First non-reserved inode */
 825        __le16  s_inode_size;           /* size of inode structure */
 826        __le16  s_block_group_nr;       /* block group # of this superblock */
 827        __le32  s_feature_compat;       /* compatible feature set */
 828/*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
 829        __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
 830/*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
 831/*78*/  char    s_volume_name[16];      /* volume name */
 832/*88*/  char    s_last_mounted[64];     /* directory where last mounted */
 833/*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
 834        /*
 835         * Performance hints.  Directory preallocation should only
 836         * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
 837         */
 838        __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
 839        __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
 840        __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
 841        /*
 842         * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
 843         */
 844/*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
 845/*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
 846        __le32  s_journal_dev;          /* device number of journal file */
 847        __le32  s_last_orphan;          /* start of list of inodes to delete */
 848        __le32  s_hash_seed[4];         /* HTREE hash seed */
 849        __u8    s_def_hash_version;     /* Default hash version to use */
 850        __u8    s_reserved_char_pad;
 851        __le16  s_desc_size;            /* size of group descriptor */
 852/*100*/ __le32  s_default_mount_opts;
 853        __le32  s_first_meta_bg;        /* First metablock block group */
 854        __le32  s_mkfs_time;            /* When the filesystem was created */
 855        __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
 856        /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
 857/*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
 858        __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
 859        __le32  s_free_blocks_count_hi; /* Free blocks count */
 860        __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
 861        __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
 862        __le32  s_flags;                /* Miscellaneous flags */
 863        __le16  s_raid_stride;          /* RAID stride */
 864        __le16  s_mmp_interval;         /* # seconds to wait in MMP checking */
 865        __le64  s_mmp_block;            /* Block for multi-mount protection */
 866        __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
 867        __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
 868        __u8    s_reserved_char_pad2;
 869        __le16  s_reserved_pad;
 870        __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
 871        __u32   s_reserved[160];        /* Padding to the end of the block */
 872};
 873
 874#ifdef __KERNEL__
 875
 876/*
 877 * run-time mount flags
 878 */
 879#define EXT4_MF_MNTDIR_SAMPLED  0x0001
 880#define EXT4_MF_FS_ABORTED      0x0002  /* Fatal error detected */
 881
 882/*
 883 * fourth extended-fs super-block data in memory
 884 */
 885struct ext4_sb_info {
 886        unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
 887        unsigned long s_inodes_per_block;/* Number of inodes per block */
 888        unsigned long s_blocks_per_group;/* Number of blocks in a group */
 889        unsigned long s_inodes_per_group;/* Number of inodes in a group */
 890        unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
 891        unsigned long s_gdb_count;      /* Number of group descriptor blocks */
 892        unsigned long s_desc_per_block; /* Number of group descriptors per block */
 893        ext4_group_t s_groups_count;    /* Number of groups in the fs */
 894        ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
 895        unsigned long s_overhead_last;  /* Last calculated overhead */
 896        unsigned long s_blocks_last;    /* Last seen block count */
 897        loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
 898        struct buffer_head * s_sbh;     /* Buffer containing the super block */
 899        struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
 900        struct buffer_head **s_group_desc;
 901        unsigned int s_mount_opt;
 902        unsigned int s_mount_flags;
 903        ext4_fsblk_t s_sb_block;
 904        uid_t s_resuid;
 905        gid_t s_resgid;
 906        unsigned short s_mount_state;
 907        unsigned short s_pad;
 908        int s_addr_per_block_bits;
 909        int s_desc_per_block_bits;
 910        int s_inode_size;
 911        int s_first_ino;
 912        unsigned int s_inode_readahead_blks;
 913        unsigned int s_inode_goal;
 914        spinlock_t s_next_gen_lock;
 915        u32 s_next_generation;
 916        u32 s_hash_seed[4];
 917        int s_def_hash_version;
 918        int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
 919        struct percpu_counter s_freeblocks_counter;
 920        struct percpu_counter s_freeinodes_counter;
 921        struct percpu_counter s_dirs_counter;
 922        struct percpu_counter s_dirtyblocks_counter;
 923        struct blockgroup_lock *s_blockgroup_lock;
 924        struct proc_dir_entry *s_proc;
 925        struct kobject s_kobj;
 926        struct completion s_kobj_unregister;
 927
 928        /* Journaling */
 929        struct inode *s_journal_inode;
 930        struct journal_s *s_journal;
 931        struct list_head s_orphan;
 932        struct mutex s_orphan_lock;
 933        struct mutex s_resize_lock;
 934        unsigned long s_commit_interval;
 935        u32 s_max_batch_time;
 936        u32 s_min_batch_time;
 937        struct block_device *journal_bdev;
 938#ifdef CONFIG_JBD2_DEBUG
 939        struct timer_list turn_ro_timer;        /* For turning read-only (crash simulation) */
 940        wait_queue_head_t ro_wait_queue;        /* For people waiting for the fs to go read-only */
 941#endif
 942#ifdef CONFIG_QUOTA
 943        char *s_qf_names[MAXQUOTAS];            /* Names of quota files with journalled quota */
 944        int s_jquota_fmt;                       /* Format of quota to use */
 945#endif
 946        unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
 947        struct rb_root system_blks;
 948
 949#ifdef EXTENTS_STATS
 950        /* ext4 extents stats */
 951        unsigned long s_ext_min;
 952        unsigned long s_ext_max;
 953        unsigned long s_depth_max;
 954        spinlock_t s_ext_stats_lock;
 955        unsigned long s_ext_blocks;
 956        unsigned long s_ext_extents;
 957#endif
 958
 959        /* for buddy allocator */
 960        struct ext4_group_info ***s_group_info;
 961        struct inode *s_buddy_cache;
 962        long s_blocks_reserved;
 963        spinlock_t s_reserve_lock;
 964        spinlock_t s_md_lock;
 965        tid_t s_last_transaction;
 966        unsigned short *s_mb_offsets;
 967        unsigned int *s_mb_maxs;
 968
 969        /* tunables */
 970        unsigned long s_stripe;
 971        unsigned int s_mb_stream_request;
 972        unsigned int s_mb_max_to_scan;
 973        unsigned int s_mb_min_to_scan;
 974        unsigned int s_mb_stats;
 975        unsigned int s_mb_order2_reqs;
 976        unsigned int s_mb_group_prealloc;
 977        unsigned int s_max_writeback_mb_bump;
 978        /* where last allocation was done - for stream allocation */
 979        unsigned long s_mb_last_group;
 980        unsigned long s_mb_last_start;
 981
 982        /* stats for buddy allocator */
 983        spinlock_t s_mb_pa_lock;
 984        atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
 985        atomic_t s_bal_success; /* we found long enough chunks */
 986        atomic_t s_bal_allocated;       /* in blocks */
 987        atomic_t s_bal_ex_scanned;      /* total extents scanned */
 988        atomic_t s_bal_goals;   /* goal hits */
 989        atomic_t s_bal_breaks;  /* too long searches */
 990        atomic_t s_bal_2orders; /* 2^order hits */
 991        spinlock_t s_bal_lock;
 992        unsigned long s_mb_buddies_generated;
 993        unsigned long long s_mb_generation_time;
 994        atomic_t s_mb_lost_chunks;
 995        atomic_t s_mb_preallocated;
 996        atomic_t s_mb_discarded;
 997        atomic_t s_lock_busy;
 998
 999        /* locality groups */
1000        struct ext4_locality_group *s_locality_groups;
1001
1002        /* for write statistics */
1003        unsigned long s_sectors_written_start;
1004        u64 s_kbytes_written;
1005
1006        unsigned int s_log_groups_per_flex;
1007        struct flex_groups *s_flex_groups;
1008
1009        /* workqueue for dio unwritten */
1010        struct workqueue_struct *dio_unwritten_wq;
1011};
1012
1013static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1014{
1015        return sb->s_fs_info;
1016}
1017static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1018{
1019        return container_of(inode, struct ext4_inode_info, vfs_inode);
1020}
1021
1022static inline struct timespec ext4_current_time(struct inode *inode)
1023{
1024        return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1025                current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1026}
1027
1028static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1029{
1030        return ino == EXT4_ROOT_INO ||
1031                ino == EXT4_JOURNAL_INO ||
1032                ino == EXT4_RESIZE_INO ||
1033                (ino >= EXT4_FIRST_INO(sb) &&
1034                 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1035}
1036#else
1037/* Assume that user mode programs are passing in an ext4fs superblock, not
1038 * a kernel struct super_block.  This will allow us to call the feature-test
1039 * macros from user land. */
1040#define EXT4_SB(sb)     (sb)
1041#endif
1042
1043#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1044
1045/*
1046 * Codes for operating systems
1047 */
1048#define EXT4_OS_LINUX           0
1049#define EXT4_OS_HURD            1
1050#define EXT4_OS_MASIX           2
1051#define EXT4_OS_FREEBSD         3
1052#define EXT4_OS_LITES           4
1053
1054/*
1055 * Revision levels
1056 */
1057#define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1058#define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1059
1060#define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1061#define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1062
1063#define EXT4_GOOD_OLD_INODE_SIZE 128
1064
1065/*
1066 * Feature set definitions
1067 */
1068
1069#define EXT4_HAS_COMPAT_FEATURE(sb,mask)                        \
1070        ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1071#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)                     \
1072        ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1073#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)                      \
1074        ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1075#define EXT4_SET_COMPAT_FEATURE(sb,mask)                        \
1076        EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1077#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)                     \
1078        EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1079#define EXT4_SET_INCOMPAT_FEATURE(sb,mask)                      \
1080        EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1081#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)                      \
1082        EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1083#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)                   \
1084        EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1085#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)                    \
1086        EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1087
1088#define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1089#define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1090#define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1091#define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1092#define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1093#define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1094
1095#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1096#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1097#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1098#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1099#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1100#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1101#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1102
1103#define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1104#define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1105#define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1106#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1107#define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1108#define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1109#define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1110#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1111#define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1112
1113#define EXT4_FEATURE_COMPAT_SUPP        EXT2_FEATURE_COMPAT_EXT_ATTR
1114#define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1115                                         EXT4_FEATURE_INCOMPAT_RECOVER| \
1116                                         EXT4_FEATURE_INCOMPAT_META_BG| \
1117                                         EXT4_FEATURE_INCOMPAT_EXTENTS| \
1118                                         EXT4_FEATURE_INCOMPAT_64BIT| \
1119                                         EXT4_FEATURE_INCOMPAT_FLEX_BG)
1120#define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1121                                         EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1122                                         EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1123                                         EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1124                                         EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1125                                         EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1126                                         EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1127
1128/*
1129 * Default values for user and/or group using reserved blocks
1130 */
1131#define EXT4_DEF_RESUID         0
1132#define EXT4_DEF_RESGID         0
1133
1134#define EXT4_DEF_INODE_READAHEAD_BLKS   32
1135
1136/*
1137 * Default mount options
1138 */
1139#define EXT4_DEFM_DEBUG         0x0001
1140#define EXT4_DEFM_BSDGROUPS     0x0002
1141#define EXT4_DEFM_XATTR_USER    0x0004
1142#define EXT4_DEFM_ACL           0x0008
1143#define EXT4_DEFM_UID16         0x0010
1144#define EXT4_DEFM_JMODE         0x0060
1145#define EXT4_DEFM_JMODE_DATA    0x0020
1146#define EXT4_DEFM_JMODE_ORDERED 0x0040
1147#define EXT4_DEFM_JMODE_WBACK   0x0060
1148
1149/*
1150 * Default journal batch times
1151 */
1152#define EXT4_DEF_MIN_BATCH_TIME 0
1153#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1154
1155/*
1156 * Minimum number of groups in a flexgroup before we separate out
1157 * directories into the first block group of a flexgroup
1158 */
1159#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1160
1161/*
1162 * Structure of a directory entry
1163 */
1164#define EXT4_NAME_LEN 255
1165
1166struct ext4_dir_entry {
1167        __le32  inode;                  /* Inode number */
1168        __le16  rec_len;                /* Directory entry length */
1169        __le16  name_len;               /* Name length */
1170        char    name[EXT4_NAME_LEN];    /* File name */
1171};
1172
1173/*
1174 * The new version of the directory entry.  Since EXT4 structures are
1175 * stored in intel byte order, and the name_len field could never be
1176 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1177 * file_type field.
1178 */
1179struct ext4_dir_entry_2 {
1180        __le32  inode;                  /* Inode number */
1181        __le16  rec_len;                /* Directory entry length */
1182        __u8    name_len;               /* Name length */
1183        __u8    file_type;
1184        char    name[EXT4_NAME_LEN];    /* File name */
1185};
1186
1187/*
1188 * Ext4 directory file types.  Only the low 3 bits are used.  The
1189 * other bits are reserved for now.
1190 */
1191#define EXT4_FT_UNKNOWN         0
1192#define EXT4_FT_REG_FILE        1
1193#define EXT4_FT_DIR             2
1194#define EXT4_FT_CHRDEV          3
1195#define EXT4_FT_BLKDEV          4
1196#define EXT4_FT_FIFO            5
1197#define EXT4_FT_SOCK            6
1198#define EXT4_FT_SYMLINK         7
1199
1200#define EXT4_FT_MAX             8
1201
1202/*
1203 * EXT4_DIR_PAD defines the directory entries boundaries
1204 *
1205 * NOTE: It must be a multiple of 4
1206 */
1207#define EXT4_DIR_PAD                    4
1208#define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
1209#define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
1210                                         ~EXT4_DIR_ROUND)
1211#define EXT4_MAX_REC_LEN                ((1<<16)-1)
1212
1213/*
1214 * Hash Tree Directory indexing
1215 * (c) Daniel Phillips, 2001
1216 */
1217
1218#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1219                                      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1220                      (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1221#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1222#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1223
1224/* Legal values for the dx_root hash_version field: */
1225
1226#define DX_HASH_LEGACY          0
1227#define DX_HASH_HALF_MD4        1
1228#define DX_HASH_TEA             2
1229#define DX_HASH_LEGACY_UNSIGNED 3
1230#define DX_HASH_HALF_MD4_UNSIGNED       4
1231#define DX_HASH_TEA_UNSIGNED            5
1232
1233#ifdef __KERNEL__
1234
1235/* hash info structure used by the directory hash */
1236struct dx_hash_info
1237{
1238        u32             hash;
1239        u32             minor_hash;
1240        int             hash_version;
1241        u32             *seed;
1242};
1243
1244#define EXT4_HTREE_EOF  0x7fffffff
1245
1246/*
1247 * Control parameters used by ext4_htree_next_block
1248 */
1249#define HASH_NB_ALWAYS          1
1250
1251
1252/*
1253 * Describe an inode's exact location on disk and in memory
1254 */
1255struct ext4_iloc
1256{
1257        struct buffer_head *bh;
1258        unsigned long offset;
1259        ext4_group_t block_group;
1260};
1261
1262static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1263{
1264        return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1265}
1266
1267/*
1268 * This structure is stuffed into the struct file's private_data field
1269 * for directories.  It is where we put information so that we can do
1270 * readdir operations in hash tree order.
1271 */
1272struct dir_private_info {
1273        struct rb_root  root;
1274        struct rb_node  *curr_node;
1275        struct fname    *extra_fname;
1276        loff_t          last_pos;
1277        __u32           curr_hash;
1278        __u32           curr_minor_hash;
1279        __u32           next_hash;
1280};
1281
1282/* calculate the first block number of the group */
1283static inline ext4_fsblk_t
1284ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1285{
1286        return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1287                le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1288}
1289
1290/*
1291 * Special error return code only used by dx_probe() and its callers.
1292 */
1293#define ERR_BAD_DX_DIR  -75000
1294
1295void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1296                        ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1297
1298extern struct proc_dir_entry *ext4_proc_root;
1299
1300/*
1301 * Function prototypes
1302 */
1303
1304/*
1305 * Ok, these declarations are also in <linux/kernel.h> but none of the
1306 * ext4 source programs needs to include it so they are duplicated here.
1307 */
1308# define NORET_TYPE     /**/
1309# define ATTRIB_NORET   __attribute__((noreturn))
1310# define NORET_AND      noreturn,
1311
1312/* bitmap.c */
1313extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1314
1315/* balloc.c */
1316extern unsigned int ext4_block_group(struct super_block *sb,
1317                        ext4_fsblk_t blocknr);
1318extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1319                        ext4_fsblk_t blocknr);
1320extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1321extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1322                        ext4_group_t group);
1323extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1324                        ext4_fsblk_t goal, unsigned long *count, int *errp);
1325extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1326extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1327extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1328                        ext4_fsblk_t block, unsigned long count, int metadata);
1329extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1330                                ext4_fsblk_t block, unsigned long count);
1331extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1332extern void ext4_check_blocks_bitmap(struct super_block *);
1333extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1334                                                    ext4_group_t block_group,
1335                                                    struct buffer_head ** bh);
1336extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1337struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1338                                      ext4_group_t block_group);
1339extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1340                                       struct buffer_head *bh,
1341                                       ext4_group_t group,
1342                                       struct ext4_group_desc *desc);
1343#define ext4_free_blocks_after_init(sb, group, desc)                    \
1344                ext4_init_block_bitmap(sb, NULL, group, desc)
1345
1346/* dir.c */
1347extern int ext4_check_dir_entry(const char *, struct inode *,
1348                                struct ext4_dir_entry_2 *,
1349                                struct buffer_head *, unsigned int);
1350extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1351                                    __u32 minor_hash,
1352                                    struct ext4_dir_entry_2 *dirent);
1353extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1354
1355/* fsync.c */
1356extern int ext4_sync_file(struct file *, struct dentry *, int);
1357
1358/* hash.c */
1359extern int ext4fs_dirhash(const char *name, int len, struct
1360                          dx_hash_info *hinfo);
1361
1362/* ialloc.c */
1363extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1364                                    const struct qstr *qstr, __u32 goal);
1365extern void ext4_free_inode(handle_t *, struct inode *);
1366extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1367extern unsigned long ext4_count_free_inodes(struct super_block *);
1368extern unsigned long ext4_count_dirs(struct super_block *);
1369extern void ext4_check_inodes_bitmap(struct super_block *);
1370extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1371                                       struct buffer_head *bh,
1372                                       ext4_group_t group,
1373                                       struct ext4_group_desc *desc);
1374extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1375
1376/* mballoc.c */
1377extern long ext4_mb_stats;
1378extern long ext4_mb_max_to_scan;
1379extern int ext4_mb_init(struct super_block *, int);
1380extern int ext4_mb_release(struct super_block *);
1381extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1382                                struct ext4_allocation_request *, int *);
1383extern int ext4_mb_reserve_blocks(struct super_block *, int);
1384extern void ext4_discard_preallocations(struct inode *);
1385extern int __init init_ext4_mballoc(void);
1386extern void exit_ext4_mballoc(void);
1387extern void ext4_mb_free_blocks(handle_t *, struct inode *,
1388                ext4_fsblk_t, unsigned long, int, unsigned long *);
1389extern int ext4_mb_add_groupinfo(struct super_block *sb,
1390                ext4_group_t i, struct ext4_group_desc *desc);
1391extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1392extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1393                                                ext4_group_t, int);
1394/* inode.c */
1395int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
1396                struct buffer_head *bh, ext4_fsblk_t blocknr);
1397struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1398                                                ext4_lblk_t, int, int *);
1399struct buffer_head *ext4_bread(handle_t *, struct inode *,
1400                                                ext4_lblk_t, int, int *);
1401int ext4_get_block(struct inode *inode, sector_t iblock,
1402                                struct buffer_head *bh_result, int create);
1403
1404extern struct inode *ext4_iget(struct super_block *, unsigned long);
1405extern int  ext4_write_inode(struct inode *, int);
1406extern int  ext4_setattr(struct dentry *, struct iattr *);
1407extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1408                                struct kstat *stat);
1409extern void ext4_delete_inode(struct inode *);
1410extern int  ext4_sync_inode(handle_t *, struct inode *);
1411extern void ext4_dirty_inode(struct inode *);
1412extern int ext4_change_inode_journal_flag(struct inode *, int);
1413extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1414extern int ext4_can_truncate(struct inode *inode);
1415extern void ext4_truncate(struct inode *);
1416extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1417extern void ext4_set_inode_flags(struct inode *);
1418extern void ext4_get_inode_flags(struct ext4_inode_info *);
1419extern int ext4_alloc_da_blocks(struct inode *inode);
1420extern void ext4_set_aops(struct inode *inode);
1421extern int ext4_writepage_trans_blocks(struct inode *);
1422extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
1423extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1424extern int ext4_block_truncate_page(handle_t *handle,
1425                struct address_space *mapping, loff_t from);
1426extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1427extern qsize_t ext4_get_reserved_space(struct inode *inode);
1428extern int flush_aio_dio_completed_IO(struct inode *inode);
1429/* ioctl.c */
1430extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1431extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1432
1433/* migrate.c */
1434extern int ext4_ext_migrate(struct inode *);
1435
1436/* namei.c */
1437extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1438extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
1439extern int ext4_orphan_add(handle_t *, struct inode *);
1440extern int ext4_orphan_del(handle_t *, struct inode *);
1441extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1442                                __u32 start_minor_hash, __u32 *next_hash);
1443
1444/* resize.c */
1445extern int ext4_group_add(struct super_block *sb,
1446                                struct ext4_new_group_data *input);
1447extern int ext4_group_extend(struct super_block *sb,
1448                                struct ext4_super_block *es,
1449                                ext4_fsblk_t n_blocks_count);
1450
1451/* super.c */
1452extern void ext4_error(struct super_block *, const char *, const char *, ...)
1453        __attribute__ ((format (printf, 3, 4)));
1454extern void __ext4_std_error(struct super_block *, const char *, int);
1455extern void ext4_abort(struct super_block *, const char *, const char *, ...)
1456        __attribute__ ((format (printf, 3, 4)));
1457extern void ext4_warning(struct super_block *, const char *, const char *, ...)
1458        __attribute__ ((format (printf, 3, 4)));
1459extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1460        __attribute__ ((format (printf, 3, 4)));
1461extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1462                                const char *, const char *, ...)
1463        __attribute__ ((format (printf, 4, 5)));
1464extern void ext4_update_dynamic_rev(struct super_block *sb);
1465extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1466                                        __u32 compat);
1467extern int ext4_update_rocompat_feature(handle_t *handle,
1468                                        struct super_block *sb, __u32 rocompat);
1469extern int ext4_update_incompat_feature(handle_t *handle,
1470                                        struct super_block *sb, __u32 incompat);
1471extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1472                                      struct ext4_group_desc *bg);
1473extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1474                                      struct ext4_group_desc *bg);
1475extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1476                                     struct ext4_group_desc *bg);
1477extern __u32 ext4_free_blks_count(struct super_block *sb,
1478                                struct ext4_group_desc *bg);
1479extern __u32 ext4_free_inodes_count(struct super_block *sb,
1480                                 struct ext4_group_desc *bg);
1481extern __u32 ext4_used_dirs_count(struct super_block *sb,
1482                                struct ext4_group_desc *bg);
1483extern __u32 ext4_itable_unused_count(struct super_block *sb,
1484                                   struct ext4_group_desc *bg);
1485extern void ext4_block_bitmap_set(struct super_block *sb,
1486                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1487extern void ext4_inode_bitmap_set(struct super_block *sb,
1488                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1489extern void ext4_inode_table_set(struct super_block *sb,
1490                                 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1491extern void ext4_free_blks_set(struct super_block *sb,
1492                               struct ext4_group_desc *bg, __u32 count);
1493extern void ext4_free_inodes_set(struct super_block *sb,
1494                                struct ext4_group_desc *bg, __u32 count);
1495extern void ext4_used_dirs_set(struct super_block *sb,
1496                                struct ext4_group_desc *bg, __u32 count);
1497extern void ext4_itable_unused_set(struct super_block *sb,
1498                                   struct ext4_group_desc *bg, __u32 count);
1499extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1500                                   struct ext4_group_desc *gdp);
1501extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1502                                       struct ext4_group_desc *gdp);
1503
1504static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1505{
1506        return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1507                le32_to_cpu(es->s_blocks_count_lo);
1508}
1509
1510static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1511{
1512        return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1513                le32_to_cpu(es->s_r_blocks_count_lo);
1514}
1515
1516static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1517{
1518        return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1519                le32_to_cpu(es->s_free_blocks_count_lo);
1520}
1521
1522static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1523                                         ext4_fsblk_t blk)
1524{
1525        es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1526        es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1527}
1528
1529static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1530                                              ext4_fsblk_t blk)
1531{
1532        es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1533        es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1534}
1535
1536static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1537                                           ext4_fsblk_t blk)
1538{
1539        es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1540        es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1541}
1542
1543static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1544{
1545        if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1546                return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1547                        le32_to_cpu(raw_inode->i_size_lo);
1548        else
1549                return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1550}
1551
1552static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1553{
1554        raw_inode->i_size_lo = cpu_to_le32(i_size);
1555        raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1556}
1557
1558static inline
1559struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1560                                            ext4_group_t group)
1561{
1562         struct ext4_group_info ***grp_info;
1563         long indexv, indexh;
1564         grp_info = EXT4_SB(sb)->s_group_info;
1565         indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1566         indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1567         return grp_info[indexv][indexh];
1568}
1569
1570/*
1571 * Reading s_groups_count requires using smp_rmb() afterwards.  See
1572 * the locking protocol documented in the comments of ext4_group_add()
1573 * in resize.c
1574 */
1575static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1576{
1577        ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
1578
1579        smp_rmb();
1580        return ngroups;
1581}
1582
1583static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1584                                             ext4_group_t block_group)
1585{
1586        return block_group >> sbi->s_log_groups_per_flex;
1587}
1588
1589static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1590{
1591        return 1 << sbi->s_log_groups_per_flex;
1592}
1593
1594#define ext4_std_error(sb, errno)                               \
1595do {                                                            \
1596        if ((errno))                                            \
1597                __ext4_std_error((sb), __func__, (errno));      \
1598} while (0)
1599
1600#ifdef CONFIG_SMP
1601/* Each CPU can accumulate percpu_counter_batch blocks in their local
1602 * counters. So we need to make sure we have free blocks more
1603 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
1604 */
1605#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1606#else
1607#define EXT4_FREEBLOCKS_WATERMARK 0
1608#endif
1609
1610static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1611{
1612        /*
1613         * XXX: replace with spinlock if seen contended -bzzz
1614         */
1615        down_write(&EXT4_I(inode)->i_data_sem);
1616        if (newsize > EXT4_I(inode)->i_disksize)
1617                EXT4_I(inode)->i_disksize = newsize;
1618        up_write(&EXT4_I(inode)->i_data_sem);
1619        return ;
1620}
1621
1622struct ext4_group_info {
1623        unsigned long   bb_state;
1624        struct rb_root  bb_free_root;
1625        ext4_grpblk_t   bb_first_free;  /* first free block */
1626        ext4_grpblk_t   bb_free;        /* total free blocks */
1627        ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
1628        struct          list_head bb_prealloc_list;
1629#ifdef DOUBLE_CHECK
1630        void            *bb_bitmap;
1631#endif
1632        struct rw_semaphore alloc_sem;
1633        ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
1634                                         * regions, index is order.
1635                                         * bb_counters[3] = 5 means
1636                                         * 5 free 8-block regions. */
1637};
1638
1639#define EXT4_GROUP_INFO_NEED_INIT_BIT   0
1640
1641#define EXT4_MB_GRP_NEED_INIT(grp)      \
1642        (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1643
1644#define EXT4_MAX_CONTENTION             8
1645#define EXT4_CONTENTION_THRESHOLD       2
1646
1647static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1648                                              ext4_group_t group)
1649{
1650        return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1651}
1652
1653/*
1654 * Returns true if the filesystem is busy enough that attempts to
1655 * access the block group locks has run into contention.
1656 */
1657static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1658{
1659        return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1660}
1661
1662static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1663{
1664        spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1665        if (spin_trylock(lock))
1666                /*
1667                 * We're able to grab the lock right away, so drop the
1668                 * lock contention counter.
1669                 */
1670                atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1671        else {
1672                /*
1673                 * The lock is busy, so bump the contention counter,
1674                 * and then wait on the spin lock.
1675                 */
1676                atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1677                                  EXT4_MAX_CONTENTION);
1678                spin_lock(lock);
1679        }
1680}
1681
1682static inline void ext4_unlock_group(struct super_block *sb,
1683                                        ext4_group_t group)
1684{
1685        spin_unlock(ext4_group_lock_ptr(sb, group));
1686}
1687
1688/*
1689 * Inodes and files operations
1690 */
1691
1692/* dir.c */
1693extern const struct file_operations ext4_dir_operations;
1694
1695/* file.c */
1696extern const struct inode_operations ext4_file_inode_operations;
1697extern const struct file_operations ext4_file_operations;
1698
1699/* namei.c */
1700extern const struct inode_operations ext4_dir_inode_operations;
1701extern const struct inode_operations ext4_special_inode_operations;
1702extern struct dentry *ext4_get_parent(struct dentry *child);
1703
1704/* symlink.c */
1705extern const struct inode_operations ext4_symlink_inode_operations;
1706extern const struct inode_operations ext4_fast_symlink_inode_operations;
1707
1708/* block_validity */
1709extern void ext4_release_system_zone(struct super_block *sb);
1710extern int ext4_setup_system_zone(struct super_block *sb);
1711extern int __init init_ext4_system_zone(void);
1712extern void exit_ext4_system_zone(void);
1713extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1714                                 ext4_fsblk_t start_blk,
1715                                 unsigned int count);
1716
1717/* extents.c */
1718extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1719extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
1720extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1721                                       int chunk);
1722extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
1723                               ext4_lblk_t iblock, unsigned int max_blocks,
1724                               struct buffer_head *bh_result, int flags);
1725extern void ext4_ext_truncate(struct inode *);
1726extern void ext4_ext_init(struct super_block *);
1727extern void ext4_ext_release(struct super_block *);
1728extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1729                          loff_t len);
1730extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1731                          loff_t len);
1732extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1733                           sector_t block, unsigned int max_blocks,
1734                           struct buffer_head *bh, int flags);
1735extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1736                        __u64 start, __u64 len);
1737/* move_extent.c */
1738extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1739                             __u64 start_orig, __u64 start_donor,
1740                             __u64 len, __u64 *moved_len);
1741
1742
1743/*
1744 * Add new method to test wether block and inode bitmaps are properly
1745 * initialized. With uninit_bg reading the block from disk is not enough
1746 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1747 */
1748#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1749
1750static inline int bitmap_uptodate(struct buffer_head *bh)
1751{
1752        return (buffer_uptodate(bh) &&
1753                        test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1754}
1755static inline void set_bitmap_uptodate(struct buffer_head *bh)
1756{
1757        set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1758}
1759
1760#endif  /* __KERNEL__ */
1761
1762#endif  /* _EXT4_H */
1763