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