linux/include/linux/fs.h
<<
>>
Prefs
   1#ifndef _LINUX_FS_H
   2#define _LINUX_FS_H
   3
   4/*
   5 * This file has definitions for some important file table
   6 * structures etc.
   7 */
   8
   9#include <linux/limits.h>
  10#include <linux/ioctl.h>
  11#include <linux/blk_types.h>
  12#include <linux/types.h>
  13
  14/*
  15 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
  16 * the file limit at runtime and only root can increase the per-process
  17 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
  18 * upper limit on files-per-process.
  19 *
  20 * Some programs (notably those using select()) may have to be 
  21 * recompiled to take full advantage of the new limits..  
  22 */
  23
  24/* Fixed constants first: */
  25#undef NR_OPEN
  26#define INR_OPEN_CUR 1024       /* Initial setting for nfile rlimits */
  27#define INR_OPEN_MAX 4096       /* Hard limit for nfile rlimits */
  28
  29#define BLOCK_SIZE_BITS 10
  30#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
  31
  32#define SEEK_SET        0       /* seek relative to beginning of file */
  33#define SEEK_CUR        1       /* seek relative to current file position */
  34#define SEEK_END        2       /* seek relative to end of file */
  35#define SEEK_DATA       3       /* seek to the next data */
  36#define SEEK_HOLE       4       /* seek to the next hole */
  37#define SEEK_MAX        SEEK_HOLE
  38
  39struct fstrim_range {
  40        __u64 start;
  41        __u64 len;
  42        __u64 minlen;
  43};
  44
  45/* And dynamically-tunable limits and defaults: */
  46struct files_stat_struct {
  47        unsigned long nr_files;         /* read only */
  48        unsigned long nr_free_files;    /* read only */
  49        unsigned long max_files;                /* tunable */
  50};
  51
  52struct inodes_stat_t {
  53        int nr_inodes;
  54        int nr_unused;
  55        int dummy[5];           /* padding for sysctl ABI compatibility */
  56};
  57
  58
  59#define NR_FILE  8192   /* this can well be larger on a larger system */
  60
  61#define MAY_EXEC                0x00000001
  62#define MAY_WRITE               0x00000002
  63#define MAY_READ                0x00000004
  64#define MAY_APPEND              0x00000008
  65#define MAY_ACCESS              0x00000010
  66#define MAY_OPEN                0x00000020
  67#define MAY_CHDIR               0x00000040
  68/* called from RCU mode, don't block */
  69#define MAY_NOT_BLOCK           0x00000080
  70
  71/*
  72 * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
  73 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
  74 */
  75
  76/* file is open for reading */
  77#define FMODE_READ              ((__force fmode_t)0x1)
  78/* file is open for writing */
  79#define FMODE_WRITE             ((__force fmode_t)0x2)
  80/* file is seekable */
  81#define FMODE_LSEEK             ((__force fmode_t)0x4)
  82/* file can be accessed using pread */
  83#define FMODE_PREAD             ((__force fmode_t)0x8)
  84/* file can be accessed using pwrite */
  85#define FMODE_PWRITE            ((__force fmode_t)0x10)
  86/* File is opened for execution with sys_execve / sys_uselib */
  87#define FMODE_EXEC              ((__force fmode_t)0x20)
  88/* File is opened with O_NDELAY (only set for block devices) */
  89#define FMODE_NDELAY            ((__force fmode_t)0x40)
  90/* File is opened with O_EXCL (only set for block devices) */
  91#define FMODE_EXCL              ((__force fmode_t)0x80)
  92/* File is opened using open(.., 3, ..) and is writeable only for ioctls
  93   (specialy hack for floppy.c) */
  94#define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
  95
  96/*
  97 * Don't update ctime and mtime.
  98 *
  99 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
 100 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
 101 */
 102#define FMODE_NOCMTIME          ((__force fmode_t)0x800)
 103
 104/* Expect random access pattern */
 105#define FMODE_RANDOM            ((__force fmode_t)0x1000)
 106
 107/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
 108#define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
 109
 110/* File is opened with O_PATH; almost nothing can be done with it */
 111#define FMODE_PATH              ((__force fmode_t)0x4000)
 112
 113/* File was opened by fanotify and shouldn't generate fanotify events */
 114#define FMODE_NONOTIFY          ((__force fmode_t)0x1000000)
 115
 116/*
 117 * The below are the various read and write types that we support. Some of
 118 * them include behavioral modifiers that send information down to the
 119 * block layer and IO scheduler. Terminology:
 120 *
 121 *      The block layer uses device plugging to defer IO a little bit, in
 122 *      the hope that we will see more IO very shortly. This increases
 123 *      coalescing of adjacent IO and thus reduces the number of IOs we
 124 *      have to send to the device. It also allows for better queuing,
 125 *      if the IO isn't mergeable. If the caller is going to be waiting
 126 *      for the IO, then he must ensure that the device is unplugged so
 127 *      that the IO is dispatched to the driver.
 128 *
 129 *      All IO is handled async in Linux. This is fine for background
 130 *      writes, but for reads or writes that someone waits for completion
 131 *      on, we want to notify the block layer and IO scheduler so that they
 132 *      know about it. That allows them to make better scheduling
 133 *      decisions. So when the below references 'sync' and 'async', it
 134 *      is referencing this priority hint.
 135 *
 136 * With that in mind, the available types are:
 137 *
 138 * READ                 A normal read operation. Device will be plugged.
 139 * READ_SYNC            A synchronous read. Device is not plugged, caller can
 140 *                      immediately wait on this read without caring about
 141 *                      unplugging.
 142 * READA                Used for read-ahead operations. Lower priority, and the
 143 *                      block layer could (in theory) choose to ignore this
 144 *                      request if it runs into resource problems.
 145 * WRITE                A normal async write. Device will be plugged.
 146 * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
 147 *                      the hint that someone will be waiting on this IO
 148 *                      shortly. The write equivalent of READ_SYNC.
 149 * WRITE_ODIRECT        Special case write for O_DIRECT only.
 150 * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
 151 * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
 152 *                      non-volatile media on completion.
 153 * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
 154 *                      by a cache flush and data is guaranteed to be on
 155 *                      non-volatile media on completion.
 156 *
 157 */
 158#define RW_MASK                 REQ_WRITE
 159#define RWA_MASK                REQ_RAHEAD
 160
 161#define READ                    0
 162#define WRITE                   RW_MASK
 163#define READA                   RWA_MASK
 164
 165#define READ_SYNC               (READ | REQ_SYNC)
 166#define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
 167#define WRITE_ODIRECT           (WRITE | REQ_SYNC)
 168#define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
 169#define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
 170#define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
 171
 172#define SEL_IN          1
 173#define SEL_OUT         2
 174#define SEL_EX          4
 175
 176/* public flags for file_system_type */
 177#define FS_REQUIRES_DEV 1 
 178#define FS_BINARY_MOUNTDATA 2
 179#define FS_HAS_SUBTYPE 4
 180#define FS_REVAL_DOT    16384   /* Check the paths ".", ".." for staleness */
 181#define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move()
 182                                         * during rename() internally.
 183                                         */
 184
 185/*
 186 * These are the fs-independent mount-flags: up to 32 flags are supported
 187 */
 188#define MS_RDONLY        1      /* Mount read-only */
 189#define MS_NOSUID        2      /* Ignore suid and sgid bits */
 190#define MS_NODEV         4      /* Disallow access to device special files */
 191#define MS_NOEXEC        8      /* Disallow program execution */
 192#define MS_SYNCHRONOUS  16      /* Writes are synced at once */
 193#define MS_REMOUNT      32      /* Alter flags of a mounted FS */
 194#define MS_MANDLOCK     64      /* Allow mandatory locks on an FS */
 195#define MS_DIRSYNC      128     /* Directory modifications are synchronous */
 196#define MS_NOATIME      1024    /* Do not update access times. */
 197#define MS_NODIRATIME   2048    /* Do not update directory access times */
 198#define MS_BIND         4096
 199#define MS_MOVE         8192
 200#define MS_REC          16384
 201#define MS_VERBOSE      32768   /* War is peace. Verbosity is silence.
 202                                   MS_VERBOSE is deprecated. */
 203#define MS_SILENT       32768
 204#define MS_POSIXACL     (1<<16) /* VFS does not apply the umask */
 205#define MS_UNBINDABLE   (1<<17) /* change to unbindable */
 206#define MS_PRIVATE      (1<<18) /* change to private */
 207#define MS_SLAVE        (1<<19) /* change to slave */
 208#define MS_SHARED       (1<<20) /* change to shared */
 209#define MS_RELATIME     (1<<21) /* Update atime relative to mtime/ctime. */
 210#define MS_KERNMOUNT    (1<<22) /* this is a kern_mount call */
 211#define MS_I_VERSION    (1<<23) /* Update inode I_version field */
 212#define MS_STRICTATIME  (1<<24) /* Always perform atime updates */
 213#define MS_NOSEC        (1<<28)
 214#define MS_BORN         (1<<29)
 215#define MS_ACTIVE       (1<<30)
 216#define MS_NOUSER       (1<<31)
 217
 218/*
 219 * Superblock flags that can be altered by MS_REMOUNT
 220 */
 221#define MS_RMT_MASK     (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
 222
 223/*
 224 * Old magic mount flag and mask
 225 */
 226#define MS_MGC_VAL 0xC0ED0000
 227#define MS_MGC_MSK 0xffff0000
 228
 229/* Inode flags - they have nothing to superblock flags now */
 230
 231#define S_SYNC          1       /* Writes are synced at once */
 232#define S_NOATIME       2       /* Do not update access times */
 233#define S_APPEND        4       /* Append-only file */
 234#define S_IMMUTABLE     8       /* Immutable file */
 235#define S_DEAD          16      /* removed, but still open directory */
 236#define S_NOQUOTA       32      /* Inode is not counted to quota */
 237#define S_DIRSYNC       64      /* Directory modifications are synchronous */
 238#define S_NOCMTIME      128     /* Do not update file c/mtime */
 239#define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
 240#define S_PRIVATE       512     /* Inode is fs-internal */
 241#define S_IMA           1024    /* Inode has an associated IMA struct */
 242#define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
 243#define S_NOSEC         4096    /* no suid or xattr security attributes */
 244
 245/*
 246 * Note that nosuid etc flags are inode-specific: setting some file-system
 247 * flags just means all the inodes inherit those flags by default. It might be
 248 * possible to override it selectively if you really wanted to with some
 249 * ioctl() that is not currently implemented.
 250 *
 251 * Exception: MS_RDONLY is always applied to the entire file system.
 252 *
 253 * Unfortunately, it is possible to change a filesystems flags with it mounted
 254 * with files in use.  This means that all of the inodes will not have their
 255 * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
 256 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
 257 */
 258#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
 259
 260#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
 261#define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
 262                                        ((inode)->i_flags & S_SYNC))
 263#define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
 264                                        ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
 265#define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
 266#define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
 267#define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
 268
 269#define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
 270#define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
 271#define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
 272#define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
 273
 274#define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
 275#define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
 276#define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
 277#define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
 278#define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
 279#define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
 280#define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
 281
 282/* the read-only stuff doesn't really belong here, but any other place is
 283   probably as bad and I don't want to create yet another include file. */
 284
 285#define BLKROSET   _IO(0x12,93) /* set device read-only (0 = read-write) */
 286#define BLKROGET   _IO(0x12,94) /* get read-only status (0 = read_write) */
 287#define BLKRRPART  _IO(0x12,95) /* re-read partition table */
 288#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
 289#define BLKFLSBUF  _IO(0x12,97) /* flush buffer cache */
 290#define BLKRASET   _IO(0x12,98) /* set read ahead for block device */
 291#define BLKRAGET   _IO(0x12,99) /* get current read ahead setting */
 292#define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
 293#define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
 294#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
 295#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
 296#define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
 297#if 0
 298#define BLKPG      _IO(0x12,105)/* See blkpg.h */
 299
 300/* Some people are morons.  Do not use sizeof! */
 301
 302#define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
 303#define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
 304/* This was here just to show that the number is taken -
 305   probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
 306#endif
 307/* A jump here: 108-111 have been used for various private purposes. */
 308#define BLKBSZGET  _IOR(0x12,112,size_t)
 309#define BLKBSZSET  _IOW(0x12,113,size_t)
 310#define BLKGETSIZE64 _IOR(0x12,114,size_t)      /* return device size in bytes (u64 *arg) */
 311#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
 312#define BLKTRACESTART _IO(0x12,116)
 313#define BLKTRACESTOP _IO(0x12,117)
 314#define BLKTRACETEARDOWN _IO(0x12,118)
 315#define BLKDISCARD _IO(0x12,119)
 316#define BLKIOMIN _IO(0x12,120)
 317#define BLKIOOPT _IO(0x12,121)
 318#define BLKALIGNOFF _IO(0x12,122)
 319#define BLKPBSZGET _IO(0x12,123)
 320#define BLKDISCARDZEROES _IO(0x12,124)
 321#define BLKSECDISCARD _IO(0x12,125)
 322
 323#define BMAP_IOCTL 1            /* obsolete - kept for compatibility */
 324#define FIBMAP     _IO(0x00,1)  /* bmap access */
 325#define FIGETBSZ   _IO(0x00,2)  /* get the block size used for bmap */
 326#define FIFREEZE        _IOWR('X', 119, int)    /* Freeze */
 327#define FITHAW          _IOWR('X', 120, int)    /* Thaw */
 328#define FITRIM          _IOWR('X', 121, struct fstrim_range)    /* Trim */
 329
 330#define FS_IOC_GETFLAGS                 _IOR('f', 1, long)
 331#define FS_IOC_SETFLAGS                 _IOW('f', 2, long)
 332#define FS_IOC_GETVERSION               _IOR('v', 1, long)
 333#define FS_IOC_SETVERSION               _IOW('v', 2, long)
 334#define FS_IOC_FIEMAP                   _IOWR('f', 11, struct fiemap)
 335#define FS_IOC32_GETFLAGS               _IOR('f', 1, int)
 336#define FS_IOC32_SETFLAGS               _IOW('f', 2, int)
 337#define FS_IOC32_GETVERSION             _IOR('v', 1, int)
 338#define FS_IOC32_SETVERSION             _IOW('v', 2, int)
 339
 340/*
 341 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
 342 */
 343#define FS_SECRM_FL                     0x00000001 /* Secure deletion */
 344#define FS_UNRM_FL                      0x00000002 /* Undelete */
 345#define FS_COMPR_FL                     0x00000004 /* Compress file */
 346#define FS_SYNC_FL                      0x00000008 /* Synchronous updates */
 347#define FS_IMMUTABLE_FL                 0x00000010 /* Immutable file */
 348#define FS_APPEND_FL                    0x00000020 /* writes to file may only append */
 349#define FS_NODUMP_FL                    0x00000040 /* do not dump file */
 350#define FS_NOATIME_FL                   0x00000080 /* do not update atime */
 351/* Reserved for compression usage... */
 352#define FS_DIRTY_FL                     0x00000100
 353#define FS_COMPRBLK_FL                  0x00000200 /* One or more compressed clusters */
 354#define FS_NOCOMP_FL                    0x00000400 /* Don't compress */
 355#define FS_ECOMPR_FL                    0x00000800 /* Compression error */
 356/* End compression flags --- maybe not all used */
 357#define FS_BTREE_FL                     0x00001000 /* btree format dir */
 358#define FS_INDEX_FL                     0x00001000 /* hash-indexed directory */
 359#define FS_IMAGIC_FL                    0x00002000 /* AFS directory */
 360#define FS_JOURNAL_DATA_FL              0x00004000 /* Reserved for ext3 */
 361#define FS_NOTAIL_FL                    0x00008000 /* file tail should not be merged */
 362#define FS_DIRSYNC_FL                   0x00010000 /* dirsync behaviour (directories only) */
 363#define FS_TOPDIR_FL                    0x00020000 /* Top of directory hierarchies*/
 364#define FS_EXTENT_FL                    0x00080000 /* Extents */
 365#define FS_DIRECTIO_FL                  0x00100000 /* Use direct i/o */
 366#define FS_NOCOW_FL                     0x00800000 /* Do not cow file */
 367#define FS_RESERVED_FL                  0x80000000 /* reserved for ext2 lib */
 368
 369#define FS_FL_USER_VISIBLE              0x0003DFFF /* User visible flags */
 370#define FS_FL_USER_MODIFIABLE           0x000380FF /* User modifiable flags */
 371
 372
 373#define SYNC_FILE_RANGE_WAIT_BEFORE     1
 374#define SYNC_FILE_RANGE_WRITE           2
 375#define SYNC_FILE_RANGE_WAIT_AFTER      4
 376
 377#ifdef __KERNEL__
 378
 379#include <linux/linkage.h>
 380#include <linux/wait.h>
 381#include <linux/kdev_t.h>
 382#include <linux/dcache.h>
 383#include <linux/path.h>
 384#include <linux/stat.h>
 385#include <linux/cache.h>
 386#include <linux/list.h>
 387#include <linux/radix-tree.h>
 388#include <linux/prio_tree.h>
 389#include <linux/init.h>
 390#include <linux/pid.h>
 391#include <linux/mutex.h>
 392#include <linux/capability.h>
 393#include <linux/semaphore.h>
 394#include <linux/fiemap.h>
 395#include <linux/rculist_bl.h>
 396#include <linux/atomic.h>
 397#include <linux/shrinker.h>
 398
 399#include <asm/byteorder.h>
 400
 401struct export_operations;
 402struct hd_geometry;
 403struct iovec;
 404struct nameidata;
 405struct kiocb;
 406struct kobject;
 407struct pipe_inode_info;
 408struct poll_table_struct;
 409struct kstatfs;
 410struct vm_area_struct;
 411struct vfsmount;
 412struct cred;
 413
 414extern void __init inode_init(void);
 415extern void __init inode_init_early(void);
 416extern void __init files_init(unsigned long);
 417
 418extern struct files_stat_struct files_stat;
 419extern unsigned long get_max_files(void);
 420extern int sysctl_nr_open;
 421extern struct inodes_stat_t inodes_stat;
 422extern int leases_enable, lease_break_time;
 423
 424struct buffer_head;
 425typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 426                        struct buffer_head *bh_result, int create);
 427typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 428                        ssize_t bytes, void *private, int ret,
 429                        bool is_async);
 430
 431/*
 432 * Attribute flags.  These should be or-ed together to figure out what
 433 * has been changed!
 434 */
 435#define ATTR_MODE       (1 << 0)
 436#define ATTR_UID        (1 << 1)
 437#define ATTR_GID        (1 << 2)
 438#define ATTR_SIZE       (1 << 3)
 439#define ATTR_ATIME      (1 << 4)
 440#define ATTR_MTIME      (1 << 5)
 441#define ATTR_CTIME      (1 << 6)
 442#define ATTR_ATIME_SET  (1 << 7)
 443#define ATTR_MTIME_SET  (1 << 8)
 444#define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
 445#define ATTR_ATTR_FLAG  (1 << 10)
 446#define ATTR_KILL_SUID  (1 << 11)
 447#define ATTR_KILL_SGID  (1 << 12)
 448#define ATTR_FILE       (1 << 13)
 449#define ATTR_KILL_PRIV  (1 << 14)
 450#define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
 451#define ATTR_TIMES_SET  (1 << 16)
 452
 453/*
 454 * This is the Inode Attributes structure, used for notify_change().  It
 455 * uses the above definitions as flags, to know which values have changed.
 456 * Also, in this manner, a Filesystem can look at only the values it cares
 457 * about.  Basically, these are the attributes that the VFS layer can
 458 * request to change from the FS layer.
 459 *
 460 * Derek Atkins <warlord@MIT.EDU> 94-10-20
 461 */
 462struct iattr {
 463        unsigned int    ia_valid;
 464        umode_t         ia_mode;
 465        uid_t           ia_uid;
 466        gid_t           ia_gid;
 467        loff_t          ia_size;
 468        struct timespec ia_atime;
 469        struct timespec ia_mtime;
 470        struct timespec ia_ctime;
 471
 472        /*
 473         * Not an attribute, but an auxiliary info for filesystems wanting to
 474         * implement an ftruncate() like method.  NOTE: filesystem should
 475         * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
 476         */
 477        struct file     *ia_file;
 478};
 479
 480/*
 481 * Includes for diskquotas.
 482 */
 483#include <linux/quota.h>
 484
 485/** 
 486 * enum positive_aop_returns - aop return codes with specific semantics
 487 *
 488 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
 489 *                          completed, that the page is still locked, and
 490 *                          should be considered active.  The VM uses this hint
 491 *                          to return the page to the active list -- it won't
 492 *                          be a candidate for writeback again in the near
 493 *                          future.  Other callers must be careful to unlock
 494 *                          the page if they get this return.  Returned by
 495 *                          writepage(); 
 496 *
 497 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
 498 *                      unlocked it and the page might have been truncated.
 499 *                      The caller should back up to acquiring a new page and
 500 *                      trying again.  The aop will be taking reasonable
 501 *                      precautions not to livelock.  If the caller held a page
 502 *                      reference, it should drop it before retrying.  Returned
 503 *                      by readpage().
 504 *
 505 * address_space_operation functions return these large constants to indicate
 506 * special semantics to the caller.  These are much larger than the bytes in a
 507 * page to allow for functions that return the number of bytes operated on in a
 508 * given page.
 509 */
 510
 511enum positive_aop_returns {
 512        AOP_WRITEPAGE_ACTIVATE  = 0x80000,
 513        AOP_TRUNCATED_PAGE      = 0x80001,
 514};
 515
 516#define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
 517#define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
 518#define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
 519                                                * helper code (eg buffer layer)
 520                                                * to clear GFP_FS from alloc */
 521
 522/*
 523 * oh the beauties of C type declarations.
 524 */
 525struct page;
 526struct address_space;
 527struct writeback_control;
 528
 529struct iov_iter {
 530        const struct iovec *iov;
 531        unsigned long nr_segs;
 532        size_t iov_offset;
 533        size_t count;
 534};
 535
 536size_t iov_iter_copy_from_user_atomic(struct page *page,
 537                struct iov_iter *i, unsigned long offset, size_t bytes);
 538size_t iov_iter_copy_from_user(struct page *page,
 539                struct iov_iter *i, unsigned long offset, size_t bytes);
 540void iov_iter_advance(struct iov_iter *i, size_t bytes);
 541int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
 542size_t iov_iter_single_seg_count(struct iov_iter *i);
 543
 544static inline void iov_iter_init(struct iov_iter *i,
 545                        const struct iovec *iov, unsigned long nr_segs,
 546                        size_t count, size_t written)
 547{
 548        i->iov = iov;
 549        i->nr_segs = nr_segs;
 550        i->iov_offset = 0;
 551        i->count = count + written;
 552
 553        iov_iter_advance(i, written);
 554}
 555
 556static inline size_t iov_iter_count(struct iov_iter *i)
 557{
 558        return i->count;
 559}
 560
 561/*
 562 * "descriptor" for what we're up to with a read.
 563 * This allows us to use the same read code yet
 564 * have multiple different users of the data that
 565 * we read from a file.
 566 *
 567 * The simplest case just copies the data to user
 568 * mode.
 569 */
 570typedef struct {
 571        size_t written;
 572        size_t count;
 573        union {
 574                char __user *buf;
 575                void *data;
 576        } arg;
 577        int error;
 578} read_descriptor_t;
 579
 580typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
 581                unsigned long, unsigned long);
 582
 583struct address_space_operations {
 584        int (*writepage)(struct page *page, struct writeback_control *wbc);
 585        int (*readpage)(struct file *, struct page *);
 586
 587        /* Write back some dirty pages from this mapping. */
 588        int (*writepages)(struct address_space *, struct writeback_control *);
 589
 590        /* Set a page dirty.  Return true if this dirtied it */
 591        int (*set_page_dirty)(struct page *page);
 592
 593        int (*readpages)(struct file *filp, struct address_space *mapping,
 594                        struct list_head *pages, unsigned nr_pages);
 595
 596        int (*write_begin)(struct file *, struct address_space *mapping,
 597                                loff_t pos, unsigned len, unsigned flags,
 598                                struct page **pagep, void **fsdata);
 599        int (*write_end)(struct file *, struct address_space *mapping,
 600                                loff_t pos, unsigned len, unsigned copied,
 601                                struct page *page, void *fsdata);
 602
 603        /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
 604        sector_t (*bmap)(struct address_space *, sector_t);
 605        void (*invalidatepage) (struct page *, unsigned long);
 606        int (*releasepage) (struct page *, gfp_t);
 607        void (*freepage)(struct page *);
 608        ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
 609                        loff_t offset, unsigned long nr_segs);
 610        int (*get_xip_mem)(struct address_space *, pgoff_t, int,
 611                                                void **, unsigned long *);
 612        /* migrate the contents of a page to the specified target */
 613        int (*migratepage) (struct address_space *,
 614                        struct page *, struct page *);
 615        int (*launder_page) (struct page *);
 616        int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
 617                                        unsigned long);
 618        int (*error_remove_page)(struct address_space *, struct page *);
 619};
 620
 621extern const struct address_space_operations empty_aops;
 622
 623/*
 624 * pagecache_write_begin/pagecache_write_end must be used by general code
 625 * to write into the pagecache.
 626 */
 627int pagecache_write_begin(struct file *, struct address_space *mapping,
 628                                loff_t pos, unsigned len, unsigned flags,
 629                                struct page **pagep, void **fsdata);
 630
 631int pagecache_write_end(struct file *, struct address_space *mapping,
 632                                loff_t pos, unsigned len, unsigned copied,
 633                                struct page *page, void *fsdata);
 634
 635struct backing_dev_info;
 636struct address_space {
 637        struct inode            *host;          /* owner: inode, block_device */
 638        struct radix_tree_root  page_tree;      /* radix tree of all pages */
 639        spinlock_t              tree_lock;      /* and lock protecting it */
 640        unsigned int            i_mmap_writable;/* count VM_SHARED mappings */
 641        struct prio_tree_root   i_mmap;         /* tree of private and shared mappings */
 642        struct list_head        i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
 643        struct mutex            i_mmap_mutex;   /* protect tree, count, list */
 644        /* Protected by tree_lock together with the radix tree */
 645        unsigned long           nrpages;        /* number of total pages */
 646        pgoff_t                 writeback_index;/* writeback starts here */
 647        const struct address_space_operations *a_ops;   /* methods */
 648        unsigned long           flags;          /* error bits/gfp mask */
 649        struct backing_dev_info *backing_dev_info; /* device readahead, etc */
 650        spinlock_t              private_lock;   /* for use by the address_space */
 651        struct list_head        private_list;   /* ditto */
 652        struct address_space    *assoc_mapping; /* ditto */
 653} __attribute__((aligned(sizeof(long))));
 654        /*
 655         * On most architectures that alignment is already the case; but
 656         * must be enforced here for CRIS, to let the least significant bit
 657         * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
 658         */
 659
 660struct block_device {
 661        dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
 662        int                     bd_openers;
 663        struct inode *          bd_inode;       /* will die */
 664        struct super_block *    bd_super;
 665        struct mutex            bd_mutex;       /* open/close mutex */
 666        struct list_head        bd_inodes;
 667        void *                  bd_claiming;
 668        void *                  bd_holder;
 669        int                     bd_holders;
 670        bool                    bd_write_holder;
 671#ifdef CONFIG_SYSFS
 672        struct list_head        bd_holder_disks;
 673#endif
 674        struct block_device *   bd_contains;
 675        unsigned                bd_block_size;
 676        struct hd_struct *      bd_part;
 677        /* number of times partitions within this device have been opened. */
 678        unsigned                bd_part_count;
 679        int                     bd_invalidated;
 680        struct gendisk *        bd_disk;
 681        struct list_head        bd_list;
 682        /*
 683         * Private data.  You must have bd_claim'ed the block_device
 684         * to use this.  NOTE:  bd_claim allows an owner to claim
 685         * the same device multiple times, the owner must take special
 686         * care to not mess up bd_private for that case.
 687         */
 688        unsigned long           bd_private;
 689
 690        /* The counter of freeze processes */
 691        int                     bd_fsfreeze_count;
 692        /* Mutex for freeze */
 693        struct mutex            bd_fsfreeze_mutex;
 694};
 695
 696/*
 697 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
 698 * radix trees
 699 */
 700#define PAGECACHE_TAG_DIRTY     0
 701#define PAGECACHE_TAG_WRITEBACK 1
 702#define PAGECACHE_TAG_TOWRITE   2
 703
 704int mapping_tagged(struct address_space *mapping, int tag);
 705
 706/*
 707 * Might pages of this file be mapped into userspace?
 708 */
 709static inline int mapping_mapped(struct address_space *mapping)
 710{
 711        return  !prio_tree_empty(&mapping->i_mmap) ||
 712                !list_empty(&mapping->i_mmap_nonlinear);
 713}
 714
 715/*
 716 * Might pages of this file have been modified in userspace?
 717 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
 718 * marks vma as VM_SHARED if it is shared, and the file was opened for
 719 * writing i.e. vma may be mprotected writable even if now readonly.
 720 */
 721static inline int mapping_writably_mapped(struct address_space *mapping)
 722{
 723        return mapping->i_mmap_writable != 0;
 724}
 725
 726/*
 727 * Use sequence counter to get consistent i_size on 32-bit processors.
 728 */
 729#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 730#include <linux/seqlock.h>
 731#define __NEED_I_SIZE_ORDERED
 732#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
 733#else
 734#define i_size_ordered_init(inode) do { } while (0)
 735#endif
 736
 737struct posix_acl;
 738#define ACL_NOT_CACHED ((void *)(-1))
 739
 740#define IOP_FASTPERM    0x0001
 741#define IOP_LOOKUP      0x0002
 742#define IOP_NOFOLLOW    0x0004
 743
 744/*
 745 * Keep mostly read-only and often accessed (especially for
 746 * the RCU path lookup and 'stat' data) fields at the beginning
 747 * of the 'struct inode'
 748 */
 749struct inode {
 750        umode_t                 i_mode;
 751        unsigned short          i_opflags;
 752        uid_t                   i_uid;
 753        gid_t                   i_gid;
 754        unsigned int            i_flags;
 755
 756#ifdef CONFIG_FS_POSIX_ACL
 757        struct posix_acl        *i_acl;
 758        struct posix_acl        *i_default_acl;
 759#endif
 760
 761        const struct inode_operations   *i_op;
 762        struct super_block      *i_sb;
 763        struct address_space    *i_mapping;
 764
 765#ifdef CONFIG_SECURITY
 766        void                    *i_security;
 767#endif
 768
 769        /* Stat data, not accessed from path walking */
 770        unsigned long           i_ino;
 771        /*
 772         * Filesystems may only read i_nlink directly.  They shall use the
 773         * following functions for modification:
 774         *
 775         *    (set|clear|inc|drop)_nlink
 776         *    inode_(inc|dec)_link_count
 777         */
 778        union {
 779                const unsigned int i_nlink;
 780                unsigned int __i_nlink;
 781        };
 782        dev_t                   i_rdev;
 783        struct timespec         i_atime;
 784        struct timespec         i_mtime;
 785        struct timespec         i_ctime;
 786        spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
 787        unsigned short          i_bytes;
 788        blkcnt_t                i_blocks;
 789        loff_t                  i_size;
 790
 791#ifdef __NEED_I_SIZE_ORDERED
 792        seqcount_t              i_size_seqcount;
 793#endif
 794
 795        /* Misc */
 796        unsigned long           i_state;
 797        struct mutex            i_mutex;
 798
 799        unsigned long           dirtied_when;   /* jiffies of first dirtying */
 800
 801        struct hlist_node       i_hash;
 802        struct list_head        i_wb_list;      /* backing dev IO list */
 803        struct list_head        i_lru;          /* inode LRU list */
 804        struct list_head        i_sb_list;
 805        union {
 806                struct list_head        i_dentry;
 807                struct rcu_head         i_rcu;
 808        };
 809        atomic_t                i_count;
 810        unsigned int            i_blkbits;
 811        u64                     i_version;
 812        atomic_t                i_dio_count;
 813        atomic_t                i_writecount;
 814        const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
 815        struct file_lock        *i_flock;
 816        struct address_space    i_data;
 817#ifdef CONFIG_QUOTA
 818        struct dquot            *i_dquot[MAXQUOTAS];
 819#endif
 820        struct list_head        i_devices;
 821        union {
 822                struct pipe_inode_info  *i_pipe;
 823                struct block_device     *i_bdev;
 824                struct cdev             *i_cdev;
 825        };
 826
 827        __u32                   i_generation;
 828
 829#ifdef CONFIG_FSNOTIFY
 830        __u32                   i_fsnotify_mask; /* all events this inode cares about */
 831        struct hlist_head       i_fsnotify_marks;
 832#endif
 833
 834#ifdef CONFIG_IMA
 835        atomic_t                i_readcount; /* struct files open RO */
 836#endif
 837        void                    *i_private; /* fs or device private pointer */
 838};
 839
 840static inline int inode_unhashed(struct inode *inode)
 841{
 842        return hlist_unhashed(&inode->i_hash);
 843}
 844
 845/*
 846 * inode->i_mutex nesting subclasses for the lock validator:
 847 *
 848 * 0: the object of the current VFS operation
 849 * 1: parent
 850 * 2: child/target
 851 * 3: quota file
 852 *
 853 * The locking order between these classes is
 854 * parent -> child -> normal -> xattr -> quota
 855 */
 856enum inode_i_mutex_lock_class
 857{
 858        I_MUTEX_NORMAL,
 859        I_MUTEX_PARENT,
 860        I_MUTEX_CHILD,
 861        I_MUTEX_XATTR,
 862        I_MUTEX_QUOTA
 863};
 864
 865/*
 866 * NOTE: in a 32bit arch with a preemptable kernel and
 867 * an UP compile the i_size_read/write must be atomic
 868 * with respect to the local cpu (unlike with preempt disabled),
 869 * but they don't need to be atomic with respect to other cpus like in
 870 * true SMP (so they need either to either locally disable irq around
 871 * the read or for example on x86 they can be still implemented as a
 872 * cmpxchg8b without the need of the lock prefix). For SMP compiles
 873 * and 64bit archs it makes no difference if preempt is enabled or not.
 874 */
 875static inline loff_t i_size_read(const struct inode *inode)
 876{
 877#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 878        loff_t i_size;
 879        unsigned int seq;
 880
 881        do {
 882                seq = read_seqcount_begin(&inode->i_size_seqcount);
 883                i_size = inode->i_size;
 884        } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
 885        return i_size;
 886#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 887        loff_t i_size;
 888
 889        preempt_disable();
 890        i_size = inode->i_size;
 891        preempt_enable();
 892        return i_size;
 893#else
 894        return inode->i_size;
 895#endif
 896}
 897
 898/*
 899 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
 900 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
 901 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
 902 */
 903static inline void i_size_write(struct inode *inode, loff_t i_size)
 904{
 905#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 906        write_seqcount_begin(&inode->i_size_seqcount);
 907        inode->i_size = i_size;
 908        write_seqcount_end(&inode->i_size_seqcount);
 909#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 910        preempt_disable();
 911        inode->i_size = i_size;
 912        preempt_enable();
 913#else
 914        inode->i_size = i_size;
 915#endif
 916}
 917
 918static inline unsigned iminor(const struct inode *inode)
 919{
 920        return MINOR(inode->i_rdev);
 921}
 922
 923static inline unsigned imajor(const struct inode *inode)
 924{
 925        return MAJOR(inode->i_rdev);
 926}
 927
 928extern struct block_device *I_BDEV(struct inode *inode);
 929
 930struct fown_struct {
 931        rwlock_t lock;          /* protects pid, uid, euid fields */
 932        struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
 933        enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
 934        uid_t uid, euid;        /* uid/euid of process setting the owner */
 935        int signum;             /* posix.1b rt signal to be delivered on IO */
 936};
 937
 938/*
 939 * Track a single file's readahead state
 940 */
 941struct file_ra_state {
 942        pgoff_t start;                  /* where readahead started */
 943        unsigned int size;              /* # of readahead pages */
 944        unsigned int async_size;        /* do asynchronous readahead when
 945                                           there are only # of pages ahead */
 946
 947        unsigned int ra_pages;          /* Maximum readahead window */
 948        unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
 949        loff_t prev_pos;                /* Cache last read() position */
 950};
 951
 952/*
 953 * Check if @index falls in the readahead windows.
 954 */
 955static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
 956{
 957        return (index >= ra->start &&
 958                index <  ra->start + ra->size);
 959}
 960
 961#define FILE_MNT_WRITE_TAKEN    1
 962#define FILE_MNT_WRITE_RELEASED 2
 963
 964struct file {
 965        /*
 966         * fu_list becomes invalid after file_free is called and queued via
 967         * fu_rcuhead for RCU freeing
 968         */
 969        union {
 970                struct list_head        fu_list;
 971                struct rcu_head         fu_rcuhead;
 972        } f_u;
 973        struct path             f_path;
 974#define f_dentry        f_path.dentry
 975#define f_vfsmnt        f_path.mnt
 976        const struct file_operations    *f_op;
 977
 978        /*
 979         * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
 980         * Must not be taken from IRQ context.
 981         */
 982        spinlock_t              f_lock;
 983#ifdef CONFIG_SMP
 984        int                     f_sb_list_cpu;
 985#endif
 986        atomic_long_t           f_count;
 987        unsigned int            f_flags;
 988        fmode_t                 f_mode;
 989        loff_t                  f_pos;
 990        struct fown_struct      f_owner;
 991        const struct cred       *f_cred;
 992        struct file_ra_state    f_ra;
 993
 994        u64                     f_version;
 995#ifdef CONFIG_SECURITY
 996        void                    *f_security;
 997#endif
 998        /* needed for tty driver, and maybe others */
 999        void                    *private_data;
1000
1001#ifdef CONFIG_EPOLL
1002        /* Used by fs/eventpoll.c to link all the hooks to this file */
1003        struct list_head        f_ep_links;
1004#endif /* #ifdef CONFIG_EPOLL */
1005        struct address_space    *f_mapping;
1006#ifdef CONFIG_DEBUG_WRITECOUNT
1007        unsigned long f_mnt_write_state;
1008#endif
1009};
1010
1011struct file_handle {
1012        __u32 handle_bytes;
1013        int handle_type;
1014        /* file identifier */
1015        unsigned char f_handle[0];
1016};
1017
1018#define get_file(x)     atomic_long_inc(&(x)->f_count)
1019#define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
1020#define file_count(x)   atomic_long_read(&(x)->f_count)
1021
1022#ifdef CONFIG_DEBUG_WRITECOUNT
1023static inline void file_take_write(struct file *f)
1024{
1025        WARN_ON(f->f_mnt_write_state != 0);
1026        f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1027}
1028static inline void file_release_write(struct file *f)
1029{
1030        f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1031}
1032static inline void file_reset_write(struct file *f)
1033{
1034        f->f_mnt_write_state = 0;
1035}
1036static inline void file_check_state(struct file *f)
1037{
1038        /*
1039         * At this point, either both or neither of these bits
1040         * should be set.
1041         */
1042        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1043        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1044}
1045static inline int file_check_writeable(struct file *f)
1046{
1047        if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1048                return 0;
1049        printk(KERN_WARNING "writeable file with no "
1050                            "mnt_want_write()\n");
1051        WARN_ON(1);
1052        return -EINVAL;
1053}
1054#else /* !CONFIG_DEBUG_WRITECOUNT */
1055static inline void file_take_write(struct file *filp) {}
1056static inline void file_release_write(struct file *filp) {}
1057static inline void file_reset_write(struct file *filp) {}
1058static inline void file_check_state(struct file *filp) {}
1059static inline int file_check_writeable(struct file *filp)
1060{
1061        return 0;
1062}
1063#endif /* CONFIG_DEBUG_WRITECOUNT */
1064
1065#define MAX_NON_LFS     ((1UL<<31) - 1)
1066
1067/* Page cache limit. The filesystems should put that into their s_maxbytes 
1068   limits, otherwise bad things can happen in VM. */ 
1069#if BITS_PER_LONG==32
1070#define MAX_LFS_FILESIZE        (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
1071#elif BITS_PER_LONG==64
1072#define MAX_LFS_FILESIZE        0x7fffffffffffffffUL
1073#endif
1074
1075#define FL_POSIX        1
1076#define FL_FLOCK        2
1077#define FL_ACCESS       8       /* not trying to lock, just looking */
1078#define FL_EXISTS       16      /* when unlocking, test for existence */
1079#define FL_LEASE        32      /* lease held on this file */
1080#define FL_CLOSE        64      /* unlock on close */
1081#define FL_SLEEP        128     /* A blocking lock */
1082#define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
1083#define FL_UNLOCK_PENDING       512 /* Lease is being broken */
1084
1085/*
1086 * Special return value from posix_lock_file() and vfs_lock_file() for
1087 * asynchronous locking.
1088 */
1089#define FILE_LOCK_DEFERRED 1
1090
1091/*
1092 * The POSIX file lock owner is determined by
1093 * the "struct files_struct" in the thread group
1094 * (or NULL for no owner - BSD locks).
1095 *
1096 * Lockd stuffs a "host" pointer into this.
1097 */
1098typedef struct files_struct *fl_owner_t;
1099
1100struct file_lock_operations {
1101        void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1102        void (*fl_release_private)(struct file_lock *);
1103};
1104
1105struct lock_manager_operations {
1106        int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1107        void (*lm_notify)(struct file_lock *);  /* unblock callback */
1108        int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1109        void (*lm_release_private)(struct file_lock *);
1110        void (*lm_break)(struct file_lock *);
1111        int (*lm_change)(struct file_lock **, int);
1112};
1113
1114struct lock_manager {
1115        struct list_head list;
1116};
1117
1118void locks_start_grace(struct lock_manager *);
1119void locks_end_grace(struct lock_manager *);
1120int locks_in_grace(void);
1121
1122/* that will die - we need it for nfs_lock_info */
1123#include <linux/nfs_fs_i.h>
1124
1125struct file_lock {
1126        struct file_lock *fl_next;      /* singly linked list for this inode  */
1127        struct list_head fl_link;       /* doubly linked list of all locks */
1128        struct list_head fl_block;      /* circular list of blocked processes */
1129        fl_owner_t fl_owner;
1130        unsigned int fl_flags;
1131        unsigned char fl_type;
1132        unsigned int fl_pid;
1133        struct pid *fl_nspid;
1134        wait_queue_head_t fl_wait;
1135        struct file *fl_file;
1136        loff_t fl_start;
1137        loff_t fl_end;
1138
1139        struct fasync_struct *  fl_fasync; /* for lease break notifications */
1140        /* for lease breaks: */
1141        unsigned long fl_break_time;
1142        unsigned long fl_downgrade_time;
1143
1144        const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
1145        const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1146        union {
1147                struct nfs_lock_info    nfs_fl;
1148                struct nfs4_lock_info   nfs4_fl;
1149                struct {
1150                        struct list_head link;  /* link in AFS vnode's pending_locks list */
1151                        int state;              /* state of grant or error if -ve */
1152                } afs;
1153        } fl_u;
1154};
1155
1156/* The following constant reflects the upper bound of the file/locking space */
1157#ifndef OFFSET_MAX
1158#define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
1159#define OFFSET_MAX      INT_LIMIT(loff_t)
1160#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1161#endif
1162
1163#include <linux/fcntl.h>
1164
1165extern void send_sigio(struct fown_struct *fown, int fd, int band);
1166
1167#ifdef CONFIG_FILE_LOCKING
1168extern int fcntl_getlk(struct file *, struct flock __user *);
1169extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1170                        struct flock __user *);
1171
1172#if BITS_PER_LONG == 32
1173extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1174extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1175                        struct flock64 __user *);
1176#endif
1177
1178extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1179extern int fcntl_getlease(struct file *filp);
1180
1181/* fs/locks.c */
1182void locks_free_lock(struct file_lock *fl);
1183extern void locks_init_lock(struct file_lock *);
1184extern struct file_lock * locks_alloc_lock(void);
1185extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1186extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1187extern void locks_remove_posix(struct file *, fl_owner_t);
1188extern void locks_remove_flock(struct file *);
1189extern void locks_release_private(struct file_lock *);
1190extern void posix_test_lock(struct file *, struct file_lock *);
1191extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1192extern int posix_lock_file_wait(struct file *, struct file_lock *);
1193extern int posix_unblock_lock(struct file *, struct file_lock *);
1194extern int vfs_test_lock(struct file *, struct file_lock *);
1195extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1196extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1197extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1198extern int __break_lease(struct inode *inode, unsigned int flags);
1199extern void lease_get_mtime(struct inode *, struct timespec *time);
1200extern int generic_setlease(struct file *, long, struct file_lock **);
1201extern int vfs_setlease(struct file *, long, struct file_lock **);
1202extern int lease_modify(struct file_lock **, int);
1203extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1204extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1205extern void lock_flocks(void);
1206extern void unlock_flocks(void);
1207#else /* !CONFIG_FILE_LOCKING */
1208static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1209{
1210        return -EINVAL;
1211}
1212
1213static inline int fcntl_setlk(unsigned int fd, struct file *file,
1214                              unsigned int cmd, struct flock __user *user)
1215{
1216        return -EACCES;
1217}
1218
1219#if BITS_PER_LONG == 32
1220static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1221{
1222        return -EINVAL;
1223}
1224
1225static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1226                                unsigned int cmd, struct flock64 __user *user)
1227{
1228        return -EACCES;
1229}
1230#endif
1231static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1232{
1233        return 0;
1234}
1235
1236static inline int fcntl_getlease(struct file *filp)
1237{
1238        return 0;
1239}
1240
1241static inline void locks_init_lock(struct file_lock *fl)
1242{
1243        return;
1244}
1245
1246static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1247{
1248        return;
1249}
1250
1251static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1252{
1253        return;
1254}
1255
1256static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1257{
1258        return;
1259}
1260
1261static inline void locks_remove_flock(struct file *filp)
1262{
1263        return;
1264}
1265
1266static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1267{
1268        return;
1269}
1270
1271static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1272                                  struct file_lock *conflock)
1273{
1274        return -ENOLCK;
1275}
1276
1277static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1278{
1279        return -ENOLCK;
1280}
1281
1282static inline int posix_unblock_lock(struct file *filp,
1283                                     struct file_lock *waiter)
1284{
1285        return -ENOENT;
1286}
1287
1288static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1289{
1290        return 0;
1291}
1292
1293static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1294                                struct file_lock *fl, struct file_lock *conf)
1295{
1296        return -ENOLCK;
1297}
1298
1299static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1300{
1301        return 0;
1302}
1303
1304static inline int flock_lock_file_wait(struct file *filp,
1305                                       struct file_lock *request)
1306{
1307        return -ENOLCK;
1308}
1309
1310static inline int __break_lease(struct inode *inode, unsigned int mode)
1311{
1312        return 0;
1313}
1314
1315static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1316{
1317        return;
1318}
1319
1320static inline int generic_setlease(struct file *filp, long arg,
1321                                    struct file_lock **flp)
1322{
1323        return -EINVAL;
1324}
1325
1326static inline int vfs_setlease(struct file *filp, long arg,
1327                               struct file_lock **lease)
1328{
1329        return -EINVAL;
1330}
1331
1332static inline int lease_modify(struct file_lock **before, int arg)
1333{
1334        return -EINVAL;
1335}
1336
1337static inline int lock_may_read(struct inode *inode, loff_t start,
1338                                unsigned long len)
1339{
1340        return 1;
1341}
1342
1343static inline int lock_may_write(struct inode *inode, loff_t start,
1344                                 unsigned long len)
1345{
1346        return 1;
1347}
1348
1349static inline void lock_flocks(void)
1350{
1351}
1352
1353static inline void unlock_flocks(void)
1354{
1355}
1356
1357#endif /* !CONFIG_FILE_LOCKING */
1358
1359
1360struct fasync_struct {
1361        spinlock_t              fa_lock;
1362        int                     magic;
1363        int                     fa_fd;
1364        struct fasync_struct    *fa_next; /* singly linked list */
1365        struct file             *fa_file;
1366        struct rcu_head         fa_rcu;
1367};
1368
1369#define FASYNC_MAGIC 0x4601
1370
1371/* SMP safe fasync helpers: */
1372extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1373extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1374extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1375extern struct fasync_struct *fasync_alloc(void);
1376extern void fasync_free(struct fasync_struct *);
1377
1378/* can be called from interrupts */
1379extern void kill_fasync(struct fasync_struct **, int, int);
1380
1381extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1382extern int f_setown(struct file *filp, unsigned long arg, int force);
1383extern void f_delown(struct file *filp);
1384extern pid_t f_getown(struct file *filp);
1385extern int send_sigurg(struct fown_struct *fown);
1386
1387/*
1388 *      Umount options
1389 */
1390
1391#define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1392#define MNT_DETACH      0x00000002      /* Just detach from the tree */
1393#define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1394#define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1395#define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1396
1397extern struct list_head super_blocks;
1398extern spinlock_t sb_lock;
1399
1400struct super_block {
1401        struct list_head        s_list;         /* Keep this first */
1402        dev_t                   s_dev;          /* search index; _not_ kdev_t */
1403        unsigned char           s_dirt;
1404        unsigned char           s_blocksize_bits;
1405        unsigned long           s_blocksize;
1406        loff_t                  s_maxbytes;     /* Max file size */
1407        struct file_system_type *s_type;
1408        const struct super_operations   *s_op;
1409        const struct dquot_operations   *dq_op;
1410        const struct quotactl_ops       *s_qcop;
1411        const struct export_operations *s_export_op;
1412        unsigned long           s_flags;
1413        unsigned long           s_magic;
1414        struct dentry           *s_root;
1415        struct rw_semaphore     s_umount;
1416        struct mutex            s_lock;
1417        int                     s_count;
1418        atomic_t                s_active;
1419#ifdef CONFIG_SECURITY
1420        void                    *s_security;
1421#endif
1422        const struct xattr_handler **s_xattr;
1423
1424        struct list_head        s_inodes;       /* all inodes */
1425        struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1426#ifdef CONFIG_SMP
1427        struct list_head __percpu *s_files;
1428#else
1429        struct list_head        s_files;
1430#endif
1431        /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1432        struct list_head        s_dentry_lru;   /* unused dentry lru */
1433        int                     s_nr_dentry_unused;     /* # of dentry on lru */
1434
1435        /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1436        spinlock_t              s_inode_lru_lock ____cacheline_aligned_in_smp;
1437        struct list_head        s_inode_lru;            /* unused inode lru */
1438        int                     s_nr_inodes_unused;     /* # of inodes on lru */
1439
1440        struct block_device     *s_bdev;
1441        struct backing_dev_info *s_bdi;
1442        struct mtd_info         *s_mtd;
1443        struct list_head        s_instances;
1444        struct quota_info       s_dquot;        /* Diskquota specific options */
1445
1446        int                     s_frozen;
1447        wait_queue_head_t       s_wait_unfrozen;
1448
1449        char s_id[32];                          /* Informational name */
1450        u8 s_uuid[16];                          /* UUID */
1451
1452        void                    *s_fs_info;     /* Filesystem private info */
1453        fmode_t                 s_mode;
1454
1455        /* Granularity of c/m/atime in ns.
1456           Cannot be worse than a second */
1457        u32                s_time_gran;
1458
1459        /*
1460         * The next field is for VFS *only*. No filesystems have any business
1461         * even looking at it. You had been warned.
1462         */
1463        struct mutex s_vfs_rename_mutex;        /* Kludge */
1464
1465        /*
1466         * Filesystem subtype.  If non-empty the filesystem type field
1467         * in /proc/mounts will be "type.subtype"
1468         */
1469        char *s_subtype;
1470
1471        /*
1472         * Saved mount options for lazy filesystems using
1473         * generic_show_options()
1474         */
1475        char __rcu *s_options;
1476        const struct dentry_operations *s_d_op; /* default d_op for dentries */
1477
1478        /*
1479         * Saved pool identifier for cleancache (-1 means none)
1480         */
1481        int cleancache_poolid;
1482
1483        struct shrinker s_shrink;       /* per-sb shrinker handle */
1484};
1485
1486/* superblock cache pruning functions */
1487extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1488extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1489
1490extern struct timespec current_fs_time(struct super_block *sb);
1491
1492/*
1493 * Snapshotting support.
1494 */
1495enum {
1496        SB_UNFROZEN = 0,
1497        SB_FREEZE_WRITE = 1,
1498        SB_FREEZE_TRANS = 2,
1499};
1500
1501#define vfs_check_frozen(sb, level) \
1502        wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1503
1504/*
1505 * until VFS tracks user namespaces for inodes, just make all files
1506 * belong to init_user_ns
1507 */
1508extern struct user_namespace init_user_ns;
1509#define inode_userns(inode) (&init_user_ns)
1510extern bool inode_owner_or_capable(const struct inode *inode);
1511
1512/* not quite ready to be deprecated, but... */
1513extern void lock_super(struct super_block *);
1514extern void unlock_super(struct super_block *);
1515
1516/*
1517 * VFS helper functions..
1518 */
1519extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1520extern int vfs_mkdir(struct inode *, struct dentry *, int);
1521extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1522extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1523extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1524extern int vfs_rmdir(struct inode *, struct dentry *);
1525extern int vfs_unlink(struct inode *, struct dentry *);
1526extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1527
1528/*
1529 * VFS dentry helper functions.
1530 */
1531extern void dentry_unhash(struct dentry *dentry);
1532
1533/*
1534 * VFS file helper functions.
1535 */
1536extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1537                        mode_t mode);
1538/*
1539 * VFS FS_IOC_FIEMAP helper definitions.
1540 */
1541struct fiemap_extent_info {
1542        unsigned int fi_flags;          /* Flags as passed from user */
1543        unsigned int fi_extents_mapped; /* Number of mapped extents */
1544        unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1545        struct fiemap_extent __user *fi_extents_start; /* Start of
1546                                                        fiemap_extent array */
1547};
1548int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1549                            u64 phys, u64 len, u32 flags);
1550int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1551
1552/*
1553 * File types
1554 *
1555 * NOTE! These match bits 12..15 of stat.st_mode
1556 * (ie "(i_mode >> 12) & 15").
1557 */
1558#define DT_UNKNOWN      0
1559#define DT_FIFO         1
1560#define DT_CHR          2
1561#define DT_DIR          4
1562#define DT_BLK          6
1563#define DT_REG          8
1564#define DT_LNK          10
1565#define DT_SOCK         12
1566#define DT_WHT          14
1567
1568/*
1569 * This is the "filldir" function type, used by readdir() to let
1570 * the kernel specify what kind of dirent layout it wants to have.
1571 * This allows the kernel to read directories into kernel space or
1572 * to have different dirent layouts depending on the binary type.
1573 */
1574typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1575struct block_device_operations;
1576
1577/* These macros are for out of kernel modules to test that
1578 * the kernel supports the unlocked_ioctl and compat_ioctl
1579 * fields in struct file_operations. */
1580#define HAVE_COMPAT_IOCTL 1
1581#define HAVE_UNLOCKED_IOCTL 1
1582
1583struct file_operations {
1584        struct module *owner;
1585        loff_t (*llseek) (struct file *, loff_t, int);
1586        ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1587        ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1588        ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1589        ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1590        int (*readdir) (struct file *, void *, filldir_t);
1591        unsigned int (*poll) (struct file *, struct poll_table_struct *);
1592        long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1593        long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1594        int (*mmap) (struct file *, struct vm_area_struct *);
1595        int (*open) (struct inode *, struct file *);
1596        int (*flush) (struct file *, fl_owner_t id);
1597        int (*release) (struct inode *, struct file *);
1598        int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1599        int (*aio_fsync) (struct kiocb *, int datasync);
1600        int (*fasync) (int, struct file *, int);
1601        int (*lock) (struct file *, int, struct file_lock *);
1602        ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1603        unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1604        int (*check_flags)(int);
1605        int (*flock) (struct file *, int, struct file_lock *);
1606        ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1607        ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1608        int (*setlease)(struct file *, long, struct file_lock **);
1609        long (*fallocate)(struct file *file, int mode, loff_t offset,
1610                          loff_t len);
1611};
1612
1613struct inode_operations {
1614        struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1615        void * (*follow_link) (struct dentry *, struct nameidata *);
1616        int (*permission) (struct inode *, int);
1617        struct posix_acl * (*get_acl)(struct inode *, int);
1618
1619        int (*readlink) (struct dentry *, char __user *,int);
1620        void (*put_link) (struct dentry *, struct nameidata *, void *);
1621
1622        int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1623        int (*link) (struct dentry *,struct inode *,struct dentry *);
1624        int (*unlink) (struct inode *,struct dentry *);
1625        int (*symlink) (struct inode *,struct dentry *,const char *);
1626        int (*mkdir) (struct inode *,struct dentry *,int);
1627        int (*rmdir) (struct inode *,struct dentry *);
1628        int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1629        int (*rename) (struct inode *, struct dentry *,
1630                        struct inode *, struct dentry *);
1631        void (*truncate) (struct inode *);
1632        int (*setattr) (struct dentry *, struct iattr *);
1633        int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1634        int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1635        ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1636        ssize_t (*listxattr) (struct dentry *, char *, size_t);
1637        int (*removexattr) (struct dentry *, const char *);
1638        void (*truncate_range)(struct inode *, loff_t, loff_t);
1639        int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1640                      u64 len);
1641} ____cacheline_aligned;
1642
1643struct seq_file;
1644
1645ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1646                              unsigned long nr_segs, unsigned long fast_segs,
1647                              struct iovec *fast_pointer,
1648                              struct iovec **ret_pointer,
1649                              int check_access);
1650
1651extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1652extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1653extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1654                unsigned long, loff_t *);
1655extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1656                unsigned long, loff_t *);
1657
1658struct super_operations {
1659        struct inode *(*alloc_inode)(struct super_block *sb);
1660        void (*destroy_inode)(struct inode *);
1661
1662        void (*dirty_inode) (struct inode *, int flags);
1663        int (*write_inode) (struct inode *, struct writeback_control *wbc);
1664        int (*drop_inode) (struct inode *);
1665        void (*evict_inode) (struct inode *);
1666        void (*put_super) (struct super_block *);
1667        void (*write_super) (struct super_block *);
1668        int (*sync_fs)(struct super_block *sb, int wait);
1669        int (*freeze_fs) (struct super_block *);
1670        int (*unfreeze_fs) (struct super_block *);
1671        int (*statfs) (struct dentry *, struct kstatfs *);
1672        int (*remount_fs) (struct super_block *, int *, char *);
1673        void (*umount_begin) (struct super_block *);
1674
1675        int (*show_options)(struct seq_file *, struct vfsmount *);
1676        int (*show_devname)(struct seq_file *, struct vfsmount *);
1677        int (*show_path)(struct seq_file *, struct vfsmount *);
1678        int (*show_stats)(struct seq_file *, struct vfsmount *);
1679#ifdef CONFIG_QUOTA
1680        ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1681        ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1682#endif
1683        int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1684        int (*nr_cached_objects)(struct super_block *);
1685        void (*free_cached_objects)(struct super_block *, int);
1686};
1687
1688/*
1689 * Inode state bits.  Protected by inode->i_lock
1690 *
1691 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1692 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1693 *
1694 * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1695 * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1696 * various stages of removing an inode.
1697 *
1698 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1699 *
1700 * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1701 *                      fdatasync().  i_atime is the usual cause.
1702 * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1703 *                      these changes separately from I_DIRTY_SYNC so that we
1704 *                      don't have to write inode on fdatasync() when only
1705 *                      mtime has changed in it.
1706 * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1707 * I_NEW                Serves as both a mutex and completion notification.
1708 *                      New inodes set I_NEW.  If two processes both create
1709 *                      the same inode, one of them will release its inode and
1710 *                      wait for I_NEW to be released before returning.
1711 *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1712 *                      also cause waiting on I_NEW, without I_NEW actually
1713 *                      being set.  find_inode() uses this to prevent returning
1714 *                      nearly-dead inodes.
1715 * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1716 *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1717 *                      cleared.
1718 * I_FREEING            Set when inode is about to be freed but still has dirty
1719 *                      pages or buffers attached or the inode itself is still
1720 *                      dirty.
1721 * I_CLEAR              Added by end_writeback().  In this state the inode is clean
1722 *                      and can be destroyed.  Inode keeps I_FREEING.
1723 *
1724 *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1725 *                      prohibited for many purposes.  iget() must wait for
1726 *                      the inode to be completely released, then create it
1727 *                      anew.  Other functions will just ignore such inodes,
1728 *                      if appropriate.  I_NEW is used for waiting.
1729 *
1730 * I_SYNC               Synchonized write of dirty inode data.  The bits is
1731 *                      set during data writeback, and cleared with a wakeup
1732 *                      on the bit address once it is done.
1733 *
1734 * I_REFERENCED         Marks the inode as recently references on the LRU list.
1735 *
1736 * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1737 *
1738 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1739 */
1740#define I_DIRTY_SYNC            (1 << 0)
1741#define I_DIRTY_DATASYNC        (1 << 1)
1742#define I_DIRTY_PAGES           (1 << 2)
1743#define __I_NEW                 3
1744#define I_NEW                   (1 << __I_NEW)
1745#define I_WILL_FREE             (1 << 4)
1746#define I_FREEING               (1 << 5)
1747#define I_CLEAR                 (1 << 6)
1748#define __I_SYNC                7
1749#define I_SYNC                  (1 << __I_SYNC)
1750#define I_REFERENCED            (1 << 8)
1751#define __I_DIO_WAKEUP          9
1752#define I_DIO_WAKEUP            (1 << I_DIO_WAKEUP)
1753
1754#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1755
1756extern void __mark_inode_dirty(struct inode *, int);
1757static inline void mark_inode_dirty(struct inode *inode)
1758{
1759        __mark_inode_dirty(inode, I_DIRTY);
1760}
1761
1762static inline void mark_inode_dirty_sync(struct inode *inode)
1763{
1764        __mark_inode_dirty(inode, I_DIRTY_SYNC);
1765}
1766
1767/**
1768 * set_nlink - directly set an inode's link count
1769 * @inode: inode
1770 * @nlink: new nlink (should be non-zero)
1771 *
1772 * This is a low-level filesystem helper to replace any
1773 * direct filesystem manipulation of i_nlink.
1774 */
1775static inline void set_nlink(struct inode *inode, unsigned int nlink)
1776{
1777        inode->__i_nlink = nlink;
1778}
1779
1780/**
1781 * inc_nlink - directly increment an inode's link count
1782 * @inode: inode
1783 *
1784 * This is a low-level filesystem helper to replace any
1785 * direct filesystem manipulation of i_nlink.  Currently,
1786 * it is only here for parity with dec_nlink().
1787 */
1788static inline void inc_nlink(struct inode *inode)
1789{
1790        inode->__i_nlink++;
1791}
1792
1793static inline void inode_inc_link_count(struct inode *inode)
1794{
1795        inc_nlink(inode);
1796        mark_inode_dirty(inode);
1797}
1798
1799/**
1800 * drop_nlink - directly drop an inode's link count
1801 * @inode: inode
1802 *
1803 * This is a low-level filesystem helper to replace any
1804 * direct filesystem manipulation of i_nlink.  In cases
1805 * where we are attempting to track writes to the
1806 * filesystem, a decrement to zero means an imminent
1807 * write when the file is truncated and actually unlinked
1808 * on the filesystem.
1809 */
1810static inline void drop_nlink(struct inode *inode)
1811{
1812        inode->__i_nlink--;
1813}
1814
1815/**
1816 * clear_nlink - directly zero an inode's link count
1817 * @inode: inode
1818 *
1819 * This is a low-level filesystem helper to replace any
1820 * direct filesystem manipulation of i_nlink.  See
1821 * drop_nlink() for why we care about i_nlink hitting zero.
1822 */
1823static inline void clear_nlink(struct inode *inode)
1824{
1825        inode->__i_nlink = 0;
1826}
1827
1828static inline void inode_dec_link_count(struct inode *inode)
1829{
1830        drop_nlink(inode);
1831        mark_inode_dirty(inode);
1832}
1833
1834/**
1835 * inode_inc_iversion - increments i_version
1836 * @inode: inode that need to be updated
1837 *
1838 * Every time the inode is modified, the i_version field will be incremented.
1839 * The filesystem has to be mounted with i_version flag
1840 */
1841
1842static inline void inode_inc_iversion(struct inode *inode)
1843{
1844       spin_lock(&inode->i_lock);
1845       inode->i_version++;
1846       spin_unlock(&inode->i_lock);
1847}
1848
1849extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1850static inline void file_accessed(struct file *file)
1851{
1852        if (!(file->f_flags & O_NOATIME))
1853                touch_atime(file->f_path.mnt, file->f_path.dentry);
1854}
1855
1856int sync_inode(struct inode *inode, struct writeback_control *wbc);
1857int sync_inode_metadata(struct inode *inode, int wait);
1858
1859struct file_system_type {
1860        const char *name;
1861        int fs_flags;
1862        struct dentry *(*mount) (struct file_system_type *, int,
1863                       const char *, void *);
1864        void (*kill_sb) (struct super_block *);
1865        struct module *owner;
1866        struct file_system_type * next;
1867        struct list_head fs_supers;
1868
1869        struct lock_class_key s_lock_key;
1870        struct lock_class_key s_umount_key;
1871        struct lock_class_key s_vfs_rename_key;
1872
1873        struct lock_class_key i_lock_key;
1874        struct lock_class_key i_mutex_key;
1875        struct lock_class_key i_mutex_dir_key;
1876};
1877
1878extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1879        void *data, int (*fill_super)(struct super_block *, void *, int));
1880extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1881        int flags, const char *dev_name, void *data,
1882        int (*fill_super)(struct super_block *, void *, int));
1883extern struct dentry *mount_single(struct file_system_type *fs_type,
1884        int flags, void *data,
1885        int (*fill_super)(struct super_block *, void *, int));
1886extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1887        int flags, void *data,
1888        int (*fill_super)(struct super_block *, void *, int));
1889extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1890void generic_shutdown_super(struct super_block *sb);
1891void kill_block_super(struct super_block *sb);
1892void kill_anon_super(struct super_block *sb);
1893void kill_litter_super(struct super_block *sb);
1894void deactivate_super(struct super_block *sb);
1895void deactivate_locked_super(struct super_block *sb);
1896int set_anon_super(struct super_block *s, void *data);
1897int get_anon_bdev(dev_t *);
1898void free_anon_bdev(dev_t);
1899struct super_block *sget(struct file_system_type *type,
1900                        int (*test)(struct super_block *,void *),
1901                        int (*set)(struct super_block *,void *),
1902                        void *data);
1903extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1904        const struct super_operations *ops,
1905        const struct dentry_operations *dops,
1906        unsigned long);
1907
1908static inline void sb_mark_dirty(struct super_block *sb)
1909{
1910        sb->s_dirt = 1;
1911}
1912static inline void sb_mark_clean(struct super_block *sb)
1913{
1914        sb->s_dirt = 0;
1915}
1916static inline int sb_is_dirty(struct super_block *sb)
1917{
1918        return sb->s_dirt;
1919}
1920
1921/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1922#define fops_get(fops) \
1923        (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1924#define fops_put(fops) \
1925        do { if (fops) module_put((fops)->owner); } while(0)
1926
1927extern int register_filesystem(struct file_system_type *);
1928extern int unregister_filesystem(struct file_system_type *);
1929extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1930#define kern_mount(type) kern_mount_data(type, NULL)
1931extern void kern_unmount(struct vfsmount *mnt);
1932extern int may_umount_tree(struct vfsmount *);
1933extern int may_umount(struct vfsmount *);
1934extern long do_mount(char *, char *, char *, unsigned long, void *);
1935extern struct vfsmount *collect_mounts(struct path *);
1936extern void drop_collected_mounts(struct vfsmount *);
1937extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1938                          struct vfsmount *);
1939extern int vfs_statfs(struct path *, struct kstatfs *);
1940extern int user_statfs(const char __user *, struct kstatfs *);
1941extern int fd_statfs(int, struct kstatfs *);
1942extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1943extern int freeze_super(struct super_block *super);
1944extern int thaw_super(struct super_block *super);
1945extern bool our_mnt(struct vfsmount *mnt);
1946
1947extern int current_umask(void);
1948
1949/* /sys/fs */
1950extern struct kobject *fs_kobj;
1951
1952#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1953extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1954
1955#define FLOCK_VERIFY_READ  1
1956#define FLOCK_VERIFY_WRITE 2
1957
1958#ifdef CONFIG_FILE_LOCKING
1959extern int locks_mandatory_locked(struct inode *);
1960extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1961
1962/*
1963 * Candidates for mandatory locking have the setgid bit set
1964 * but no group execute bit -  an otherwise meaningless combination.
1965 */
1966
1967static inline int __mandatory_lock(struct inode *ino)
1968{
1969        return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1970}
1971
1972/*
1973 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1974 * otherwise these will be advisory locks
1975 */
1976
1977static inline int mandatory_lock(struct inode *ino)
1978{
1979        return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1980}
1981
1982static inline int locks_verify_locked(struct inode *inode)
1983{
1984        if (mandatory_lock(inode))
1985                return locks_mandatory_locked(inode);
1986        return 0;
1987}
1988
1989static inline int locks_verify_truncate(struct inode *inode,
1990                                    struct file *filp,
1991                                    loff_t size)
1992{
1993        if (inode->i_flock && mandatory_lock(inode))
1994                return locks_mandatory_area(
1995                        FLOCK_VERIFY_WRITE, inode, filp,
1996                        size < inode->i_size ? size : inode->i_size,
1997                        (size < inode->i_size ? inode->i_size - size
1998                         : size - inode->i_size)
1999                );
2000        return 0;
2001}
2002
2003static inline int break_lease(struct inode *inode, unsigned int mode)
2004{
2005        if (inode->i_flock)
2006                return __break_lease(inode, mode);
2007        return 0;
2008}
2009#else /* !CONFIG_FILE_LOCKING */
2010static inline int locks_mandatory_locked(struct inode *inode)
2011{
2012        return 0;
2013}
2014
2015static inline int locks_mandatory_area(int rw, struct inode *inode,
2016                                       struct file *filp, loff_t offset,
2017                                       size_t count)
2018{
2019        return 0;
2020}
2021
2022static inline int __mandatory_lock(struct inode *inode)
2023{
2024        return 0;
2025}
2026
2027static inline int mandatory_lock(struct inode *inode)
2028{
2029        return 0;
2030}
2031
2032static inline int locks_verify_locked(struct inode *inode)
2033{
2034        return 0;
2035}
2036
2037static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2038                                        size_t size)
2039{
2040        return 0;
2041}
2042
2043static inline int break_lease(struct inode *inode, unsigned int mode)
2044{
2045        return 0;
2046}
2047
2048#endif /* CONFIG_FILE_LOCKING */
2049
2050/* fs/open.c */
2051
2052extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2053                       struct file *filp);
2054extern int do_fallocate(struct file *file, int mode, loff_t offset,
2055                        loff_t len);
2056extern long do_sys_open(int dfd, const char __user *filename, int flags,
2057                        int mode);
2058extern struct file *filp_open(const char *, int, int);
2059extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2060                                   const char *, int);
2061extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2062                                 const struct cred *);
2063extern int filp_close(struct file *, fl_owner_t id);
2064extern char * getname(const char __user *);
2065
2066/* fs/ioctl.c */
2067
2068extern int ioctl_preallocate(struct file *filp, void __user *argp);
2069
2070/* fs/dcache.c */
2071extern void __init vfs_caches_init_early(void);
2072extern void __init vfs_caches_init(unsigned long);
2073
2074extern struct kmem_cache *names_cachep;
2075
2076#define __getname_gfp(gfp)      kmem_cache_alloc(names_cachep, (gfp))
2077#define __getname()             __getname_gfp(GFP_KERNEL)
2078#define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2079#ifndef CONFIG_AUDITSYSCALL
2080#define putname(name)   __putname(name)
2081#else
2082extern void putname(const char *name);
2083#endif
2084
2085#ifdef CONFIG_BLOCK
2086extern int register_blkdev(unsigned int, const char *);
2087extern void unregister_blkdev(unsigned int, const char *);
2088extern struct block_device *bdget(dev_t);
2089extern struct block_device *bdgrab(struct block_device *bdev);
2090extern void bd_set_size(struct block_device *, loff_t size);
2091extern void bd_forget(struct inode *inode);
2092extern void bdput(struct block_device *);
2093extern void invalidate_bdev(struct block_device *);
2094extern int sync_blockdev(struct block_device *bdev);
2095extern struct super_block *freeze_bdev(struct block_device *);
2096extern void emergency_thaw_all(void);
2097extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2098extern int fsync_bdev(struct block_device *);
2099#else
2100static inline void bd_forget(struct inode *inode) {}
2101static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2102static inline void invalidate_bdev(struct block_device *bdev) {}
2103
2104static inline struct super_block *freeze_bdev(struct block_device *sb)
2105{
2106        return NULL;
2107}
2108
2109static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2110{
2111        return 0;
2112}
2113#endif
2114extern int sync_filesystem(struct super_block *);
2115extern const struct file_operations def_blk_fops;
2116extern const struct file_operations def_chr_fops;
2117extern const struct file_operations bad_sock_fops;
2118extern const struct file_operations def_fifo_fops;
2119#ifdef CONFIG_BLOCK
2120extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2121extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2122extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2123extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2124extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2125                                               void *holder);
2126extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2127                                              void *holder);
2128extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2129#ifdef CONFIG_SYSFS
2130extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2131extern void bd_unlink_disk_holder(struct block_device *bdev,
2132                                  struct gendisk *disk);
2133#else
2134static inline int bd_link_disk_holder(struct block_device *bdev,
2135                                      struct gendisk *disk)
2136{
2137        return 0;
2138}
2139static inline void bd_unlink_disk_holder(struct block_device *bdev,
2140                                         struct gendisk *disk)
2141{
2142}
2143#endif
2144#endif
2145
2146/* fs/char_dev.c */
2147#define CHRDEV_MAJOR_HASH_SIZE  255
2148extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2149extern int register_chrdev_region(dev_t, unsigned, const char *);
2150extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2151                             unsigned int count, const char *name,
2152                             const struct file_operations *fops);
2153extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2154                                unsigned int count, const char *name);
2155extern void unregister_chrdev_region(dev_t, unsigned);
2156extern void chrdev_show(struct seq_file *,off_t);
2157
2158static inline int register_chrdev(unsigned int major, const char *name,
2159                                  const struct file_operations *fops)
2160{
2161        return __register_chrdev(major, 0, 256, name, fops);
2162}
2163
2164static inline void unregister_chrdev(unsigned int major, const char *name)
2165{
2166        __unregister_chrdev(major, 0, 256, name);
2167}
2168
2169/* fs/block_dev.c */
2170#define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2171#define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2172
2173#ifdef CONFIG_BLOCK
2174#define BLKDEV_MAJOR_HASH_SIZE  255
2175extern const char *__bdevname(dev_t, char *buffer);
2176extern const char *bdevname(struct block_device *bdev, char *buffer);
2177extern struct block_device *lookup_bdev(const char *);
2178extern void blkdev_show(struct seq_file *,off_t);
2179
2180#else
2181#define BLKDEV_MAJOR_HASH_SIZE  0
2182#endif
2183
2184extern void init_special_inode(struct inode *, umode_t, dev_t);
2185
2186/* Invalid inode operations -- fs/bad_inode.c */
2187extern void make_bad_inode(struct inode *);
2188extern int is_bad_inode(struct inode *);
2189
2190extern const struct file_operations read_pipefifo_fops;
2191extern const struct file_operations write_pipefifo_fops;
2192extern const struct file_operations rdwr_pipefifo_fops;
2193
2194extern int fs_may_remount_ro(struct super_block *);
2195
2196#ifdef CONFIG_BLOCK
2197/*
2198 * return READ, READA, or WRITE
2199 */
2200#define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2201
2202/*
2203 * return data direction, READ or WRITE
2204 */
2205#define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2206
2207extern void check_disk_size_change(struct gendisk *disk,
2208                                   struct block_device *bdev);
2209extern int revalidate_disk(struct gendisk *);
2210extern int check_disk_change(struct block_device *);
2211extern int __invalidate_device(struct block_device *, bool);
2212extern int invalidate_partition(struct gendisk *, int);
2213#endif
2214unsigned long invalidate_mapping_pages(struct address_space *mapping,
2215                                        pgoff_t start, pgoff_t end);
2216
2217static inline void invalidate_remote_inode(struct inode *inode)
2218{
2219        if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2220            S_ISLNK(inode->i_mode))
2221                invalidate_mapping_pages(inode->i_mapping, 0, -1);
2222}
2223extern int invalidate_inode_pages2(struct address_space *mapping);
2224extern int invalidate_inode_pages2_range(struct address_space *mapping,
2225                                         pgoff_t start, pgoff_t end);
2226extern int write_inode_now(struct inode *, int);
2227extern int filemap_fdatawrite(struct address_space *);
2228extern int filemap_flush(struct address_space *);
2229extern int filemap_fdatawait(struct address_space *);
2230extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2231                                   loff_t lend);
2232extern int filemap_write_and_wait(struct address_space *mapping);
2233extern int filemap_write_and_wait_range(struct address_space *mapping,
2234                                        loff_t lstart, loff_t lend);
2235extern int __filemap_fdatawrite_range(struct address_space *mapping,
2236                                loff_t start, loff_t end, int sync_mode);
2237extern int filemap_fdatawrite_range(struct address_space *mapping,
2238                                loff_t start, loff_t end);
2239
2240extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2241                           int datasync);
2242extern int vfs_fsync(struct file *file, int datasync);
2243extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2244extern void sync_supers(void);
2245extern void emergency_sync(void);
2246extern void emergency_remount(void);
2247#ifdef CONFIG_BLOCK
2248extern sector_t bmap(struct inode *, sector_t);
2249#endif
2250extern int notify_change(struct dentry *, struct iattr *);
2251extern int inode_permission(struct inode *, int);
2252extern int generic_permission(struct inode *, int);
2253
2254static inline bool execute_ok(struct inode *inode)
2255{
2256        return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2257}
2258
2259/*
2260 * get_write_access() gets write permission for a file.
2261 * put_write_access() releases this write permission.
2262 * This is used for regular files.
2263 * We cannot support write (and maybe mmap read-write shared) accesses and
2264 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2265 * can have the following values:
2266 * 0: no writers, no VM_DENYWRITE mappings
2267 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2268 * > 0: (i_writecount) users are writing to the file.
2269 *
2270 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2271 * except for the cases where we don't hold i_writecount yet. Then we need to
2272 * use {get,deny}_write_access() - these functions check the sign and refuse
2273 * to do the change if sign is wrong.
2274 */
2275static inline int get_write_access(struct inode *inode)
2276{
2277        return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2278}
2279static inline int deny_write_access(struct file *file)
2280{
2281        struct inode *inode = file->f_path.dentry->d_inode;
2282        return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2283}
2284static inline void put_write_access(struct inode * inode)
2285{
2286        atomic_dec(&inode->i_writecount);
2287}
2288static inline void allow_write_access(struct file *file)
2289{
2290        if (file)
2291                atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2292}
2293#ifdef CONFIG_IMA
2294static inline void i_readcount_dec(struct inode *inode)
2295{
2296        BUG_ON(!atomic_read(&inode->i_readcount));
2297        atomic_dec(&inode->i_readcount);
2298}
2299static inline void i_readcount_inc(struct inode *inode)
2300{
2301        atomic_inc(&inode->i_readcount);
2302}
2303#else
2304static inline void i_readcount_dec(struct inode *inode)
2305{
2306        return;
2307}
2308static inline void i_readcount_inc(struct inode *inode)
2309{
2310        return;
2311}
2312#endif
2313extern int do_pipe_flags(int *, int);
2314extern struct file *create_read_pipe(struct file *f, int flags);
2315extern struct file *create_write_pipe(int flags);
2316extern void free_write_pipe(struct file *);
2317
2318extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2319extern struct file * open_exec(const char *);
2320 
2321/* fs/dcache.c -- generic fs support functions */
2322extern int is_subdir(struct dentry *, struct dentry *);
2323extern int path_is_under(struct path *, struct path *);
2324extern ino_t find_inode_number(struct dentry *, struct qstr *);
2325
2326#include <linux/err.h>
2327
2328/* needed for stackable file system support */
2329extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2330
2331extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2332
2333extern int inode_init_always(struct super_block *, struct inode *);
2334extern void inode_init_once(struct inode *);
2335extern void address_space_init_once(struct address_space *mapping);
2336extern void ihold(struct inode * inode);
2337extern void iput(struct inode *);
2338extern struct inode * igrab(struct inode *);
2339extern ino_t iunique(struct super_block *, ino_t);
2340extern int inode_needs_sync(struct inode *inode);
2341extern int generic_delete_inode(struct inode *inode);
2342extern int generic_drop_inode(struct inode *inode);
2343
2344extern struct inode *ilookup5_nowait(struct super_block *sb,
2345                unsigned long hashval, int (*test)(struct inode *, void *),
2346                void *data);
2347extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2348                int (*test)(struct inode *, void *), void *data);
2349extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2350
2351extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2352extern struct inode * iget_locked(struct super_block *, unsigned long);
2353extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2354extern int insert_inode_locked(struct inode *);
2355#ifdef CONFIG_DEBUG_LOCK_ALLOC
2356extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2357#else
2358static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2359#endif
2360extern void unlock_new_inode(struct inode *);
2361extern unsigned int get_next_ino(void);
2362
2363extern void __iget(struct inode * inode);
2364extern void iget_failed(struct inode *);
2365extern void end_writeback(struct inode *);
2366extern void __destroy_inode(struct inode *);
2367extern struct inode *new_inode_pseudo(struct super_block *sb);
2368extern struct inode *new_inode(struct super_block *sb);
2369extern void free_inode_nonrcu(struct inode *inode);
2370extern int should_remove_suid(struct dentry *);
2371extern int file_remove_suid(struct file *);
2372
2373extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2374static inline void insert_inode_hash(struct inode *inode)
2375{
2376        __insert_inode_hash(inode, inode->i_ino);
2377}
2378
2379extern void __remove_inode_hash(struct inode *);
2380static inline void remove_inode_hash(struct inode *inode)
2381{
2382        if (!inode_unhashed(inode))
2383                __remove_inode_hash(inode);
2384}
2385
2386extern void inode_sb_list_add(struct inode *inode);
2387
2388#ifdef CONFIG_BLOCK
2389extern void submit_bio(int, struct bio *);
2390extern int bdev_read_only(struct block_device *);
2391#endif
2392extern int set_blocksize(struct block_device *, int);
2393extern int sb_set_blocksize(struct super_block *, int);
2394extern int sb_min_blocksize(struct super_block *, int);
2395
2396extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2397extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2398extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2399int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2400extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2401extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2402                loff_t *);
2403extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2404extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2405                unsigned long *, loff_t, loff_t *, size_t, size_t);
2406extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2407                unsigned long, loff_t, loff_t *, size_t, ssize_t);
2408extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2409extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2410extern int generic_segment_checks(const struct iovec *iov,
2411                unsigned long *nr_segs, size_t *count, int access_flags);
2412
2413/* fs/block_dev.c */
2414extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2415                                unsigned long nr_segs, loff_t pos);
2416extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2417                        int datasync);
2418
2419/* fs/splice.c */
2420extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2421                struct pipe_inode_info *, size_t, unsigned int);
2422extern ssize_t default_file_splice_read(struct file *, loff_t *,
2423                struct pipe_inode_info *, size_t, unsigned int);
2424extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2425                struct file *, loff_t *, size_t, unsigned int);
2426extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2427                struct file *out, loff_t *, size_t len, unsigned int flags);
2428extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2429                size_t len, unsigned int flags);
2430
2431extern void
2432file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2433extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2434extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2435extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2436extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2437                int origin, loff_t maxsize);
2438extern int generic_file_open(struct inode * inode, struct file * filp);
2439extern int nonseekable_open(struct inode * inode, struct file * filp);
2440
2441#ifdef CONFIG_FS_XIP
2442extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2443                             loff_t *ppos);
2444extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2445extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2446                              size_t len, loff_t *ppos);
2447extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2448#else
2449static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2450{
2451        return 0;
2452}
2453#endif
2454
2455#ifdef CONFIG_BLOCK
2456typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2457                            loff_t file_offset);
2458
2459enum {
2460        /* need locking between buffered and direct access */
2461        DIO_LOCKING     = 0x01,
2462
2463        /* filesystem does not support filling holes */
2464        DIO_SKIP_HOLES  = 0x02,
2465};
2466
2467void dio_end_io(struct bio *bio, int error);
2468void inode_dio_wait(struct inode *inode);
2469void inode_dio_done(struct inode *inode);
2470
2471ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2472        struct block_device *bdev, const struct iovec *iov, loff_t offset,
2473        unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2474        dio_submit_t submit_io, int flags);
2475
2476static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2477                struct inode *inode, const struct iovec *iov, loff_t offset,
2478                unsigned long nr_segs, get_block_t get_block)
2479{
2480        return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2481                                    offset, nr_segs, get_block, NULL, NULL,
2482                                    DIO_LOCKING | DIO_SKIP_HOLES);
2483}
2484#else
2485static inline void inode_dio_wait(struct inode *inode)
2486{
2487}
2488#endif
2489
2490extern const struct file_operations generic_ro_fops;
2491
2492#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2493
2494extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2495extern int vfs_follow_link(struct nameidata *, const char *);
2496extern int page_readlink(struct dentry *, char __user *, int);
2497extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2498extern void page_put_link(struct dentry *, struct nameidata *, void *);
2499extern int __page_symlink(struct inode *inode, const char *symname, int len,
2500                int nofs);
2501extern int page_symlink(struct inode *inode, const char *symname, int len);
2502extern const struct inode_operations page_symlink_inode_operations;
2503extern int generic_readlink(struct dentry *, char __user *, int);
2504extern void generic_fillattr(struct inode *, struct kstat *);
2505extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2506void __inode_add_bytes(struct inode *inode, loff_t bytes);
2507void inode_add_bytes(struct inode *inode, loff_t bytes);
2508void inode_sub_bytes(struct inode *inode, loff_t bytes);
2509loff_t inode_get_bytes(struct inode *inode);
2510void inode_set_bytes(struct inode *inode, loff_t bytes);
2511
2512extern int vfs_readdir(struct file *, filldir_t, void *);
2513
2514extern int vfs_stat(const char __user *, struct kstat *);
2515extern int vfs_lstat(const char __user *, struct kstat *);
2516extern int vfs_fstat(unsigned int, struct kstat *);
2517extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2518
2519extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2520                    unsigned long arg);
2521extern int __generic_block_fiemap(struct inode *inode,
2522                                  struct fiemap_extent_info *fieinfo,
2523                                  loff_t start, loff_t len,
2524                                  get_block_t *get_block);
2525extern int generic_block_fiemap(struct inode *inode,
2526                                struct fiemap_extent_info *fieinfo, u64 start,
2527                                u64 len, get_block_t *get_block);
2528
2529extern void get_filesystem(struct file_system_type *fs);
2530extern void put_filesystem(struct file_system_type *fs);
2531extern struct file_system_type *get_fs_type(const char *name);
2532extern struct super_block *get_super(struct block_device *);
2533extern struct super_block *get_active_super(struct block_device *bdev);
2534extern struct super_block *user_get_super(dev_t);
2535extern void drop_super(struct super_block *sb);
2536extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2537extern void iterate_supers_type(struct file_system_type *,
2538                                void (*)(struct super_block *, void *), void *);
2539
2540extern int dcache_dir_open(struct inode *, struct file *);
2541extern int dcache_dir_close(struct inode *, struct file *);
2542extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2543extern int dcache_readdir(struct file *, void *, filldir_t);
2544extern int simple_setattr(struct dentry *, struct iattr *);
2545extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2546extern int simple_statfs(struct dentry *, struct kstatfs *);
2547extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2548extern int simple_unlink(struct inode *, struct dentry *);
2549extern int simple_rmdir(struct inode *, struct dentry *);
2550extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2551extern int noop_fsync(struct file *, loff_t, loff_t, int);
2552extern int simple_empty(struct dentry *);
2553extern int simple_readpage(struct file *file, struct page *page);
2554extern int simple_write_begin(struct file *file, struct address_space *mapping,
2555                        loff_t pos, unsigned len, unsigned flags,
2556                        struct page **pagep, void **fsdata);
2557extern int simple_write_end(struct file *file, struct address_space *mapping,
2558                        loff_t pos, unsigned len, unsigned copied,
2559                        struct page *page, void *fsdata);
2560
2561extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2562extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2563extern const struct file_operations simple_dir_operations;
2564extern const struct inode_operations simple_dir_inode_operations;
2565struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2566struct dentry *d_alloc_name(struct dentry *, const char *);
2567extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2568extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2569extern void simple_release_fs(struct vfsmount **mount, int *count);
2570
2571extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2572                        loff_t *ppos, const void *from, size_t available);
2573extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2574                const void __user *from, size_t count);
2575
2576extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2577
2578extern int generic_check_addressable(unsigned, u64);
2579
2580#ifdef CONFIG_MIGRATION
2581extern int buffer_migrate_page(struct address_space *,
2582                                struct page *, struct page *);
2583#else
2584#define buffer_migrate_page NULL
2585#endif
2586
2587extern int inode_change_ok(const struct inode *, struct iattr *);
2588extern int inode_newsize_ok(const struct inode *, loff_t offset);
2589extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2590
2591extern void file_update_time(struct file *file);
2592
2593extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2594extern void save_mount_options(struct super_block *sb, char *options);
2595extern void replace_mount_options(struct super_block *sb, char *options);
2596
2597static inline ino_t parent_ino(struct dentry *dentry)
2598{
2599        ino_t res;
2600
2601        /*
2602         * Don't strictly need d_lock here? If the parent ino could change
2603         * then surely we'd have a deeper race in the caller?
2604         */
2605        spin_lock(&dentry->d_lock);
2606        res = dentry->d_parent->d_inode->i_ino;
2607        spin_unlock(&dentry->d_lock);
2608        return res;
2609}
2610
2611/* Transaction based IO helpers */
2612
2613/*
2614 * An argresp is stored in an allocated page and holds the
2615 * size of the argument or response, along with its content
2616 */
2617struct simple_transaction_argresp {
2618        ssize_t size;
2619        char data[0];
2620};
2621
2622#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2623
2624char *simple_transaction_get(struct file *file, const char __user *buf,
2625                                size_t size);
2626ssize_t simple_transaction_read(struct file *file, char __user *buf,
2627                                size_t size, loff_t *pos);
2628int simple_transaction_release(struct inode *inode, struct file *file);
2629
2630void simple_transaction_set(struct file *file, size_t n);
2631
2632/*
2633 * simple attribute files
2634 *
2635 * These attributes behave similar to those in sysfs:
2636 *
2637 * Writing to an attribute immediately sets a value, an open file can be
2638 * written to multiple times.
2639 *
2640 * Reading from an attribute creates a buffer from the value that might get
2641 * read with multiple read calls. When the attribute has been read
2642 * completely, no further read calls are possible until the file is opened
2643 * again.
2644 *
2645 * All attributes contain a text representation of a numeric value
2646 * that are accessed with the get() and set() functions.
2647 */
2648#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2649static int __fops ## _open(struct inode *inode, struct file *file)      \
2650{                                                                       \
2651        __simple_attr_check_format(__fmt, 0ull);                        \
2652        return simple_attr_open(inode, file, __get, __set, __fmt);      \
2653}                                                                       \
2654static const struct file_operations __fops = {                          \
2655        .owner   = THIS_MODULE,                                         \
2656        .open    = __fops ## _open,                                     \
2657        .release = simple_attr_release,                                 \
2658        .read    = simple_attr_read,                                    \
2659        .write   = simple_attr_write,                                   \
2660        .llseek  = generic_file_llseek,                                 \
2661};
2662
2663static inline __printf(1, 2)
2664void __simple_attr_check_format(const char *fmt, ...)
2665{
2666        /* don't do anything, just let the compiler check the arguments; */
2667}
2668
2669int simple_attr_open(struct inode *inode, struct file *file,
2670                     int (*get)(void *, u64 *), int (*set)(void *, u64),
2671                     const char *fmt);
2672int simple_attr_release(struct inode *inode, struct file *file);
2673ssize_t simple_attr_read(struct file *file, char __user *buf,
2674                         size_t len, loff_t *ppos);
2675ssize_t simple_attr_write(struct file *file, const char __user *buf,
2676                          size_t len, loff_t *ppos);
2677
2678struct ctl_table;
2679int proc_nr_files(struct ctl_table *table, int write,
2680                  void __user *buffer, size_t *lenp, loff_t *ppos);
2681int proc_nr_dentry(struct ctl_table *table, int write,
2682                  void __user *buffer, size_t *lenp, loff_t *ppos);
2683int proc_nr_inodes(struct ctl_table *table, int write,
2684                   void __user *buffer, size_t *lenp, loff_t *ppos);
2685int __init get_filesystem_list(char *buf);
2686
2687#define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2688#define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2689
2690#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2691#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2692                                            (flag & __FMODE_NONOTIFY)))
2693
2694static inline int is_sxid(mode_t mode)
2695{
2696        return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2697}
2698
2699static inline void inode_has_no_xattr(struct inode *inode)
2700{
2701        if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2702                inode->i_flags |= S_NOSEC;
2703}
2704
2705#endif /* __KERNEL__ */
2706#endif /* _LINUX_FS_H */
2707
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.