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
  12/*
  13 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
  14 * the file limit at runtime and only root can increase the per-process
  15 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
  16 * upper limit on files-per-process.
  17 *
  18 * Some programs (notably those using select()) may have to be 
  19 * recompiled to take full advantage of the new limits..  
  20 */
  21
  22/* Fixed constants first: */
  23#undef NR_OPEN
  24extern int sysctl_nr_open;
  25#define INR_OPEN 1024           /* Initial setting for nfile rlimits */
  26
  27#define BLOCK_SIZE_BITS 10
  28#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
  29
  30#define SEEK_SET        0       /* seek relative to beginning of file */
  31#define SEEK_CUR        1       /* seek relative to current file position */
  32#define SEEK_END        2       /* seek relative to end of file */
  33#define SEEK_MAX        SEEK_END
  34
  35/* And dynamically-tunable limits and defaults: */
  36struct files_stat_struct {
  37        int nr_files;           /* read only */
  38        int nr_free_files;      /* read only */
  39        int max_files;          /* tunable */
  40};
  41extern struct files_stat_struct files_stat;
  42extern int get_max_files(void);
  43
  44struct inodes_stat_t {
  45        int nr_inodes;
  46        int nr_unused;
  47        int dummy[5];           /* padding for sysctl ABI compatibility */
  48};
  49extern struct inodes_stat_t inodes_stat;
  50
  51extern int leases_enable, lease_break_time;
  52
  53#ifdef CONFIG_DNOTIFY
  54extern int dir_notify_enable;
  55#endif
  56
  57#define NR_FILE  8192   /* this can well be larger on a larger system */
  58
  59#define MAY_EXEC 1
  60#define MAY_WRITE 2
  61#define MAY_READ 4
  62#define MAY_APPEND 8
  63#define MAY_ACCESS 16
  64#define MAY_OPEN 32
  65
  66#define FMODE_READ 1
  67#define FMODE_WRITE 2
  68
  69/* Internal kernel extensions */
  70#define FMODE_LSEEK     4
  71#define FMODE_PREAD     8
  72#define FMODE_PWRITE    FMODE_PREAD     /* These go hand in hand */
  73
  74/* File is being opened for execution. Primary users of this flag are
  75   distributed filesystems that can use it to achieve correct ETXTBUSY
  76   behavior for cross-node execution/opening_for_writing of files */
  77#define FMODE_EXEC      16
  78
  79#define RW_MASK         1
  80#define RWA_MASK        2
  81#define READ 0
  82#define WRITE 1
  83#define READA 2         /* read-ahead  - don't block if no resources */
  84#define SWRITE 3        /* for ll_rw_block() - wait for buffer lock */
  85#define READ_SYNC       (READ | (1 << BIO_RW_SYNC))
  86#define READ_META       (READ | (1 << BIO_RW_META))
  87#define WRITE_SYNC      (WRITE | (1 << BIO_RW_SYNC))
  88#define SWRITE_SYNC     (SWRITE | (1 << BIO_RW_SYNC))
  89#define WRITE_BARRIER   ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
  90
  91#define SEL_IN          1
  92#define SEL_OUT         2
  93#define SEL_EX          4
  94
  95/* public flags for file_system_type */
  96#define FS_REQUIRES_DEV 1 
  97#define FS_BINARY_MOUNTDATA 2
  98#define FS_HAS_SUBTYPE 4
  99#define FS_REVAL_DOT    16384   /* Check the paths ".", ".." for staleness */
 100#define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move()
 101                                         * during rename() internally.
 102                                         */
 103
 104/*
 105 * These are the fs-independent mount-flags: up to 32 flags are supported
 106 */
 107#define MS_RDONLY        1      /* Mount read-only */
 108#define MS_NOSUID        2      /* Ignore suid and sgid bits */
 109#define MS_NODEV         4      /* Disallow access to device special files */
 110#define MS_NOEXEC        8      /* Disallow program execution */
 111#define MS_SYNCHRONOUS  16      /* Writes are synced at once */
 112#define MS_REMOUNT      32      /* Alter flags of a mounted FS */
 113#define MS_MANDLOCK     64      /* Allow mandatory locks on an FS */
 114#define MS_DIRSYNC      128     /* Directory modifications are synchronous */
 115#define MS_NOATIME      1024    /* Do not update access times. */
 116#define MS_NODIRATIME   2048    /* Do not update directory access times */
 117#define MS_BIND         4096
 118#define MS_MOVE         8192
 119#define MS_REC          16384
 120#define MS_VERBOSE      32768   /* War is peace. Verbosity is silence.
 121                                   MS_VERBOSE is deprecated. */
 122#define MS_SILENT       32768
 123#define MS_POSIXACL     (1<<16) /* VFS does not apply the umask */
 124#define MS_UNBINDABLE   (1<<17) /* change to unbindable */
 125#define MS_PRIVATE      (1<<18) /* change to private */
 126#define MS_SLAVE        (1<<19) /* change to slave */
 127#define MS_SHARED       (1<<20) /* change to shared */
 128#define MS_RELATIME     (1<<21) /* Update atime relative to mtime/ctime. */
 129#define MS_KERNMOUNT    (1<<22) /* this is a kern_mount call */
 130#define MS_I_VERSION    (1<<23) /* Update inode I_version field */
 131#define MS_ACTIVE       (1<<30)
 132#define MS_NOUSER       (1<<31)
 133
 134/*
 135 * Superblock flags that can be altered by MS_REMOUNT
 136 */
 137#define MS_RMT_MASK     (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
 138
 139/*
 140 * Old magic mount flag and mask
 141 */
 142#define MS_MGC_VAL 0xC0ED0000
 143#define MS_MGC_MSK 0xffff0000
 144
 145/* Inode flags - they have nothing to superblock flags now */
 146
 147#define S_SYNC          1       /* Writes are synced at once */
 148#define S_NOATIME       2       /* Do not update access times */
 149#define S_APPEND        4       /* Append-only file */
 150#define S_IMMUTABLE     8       /* Immutable file */
 151#define S_DEAD          16      /* removed, but still open directory */
 152#define S_NOQUOTA       32      /* Inode is not counted to quota */
 153#define S_DIRSYNC       64      /* Directory modifications are synchronous */
 154#define S_NOCMTIME      128     /* Do not update file c/mtime */
 155#define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
 156#define S_PRIVATE       512     /* Inode is fs-internal */
 157
 158/*
 159 * Note that nosuid etc flags are inode-specific: setting some file-system
 160 * flags just means all the inodes inherit those flags by default. It might be
 161 * possible to override it selectively if you really wanted to with some
 162 * ioctl() that is not currently implemented.
 163 *
 164 * Exception: MS_RDONLY is always applied to the entire file system.
 165 *
 166 * Unfortunately, it is possible to change a filesystems flags with it mounted
 167 * with files in use.  This means that all of the inodes will not have their
 168 * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
 169 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
 170 */
 171#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
 172
 173#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
 174#define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
 175                                        ((inode)->i_flags & S_SYNC))
 176#define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
 177                                        ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
 178#define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
 179#define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
 180#define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
 181
 182#define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
 183#define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
 184#define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
 185#define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
 186
 187#define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
 188#define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
 189#define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
 190#define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
 191
 192/* the read-only stuff doesn't really belong here, but any other place is
 193   probably as bad and I don't want to create yet another include file. */
 194
 195#define BLKROSET   _IO(0x12,93) /* set device read-only (0 = read-write) */
 196#define BLKROGET   _IO(0x12,94) /* get read-only status (0 = read_write) */
 197#define BLKRRPART  _IO(0x12,95) /* re-read partition table */
 198#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
 199#define BLKFLSBUF  _IO(0x12,97) /* flush buffer cache */
 200#define BLKRASET   _IO(0x12,98) /* set read ahead for block device */
 201#define BLKRAGET   _IO(0x12,99) /* get current read ahead setting */
 202#define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
 203#define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
 204#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
 205#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
 206#define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
 207#if 0
 208#define BLKPG      _IO(0x12,105)/* See blkpg.h */
 209
 210/* Some people are morons.  Do not use sizeof! */
 211
 212#define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
 213#define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
 214/* This was here just to show that the number is taken -
 215   probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
 216#endif
 217/* A jump here: 108-111 have been used for various private purposes. */
 218#define BLKBSZGET  _IOR(0x12,112,size_t)
 219#define BLKBSZSET  _IOW(0x12,113,size_t)
 220#define BLKGETSIZE64 _IOR(0x12,114,size_t)      /* return device size in bytes (u64 *arg) */
 221#define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
 222#define BLKTRACESTART _IO(0x12,116)
 223#define BLKTRACESTOP _IO(0x12,117)
 224#define BLKTRACETEARDOWN _IO(0x12,118)
 225
 226#define BMAP_IOCTL 1            /* obsolete - kept for compatibility */
 227#define FIBMAP     _IO(0x00,1)  /* bmap access */
 228#define FIGETBSZ   _IO(0x00,2)  /* get the block size used for bmap */
 229
 230#define FS_IOC_GETFLAGS                 _IOR('f', 1, long)
 231#define FS_IOC_SETFLAGS                 _IOW('f', 2, long)
 232#define FS_IOC_GETVERSION               _IOR('v', 1, long)
 233#define FS_IOC_SETVERSION               _IOW('v', 2, long)
 234#define FS_IOC32_GETFLAGS               _IOR('f', 1, int)
 235#define FS_IOC32_SETFLAGS               _IOW('f', 2, int)
 236#define FS_IOC32_GETVERSION             _IOR('v', 1, int)
 237#define FS_IOC32_SETVERSION             _IOW('v', 2, int)
 238
 239/*
 240 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
 241 */
 242#define FS_SECRM_FL                     0x00000001 /* Secure deletion */
 243#define FS_UNRM_FL                      0x00000002 /* Undelete */
 244#define FS_COMPR_FL                     0x00000004 /* Compress file */
 245#define FS_SYNC_FL                      0x00000008 /* Synchronous updates */
 246#define FS_IMMUTABLE_FL                 0x00000010 /* Immutable file */
 247#define FS_APPEND_FL                    0x00000020 /* writes to file may only append */
 248#define FS_NODUMP_FL                    0x00000040 /* do not dump file */
 249#define FS_NOATIME_FL                   0x00000080 /* do not update atime */
 250/* Reserved for compression usage... */
 251#define FS_DIRTY_FL                     0x00000100
 252#define FS_COMPRBLK_FL                  0x00000200 /* One or more compressed clusters */
 253#define FS_NOCOMP_FL                    0x00000400 /* Don't compress */
 254#define FS_ECOMPR_FL                    0x00000800 /* Compression error */
 255/* End compression flags --- maybe not all used */
 256#define FS_BTREE_FL                     0x00001000 /* btree format dir */
 257#define FS_INDEX_FL                     0x00001000 /* hash-indexed directory */
 258#define FS_IMAGIC_FL                    0x00002000 /* AFS directory */
 259#define FS_JOURNAL_DATA_FL              0x00004000 /* Reserved for ext3 */
 260#define FS_NOTAIL_FL                    0x00008000 /* file tail should not be merged */
 261#define FS_DIRSYNC_FL                   0x00010000 /* dirsync behaviour (directories only) */
 262#define FS_TOPDIR_FL                    0x00020000 /* Top of directory hierarchies*/
 263#define FS_EXTENT_FL                    0x00080000 /* Extents */
 264#define FS_DIRECTIO_FL                  0x00100000 /* Use direct i/o */
 265#define FS_RESERVED_FL                  0x80000000 /* reserved for ext2 lib */
 266
 267#define FS_FL_USER_VISIBLE              0x0003DFFF /* User visible flags */
 268#define FS_FL_USER_MODIFIABLE           0x000380FF /* User modifiable flags */
 269
 270
 271#define SYNC_FILE_RANGE_WAIT_BEFORE     1
 272#define SYNC_FILE_RANGE_WRITE           2
 273#define SYNC_FILE_RANGE_WAIT_AFTER      4
 274
 275#ifdef __KERNEL__
 276
 277#include <linux/linkage.h>
 278#include <linux/wait.h>
 279#include <linux/types.h>
 280#include <linux/kdev_t.h>
 281#include <linux/dcache.h>
 282#include <linux/path.h>
 283#include <linux/stat.h>
 284#include <linux/cache.h>
 285#include <linux/kobject.h>
 286#include <linux/list.h>
 287#include <linux/radix-tree.h>
 288#include <linux/prio_tree.h>
 289#include <linux/init.h>
 290#include <linux/pid.h>
 291#include <linux/mutex.h>
 292#include <linux/capability.h>
 293#include <linux/semaphore.h>
 294
 295#include <asm/atomic.h>
 296#include <asm/byteorder.h>
 297
 298struct export_operations;
 299struct hd_geometry;
 300struct iovec;
 301struct nameidata;
 302struct kiocb;
 303struct pipe_inode_info;
 304struct poll_table_struct;
 305struct kstatfs;
 306struct vm_area_struct;
 307struct vfsmount;
 308
 309extern void __init inode_init(void);
 310extern void __init inode_init_early(void);
 311extern void __init files_init(unsigned long);
 312
 313struct buffer_head;
 314typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 315                        struct buffer_head *bh_result, int create);
 316typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 317                        ssize_t bytes, void *private);
 318
 319/*
 320 * Attribute flags.  These should be or-ed together to figure out what
 321 * has been changed!
 322 */
 323#define ATTR_MODE       (1 << 0)
 324#define ATTR_UID        (1 << 1)
 325#define ATTR_GID        (1 << 2)
 326#define ATTR_SIZE       (1 << 3)
 327#define ATTR_ATIME      (1 << 4)
 328#define ATTR_MTIME      (1 << 5)
 329#define ATTR_CTIME      (1 << 6)
 330#define ATTR_ATIME_SET  (1 << 7)
 331#define ATTR_MTIME_SET  (1 << 8)
 332#define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
 333#define ATTR_ATTR_FLAG  (1 << 10)
 334#define ATTR_KILL_SUID  (1 << 11)
 335#define ATTR_KILL_SGID  (1 << 12)
 336#define ATTR_FILE       (1 << 13)
 337#define ATTR_KILL_PRIV  (1 << 14)
 338#define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
 339#define ATTR_TIMES_SET  (1 << 16)
 340
 341/*
 342 * This is the Inode Attributes structure, used for notify_change().  It
 343 * uses the above definitions as flags, to know which values have changed.
 344 * Also, in this manner, a Filesystem can look at only the values it cares
 345 * about.  Basically, these are the attributes that the VFS layer can
 346 * request to change from the FS layer.
 347 *
 348 * Derek Atkins <warlord@MIT.EDU> 94-10-20
 349 */
 350struct iattr {
 351        unsigned int    ia_valid;
 352        umode_t         ia_mode;
 353        uid_t           ia_uid;
 354        gid_t           ia_gid;
 355        loff_t          ia_size;
 356        struct timespec ia_atime;
 357        struct timespec ia_mtime;
 358        struct timespec ia_ctime;
 359
 360        /*
 361         * Not an attribute, but an auxilary info for filesystems wanting to
 362         * implement an ftruncate() like method.  NOTE: filesystem should
 363         * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
 364         */
 365        struct file     *ia_file;
 366};
 367
 368/*
 369 * Includes for diskquotas.
 370 */
 371#include <linux/quota.h>
 372
 373/** 
 374 * enum positive_aop_returns - aop return codes with specific semantics
 375 *
 376 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
 377 *                          completed, that the page is still locked, and
 378 *                          should be considered active.  The VM uses this hint
 379 *                          to return the page to the active list -- it won't
 380 *                          be a candidate for writeback again in the near
 381 *                          future.  Other callers must be careful to unlock
 382 *                          the page if they get this return.  Returned by
 383 *                          writepage(); 
 384 *
 385 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
 386 *                      unlocked it and the page might have been truncated.
 387 *                      The caller should back up to acquiring a new page and
 388 *                      trying again.  The aop will be taking reasonable
 389 *                      precautions not to livelock.  If the caller held a page
 390 *                      reference, it should drop it before retrying.  Returned
 391 *                      by readpage().
 392 *
 393 * address_space_operation functions return these large constants to indicate
 394 * special semantics to the caller.  These are much larger than the bytes in a
 395 * page to allow for functions that return the number of bytes operated on in a
 396 * given page.
 397 */
 398
 399enum positive_aop_returns {
 400        AOP_WRITEPAGE_ACTIVATE  = 0x80000,
 401        AOP_TRUNCATED_PAGE      = 0x80001,
 402};
 403
 404#define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
 405#define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
 406
 407/*
 408 * oh the beauties of C type declarations.
 409 */
 410struct page;
 411struct address_space;
 412struct writeback_control;
 413
 414struct iov_iter {
 415        const struct iovec *iov;
 416        unsigned long nr_segs;
 417        size_t iov_offset;
 418        size_t count;
 419};
 420
 421size_t iov_iter_copy_from_user_atomic(struct page *page,
 422                struct iov_iter *i, unsigned long offset, size_t bytes);
 423size_t iov_iter_copy_from_user(struct page *page,
 424                struct iov_iter *i, unsigned long offset, size_t bytes);
 425void iov_iter_advance(struct iov_iter *i, size_t bytes);
 426int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
 427size_t iov_iter_single_seg_count(struct iov_iter *i);
 428
 429static inline void iov_iter_init(struct iov_iter *i,
 430                        const struct iovec *iov, unsigned long nr_segs,
 431                        size_t count, size_t written)
 432{
 433        i->iov = iov;
 434        i->nr_segs = nr_segs;
 435        i->iov_offset = 0;
 436        i->count = count + written;
 437
 438        iov_iter_advance(i, written);
 439}
 440
 441static inline size_t iov_iter_count(struct iov_iter *i)
 442{
 443        return i->count;
 444}
 445
 446/*
 447 * "descriptor" for what we're up to with a read.
 448 * This allows us to use the same read code yet
 449 * have multiple different users of the data that
 450 * we read from a file.
 451 *
 452 * The simplest case just copies the data to user
 453 * mode.
 454 */
 455typedef struct {
 456        size_t written;
 457        size_t count;
 458        union {
 459                char __user *buf;
 460                void *data;
 461        } arg;
 462        int error;
 463} read_descriptor_t;
 464
 465typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
 466                unsigned long, unsigned long);
 467
 468struct address_space_operations {
 469        int (*writepage)(struct page *page, struct writeback_control *wbc);
 470        int (*readpage)(struct file *, struct page *);
 471        void (*sync_page)(struct page *);
 472
 473        /* Write back some dirty pages from this mapping. */
 474        int (*writepages)(struct address_space *, struct writeback_control *);
 475
 476        /* Set a page dirty.  Return true if this dirtied it */
 477        int (*set_page_dirty)(struct page *page);
 478
 479        int (*readpages)(struct file *filp, struct address_space *mapping,
 480                        struct list_head *pages, unsigned nr_pages);
 481
 482        /*
 483         * ext3 requires that a successful prepare_write() call be followed
 484         * by a commit_write() call - they must be balanced
 485         */
 486        int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
 487        int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
 488
 489        int (*write_begin)(struct file *, struct address_space *mapping,
 490                                loff_t pos, unsigned len, unsigned flags,
 491                                struct page **pagep, void **fsdata);
 492        int (*write_end)(struct file *, struct address_space *mapping,
 493                                loff_t pos, unsigned len, unsigned copied,
 494                                struct page *page, void *fsdata);
 495
 496        /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
 497        sector_t (*bmap)(struct address_space *, sector_t);
 498        void (*invalidatepage) (struct page *, unsigned long);
 499        int (*releasepage) (struct page *, gfp_t);
 500        ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
 501                        loff_t offset, unsigned long nr_segs);
 502        int (*get_xip_mem)(struct address_space *, pgoff_t, int,
 503                                                void **, unsigned long *);
 504        /* migrate the contents of a page to the specified target */
 505        int (*migratepage) (struct address_space *,
 506                        struct page *, struct page *);
 507        int (*launder_page) (struct page *);
 508        int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
 509                                        unsigned long);
 510};
 511
 512/*
 513 * pagecache_write_begin/pagecache_write_end must be used by general code
 514 * to write into the pagecache.
 515 */
 516int pagecache_write_begin(struct file *, struct address_space *mapping,
 517                                loff_t pos, unsigned len, unsigned flags,
 518                                struct page **pagep, void **fsdata);
 519
 520int pagecache_write_end(struct file *, struct address_space *mapping,
 521                                loff_t pos, unsigned len, unsigned copied,
 522                                struct page *page, void *fsdata);
 523
 524struct backing_dev_info;
 525struct address_space {
 526        struct inode            *host;          /* owner: inode, block_device */
 527        struct radix_tree_root  page_tree;      /* radix tree of all pages */
 528        spinlock_t              tree_lock;      /* and lock protecting it */
 529        unsigned int            i_mmap_writable;/* count VM_SHARED mappings */
 530        struct prio_tree_root   i_mmap;         /* tree of private and shared mappings */
 531        struct list_head        i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
 532        spinlock_t              i_mmap_lock;    /* protect tree, count, list */
 533        unsigned int            truncate_count; /* Cover race condition with truncate */
 534        unsigned long           nrpages;        /* number of total pages */
 535        pgoff_t                 writeback_index;/* writeback starts here */
 536        const struct address_space_operations *a_ops;   /* methods */
 537        unsigned long           flags;          /* error bits/gfp mask */
 538        struct backing_dev_info *backing_dev_info; /* device readahead, etc */
 539        spinlock_t              private_lock;   /* for use by the address_space */
 540        struct list_head        private_list;   /* ditto */
 541        struct address_space    *assoc_mapping; /* ditto */
 542} __attribute__((aligned(sizeof(long))));
 543        /*
 544         * On most architectures that alignment is already the case; but
 545         * must be enforced here for CRIS, to let the least signficant bit
 546         * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
 547         */
 548
 549struct block_device {
 550        dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
 551        struct inode *          bd_inode;       /* will die */
 552        int                     bd_openers;
 553        struct mutex            bd_mutex;       /* open/close mutex */
 554        struct semaphore        bd_mount_sem;
 555        struct list_head        bd_inodes;
 556        void *                  bd_holder;
 557        int                     bd_holders;
 558#ifdef CONFIG_SYSFS
 559        struct list_head        bd_holder_list;
 560#endif
 561        struct block_device *   bd_contains;
 562        unsigned                bd_block_size;
 563        struct hd_struct *      bd_part;
 564        /* number of times partitions within this device have been opened. */
 565        unsigned                bd_part_count;
 566        int                     bd_invalidated;
 567        struct gendisk *        bd_disk;
 568        struct list_head        bd_list;
 569        struct backing_dev_info *bd_inode_backing_dev_info;
 570        /*
 571         * Private data.  You must have bd_claim'ed the block_device
 572         * to use this.  NOTE:  bd_claim allows an owner to claim
 573         * the same device multiple times, the owner must take special
 574         * care to not mess up bd_private for that case.
 575         */
 576        unsigned long           bd_private;
 577};
 578
 579/*
 580 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
 581 * radix trees
 582 */
 583#define PAGECACHE_TAG_DIRTY     0
 584#define PAGECACHE_TAG_WRITEBACK 1
 585
 586int mapping_tagged(struct address_space *mapping, int tag);
 587
 588/*
 589 * Might pages of this file be mapped into userspace?
 590 */
 591static inline int mapping_mapped(struct address_space *mapping)
 592{
 593        return  !prio_tree_empty(&mapping->i_mmap) ||
 594                !list_empty(&mapping->i_mmap_nonlinear);
 595}
 596
 597/*
 598 * Might pages of this file have been modified in userspace?
 599 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
 600 * marks vma as VM_SHARED if it is shared, and the file was opened for
 601 * writing i.e. vma may be mprotected writable even if now readonly.
 602 */
 603static inline int mapping_writably_mapped(struct address_space *mapping)
 604{
 605        return mapping->i_mmap_writable != 0;
 606}
 607
 608/*
 609 * Use sequence counter to get consistent i_size on 32-bit processors.
 610 */
 611#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 612#include <linux/seqlock.h>
 613#define __NEED_I_SIZE_ORDERED
 614#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
 615#else
 616#define i_size_ordered_init(inode) do { } while (0)
 617#endif
 618
 619struct inode {
 620        struct hlist_node       i_hash;
 621        struct list_head        i_list;
 622        struct list_head        i_sb_list;
 623        struct list_head        i_dentry;
 624        unsigned long           i_ino;
 625        atomic_t                i_count;
 626        unsigned int            i_nlink;
 627        uid_t                   i_uid;
 628        gid_t                   i_gid;
 629        dev_t                   i_rdev;
 630        u64                     i_version;
 631        loff_t                  i_size;
 632#ifdef __NEED_I_SIZE_ORDERED
 633        seqcount_t              i_size_seqcount;
 634#endif
 635        struct timespec         i_atime;
 636        struct timespec         i_mtime;
 637        struct timespec         i_ctime;
 638        unsigned int            i_blkbits;
 639        blkcnt_t                i_blocks;
 640        unsigned short          i_bytes;
 641        umode_t                 i_mode;
 642        spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
 643        struct mutex            i_mutex;
 644        struct rw_semaphore     i_alloc_sem;
 645        const struct inode_operations   *i_op;
 646        const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
 647        struct super_block      *i_sb;
 648        struct file_lock        *i_flock;
 649        struct address_space    *i_mapping;
 650        struct address_space    i_data;
 651#ifdef CONFIG_QUOTA
 652        struct dquot            *i_dquot[MAXQUOTAS];
 653#endif
 654        struct list_head        i_devices;
 655        union {
 656                struct pipe_inode_info  *i_pipe;
 657                struct block_device     *i_bdev;
 658                struct cdev             *i_cdev;
 659        };
 660        int                     i_cindex;
 661
 662        __u32                   i_generation;
 663
 664#ifdef CONFIG_DNOTIFY
 665        unsigned long           i_dnotify_mask; /* Directory notify events */
 666        struct dnotify_struct   *i_dnotify; /* for directory notifications */
 667#endif
 668
 669#ifdef CONFIG_INOTIFY
 670        struct list_head        inotify_watches; /* watches on this inode */
 671        struct mutex            inotify_mutex;  /* protects the watches list */
 672#endif
 673
 674        unsigned long           i_state;
 675        unsigned long           dirtied_when;   /* jiffies of first dirtying */
 676
 677        unsigned int            i_flags;
 678
 679        atomic_t                i_writecount;
 680#ifdef CONFIG_SECURITY
 681        void                    *i_security;
 682#endif
 683        void                    *i_private; /* fs or device private pointer */
 684};
 685
 686/*
 687 * inode->i_mutex nesting subclasses for the lock validator:
 688 *
 689 * 0: the object of the current VFS operation
 690 * 1: parent
 691 * 2: child/target
 692 * 3: quota file
 693 *
 694 * The locking order between these classes is
 695 * parent -> child -> normal -> xattr -> quota
 696 */
 697enum inode_i_mutex_lock_class
 698{
 699        I_MUTEX_NORMAL,
 700        I_MUTEX_PARENT,
 701        I_MUTEX_CHILD,
 702        I_MUTEX_XATTR,
 703        I_MUTEX_QUOTA
 704};
 705
 706extern void inode_double_lock(struct inode *inode1, struct inode *inode2);
 707extern void inode_double_unlock(struct inode *inode1, struct inode *inode2);
 708
 709/*
 710 * NOTE: in a 32bit arch with a preemptable kernel and
 711 * an UP compile the i_size_read/write must be atomic
 712 * with respect to the local cpu (unlike with preempt disabled),
 713 * but they don't need to be atomic with respect to other cpus like in
 714 * true SMP (so they need either to either locally disable irq around
 715 * the read or for example on x86 they can be still implemented as a
 716 * cmpxchg8b without the need of the lock prefix). For SMP compiles
 717 * and 64bit archs it makes no difference if preempt is enabled or not.
 718 */
 719static inline loff_t i_size_read(const struct inode *inode)
 720{
 721#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 722        loff_t i_size;
 723        unsigned int seq;
 724
 725        do {
 726                seq = read_seqcount_begin(&inode->i_size_seqcount);
 727                i_size = inode->i_size;
 728        } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
 729        return i_size;
 730#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 731        loff_t i_size;
 732
 733        preempt_disable();
 734        i_size = inode->i_size;
 735        preempt_enable();
 736        return i_size;
 737#else
 738        return inode->i_size;
 739#endif
 740}
 741
 742/*
 743 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
 744 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
 745 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
 746 */
 747static inline void i_size_write(struct inode *inode, loff_t i_size)
 748{
 749#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 750        write_seqcount_begin(&inode->i_size_seqcount);
 751        inode->i_size = i_size;
 752        write_seqcount_end(&inode->i_size_seqcount);
 753#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 754        preempt_disable();
 755        inode->i_size = i_size;
 756        preempt_enable();
 757#else
 758        inode->i_size = i_size;
 759#endif
 760}
 761
 762static inline unsigned iminor(const struct inode *inode)
 763{
 764        return MINOR(inode->i_rdev);
 765}
 766
 767static inline unsigned imajor(const struct inode *inode)
 768{
 769        return MAJOR(inode->i_rdev);
 770}
 771
 772extern struct block_device *I_BDEV(struct inode *inode);
 773
 774struct fown_struct {
 775        rwlock_t lock;          /* protects pid, uid, euid fields */
 776        struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
 777        enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
 778        uid_t uid, euid;        /* uid/euid of process setting the owner */
 779        int signum;             /* posix.1b rt signal to be delivered on IO */
 780};
 781
 782/*
 783 * Track a single file's readahead state
 784 */
 785struct file_ra_state {
 786        pgoff_t start;                  /* where readahead started */
 787        unsigned int size;              /* # of readahead pages */
 788        unsigned int async_size;        /* do asynchronous readahead when
 789                                           there are only # of pages ahead */
 790
 791        unsigned int ra_pages;          /* Maximum readahead window */
 792        int mmap_miss;                  /* Cache miss stat for mmap accesses */
 793        loff_t prev_pos;                /* Cache last read() position */
 794};
 795
 796/*
 797 * Check if @index falls in the readahead windows.
 798 */
 799static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
 800{
 801        return (index >= ra->start &&
 802                index <  ra->start + ra->size);
 803}
 804
 805#define FILE_MNT_WRITE_TAKEN    1
 806#define FILE_MNT_WRITE_RELEASED 2
 807
 808struct file {
 809        /*
 810         * fu_list becomes invalid after file_free is called and queued via
 811         * fu_rcuhead for RCU freeing
 812         */
 813        union {
 814                struct list_head        fu_list;
 815                struct rcu_head         fu_rcuhead;
 816        } f_u;
 817        struct path             f_path;
 818#define f_dentry        f_path.dentry
 819#define f_vfsmnt        f_path.mnt
 820        const struct file_operations    *f_op;
 821        atomic_long_t           f_count;
 822        unsigned int            f_flags;
 823        mode_t                  f_mode;
 824        loff_t                  f_pos;
 825        struct fown_struct      f_owner;
 826        unsigned int            f_uid, f_gid;
 827        struct file_ra_state    f_ra;
 828
 829        u64                     f_version;
 830#ifdef CONFIG_SECURITY
 831        void                    *f_security;
 832#endif
 833        /* needed for tty driver, and maybe others */
 834        void                    *private_data;
 835
 836#ifdef CONFIG_EPOLL
 837        /* Used by fs/eventpoll.c to link all the hooks to this file */
 838        struct list_head        f_ep_links;
 839        spinlock_t              f_ep_lock;
 840#endif /* #ifdef CONFIG_EPOLL */
 841        struct address_space    *f_mapping;
 842#ifdef CONFIG_DEBUG_WRITECOUNT
 843        unsigned long f_mnt_write_state;
 844#endif
 845};
 846extern spinlock_t files_lock;
 847#define file_list_lock() spin_lock(&files_lock);
 848#define file_list_unlock() spin_unlock(&files_lock);
 849
 850#define get_file(x)     atomic_long_inc(&(x)->f_count)
 851#define file_count(x)   atomic_long_read(&(x)->f_count)
 852
 853#ifdef CONFIG_DEBUG_WRITECOUNT
 854static inline void file_take_write(struct file *f)
 855{
 856        WARN_ON(f->f_mnt_write_state != 0);
 857        f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
 858}
 859static inline void file_release_write(struct file *f)
 860{
 861        f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
 862}
 863static inline void file_reset_write(struct file *f)
 864{
 865        f->f_mnt_write_state = 0;
 866}
 867static inline void file_check_state(struct file *f)
 868{
 869        /*
 870         * At this point, either both or neither of these bits
 871         * should be set.
 872         */
 873        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
 874        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
 875}
 876static inline int file_check_writeable(struct file *f)
 877{
 878        if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
 879                return 0;
 880        printk(KERN_WARNING "writeable file with no "
 881                            "mnt_want_write()\n");
 882        WARN_ON(1);
 883        return -EINVAL;
 884}
 885#else /* !CONFIG_DEBUG_WRITECOUNT */
 886static inline void file_take_write(struct file *filp) {}
 887static inline void file_release_write(struct file *filp) {}
 888static inline void file_reset_write(struct file *filp) {}
 889static inline void file_check_state(struct file *filp) {}
 890static inline int file_check_writeable(struct file *filp)
 891{
 892        return 0;
 893}
 894#endif /* CONFIG_DEBUG_WRITECOUNT */
 895
 896#define MAX_NON_LFS     ((1UL<<31) - 1)
 897
 898/* Page cache limit. The filesystems should put that into their s_maxbytes 
 899   limits, otherwise bad things can happen in VM. */ 
 900#if BITS_PER_LONG==32
 901#define MAX_LFS_FILESIZE        (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
 902#elif BITS_PER_LONG==64
 903#define MAX_LFS_FILESIZE        0x7fffffffffffffffUL
 904#endif
 905
 906#define FL_POSIX        1
 907#define FL_FLOCK        2
 908#define FL_ACCESS       8       /* not trying to lock, just looking */
 909#define FL_EXISTS       16      /* when unlocking, test for existence */
 910#define FL_LEASE        32      /* lease held on this file */
 911#define FL_CLOSE        64      /* unlock on close */
 912#define FL_SLEEP        128     /* A blocking lock */
 913
 914/*
 915 * Special return value from posix_lock_file() and vfs_lock_file() for
 916 * asynchronous locking.
 917 */
 918#define FILE_LOCK_DEFERRED 1
 919
 920/*
 921 * The POSIX file lock owner is determined by
 922 * the "struct files_struct" in the thread group
 923 * (or NULL for no owner - BSD locks).
 924 *
 925 * Lockd stuffs a "host" pointer into this.
 926 */
 927typedef struct files_struct *fl_owner_t;
 928
 929struct file_lock_operations {
 930        void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
 931        void (*fl_release_private)(struct file_lock *);
 932};
 933
 934struct lock_manager_operations {
 935        int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
 936        void (*fl_notify)(struct file_lock *);  /* unblock callback */
 937        int (*fl_grant)(struct file_lock *, struct file_lock *, int);
 938        void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
 939        void (*fl_release_private)(struct file_lock *);
 940        void (*fl_break)(struct file_lock *);
 941        int (*fl_mylease)(struct file_lock *, struct file_lock *);
 942        int (*fl_change)(struct file_lock **, int);
 943};
 944
 945/* that will die - we need it for nfs_lock_info */
 946#include <linux/nfs_fs_i.h>
 947
 948struct file_lock {
 949        struct file_lock *fl_next;      /* singly linked list for this inode  */
 950        struct list_head fl_link;       /* doubly linked list of all locks */
 951        struct list_head fl_block;      /* circular list of blocked processes */
 952        fl_owner_t fl_owner;
 953        unsigned char fl_flags;
 954        unsigned char fl_type;
 955        unsigned int fl_pid;
 956        struct pid *fl_nspid;
 957        wait_queue_head_t fl_wait;
 958        struct file *fl_file;
 959        loff_t fl_start;
 960        loff_t fl_end;
 961
 962        struct fasync_struct *  fl_fasync; /* for lease break notifications */
 963        unsigned long fl_break_time;    /* for nonblocking lease breaks */
 964
 965        struct file_lock_operations *fl_ops;    /* Callbacks for filesystems */
 966        struct lock_manager_operations *fl_lmops;       /* Callbacks for lockmanagers */
 967        union {
 968                struct nfs_lock_info    nfs_fl;
 969                struct nfs4_lock_info   nfs4_fl;
 970                struct {
 971                        struct list_head link;  /* link in AFS vnode's pending_locks list */
 972                        int state;              /* state of grant or error if -ve */
 973                } afs;
 974        } fl_u;
 975};
 976
 977/* The following constant reflects the upper bound of the file/locking space */
 978#ifndef OFFSET_MAX
 979#define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
 980#define OFFSET_MAX      INT_LIMIT(loff_t)
 981#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
 982#endif
 983
 984#include <linux/fcntl.h>
 985
 986extern int fcntl_getlk(struct file *, struct flock __user *);
 987extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
 988                        struct flock __user *);
 989
 990#if BITS_PER_LONG == 32
 991extern int fcntl_getlk64(struct file *, struct flock64 __user *);
 992extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
 993                        struct flock64 __user *);
 994#endif
 995
 996extern void send_sigio(struct fown_struct *fown, int fd, int band);
 997extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
 998extern int fcntl_getlease(struct file *filp);
 999
1000/* fs/sync.c */
1001extern int do_sync_mapping_range(struct address_space *mapping, loff_t offset,
1002                        loff_t endbyte, unsigned int flags);
1003
1004/* fs/locks.c */
1005extern void locks_init_lock(struct file_lock *);
1006extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1007extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1008extern void locks_remove_posix(struct file *, fl_owner_t);
1009extern void locks_remove_flock(struct file *);
1010extern void posix_test_lock(struct file *, struct file_lock *);
1011extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1012extern int posix_lock_file_wait(struct file *, struct file_lock *);
1013extern int posix_unblock_lock(struct file *, struct file_lock *);
1014extern int vfs_test_lock(struct file *, struct file_lock *);
1015extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1016extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1017extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1018extern int __break_lease(struct inode *inode, unsigned int flags);
1019extern void lease_get_mtime(struct inode *, struct timespec *time);
1020extern int generic_setlease(struct file *, long, struct file_lock **);
1021extern int vfs_setlease(struct file *, long, struct file_lock **);
1022extern int lease_modify(struct file_lock **, int);
1023extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1024extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1025extern struct seq_operations locks_seq_operations;
1026
1027struct fasync_struct {
1028        int     magic;
1029        int     fa_fd;
1030        struct  fasync_struct   *fa_next; /* singly linked list */
1031        struct  file            *fa_file;
1032};
1033
1034#define FASYNC_MAGIC 0x4601
1035
1036/* SMP safe fasync helpers: */
1037extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1038/* can be called from interrupts */
1039extern void kill_fasync(struct fasync_struct **, int, int);
1040/* only for net: no internal synchronization */
1041extern void __kill_fasync(struct fasync_struct *, int, int);
1042
1043extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1044extern int f_setown(struct file *filp, unsigned long arg, int force);
1045extern void f_delown(struct file *filp);
1046extern pid_t f_getown(struct file *filp);
1047extern int send_sigurg(struct fown_struct *fown);
1048
1049/*
1050 *      Umount options
1051 */
1052
1053#define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1054#define MNT_DETACH      0x00000002      /* Just detach from the tree */
1055#define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1056
1057extern struct list_head super_blocks;
1058extern spinlock_t sb_lock;
1059
1060#define sb_entry(list)  list_entry((list), struct super_block, s_list)
1061#define S_BIAS (1<<30)
1062struct super_block {
1063        struct list_head        s_list;         /* Keep this first */
1064        dev_t                   s_dev;          /* search index; _not_ kdev_t */
1065        unsigned long           s_blocksize;
1066        unsigned char           s_blocksize_bits;
1067        unsigned char           s_dirt;
1068        unsigned long long      s_maxbytes;     /* Max file size */
1069        struct file_system_type *s_type;
1070        const struct super_operations   *s_op;
1071        struct dquot_operations *dq_op;
1072        struct quotactl_ops     *s_qcop;
1073        const struct export_operations *s_export_op;
1074        unsigned long           s_flags;
1075        unsigned long           s_magic;
1076        struct dentry           *s_root;
1077        struct rw_semaphore     s_umount;
1078        struct mutex            s_lock;
1079        int                     s_count;
1080        int                     s_syncing;
1081        int                     s_need_sync_fs;
1082        atomic_t                s_active;
1083#ifdef CONFIG_SECURITY
1084        void                    *s_security;
1085#endif
1086        struct xattr_handler    **s_xattr;
1087
1088        struct list_head        s_inodes;       /* all inodes */
1089        struct list_head        s_dirty;        /* dirty inodes */
1090        struct list_head        s_io;           /* parked for writeback */
1091        struct list_head        s_more_io;      /* parked for more writeback */
1092        struct hlist_head       s_anon;         /* anonymous dentries for (nfs) exporting */
1093        struct list_head        s_files;
1094        /* s_dentry_lru and s_nr_dentry_unused are protected by dcache_lock */
1095        struct list_head        s_dentry_lru;   /* unused dentry lru */
1096        int                     s_nr_dentry_unused;     /* # of dentry on lru */
1097
1098        struct block_device     *s_bdev;
1099        struct mtd_info         *s_mtd;
1100        struct list_head        s_instances;
1101        struct quota_info       s_dquot;        /* Diskquota specific options */
1102
1103        int                     s_frozen;
1104        wait_queue_head_t       s_wait_unfrozen;
1105
1106        char s_id[32];                          /* Informational name */
1107
1108        void                    *s_fs_info;     /* Filesystem private info */
1109
1110        /*
1111         * The next field is for VFS *only*. No filesystems have any business
1112         * even looking at it. You had been warned.
1113         */
1114        struct mutex s_vfs_rename_mutex;        /* Kludge */
1115
1116        /* Granularity of c/m/atime in ns.
1117           Cannot be worse than a second */
1118        u32                s_time_gran;
1119
1120        /*
1121         * Filesystem subtype.  If non-empty the filesystem type field
1122         * in /proc/mounts will be "type.subtype"
1123         */
1124        char *s_subtype;
1125
1126        /*
1127         * Saved mount options for lazy filesystems using
1128         * generic_show_options()
1129         */
1130        char *s_options;
1131};
1132
1133extern struct timespec current_fs_time(struct super_block *sb);
1134
1135/*
1136 * Snapshotting support.
1137 */
1138enum {
1139        SB_UNFROZEN = 0,
1140        SB_FREEZE_WRITE = 1,
1141        SB_FREEZE_TRANS = 2,
1142};
1143
1144#define vfs_check_frozen(sb, level) \
1145        wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1146
1147#define get_fs_excl() atomic_inc(&current->fs_excl)
1148#define put_fs_excl() atomic_dec(&current->fs_excl)
1149#define has_fs_excl() atomic_read(&current->fs_excl)
1150
1151#define is_owner_or_cap(inode)  \
1152        ((current->fsuid == (inode)->i_uid) || capable(CAP_FOWNER))
1153
1154/* not quite ready to be deprecated, but... */
1155extern void lock_super(struct super_block *);
1156extern void unlock_super(struct super_block *);
1157
1158/*
1159 * VFS helper functions..
1160 */
1161extern int vfs_permission(struct nameidata *, int);
1162extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1163extern int vfs_mkdir(struct inode *, struct dentry *, int);
1164extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1165extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1166extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1167extern int vfs_rmdir(struct inode *, struct dentry *);
1168extern int vfs_unlink(struct inode *, struct dentry *);
1169extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1170
1171/*
1172 * VFS dentry helper functions.
1173 */
1174extern void dentry_unhash(struct dentry *dentry);
1175
1176/*
1177 * VFS file helper functions.
1178 */
1179extern int file_permission(struct file *, int);
1180
1181/*
1182 * File types
1183 *
1184 * NOTE! These match bits 12..15 of stat.st_mode
1185 * (ie "(i_mode >> 12) & 15").
1186 */
1187#define DT_UNKNOWN      0
1188#define DT_FIFO         1
1189#define DT_CHR          2
1190#define DT_DIR          4
1191#define DT_BLK          6
1192#define DT_REG          8
1193#define DT_LNK          10
1194#define DT_SOCK         12
1195#define DT_WHT          14
1196
1197#define OSYNC_METADATA  (1<<0)
1198#define OSYNC_DATA      (1<<1)
1199#define OSYNC_INODE     (1<<2)
1200int generic_osync_inode(struct inode *, struct address_space *, int);
1201
1202/*
1203 * This is the "filldir" function type, used by readdir() to let
1204 * the kernel specify what kind of dirent layout it wants to have.
1205 * This allows the kernel to read directories into kernel space or
1206 * to have different dirent layouts depending on the binary type.
1207 */
1208typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1209
1210struct block_device_operations {
1211        int (*open) (struct inode *, struct file *);
1212        int (*release) (struct inode *, struct file *);
1213        int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
1214        long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
1215        long (*compat_ioctl) (struct file *, unsigned, unsigned long);
1216        int (*direct_access) (struct block_device *, sector_t,
1217                                                void **, unsigned long *);
1218        int (*media_changed) (struct gendisk *);
1219        int (*revalidate_disk) (struct gendisk *);
1220        int (*getgeo)(struct block_device *, struct hd_geometry *);
1221        struct module *owner;
1222};
1223
1224/* These macros are for out of kernel modules to test that
1225 * the kernel supports the unlocked_ioctl and compat_ioctl
1226 * fields in struct file_operations. */
1227#define HAVE_COMPAT_IOCTL 1
1228#define HAVE_UNLOCKED_IOCTL 1
1229
1230/*
1231 * NOTE:
1232 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1233 * can be called without the big kernel lock held in all filesystems.
1234 */
1235struct file_operations {
1236        struct module *owner;
1237        loff_t (*llseek) (struct file *, loff_t, int);
1238        ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1239        ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1240        ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1241        ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1242        int (*readdir) (struct file *, void *, filldir_t);
1243        unsigned int (*poll) (struct file *, struct poll_table_struct *);
1244        int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1245        long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1246        long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1247        int (*mmap) (struct file *, struct vm_area_struct *);
1248        int (*open) (struct inode *, struct file *);
1249        int (*flush) (struct file *, fl_owner_t id);
1250        int (*release) (struct inode *, struct file *);
1251        int (*fsync) (struct file *, struct dentry *, int datasync);
1252        int (*aio_fsync) (struct kiocb *, int datasync);
1253        int (*fasync) (int, struct file *, int);
1254        int (*lock) (struct file *, int, struct file_lock *);
1255        ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1256        unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1257        int (*check_flags)(int);
1258        int (*dir_notify)(struct file *filp, unsigned long arg);
1259        int (*flock) (struct file *, int, struct file_lock *);
1260        ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1261        ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1262        int (*setlease)(struct file *, long, struct file_lock **);
1263};
1264
1265struct inode_operations {
1266        int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1267        struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1268        int (*link) (struct dentry *,struct inode *,struct dentry *);
1269        int (*unlink) (struct inode *,struct dentry *);
1270        int (*symlink) (struct inode *,struct dentry *,const char *);
1271        int (*mkdir) (struct inode *,struct dentry *,int);
1272        int (*rmdir) (struct inode *,struct dentry *);
1273        int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1274        int (*rename) (struct inode *, struct dentry *,
1275                        struct inode *, struct dentry *);
1276        int (*readlink) (struct dentry *, char __user *,int);
1277        void * (*follow_link) (struct dentry *, struct nameidata *);
1278        void (*put_link) (struct dentry *, struct nameidata *, void *);
1279        void (*truncate) (struct inode *);
1280        int (*permission) (struct inode *, int);
1281        int (*setattr) (struct dentry *, struct iattr *);
1282        int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1283        int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1284        ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1285        ssize_t (*listxattr) (struct dentry *, char *, size_t);
1286        int (*removexattr) (struct dentry *, const char *);
1287        void (*truncate_range)(struct inode *, loff_t, loff_t);
1288        long (*fallocate)(struct inode *inode, int mode, loff_t offset,
1289                          loff_t len);
1290};
1291
1292struct seq_file;
1293
1294ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1295                                unsigned long nr_segs, unsigned long fast_segs,
1296                                struct iovec *fast_pointer,
1297                                struct iovec **ret_pointer);
1298
1299extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1300extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1301extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1302                unsigned long, loff_t *);
1303extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1304                unsigned long, loff_t *);
1305
1306struct super_operations {
1307        struct inode *(*alloc_inode)(struct super_block *sb);
1308        void (*destroy_inode)(struct inode *);
1309
1310        void (*dirty_inode) (struct inode *);
1311        int (*write_inode) (struct inode *, int);
1312        void (*drop_inode) (struct inode *);
1313        void (*delete_inode) (struct inode *);
1314        void (*put_super) (struct super_block *);
1315        void (*write_super) (struct super_block *);
1316        int (*sync_fs)(struct super_block *sb, int wait);
1317        void (*write_super_lockfs) (struct super_block *);
1318        void (*unlockfs) (struct super_block *);
1319        int (*statfs) (struct dentry *, struct kstatfs *);
1320        int (*remount_fs) (struct super_block *, int *, char *);
1321        void (*clear_inode) (struct inode *);
1322        void (*umount_begin) (struct super_block *);
1323
1324        int (*show_options)(struct seq_file *, struct vfsmount *);
1325        int (*show_stats)(struct seq_file *, struct vfsmount *);
1326#ifdef CONFIG_QUOTA
1327        ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1328        ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1329#endif
1330};
1331
1332/*
1333 * Inode state bits.  Protected by inode_lock.
1334 *
1335 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1336 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1337 *
1338 * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1339 * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1340 * various stages of removing an inode.
1341 *
1342 * Two bits are used for locking and completion notification, I_LOCK and I_SYNC.
1343 *
1344 * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1345 *                      fdatasync().  i_atime is the usual cause.
1346 * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1347 *                      these changes separately from I_DIRTY_SYNC so that we
1348 *                      don't have to write inode on fdatasync() when only
1349 *                      mtime has changed in it.
1350 * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1351 * I_NEW                get_new_inode() sets i_state to I_LOCK|I_NEW.  Both
1352 *                      are cleared by unlock_new_inode(), called from iget().
1353 * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1354 *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1355 *                      cleared.
1356 * I_FREEING            Set when inode is about to be freed but still has dirty
1357 *                      pages or buffers attached or the inode itself is still
1358 *                      dirty.
1359 * I_CLEAR              Set by clear_inode().  In this state the inode is clean
1360 *                      and can be destroyed.
1361 *
1362 *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1363 *                      prohibited for many purposes.  iget() must wait for
1364 *                      the inode to be completely released, then create it
1365 *                      anew.  Other functions will just ignore such inodes,
1366 *                      if appropriate.  I_LOCK is used for waiting.
1367 *
1368 * I_LOCK               Serves as both a mutex and completion notification.
1369 *                      New inodes set I_LOCK.  If two processes both create
1370 *                      the same inode, one of them will release its inode and
1371 *                      wait for I_LOCK to be released before returning.
1372 *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1373 *                      also cause waiting on I_LOCK, without I_LOCK actually
1374 *                      being set.  find_inode() uses this to prevent returning
1375 *                      nearly-dead inodes.
1376 * I_SYNC               Similar to I_LOCK, but limited in scope to writeback
1377 *                      of inode dirty data.  Having a separate lock for this
1378 *                      purpose reduces latency and prevents some filesystem-
1379 *                      specific deadlocks.
1380 *
1381 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1382 * Q: igrab() only checks on (I_FREEING|I_WILL_FREE).  Should it also check on
1383 *    I_CLEAR?  If not, why?
1384 */
1385#define I_DIRTY_SYNC            1
1386#define I_DIRTY_DATASYNC        2
1387#define I_DIRTY_PAGES           4
1388#define I_NEW                   8
1389#define I_WILL_FREE             16
1390#define I_FREEING               32
1391#define I_CLEAR                 64
1392#define __I_LOCK                7
1393#define I_LOCK                  (1 << __I_LOCK)
1394#define __I_SYNC                8
1395#define I_SYNC                  (1 << __I_SYNC)
1396
1397#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1398
1399extern void __mark_inode_dirty(struct inode *, int);
1400static inline void mark_inode_dirty(struct inode *inode)
1401{
1402        __mark_inode_dirty(inode, I_DIRTY);
1403}
1404
1405static inline void mark_inode_dirty_sync(struct inode *inode)
1406{
1407        __mark_inode_dirty(inode, I_DIRTY_SYNC);
1408}
1409
1410/**
1411 * inc_nlink - directly increment an inode's link count
1412 * @inode: inode
1413 *
1414 * This is a low-level filesystem helper to replace any
1415 * direct filesystem manipulation of i_nlink.  Currently,
1416 * it is only here for parity with dec_nlink().
1417 */
1418static inline void inc_nlink(struct inode *inode)
1419{
1420        inode->i_nlink++;
1421}
1422
1423static inline void inode_inc_link_count(struct inode *inode)
1424{
1425        inc_nlink(inode);
1426        mark_inode_dirty(inode);
1427}
1428
1429/**
1430 * drop_nlink - directly drop an inode's link count
1431 * @inode: inode
1432 *
1433 * This is a low-level filesystem helper to replace any
1434 * direct filesystem manipulation of i_nlink.  In cases
1435 * where we are attempting to track writes to the
1436 * filesystem, a decrement to zero means an imminent
1437 * write when the file is truncated and actually unlinked
1438 * on the filesystem.
1439 */
1440static inline void drop_nlink(struct inode *inode)
1441{
1442        inode->i_nlink--;
1443}
1444
1445/**
1446 * clear_nlink - directly zero an inode's link count
1447 * @inode: inode
1448 *
1449 * This is a low-level filesystem helper to replace any
1450 * direct filesystem manipulation of i_nlink.  See
1451 * drop_nlink() for why we care about i_nlink hitting zero.
1452 */
1453static inline void clear_nlink(struct inode *inode)
1454{
1455        inode->i_nlink = 0;
1456}
1457
1458static inline void inode_dec_link_count(struct inode *inode)
1459{
1460        drop_nlink(inode);
1461        mark_inode_dirty(inode);
1462}
1463
1464/**
1465 * inode_inc_iversion - increments i_version
1466 * @inode: inode that need to be updated
1467 *
1468 * Every time the inode is modified, the i_version field will be incremented.
1469 * The filesystem has to be mounted with i_version flag
1470 */
1471
1472static inline void inode_inc_iversion(struct inode *inode)
1473{
1474       spin_lock(&inode->i_lock);
1475       inode->i_version++;
1476       spin_unlock(&inode->i_lock);
1477}
1478
1479extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1480static inline void file_accessed(struct file *file)
1481{
1482        if (!(file->f_flags & O_NOATIME))
1483                touch_atime(file->f_path.mnt, file->f_path.dentry);
1484}
1485
1486int sync_inode(struct inode *inode, struct writeback_control *wbc);
1487
1488struct file_system_type {
1489        const char *name;
1490        int fs_flags;
1491        int (*get_sb) (struct file_system_type *, int,
1492                       const char *, void *, struct vfsmount *);
1493        void (*kill_sb) (struct super_block *);
1494        struct module *owner;
1495        struct file_system_type * next;
1496        struct list_head fs_supers;
1497
1498        struct lock_class_key s_lock_key;
1499        struct lock_class_key s_umount_key;
1500
1501        struct lock_class_key i_lock_key;
1502        struct lock_class_key i_mutex_key;
1503        struct lock_class_key i_mutex_dir_key;
1504        struct lock_class_key i_alloc_sem_key;
1505};
1506
1507extern int get_sb_bdev(struct file_system_type *fs_type,
1508        int flags, const char *dev_name, void *data,
1509        int (*fill_super)(struct super_block *, void *, int),
1510        struct vfsmount *mnt);
1511extern int get_sb_single(struct file_system_type *fs_type,
1512        int flags, void *data,
1513        int (*fill_super)(struct super_block *, void *, int),
1514        struct vfsmount *mnt);
1515extern int get_sb_nodev(struct file_system_type *fs_type,
1516        int flags, void *data,
1517        int (*fill_super)(struct super_block *, void *, int),
1518        struct vfsmount *mnt);
1519void generic_shutdown_super(struct super_block *sb);
1520void kill_block_super(struct super_block *sb);
1521void kill_anon_super(struct super_block *sb);
1522void kill_litter_super(struct super_block *sb);
1523void deactivate_super(struct super_block *sb);
1524int set_anon_super(struct super_block *s, void *data);
1525struct super_block *sget(struct file_system_type *type,
1526                        int (*test)(struct super_block *,void *),
1527                        int (*set)(struct super_block *,void *),
1528                        void *data);
1529extern int get_sb_pseudo(struct file_system_type *, char *,
1530        const struct super_operations *ops, unsigned long,
1531        struct vfsmount *mnt);
1532extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1533int __put_super_and_need_restart(struct super_block *sb);
1534void unnamed_dev_init(void);
1535
1536/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1537#define fops_get(fops) \
1538        (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1539#define fops_put(fops) \
1540        do { if (fops) module_put((fops)->owner); } while(0)
1541
1542extern int register_filesystem(struct file_system_type *);
1543extern int unregister_filesystem(struct file_system_type *);
1544extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1545#define kern_mount(type) kern_mount_data(type, NULL)
1546extern int may_umount_tree(struct vfsmount *);
1547extern int may_umount(struct vfsmount *);
1548extern long do_mount(char *, char *, char *, unsigned long, void *);
1549extern struct vfsmount *collect_mounts(struct vfsmount *, struct dentry *);
1550extern void drop_collected_mounts(struct vfsmount *);
1551
1552extern int vfs_statfs(struct dentry *, struct kstatfs *);
1553
1554/* /sys/fs */
1555extern struct kobject *fs_kobj;
1556
1557#define FLOCK_VERIFY_READ  1
1558#define FLOCK_VERIFY_WRITE 2
1559
1560extern int locks_mandatory_locked(struct inode *);
1561extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1562
1563/*
1564 * Candidates for mandatory locking have the setgid bit set
1565 * but no group execute bit -  an otherwise meaningless combination.
1566 */
1567
1568static inline int __mandatory_lock(struct inode *ino)
1569{
1570        return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1571}
1572
1573/*
1574 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1575 * otherwise these will be advisory locks
1576 */
1577
1578static inline int mandatory_lock(struct inode *ino)
1579{
1580        return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1581}
1582
1583static inline int locks_verify_locked(struct inode *inode)
1584{
1585        if (mandatory_lock(inode))
1586                return locks_mandatory_locked(inode);
1587        return 0;
1588}
1589
1590extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1591
1592static inline int locks_verify_truncate(struct inode *inode,
1593                                    struct file *filp,
1594                                    loff_t size)
1595{
1596        if (inode->i_flock && mandatory_lock(inode))
1597                return locks_mandatory_area(
1598                        FLOCK_VERIFY_WRITE, inode, filp,
1599                        size < inode->i_size ? size : inode->i_size,
1600                        (size < inode->i_size ? inode->i_size - size
1601                         : size - inode->i_size)
1602                );
1603        return 0;
1604}
1605
1606static inline int break_lease(struct inode *inode, unsigned int mode)
1607{
1608        if (inode->i_flock)
1609                return __break_lease(inode, mode);
1610        return 0;
1611}
1612
1613/* fs/open.c */
1614
1615extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1616                       struct file *filp);
1617extern long do_sys_open(int dfd, const char __user *filename, int flags,
1618                        int mode);
1619extern struct file *filp_open(const char *, int, int);
1620extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1621extern int filp_close(struct file *, fl_owner_t id);
1622extern char * getname(const char __user *);
1623
1624/* fs/dcache.c */
1625extern void __init vfs_caches_init_early(void);
1626extern void __init vfs_caches_init(unsigned long);
1627
1628extern struct kmem_cache *names_cachep;
1629
1630#define __getname()     kmem_cache_alloc(names_cachep, GFP_KERNEL)
1631#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1632#ifndef CONFIG_AUDITSYSCALL
1633#define putname(name)   __putname(name)
1634#else
1635extern void putname(const char *name);
1636#endif
1637
1638#ifdef CONFIG_BLOCK
1639extern int register_blkdev(unsigned int, const char *);
1640extern void unregister_blkdev(unsigned int, const char *);
1641extern struct block_device *bdget(dev_t);
1642extern void bd_set_size(struct block_device *, loff_t size);
1643extern void bd_forget(struct inode *inode);
1644extern void bdput(struct block_device *);
1645extern struct block_device *open_by_devnum(dev_t, unsigned);
1646#else
1647static inline void bd_forget(struct inode *inode) {}
1648#endif
1649extern const struct file_operations def_blk_fops;
1650extern const struct file_operations def_chr_fops;
1651extern const struct file_operations bad_sock_fops;
1652extern const struct file_operations def_fifo_fops;
1653#ifdef CONFIG_BLOCK
1654extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1655extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1656extern int blkdev_driver_ioctl(struct inode *inode, struct file *file,
1657                               struct gendisk *disk, unsigned cmd,
1658                               unsigned long arg);
1659extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1660extern int blkdev_get(struct block_device *, mode_t, unsigned);
1661extern int blkdev_put(struct block_device *);
1662extern int bd_claim(struct block_device *, void *);
1663extern void bd_release(struct block_device *);
1664#ifdef CONFIG_SYSFS
1665extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1666extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1667#else
1668#define bd_claim_by_disk(bdev, holder, disk)    bd_claim(bdev, holder)
1669#define bd_release_from_disk(bdev, disk)        bd_release(bdev)
1670#endif
1671#endif
1672
1673/* fs/char_dev.c */
1674#define CHRDEV_MAJOR_HASH_SIZE  255
1675extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1676extern int register_chrdev_region(dev_t, unsigned, const char *);
1677extern int register_chrdev(unsigned int, const char *,
1678                           const struct file_operations *);
1679extern void unregister_chrdev(unsigned int, const char *);
1680extern void unregister_chrdev_region(dev_t, unsigned);
1681extern void chrdev_show(struct seq_file *,off_t);
1682
1683/* fs/block_dev.c */
1684#define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
1685
1686#ifdef CONFIG_BLOCK
1687#define BLKDEV_MAJOR_HASH_SIZE  255
1688extern const char *__bdevname(dev_t, char *buffer);
1689extern const char *bdevname(struct block_device *bdev, char *buffer);
1690extern struct block_device *lookup_bdev(const char *);
1691extern struct block_device *open_bdev_excl(const char *, int, void *);
1692extern void close_bdev_excl(struct block_device *);
1693extern void blkdev_show(struct seq_file *,off_t);
1694#else
1695#define BLKDEV_MAJOR_HASH_SIZE  0
1696#endif
1697
1698extern void init_special_inode(struct inode *, umode_t, dev_t);
1699
1700/* Invalid inode operations -- fs/bad_inode.c */
1701extern void make_bad_inode(struct inode *);
1702extern int is_bad_inode(struct inode *);
1703
1704extern const struct file_operations read_pipefifo_fops;
1705extern const struct file_operations write_pipefifo_fops;
1706extern const struct file_operations rdwr_pipefifo_fops;
1707
1708extern int fs_may_remount_ro(struct super_block *);
1709
1710#ifdef CONFIG_BLOCK
1711/*
1712 * return READ, READA, or WRITE
1713 */
1714#define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1715
1716/*
1717 * return data direction, READ or WRITE
1718 */
1719#define bio_data_dir(bio)       ((bio)->bi_rw & 1)
1720
1721extern int check_disk_change(struct block_device *);
1722extern int __invalidate_device(struct block_device *);
1723extern int invalidate_partition(struct gendisk *, int);
1724#endif
1725extern int invalidate_inodes(struct super_block *);
1726unsigned long __invalidate_mapping_pages(struct address_space *mapping,
1727                                        pgoff_t start, pgoff_t end,
1728                                        bool be_atomic);
1729unsigned long invalidate_mapping_pages(struct address_space *mapping,
1730                                        pgoff_t start, pgoff_t end);
1731
1732static inline unsigned long __deprecated
1733invalidate_inode_pages(struct address_space *mapping)
1734{
1735        return invalidate_mapping_pages(mapping, 0, ~0UL);
1736}
1737
1738static inline void invalidate_remote_inode(struct inode *inode)
1739{
1740        if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1741            S_ISLNK(inode->i_mode))
1742                invalidate_mapping_pages(inode->i_mapping, 0, -1);
1743}
1744extern int invalidate_inode_pages2(struct address_space *mapping);
1745extern int invalidate_inode_pages2_range(struct address_space *mapping,
1746                                         pgoff_t start, pgoff_t end);
1747extern void generic_sync_sb_inodes(struct super_block *sb,
1748                                struct writeback_control *wbc);
1749extern int write_inode_now(struct inode *, int);
1750extern int filemap_fdatawrite(struct address_space *);
1751extern int filemap_flush(struct address_space *);
1752extern int filemap_fdatawait(struct address_space *);
1753extern int filemap_write_and_wait(struct address_space *mapping);
1754extern int filemap_write_and_wait_range(struct address_space *mapping,
1755                                        loff_t lstart, loff_t lend);
1756extern int wait_on_page_writeback_range(struct address_space *mapping,
1757                                pgoff_t start, pgoff_t end);
1758extern int __filemap_fdatawrite_range(struct address_space *mapping,
1759                                loff_t start, loff_t end, int sync_mode);
1760extern int filemap_fdatawrite_range(struct address_space *mapping,
1761                                loff_t start, loff_t end);
1762
1763extern long do_fsync(struct file *file, int datasync);
1764extern void sync_supers(void);
1765extern void sync_filesystems(int wait);
1766extern void __fsync_super(struct super_block *sb);
1767extern void emergency_sync(void);
1768extern void emergency_remount(void);
1769extern int do_remount_sb(struct super_block *sb, int flags,
1770                         void *data, int force);
1771#ifdef CONFIG_BLOCK
1772extern sector_t bmap(struct inode *, sector_t);
1773#endif
1774extern int notify_change(struct dentry *, struct iattr *);
1775extern int inode_permission(struct inode *, int);
1776extern int generic_permission(struct inode *, int,
1777                int (*check_acl)(struct inode *, int));
1778
1779extern int get_write_access(struct inode *);
1780extern int deny_write_access(struct file *);
1781static inline void put_write_access(struct inode * inode)
1782{
1783        atomic_dec(&inode->i_writecount);
1784}
1785static inline void allow_write_access(struct file *file)
1786{
1787        if (file)
1788                atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
1789}
1790extern int do_pipe(int *);
1791extern int do_pipe_flags(int *, int);
1792extern struct file *create_read_pipe(struct file *f, int flags);
1793extern struct file *create_write_pipe(int flags);
1794extern void free_write_pipe(struct file *);
1795
1796extern struct file *do_filp_open(int dfd, const char *pathname,
1797                int open_flag, int mode);
1798extern int may_open(struct nameidata *, int, int);
1799
1800extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1801extern struct file * open_exec(const char *);
1802 
1803/* fs/dcache.c -- generic fs support functions */
1804extern int is_subdir(struct dentry *, struct dentry *);
1805extern ino_t find_inode_number(struct dentry *, struct qstr *);
1806
1807#include <linux/err.h>
1808
1809/* needed for stackable file system support */
1810extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1811
1812extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1813
1814extern void inode_init_once(struct inode *);
1815extern void iput(struct inode *);
1816extern struct inode * igrab(struct inode *);
1817extern ino_t iunique(struct super_block *, ino_t);
1818extern int inode_needs_sync(struct inode *inode);
1819extern void generic_delete_inode(struct inode *inode);
1820extern void generic_drop_inode(struct inode *inode);
1821
1822extern struct inode *ilookup5_nowait(struct super_block *sb,
1823                unsigned long hashval, int (*test)(struct inode *, void *),
1824                void *data);
1825extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1826                int (*test)(struct inode *, void *), void *data);
1827extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1828
1829extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1830extern struct inode * iget_locked(struct super_block *, unsigned long);
1831extern void unlock_new_inode(struct inode *);
1832
1833extern void __iget(struct inode * inode);
1834extern void iget_failed(struct inode *);
1835extern void clear_inode(struct inode *);
1836extern void destroy_inode(struct inode *);
1837extern struct inode *new_inode(struct super_block *);
1838extern int should_remove_suid(struct dentry *);
1839extern int file_remove_suid(struct file *);
1840
1841extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1842extern void remove_inode_hash(struct inode *);
1843static inline void insert_inode_hash(struct inode *inode) {
1844        __insert_inode_hash(inode, inode->i_ino);
1845}
1846
1847extern struct file * get_empty_filp(void);
1848extern void file_move(struct file *f, struct list_head *list);
1849extern void file_kill(struct file *f);
1850#ifdef CONFIG_BLOCK
1851struct bio;
1852extern void submit_bio(int, struct bio *);
1853extern int bdev_read_only(struct block_device *);
1854#endif
1855extern int set_blocksize(struct block_device *, int);
1856extern int sb_set_blocksize(struct super_block *, int);
1857extern int sb_min_blocksize(struct super_block *, int);
1858extern int sb_has_dirty_inodes(struct super_block *);
1859
1860extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1861extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1862extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1863int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1864extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1865extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1866extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1867                unsigned long, loff_t);
1868extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1869                unsigned long *, loff_t, loff_t *, size_t, size_t);
1870extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1871                unsigned long, loff_t, loff_t *, size_t, ssize_t);
1872extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1873extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1874extern int generic_segment_checks(const struct iovec *iov,
1875                unsigned long *nr_segs, size_t *count, int access_flags);
1876
1877/* fs/splice.c */
1878extern ssize_t generic_file_splice_read(struct file *, loff_t *,
1879                struct pipe_inode_info *, size_t, unsigned int);
1880extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
1881                struct file *, loff_t *, size_t, unsigned int);
1882extern ssize_t generic_file_splice_write_nolock(struct pipe_inode_info *,
1883                struct file *, loff_t *, size_t, unsigned int);
1884extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
1885                struct file *out, loff_t *, size_t len, unsigned int flags);
1886extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
1887                size_t len, unsigned int flags);
1888
1889extern void
1890file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1891extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1892extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1893extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
1894                        int origin);
1895extern int generic_file_open(struct inode * inode, struct file * filp);
1896extern int nonseekable_open(struct inode * inode, struct file * filp);
1897
1898#ifdef CONFIG_FS_XIP
1899extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1900                             loff_t *ppos);
1901extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
1902extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1903                              size_t len, loff_t *ppos);
1904extern int xip_truncate_page(struct address_space *mapping, loff_t from);
1905#else
1906static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1907{
1908        return 0;
1909}
1910#endif
1911
1912#ifdef CONFIG_BLOCK
1913ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1914        struct block_device *bdev, const struct iovec *iov, loff_t offset,
1915        unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1916        int lock_type);
1917
1918enum {
1919        DIO_LOCKING = 1, /* need locking between buffered and direct access */
1920        DIO_NO_LOCKING,  /* bdev; no locking at all between buffered/direct */
1921        DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1922};
1923
1924static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1925        struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1926        loff_t offset, unsigned long nr_segs, get_block_t get_block,
1927        dio_iodone_t end_io)
1928{
1929        return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1930                                nr_segs, get_block, end_io, DIO_LOCKING);
1931}
1932
1933static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1934        struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1935        loff_t offset, unsigned long nr_segs, get_block_t get_block,
1936        dio_iodone_t end_io)
1937{
1938        return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1939                                nr_segs, get_block, end_io, DIO_NO_LOCKING);
1940}
1941
1942static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1943        struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1944        loff_t offset, unsigned long nr_segs, get_block_t get_block,
1945        dio_iodone_t end_io)
1946{
1947        return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1948                                nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1949}
1950#endif
1951
1952extern const struct file_operations generic_ro_fops;
1953
1954#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1955
1956extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1957extern int vfs_follow_link(struct nameidata *, const char *);
1958extern int page_readlink(struct dentry *, char __user *, int);
1959extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1960extern void page_put_link(struct dentry *, struct nameidata *, void *);
1961extern int __page_symlink(struct inode *inode, const char *symname, int len,
1962                gfp_t gfp_mask);
1963extern int page_symlink(struct inode *inode, const char *symname, int len);
1964extern const struct inode_operations page_symlink_inode_operations;
1965extern int generic_readlink(struct dentry *, char __user *, int);
1966extern void generic_fillattr(struct inode *, struct kstat *);
1967extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1968void inode_add_bytes(struct inode *inode, loff_t bytes);
1969void inode_sub_bytes(struct inode *inode, loff_t bytes);
1970loff_t inode_get_bytes(struct inode *inode);
1971void inode_set_bytes(struct inode *inode, loff_t bytes);
1972
1973extern int vfs_readdir(struct file *, filldir_t, void *);
1974
1975extern int vfs_stat(char __user *, struct kstat *);
1976extern int vfs_lstat(char __user *, struct kstat *);
1977extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1978extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1979extern int vfs_fstat(unsigned int, struct kstat *);
1980
1981extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
1982                    unsigned long arg);
1983
1984extern void get_filesystem(struct file_system_type *fs);
1985extern void put_filesystem(struct file_system_type *fs);
1986extern struct file_system_type *get_fs_type(const char *name);
1987extern struct super_block *get_super(struct block_device *);
1988extern struct super_block *user_get_super(dev_t);
1989extern void drop_super(struct super_block *sb);
1990
1991extern int dcache_dir_open(struct inode *, struct file *);
1992extern int dcache_dir_close(struct inode *, struct file *);
1993extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1994extern int dcache_readdir(struct file *, void *, filldir_t);
1995extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1996extern int simple_statfs(struct dentry *, struct kstatfs *);
1997extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1998extern int simple_unlink(struct inode *, struct dentry *);
1999extern int simple_rmdir(struct inode *, struct dentry *);
2000extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2001extern int simple_sync_file(struct file *, struct dentry *, int);
2002extern int simple_empty(struct dentry *);
2003extern int simple_readpage(struct file *file, struct page *page);
2004extern int simple_prepare_write(struct file *file, struct page *page,
2005                        unsigned offset, unsigned to);
2006extern int simple_write_begin(struct file *file, struct address_space *mapping,
2007                        loff_t pos, unsigned len, unsigned flags,
2008                        struct page **pagep, void **fsdata);
2009extern int simple_write_end(struct file *file, struct address_space *mapping,
2010                        loff_t pos, unsigned len, unsigned copied,
2011                        struct page *page, void *fsdata);
2012
2013extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2014extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2015extern const struct file_operations simple_dir_operations;
2016extern const struct inode_operations simple_dir_inode_operations;
2017struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2018struct dentry *d_alloc_name(struct dentry *, const char *);
2019extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
2020extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2021extern void simple_release_fs(struct vfsmount **mount, int *count);
2022
2023extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2024                        loff_t *ppos, const void *from, size_t available);
2025
2026#ifdef CONFIG_MIGRATION
2027extern int buffer_migrate_page(struct address_space *,
2028                                struct page *, struct page *);
2029#else
2030#define buffer_migrate_page NULL
2031#endif
2032
2033extern int inode_change_ok(struct inode *, struct iattr *);
2034extern int __must_check inode_setattr(struct inode *, struct iattr *);
2035
2036extern void file_update_time(struct file *file);
2037
2038extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2039extern void save_mount_options(struct super_block *sb, char *options);
2040
2041static inline ino_t parent_ino(struct dentry *dentry)
2042{
2043        ino_t res;
2044
2045        spin_lock(&dentry->d_lock);
2046        res = dentry->d_parent->d_inode->i_ino;
2047        spin_unlock(&dentry->d_lock);
2048        return res;
2049}
2050
2051/* Transaction based IO helpers */
2052
2053/*
2054 * An argresp is stored in an allocated page and holds the
2055 * size of the argument or response, along with its content
2056 */
2057struct simple_transaction_argresp {
2058        ssize_t size;
2059        char data[0];
2060};
2061
2062#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2063
2064char *simple_transaction_get(struct file *file, const char __user *buf,
2065                                size_t size);
2066ssize_t simple_transaction_read(struct file *file, char __user *buf,
2067                                size_t size, loff_t *pos);
2068int simple_transaction_release(struct inode *inode, struct file *file);
2069
2070static inline void simple_transaction_set(struct file *file, size_t n)
2071{
2072        struct simple_transaction_argresp *ar = file->private_data;
2073
2074        BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
2075
2076        /*
2077         * The barrier ensures that ar->size will really remain zero until
2078         * ar->data is ready for reading.
2079         */
2080        smp_mb();
2081        ar->size = n;
2082}
2083
2084/*
2085 * simple attribute files
2086 *
2087 * These attributes behave similar to those in sysfs:
2088 *
2089 * Writing to an attribute immediately sets a value, an open file can be
2090 * written to multiple times.
2091 *
2092 * Reading from an attribute creates a buffer from the value that might get
2093 * read with multiple read calls. When the attribute has been read
2094 * completely, no further read calls are possible until the file is opened
2095 * again.
2096 *
2097 * All attributes contain a text representation of a numeric value
2098 * that are accessed with the get() and set() functions.
2099 */
2100#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2101static int __fops ## _open(struct inode *inode, struct file *file)      \
2102{                                                                       \
2103        __simple_attr_check_format(__fmt, 0ull);                        \
2104        return simple_attr_open(inode, file, __get, __set, __fmt);      \
2105}                                                                       \
2106static struct file_operations __fops = {                                \
2107        .owner   = THIS_MODULE,                                         \
2108        .open    = __fops ## _open,                                     \
2109        .release = simple_attr_release,                                 \
2110        .read    = simple_attr_read,                                    \
2111        .write   = simple_attr_write,                                   \
2112};
2113
2114static inline void __attribute__((format(printf, 1, 2)))
2115__simple_attr_check_format(const char *fmt, ...)
2116{
2117        /* don't do anything, just let the compiler check the arguments; */
2118}
2119
2120int simple_attr_open(struct inode *inode, struct file *file,
2121                     int (*get)(void *, u64 *), int (*set)(void *, u64),
2122                     const char *fmt);
2123int simple_attr_release(struct inode *inode, struct file *file);
2124ssize_t simple_attr_read(struct file *file, char __user *buf,
2125                         size_t len, loff_t *ppos);
2126ssize_t simple_attr_write(struct file *file, const char __user *buf,
2127                          size_t len, loff_t *ppos);
2128
2129
2130#ifdef CONFIG_SECURITY
2131static inline char *alloc_secdata(void)
2132{
2133        return (char *)get_zeroed_page(GFP_KERNEL);
2134}
2135
2136static inline void free_secdata(void *secdata)
2137{
2138        free_page((unsigned long)secdata);
2139}
2140#else
2141static inline char *alloc_secdata(void)
2142{
2143        return (char *)1;
2144}
2145
2146static inline void free_secdata(void *secdata)
2147{ }
2148#endif  /* CONFIG_SECURITY */
2149
2150struct ctl_table;
2151int proc_nr_files(struct ctl_table *table, int write, struct file *filp,
2152                  void __user *buffer, size_t *lenp, loff_t *ppos);
2153
2154int get_filesystem_list(char * buf);
2155
2156#endif /* __KERNEL__ */
2157#endif /* _LINUX_FS_H */
2158
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.