linux-bk/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/config.h>
  10#include <linux/linkage.h>
  11#include <linux/limits.h>
  12#include <linux/wait.h>
  13#include <linux/types.h>
  14#include <linux/kdev_t.h>
  15#include <linux/ioctl.h>
  16#include <linux/dcache.h>
  17#include <linux/stat.h>
  18#include <linux/cache.h>
  19#include <linux/prio_tree.h>
  20#include <linux/kobject.h>
  21#include <asm/atomic.h>
  22
  23struct iovec;
  24struct nameidata;
  25struct pipe_inode_info;
  26struct poll_table_struct;
  27struct kstatfs;
  28struct vm_area_struct;
  29struct vfsmount;
  30
  31/*
  32 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
  33 * the file limit at runtime and only root can increase the per-process
  34 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
  35 * upper limit on files-per-process.
  36 *
  37 * Some programs (notably those using select()) may have to be 
  38 * recompiled to take full advantage of the new limits..  
  39 */
  40
  41/* Fixed constants first: */
  42#undef NR_OPEN
  43#define NR_OPEN (1024*1024)     /* Absolute upper limit on fd num */
  44#define INR_OPEN 1024           /* Initial setting for nfile rlimits */
  45
  46#define BLOCK_SIZE_BITS 10
  47#define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
  48
  49/* And dynamically-tunable limits and defaults: */
  50struct files_stat_struct {
  51        int nr_files;           /* read only */
  52        int nr_free_files;      /* read only */
  53        int max_files;          /* tunable */
  54};
  55extern struct files_stat_struct files_stat;
  56
  57struct inodes_stat_t {
  58        int nr_inodes;
  59        int nr_unused;
  60        int dummy[5];
  61};
  62extern struct inodes_stat_t inodes_stat;
  63
  64extern int leases_enable, dir_notify_enable, lease_break_time;
  65
  66#define NR_FILE  8192   /* this can well be larger on a larger system */
  67#define NR_RESERVED_FILES 10 /* reserved for root */
  68#define NR_SUPER 256
  69
  70#define MAY_EXEC 1
  71#define MAY_WRITE 2
  72#define MAY_READ 4
  73#define MAY_APPEND 8
  74
  75#define FMODE_READ 1
  76#define FMODE_WRITE 2
  77
  78/* Internal kernel extensions */
  79#define FMODE_LSEEK     4
  80#define FMODE_PREAD     8
  81#define FMODE_PWRITE    FMODE_PREAD     /* These go hand in hand */
  82
  83#define RW_MASK         1
  84#define RWA_MASK        2
  85#define READ 0
  86#define WRITE 1
  87#define READA 2         /* read-ahead  - don't block if no resources */
  88#define SPECIAL 4       /* For non-blockdevice requests in request queue */
  89#define READ_SYNC       (READ | (1 << BIO_RW_SYNC))
  90#define WRITE_SYNC      (WRITE | (1 << BIO_RW_SYNC))
  91#define WRITE_BARRIER   ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
  92
  93#define SEL_IN          1
  94#define SEL_OUT         2
  95#define SEL_EX          4
  96
  97/* public flags for file_system_type */
  98#define FS_REQUIRES_DEV 1 
  99#define FS_BINARY_MOUNTDATA 2
 100#define FS_REVAL_DOT    16384   /* Check the paths ".", ".." for staleness */
 101#define FS_ODD_RENAME   32768   /* Temporary stuff; will go away as soon
 102                                  * as nfs_rename() will be cleaned up
 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
 121#define MS_POSIXACL     (1<<16) /* VFS does not apply the umask */
 122#define MS_ONE_SECOND   (1<<17) /* fs has 1 sec a/m/ctime resolution */
 123#define MS_ACTIVE       (1<<30)
 124#define MS_NOUSER       (1<<31)
 125
 126/*
 127 * Superblock flags that can be altered by MS_REMOUNT
 128 */
 129#define MS_RMT_MASK     (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_NOATIME|\
 130                         MS_NODIRATIME)
 131
 132/*
 133 * Old magic mount flag and mask
 134 */
 135#define MS_MGC_VAL 0xC0ED0000
 136#define MS_MGC_MSK 0xffff0000
 137
 138/* Inode flags - they have nothing to superblock flags now */
 139
 140#define S_SYNC          1       /* Writes are synced at once */
 141#define S_NOATIME       2       /* Do not update access times */
 142#define S_APPEND        4       /* Append-only file */
 143#define S_IMMUTABLE     8       /* Immutable file */
 144#define S_DEAD          16      /* removed, but still open directory */
 145#define S_NOQUOTA       32      /* Inode is not counted to quota */
 146#define S_DIRSYNC       64      /* Directory modifications are synchronous */
 147#define S_NOCMTIME      128     /* Do not update file c/mtime */
 148#define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
 149
 150/*
 151 * Note that nosuid etc flags are inode-specific: setting some file-system
 152 * flags just means all the inodes inherit those flags by default. It might be
 153 * possible to override it selectively if you really wanted to with some
 154 * ioctl() that is not currently implemented.
 155 *
 156 * Exception: MS_RDONLY is always applied to the entire file system.
 157 *
 158 * Unfortunately, it is possible to change a filesystems flags with it mounted
 159 * with files in use.  This means that all of the inodes will not have their
 160 * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
 161 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
 162 */
 163#define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
 164
 165#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
 166#define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
 167                                        ((inode)->i_flags & S_SYNC))
 168#define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
 169                                        ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
 170#define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
 171
 172#define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
 173#define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
 174#define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
 175#define IS_NOATIME(inode)       (__IS_FLG(inode, MS_NOATIME) || ((inode)->i_flags & S_NOATIME))
 176#define IS_NODIRATIME(inode)    __IS_FLG(inode, MS_NODIRATIME)
 177#define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
 178#define IS_ONE_SECOND(inode)    __IS_FLG(inode, MS_ONE_SECOND)
 179
 180#define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
 181#define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
 182#define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
 183
 184/* the read-only stuff doesn't really belong here, but any other place is
 185   probably as bad and I don't want to create yet another include file. */
 186
 187#define BLKROSET   _IO(0x12,93) /* set device read-only (0 = read-write) */
 188#define BLKROGET   _IO(0x12,94) /* get read-only status (0 = read_write) */
 189#define BLKRRPART  _IO(0x12,95) /* re-read partition table */
 190#define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
 191#define BLKFLSBUF  _IO(0x12,97) /* flush buffer cache */
 192#define BLKRASET   _IO(0x12,98) /* set read ahead for block device */
 193#define BLKRAGET   _IO(0x12,99) /* get current read ahead setting */
 194#define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
 195#define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
 196#define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
 197#define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
 198#define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
 199#if 0
 200#define BLKPG      _IO(0x12,105)/* See blkpg.h */
 201
 202/* Some people are morons.  Do not use sizeof! */
 203
 204#define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
 205#define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
 206/* This was here just to show that the number is taken -
 207   probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
 208#endif
 209/* A jump here: 108-111 have been used for various private purposes. */
 210#define BLKBSZGET  _IOR(0x12,112,size_t)
 211#define BLKBSZSET  _IOW(0x12,113,size_t)
 212#define BLKGETSIZE64 _IOR(0x12,114,size_t)      /* return device size in bytes (u64 *arg) */
 213
 214#define BMAP_IOCTL 1            /* obsolete - kept for compatibility */
 215#define FIBMAP     _IO(0x00,1)  /* bmap access */
 216#define FIGETBSZ   _IO(0x00,2)  /* get the block size used for bmap */
 217
 218#ifdef __KERNEL__
 219
 220#include <linux/list.h>
 221#include <linux/radix-tree.h>
 222#include <linux/audit.h>
 223#include <linux/init.h>
 224#include <asm/semaphore.h>
 225#include <asm/byteorder.h>
 226
 227/* Used to be a macro which just called the function, now just a function */
 228extern void update_atime (struct inode *);
 229
 230extern void __init inode_init(unsigned long);
 231extern void __init inode_init_early(void);
 232extern void __init mnt_init(unsigned long);
 233extern void __init files_init(unsigned long);
 234
 235struct buffer_head;
 236typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 237                        struct buffer_head *bh_result, int create);
 238typedef int (get_blocks_t)(struct inode *inode, sector_t iblock,
 239                        unsigned long max_blocks,
 240                        struct buffer_head *bh_result, int create);
 241typedef void (dio_iodone_t)(struct inode *inode, loff_t offset,
 242                        ssize_t bytes, void *private);
 243
 244/*
 245 * Attribute flags.  These should be or-ed together to figure out what
 246 * has been changed!
 247 */
 248#define ATTR_MODE       1
 249#define ATTR_UID        2
 250#define ATTR_GID        4
 251#define ATTR_SIZE       8
 252#define ATTR_ATIME      16
 253#define ATTR_MTIME      32
 254#define ATTR_CTIME      64
 255#define ATTR_ATIME_SET  128
 256#define ATTR_MTIME_SET  256
 257#define ATTR_FORCE      512     /* Not a change, but a change it */
 258#define ATTR_ATTR_FLAG  1024
 259#define ATTR_KILL_SUID  2048
 260#define ATTR_KILL_SGID  4096
 261
 262/*
 263 * This is the Inode Attributes structure, used for notify_change().  It
 264 * uses the above definitions as flags, to know which values have changed.
 265 * Also, in this manner, a Filesystem can look at only the values it cares
 266 * about.  Basically, these are the attributes that the VFS layer can
 267 * request to change from the FS layer.
 268 *
 269 * Derek Atkins <warlord@MIT.EDU> 94-10-20
 270 */
 271struct iattr {
 272        unsigned int    ia_valid;
 273        umode_t         ia_mode;
 274        uid_t           ia_uid;
 275        gid_t           ia_gid;
 276        loff_t          ia_size;
 277        struct timespec ia_atime;
 278        struct timespec ia_mtime;
 279        struct timespec ia_ctime;
 280        unsigned int    ia_attr_flags;
 281};
 282
 283/*
 284 * This is the inode attributes flag definitions
 285 */
 286#define ATTR_FLAG_SYNCRONOUS    1       /* Syncronous write */
 287#define ATTR_FLAG_NOATIME       2       /* Don't update atime */
 288#define ATTR_FLAG_APPEND        4       /* Append-only file */
 289#define ATTR_FLAG_IMMUTABLE     8       /* Immutable file */
 290#define ATTR_FLAG_NODIRATIME    16      /* Don't update atime for directory */
 291
 292/*
 293 * Includes for diskquotas.
 294 */
 295#include <linux/quota.h>
 296
 297/*
 298 * oh the beauties of C type declarations.
 299 */
 300struct page;
 301struct address_space;
 302struct writeback_control;
 303struct kiocb;
 304
 305struct address_space_operations {
 306        int (*writepage)(struct page *page, struct writeback_control *wbc);
 307        int (*readpage)(struct file *, struct page *);
 308        int (*sync_page)(struct page *);
 309
 310        /* Write back some dirty pages from this mapping. */
 311        int (*writepages)(struct address_space *, struct writeback_control *);
 312
 313        /* Set a page dirty */
 314        int (*set_page_dirty)(struct page *page);
 315
 316        int (*readpages)(struct file *filp, struct address_space *mapping,
 317                        struct list_head *pages, unsigned nr_pages);
 318
 319        /*
 320         * ext3 requires that a successful prepare_write() call be followed
 321         * by a commit_write() call - they must be balanced
 322         */
 323        int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
 324        int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
 325        /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
 326        sector_t (*bmap)(struct address_space *, sector_t);
 327        int (*invalidatepage) (struct page *, unsigned long);
 328        int (*releasepage) (struct page *, int);
 329        ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
 330                        loff_t offset, unsigned long nr_segs);
 331};
 332
 333struct backing_dev_info;
 334struct address_space {
 335        struct inode            *host;          /* owner: inode, block_device */
 336        struct radix_tree_root  page_tree;      /* radix tree of all pages */
 337        spinlock_t              tree_lock;      /* and spinlock protecting it */
 338        unsigned int            i_mmap_writable;/* count VM_SHARED mappings */
 339        struct prio_tree_root   i_mmap;         /* tree of private and shared mappings */
 340        struct list_head        i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
 341        spinlock_t              i_mmap_lock;    /* protect tree, count, list */
 342        atomic_t                truncate_count; /* Cover race condition with truncate */
 343        unsigned long           nrpages;        /* number of total pages */
 344        pgoff_t                 writeback_index;/* writeback starts here */
 345        struct address_space_operations *a_ops; /* methods */
 346        unsigned long           flags;          /* error bits/gfp mask */
 347        struct backing_dev_info *backing_dev_info; /* device readahead, etc */
 348        spinlock_t              private_lock;   /* for use by the address_space */
 349        struct list_head        private_list;   /* ditto */
 350        struct address_space    *assoc_mapping; /* ditto */
 351};
 352
 353struct block_device {
 354        dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
 355        struct inode *          bd_inode;       /* will die */
 356        int                     bd_openers;
 357        struct semaphore        bd_sem; /* open/close mutex */
 358        struct semaphore        bd_mount_sem;   /* mount mutex */
 359        struct list_head        bd_inodes;
 360        void *                  bd_holder;
 361        int                     bd_holders;
 362        struct block_device *   bd_contains;
 363        unsigned                bd_block_size;
 364        struct hd_struct *      bd_part;
 365        /* number of times partitions within this device have been opened. */
 366        unsigned                bd_part_count;
 367        int                     bd_invalidated;
 368        struct gendisk *        bd_disk;
 369        struct list_head        bd_list;
 370        struct backing_dev_info *bd_inode_backing_dev_info;
 371        /*
 372         * Private data.  You must have bd_claim'ed the block_device
 373         * to use this.  NOTE:  bd_claim allows an owner to claim
 374         * the same device multiple times, the owner must take special
 375         * care to not mess up bd_private for that case.
 376         */
 377        unsigned long           bd_private;
 378};
 379
 380/*
 381 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
 382 * radix trees
 383 */
 384#define PAGECACHE_TAG_DIRTY     0
 385#define PAGECACHE_TAG_WRITEBACK 1
 386
 387int mapping_tagged(struct address_space *mapping, int tag);
 388
 389/*
 390 * Might pages of this file be mapped into userspace?
 391 */
 392static inline int mapping_mapped(struct address_space *mapping)
 393{
 394        return  !prio_tree_empty(&mapping->i_mmap) ||
 395                !list_empty(&mapping->i_mmap_nonlinear);
 396}
 397
 398/*
 399 * Might pages of this file have been modified in userspace?
 400 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
 401 * marks vma as VM_SHARED if it is shared, and the file was opened for
 402 * writing i.e. vma may be mprotected writable even if now readonly.
 403 */
 404static inline int mapping_writably_mapped(struct address_space *mapping)
 405{
 406        return mapping->i_mmap_writable != 0;
 407}
 408
 409/*
 410 * Use sequence counter to get consistent i_size on 32-bit processors.
 411 */
 412#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 413#include <linux/seqlock.h>
 414#define __NEED_I_SIZE_ORDERED
 415#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
 416#else
 417#define i_size_ordered_init(inode) do { } while (0)
 418#endif
 419
 420struct inode {
 421        struct hlist_node       i_hash;
 422        struct list_head        i_list;
 423        struct list_head        i_dentry;
 424        unsigned long           i_ino;
 425        atomic_t                i_count;
 426        umode_t                 i_mode;
 427        unsigned int            i_nlink;
 428        uid_t                   i_uid;
 429        gid_t                   i_gid;
 430        dev_t                   i_rdev;
 431        loff_t                  i_size;
 432        struct timespec         i_atime;
 433        struct timespec         i_mtime;
 434        struct timespec         i_ctime;
 435        unsigned int            i_blkbits;
 436        unsigned long           i_blksize;
 437        unsigned long           i_version;
 438        unsigned long           i_blocks;
 439        unsigned short          i_bytes;
 440        unsigned char           i_sock;
 441        spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
 442        struct semaphore        i_sem;
 443        struct rw_semaphore     i_alloc_sem;
 444        struct inode_operations *i_op;
 445        struct file_operations  *i_fop; /* former ->i_op->default_file_ops */
 446        struct super_block      *i_sb;
 447        struct file_lock        *i_flock;
 448        struct address_space    *i_mapping;
 449        struct address_space    i_data;
 450#ifdef CONFIG_QUOTA
 451        struct dquot            *i_dquot[MAXQUOTAS];
 452#endif
 453        /* These three should probably be a union */
 454        struct list_head        i_devices;
 455        struct pipe_inode_info  *i_pipe;
 456        struct block_device     *i_bdev;
 457        struct cdev             *i_cdev;
 458        int                     i_cindex;
 459
 460        __u32                   i_generation;
 461
 462        unsigned long           i_dnotify_mask; /* Directory notify events */
 463        struct dnotify_struct   *i_dnotify; /* for directory notifications */
 464
 465        unsigned long           i_state;
 466        unsigned long           dirtied_when;   /* jiffies of first dirtying */
 467
 468        unsigned int            i_flags;
 469
 470        atomic_t                i_writecount;
 471        void                    *i_security;
 472        union {
 473                void            *generic_ip;
 474        } u;
 475#ifdef __NEED_I_SIZE_ORDERED
 476        seqcount_t              i_size_seqcount;
 477#endif
 478};
 479
 480/*
 481 * NOTE: in a 32bit arch with a preemptable kernel and
 482 * an UP compile the i_size_read/write must be atomic
 483 * with respect to the local cpu (unlike with preempt disabled),
 484 * but they don't need to be atomic with respect to other cpus like in
 485 * true SMP (so they need either to either locally disable irq around
 486 * the read or for example on x86 they can be still implemented as a
 487 * cmpxchg8b without the need of the lock prefix). For SMP compiles
 488 * and 64bit archs it makes no difference if preempt is enabled or not.
 489 */
 490static inline loff_t i_size_read(struct inode *inode)
 491{
 492#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 493        loff_t i_size;
 494        unsigned int seq;
 495
 496        do {
 497                seq = read_seqcount_begin(&inode->i_size_seqcount);
 498                i_size = inode->i_size;
 499        } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
 500        return i_size;
 501#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 502        loff_t i_size;
 503
 504        preempt_disable();
 505        i_size = inode->i_size;
 506        preempt_enable();
 507        return i_size;
 508#else
 509        return inode->i_size;
 510#endif
 511}
 512
 513
 514static inline void i_size_write(struct inode *inode, loff_t i_size)
 515{
 516#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 517        write_seqcount_begin(&inode->i_size_seqcount);
 518        inode->i_size = i_size;
 519        write_seqcount_end(&inode->i_size_seqcount);
 520#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 521        preempt_disable();
 522        inode->i_size = i_size;
 523        preempt_enable();
 524#else
 525        inode->i_size = i_size;
 526#endif
 527}
 528
 529static inline unsigned iminor(struct inode *inode)
 530{
 531        return MINOR(inode->i_rdev);
 532}
 533
 534static inline unsigned imajor(struct inode *inode)
 535{
 536        return MAJOR(inode->i_rdev);
 537}
 538
 539extern struct block_device *I_BDEV(struct inode *inode);
 540
 541struct fown_struct {
 542        rwlock_t lock;          /* protects pid, uid, euid fields */
 543        int pid;                /* pid or -pgrp where SIGIO should be sent */
 544        uid_t uid, euid;        /* uid/euid of process setting the owner */
 545        void *security;
 546        int signum;             /* posix.1b rt signal to be delivered on IO */
 547};
 548
 549/*
 550 * Track a single file's readahead state
 551 */
 552struct file_ra_state {
 553        unsigned long start;            /* Current window */
 554        unsigned long size;
 555        unsigned long next_size;        /* Next window size */
 556        unsigned long prev_page;        /* Cache last read() position */
 557        unsigned long ahead_start;      /* Ahead window */
 558        unsigned long ahead_size;
 559        unsigned long currnt_wnd_hit;   /* locality in the current window */
 560        unsigned long average;          /* size of next current window */
 561        unsigned long ra_pages;         /* Maximum readahead window */
 562        unsigned long mmap_hit;         /* Cache hit stat for mmap accesses */
 563        unsigned long mmap_miss;        /* Cache miss stat for mmap accesses */
 564};
 565
 566struct file {
 567        struct list_head        f_list;
 568        struct dentry           *f_dentry;
 569        struct vfsmount         *f_vfsmnt;
 570        struct file_operations  *f_op;
 571        atomic_t                f_count;
 572        unsigned int            f_flags;
 573        mode_t                  f_mode;
 574        int                     f_error;
 575        loff_t                  f_pos;
 576        struct fown_struct      f_owner;
 577        unsigned int            f_uid, f_gid;
 578        struct file_ra_state    f_ra;
 579
 580        unsigned long           f_version;
 581        void                    *f_security;
 582
 583        /* needed for tty driver, and maybe others */
 584        void                    *private_data;
 585
 586#ifdef CONFIG_EPOLL
 587        /* Used by fs/eventpoll.c to link all the hooks to this file */
 588        struct list_head        f_ep_links;
 589        spinlock_t              f_ep_lock;
 590#endif /* #ifdef CONFIG_EPOLL */
 591        struct address_space    *f_mapping;
 592};
 593extern spinlock_t files_lock;
 594#define file_list_lock() spin_lock(&files_lock);
 595#define file_list_unlock() spin_unlock(&files_lock);
 596
 597#define get_file(x)     atomic_inc(&(x)->f_count)
 598#define file_count(x)   atomic_read(&(x)->f_count)
 599
 600#define MAX_NON_LFS     ((1UL<<31) - 1)
 601
 602/* Page cache limit. The filesystems should put that into their s_maxbytes 
 603   limits, otherwise bad things can happen in VM. */ 
 604#if BITS_PER_LONG==32
 605#define MAX_LFS_FILESIZE        (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
 606#elif BITS_PER_LONG==64
 607#define MAX_LFS_FILESIZE        0x7fffffffffffffffUL
 608#endif
 609
 610#define FL_POSIX        1
 611#define FL_FLOCK        2
 612#define FL_ACCESS       8       /* not trying to lock, just looking */
 613#define FL_LOCKD        16      /* lock held by rpc.lockd */
 614#define FL_LEASE        32      /* lease held on this file */
 615#define FL_SLEEP        128     /* A blocking lock */
 616
 617/*
 618 * The POSIX file lock owner is determined by
 619 * the "struct files_struct" in the thread group
 620 * (or NULL for no owner - BSD locks).
 621 *
 622 * Lockd stuffs a "host" pointer into this.
 623 */
 624typedef struct files_struct *fl_owner_t;
 625
 626struct file_lock_operations {
 627        void (*fl_insert)(struct file_lock *);  /* lock insertion callback */
 628        void (*fl_remove)(struct file_lock *);  /* lock removal callback */
 629        void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
 630        void (*fl_release_private)(struct file_lock *);
 631};
 632
 633struct lock_manager_operations {
 634        int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
 635        void (*fl_notify)(struct file_lock *);  /* unblock callback */
 636};
 637
 638/* that will die - we need it for nfs_lock_info */
 639#include <linux/nfs_fs_i.h>
 640
 641struct file_lock {
 642        struct file_lock *fl_next;      /* singly linked list for this inode  */
 643        struct list_head fl_link;       /* doubly linked list of all locks */
 644        struct list_head fl_block;      /* circular list of blocked processes */
 645        fl_owner_t fl_owner;
 646        unsigned int fl_pid;
 647        wait_queue_head_t fl_wait;
 648        struct file *fl_file;
 649        unsigned char fl_flags;
 650        unsigned char fl_type;
 651        loff_t fl_start;
 652        loff_t fl_end;
 653
 654        struct fasync_struct *  fl_fasync; /* for lease break notifications */
 655        unsigned long fl_break_time;    /* for nonblocking lease breaks */
 656
 657        struct file_lock_operations *fl_ops;    /* Callbacks for filesystems */
 658        struct lock_manager_operations *fl_lmops;       /* Callbacks for lockmanagers */
 659        union {
 660                struct nfs_lock_info    nfs_fl;
 661        } fl_u;
 662};
 663
 664/* The following constant reflects the upper bound of the file/locking space */
 665#ifndef OFFSET_MAX
 666#define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
 667#define OFFSET_MAX      INT_LIMIT(loff_t)
 668#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
 669#endif
 670
 671extern struct list_head file_lock_list;
 672
 673#include <linux/fcntl.h>
 674
 675extern int fcntl_getlk(struct file *, struct flock __user *);
 676extern int fcntl_setlk(struct file *, unsigned int, struct flock __user *);
 677
 678#if BITS_PER_LONG == 32
 679extern int fcntl_getlk64(struct file *, struct flock64 __user *);
 680extern int fcntl_setlk64(struct file *, unsigned int, struct flock64 __user *);
 681#endif
 682
 683extern void send_sigio(struct fown_struct *fown, int fd, int band);
 684extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
 685extern int fcntl_getlease(struct file *filp);
 686
 687/* fs/locks.c */
 688extern void locks_init_lock(struct file_lock *);
 689extern void locks_copy_lock(struct file_lock *, struct file_lock *);
 690extern void locks_remove_posix(struct file *, fl_owner_t);
 691extern void locks_remove_flock(struct file *);
 692extern struct file_lock *posix_test_lock(struct file *, struct file_lock *);
 693extern int posix_lock_file(struct file *, struct file_lock *);
 694extern int posix_lock_file_wait(struct file *, struct file_lock *);
 695extern void posix_block_lock(struct file_lock *, struct file_lock *);
 696extern void posix_unblock_lock(struct file *, struct file_lock *);
 697extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
 698extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
 699extern int __break_lease(struct inode *inode, unsigned int flags);
 700extern void lease_get_mtime(struct inode *, struct timespec *time);
 701extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
 702extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
 703extern void steal_locks(fl_owner_t from);
 704
 705struct fasync_struct {
 706        int     magic;
 707        int     fa_fd;
 708        struct  fasync_struct   *fa_next; /* singly linked list */
 709        struct  file            *fa_file;
 710};
 711
 712#define FASYNC_MAGIC 0x4601
 713
 714/* SMP safe fasync helpers: */
 715extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
 716/* can be called from interrupts */
 717extern void kill_fasync(struct fasync_struct **, int, int);
 718/* only for net: no internal synchronization */
 719extern void __kill_fasync(struct fasync_struct *, int, int);
 720
 721extern int f_setown(struct file *filp, unsigned long arg, int force);
 722extern void f_delown(struct file *filp);
 723extern int send_sigurg(struct fown_struct *fown);
 724
 725/*
 726 *      Umount options
 727 */
 728
 729#define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
 730#define MNT_DETACH      0x00000002      /* Just detach from the tree */
 731#define MNT_EXPIRE      0x00000004      /* Mark for expiry */
 732
 733extern struct list_head super_blocks;
 734extern spinlock_t sb_lock;
 735
 736#define sb_entry(list)  list_entry((list), struct super_block, s_list)
 737#define S_BIAS (1<<30)
 738struct super_block {
 739        struct list_head        s_list;         /* Keep this first */
 740        dev_t                   s_dev;          /* search index; _not_ kdev_t */
 741        unsigned long           s_blocksize;
 742        unsigned long           s_old_blocksize;
 743        unsigned char           s_blocksize_bits;
 744        unsigned char           s_dirt;
 745        unsigned long long      s_maxbytes;     /* Max file size */
 746        struct file_system_type *s_type;
 747        struct super_operations *s_op;
 748        struct dquot_operations *dq_op;
 749        struct quotactl_ops     *s_qcop;
 750        struct export_operations *s_export_op;
 751        unsigned long           s_flags;
 752        unsigned long           s_magic;
 753        struct dentry           *s_root;
 754        struct rw_semaphore     s_umount;
 755        struct semaphore        s_lock;
 756        int                     s_count;
 757        int                     s_syncing;
 758        int                     s_need_sync_fs;
 759        atomic_t                s_active;
 760        void                    *s_security;
 761
 762        struct list_head        s_dirty;        /* dirty inodes */
 763        struct list_head        s_io;           /* parked for writeback */
 764        struct hlist_head       s_anon;         /* anonymous dentries for (nfs) exporting */
 765        struct list_head        s_files;
 766
 767        struct block_device     *s_bdev;
 768        struct list_head        s_instances;
 769        struct quota_info       s_dquot;        /* Diskquota specific options */
 770
 771        int                     s_frozen;
 772        wait_queue_head_t       s_wait_unfrozen;
 773
 774        char s_id[32];                          /* Informational name */
 775
 776        void                    *s_fs_info;     /* Filesystem private info */
 777
 778        /*
 779         * The next field is for VFS *only*. No filesystems have any business
 780         * even looking at it. You had been warned.
 781         */
 782        struct semaphore s_vfs_rename_sem;      /* Kludge */
 783};
 784
 785/*
 786 * Snapshotting support.
 787 */
 788enum {
 789        SB_UNFROZEN = 0,
 790        SB_FREEZE_WRITE = 1,
 791        SB_FREEZE_TRANS = 2,
 792};
 793
 794#define vfs_check_frozen(sb, level) \
 795        wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
 796
 797/*
 798 * Superblock locking.
 799 */
 800static inline void lock_super(struct super_block * sb)
 801{
 802        down(&sb->s_lock);
 803}
 804
 805static inline void unlock_super(struct super_block * sb)
 806{
 807        up(&sb->s_lock);
 808}
 809
 810/*
 811 * VFS helper functions..
 812 */
 813extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
 814extern int vfs_mkdir(struct inode *, struct dentry *, int);
 815extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
 816extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
 817extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
 818extern int vfs_rmdir(struct inode *, struct dentry *);
 819extern int vfs_unlink(struct inode *, struct dentry *);
 820extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
 821
 822/*
 823 * VFS dentry helper functions.
 824 */
 825extern void dentry_unhash(struct dentry *dentry);
 826
 827/*
 828 * File types
 829 *
 830 * NOTE! These match bits 12..15 of stat.st_mode
 831 * (ie "(i_mode >> 12) & 15").
 832 */
 833#define DT_UNKNOWN      0
 834#define DT_FIFO         1
 835#define DT_CHR          2
 836#define DT_DIR          4
 837#define DT_BLK          6
 838#define DT_REG          8
 839#define DT_LNK          10
 840#define DT_SOCK         12
 841#define DT_WHT          14
 842
 843#define OSYNC_METADATA  (1<<0)
 844#define OSYNC_DATA      (1<<1)
 845#define OSYNC_INODE     (1<<2)
 846int generic_osync_inode(struct inode *, struct address_space *, int);
 847
 848/*
 849 * This is the "filldir" function type, used by readdir() to let
 850 * the kernel specify what kind of dirent layout it wants to have.
 851 * This allows the kernel to read directories into kernel space or
 852 * to have different dirent layouts depending on the binary type.
 853 */
 854typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
 855
 856struct block_device_operations {
 857        int (*open) (struct inode *, struct file *);
 858        int (*release) (struct inode *, struct file *);
 859        int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
 860        int (*media_changed) (struct gendisk *);
 861        int (*revalidate_disk) (struct gendisk *);
 862        struct module *owner;
 863};
 864
 865/*
 866 * "descriptor" for what we're up to with a read for sendfile().
 867 * This allows us to use the same read code yet
 868 * have multiple different users of the data that
 869 * we read from a file.
 870 *
 871 * The simplest case just copies the data to user
 872 * mode.
 873 */
 874typedef struct {
 875        size_t written;
 876        size_t count;
 877        union {
 878                char __user * buf;
 879                void *data;
 880        } arg;
 881        int error;
 882} read_descriptor_t;
 883
 884typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
 885
 886/*
 887 * NOTE:
 888 * read, write, poll, fsync, readv, writev can be called
 889 *   without the big kernel lock held in all filesystems.
 890 */
 891struct file_operations {
 892        struct module *owner;
 893        loff_t (*llseek) (struct file *, loff_t, int);
 894        ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
 895        ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
 896        ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
 897        ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
 898        int (*readdir) (struct file *, void *, filldir_t);
 899        unsigned int (*poll) (struct file *, struct poll_table_struct *);
 900        int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
 901        int (*mmap) (struct file *, struct vm_area_struct *);
 902        int (*open) (struct inode *, struct file *);
 903        int (*flush) (struct file *);
 904        int (*release) (struct inode *, struct file *);
 905        int (*fsync) (struct file *, struct dentry *, int datasync);
 906        int (*aio_fsync) (struct kiocb *, int datasync);
 907        int (*fasync) (int, struct file *, int);
 908        int (*lock) (struct file *, int, struct file_lock *);
 909        ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
 910        ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
 911        ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
 912        ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
 913        unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
 914        int (*check_flags)(int);
 915        int (*dir_notify)(struct file *filp, unsigned long arg);
 916        int (*flock) (struct file *, int, struct file_lock *);
 917};
 918
 919struct inode_operations {
 920        int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
 921        struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
 922        int (*link) (struct dentry *,struct inode *,struct dentry *);
 923        int (*unlink) (struct inode *,struct dentry *);
 924        int (*symlink) (struct inode *,struct dentry *,const char *);
 925        int (*mkdir) (struct inode *,struct dentry *,int);
 926        int (*rmdir) (struct inode *,struct dentry *);
 927        int (*mknod) (struct inode *,struct dentry *,int,dev_t);
 928        int (*rename) (struct inode *, struct dentry *,
 929                        struct inode *, struct dentry *);
 930        int (*readlink) (struct dentry *, char __user *,int);
 931        int (*follow_link) (struct dentry *, struct nameidata *);
 932        void (*put_link) (struct dentry *, struct nameidata *);
 933        void (*truncate) (struct inode *);
 934        int (*permission) (struct inode *, int, struct nameidata *);
 935        int (*setattr) (struct dentry *, struct iattr *);
 936        int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
 937        int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
 938        ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
 939        ssize_t (*listxattr) (struct dentry *, char *, size_t);
 940        int (*removexattr) (struct dentry *, const char *);
 941};
 942
 943struct seq_file;
 944
 945extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
 946extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
 947extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
 948                unsigned long, loff_t *);
 949extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
 950                unsigned long, loff_t *);
 951
 952/*
 953 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
 954 * without the big kernel lock held in all filesystems.
 955 */
 956struct super_operations {
 957        struct inode *(*alloc_inode)(struct super_block *sb);
 958        void (*destroy_inode)(struct inode *);
 959
 960        void (*read_inode) (struct inode *);
 961  
 962        void (*dirty_inode) (struct inode *);
 963        int (*write_inode) (struct inode *, int);
 964        void (*put_inode) (struct inode *);
 965        void (*drop_inode) (struct inode *);
 966        void (*delete_inode) (struct inode *);
 967        void (*put_super) (struct super_block *);
 968        void (*write_super) (struct super_block *);
 969        int (*sync_fs)(struct super_block *sb, int wait);
 970        void (*write_super_lockfs) (struct super_block *);
 971        void (*unlockfs) (struct super_block *);
 972        int (*statfs) (struct super_block *, struct kstatfs *);
 973        int (*remount_fs) (struct super_block *, int *, char *);
 974        void (*clear_inode) (struct inode *);
 975        void (*umount_begin) (struct super_block *);
 976
 977        int (*show_options)(struct seq_file *, struct vfsmount *);
 978};
 979
 980/* Inode state bits.  Protected by inode_lock. */
 981#define I_DIRTY_SYNC            1 /* Not dirty enough for O_DATASYNC */
 982#define I_DIRTY_DATASYNC        2 /* Data-related inode changes pending */
 983#define I_DIRTY_PAGES           4 /* Data-related inode changes pending */
 984#define I_LOCK                  8
 985#define I_FREEING               16
 986#define I_CLEAR                 32
 987#define I_NEW                   64
 988
 989#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
 990
 991extern void __mark_inode_dirty(struct inode *, int);
 992static inline void mark_inode_dirty(struct inode *inode)
 993{
 994        __mark_inode_dirty(inode, I_DIRTY);
 995}
 996
 997static inline void mark_inode_dirty_sync(struct inode *inode)
 998{
 999        __mark_inode_dirty(inode, I_DIRTY_SYNC);
1000}
1001
1002static inline void touch_atime(struct vfsmount *mnt, struct dentry *dentry)
1003{
1004        /* per-mountpoint checks will go here */
1005        update_atime(dentry->d_inode);
1006}
1007
1008static inline void file_accessed(struct file *file)
1009{
1010        if (!(file->f_flags & O_NOATIME))
1011                touch_atime(file->f_vfsmnt, file->f_dentry);
1012}
1013
1014int sync_inode(struct inode *inode, struct writeback_control *wbc);
1015
1016/**
1017 * &export_operations - for nfsd to communicate with file systems
1018 * decode_fh:      decode a file handle fragment and return a &struct dentry
1019 * encode_fh:      encode a file handle fragment from a dentry
1020 * get_name:       find the name for a given inode in a given directory
1021 * get_parent:     find the parent of a given directory
1022 * get_dentry:     find a dentry for the inode given a file handle sub-fragment
1023 *
1024 * Description:
1025 *    The export_operations structure provides a means for nfsd to communicate
1026 *    with a particular exported file system  - particularly enabling nfsd and
1027 *    the filesystem to co-operate when dealing with file handles.
1028 *
1029 *    export_operations contains two basic operation for dealing with file handles,
1030 *    decode_fh() and encode_fh(), and allows for some other operations to be defined
1031 *    which standard helper routines use to get specific information from the
1032 *    filesystem.
1033 *
1034 *    nfsd encodes information use to determine which filesystem a filehandle
1035 *    applies to in the initial part of the file handle.  The remainder, termed a
1036 *    file handle fragment, is controlled completely by the filesystem.
1037 *    The standard helper routines assume that this fragment will contain one or two
1038 *    sub-fragments, one which identifies the file, and one which may be used to
1039 *    identify the (a) directory containing the file.
1040 *
1041 *    In some situations, nfsd needs to get a dentry which is connected into a
1042 *    specific part of the file tree.  To allow for this, it passes the function
1043 *    acceptable() together with a @context which can be used to see if the dentry
1044 *    is acceptable.  As there can be multiple dentrys for a given file, the filesystem
1045 *    should check each one for acceptability before looking for the next.  As soon
1046 *    as an acceptable one is found, it should be returned.
1047 *
1048 * decode_fh:
1049 *    @decode_fh is given a &struct super_block (@sb), a file handle fragment (@fh, @fh_len)
1050 *    and an acceptability testing function (@acceptable, @context).  It should return
1051 *    a &struct dentry which refers to the same file that the file handle fragment refers
1052 *    to,  and which passes the acceptability test.  If it cannot, it should return
1053 *    a %NULL pointer if the file was found but no acceptable &dentries were available, or
1054 *    a %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or %ENOMEM).
1055 *
1056 * encode_fh:
1057 *    @encode_fh should store in the file handle fragment @fh (using at most @max_len bytes)
1058 *    information that can be used by @decode_fh to recover the file refered to by the
1059 *    &struct dentry @de.  If the @connectable flag is set, the encode_fh() should store
1060 *    sufficient information so that a good attempt can be made to find not only
1061 *    the file but also it's place in the filesystem.   This typically means storing
1062 *    a reference to de->d_parent in the filehandle fragment.
1063 *    encode_fh() should return the number of bytes stored or a negative error code
1064 *    such as %-ENOSPC
1065 *
1066 * get_name:
1067 *    @get_name should find a name for the given @child in the given @parent directory.
1068 *    The name should be stored in the @name (with the understanding that it is already
1069 *    pointing to a a %NAME_MAX+1 sized buffer.   get_name() should return %0 on success,
1070 *    a negative error code or error.
1071 *    @get_name will be called without @parent->i_sem held.
1072 *
1073 * get_parent:
1074 *    @get_parent should find the parent directory for the given @child which is also
1075 *    a directory.  In the event that it cannot be found, or storage space cannot be
1076 *    allocated, a %ERR_PTR should be returned.
1077 *
1078 * get_dentry:
1079 *    Given a &super_block (@sb) and a pointer to a file-system specific inode identifier,
1080 *    possibly an inode number, (@inump) get_dentry() should find the identified inode and
1081 *    return a dentry for that inode.
1082 *    Any suitable dentry can be returned including, if necessary, a new dentry created
1083 *    with d_alloc_root.  The caller can then find any other extant dentrys by following the
1084 *    d_alias links.  If a new dentry was created using d_alloc_root, DCACHE_NFSD_DISCONNECTED
1085 *    should be set, and the dentry should be d_rehash()ed.
1086 *
1087 *    If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code can be returned.
1088 *    The @inump will be whatever was passed to nfsd_find_fh_dentry() in either the
1089 *    @obj or @parent parameters.
1090 *
1091 * Locking rules:
1092 *  get_parent is called with child->d_inode->i_sem down
1093 *  get_name is not (which is possibly inconsistent)
1094 */
1095
1096struct export_operations {
1097        struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1098                         int (*acceptable)(void *context, struct dentry *de),
1099                         void *context);
1100        int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1101                         int connectable);
1102
1103        /* the following are only called from the filesystem itself */
1104        int (*get_name)(struct dentry *parent, char *name,
1105                        struct dentry *child);
1106        struct dentry * (*get_parent)(struct dentry *child);
1107        struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1108
1109        /* This is set by the exporting module to a standard helper */
1110        struct dentry * (*find_exported_dentry)(
1111                struct super_block *sb, void *obj, void *parent,
1112                int (*acceptable)(void *context, struct dentry *de),
1113                void *context);
1114
1115
1116};
1117
1118
1119struct file_system_type {
1120        const char *name;
1121        int fs_flags;
1122        struct super_block *(*get_sb) (struct file_system_type *, int,
1123                                       const char *, void *);
1124        void (*kill_sb) (struct super_block *);
1125        struct module *owner;
1126        struct file_system_type * next;
1127        struct list_head fs_supers;
1128};
1129
1130struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1131        int flags, const char *dev_name, void *data,
1132        int (*fill_super)(struct super_block *, void *, int));
1133struct super_block *get_sb_single(struct file_system_type *fs_type,
1134        int flags, void *data,
1135        int (*fill_super)(struct super_block *, void *, int));
1136struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1137        int flags, void *data,
1138        int (*fill_super)(struct super_block *, void *, int));
1139void generic_shutdown_super(struct super_block *sb);
1140void kill_block_super(struct super_block *sb);
1141void kill_anon_super(struct super_block *sb);
1142void kill_litter_super(struct super_block *sb);
1143void deactivate_super(struct super_block *sb);
1144int set_anon_super(struct super_block *s, void *data);
1145struct super_block *sget(struct file_system_type *type,
1146                        int (*test)(struct super_block *,void *),
1147                        int (*set)(struct super_block *,void *),
1148                        void *data);
1149struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1150                        struct super_operations *ops, unsigned long);
1151int __put_super(struct super_block *sb);
1152int __put_super_and_need_restart(struct super_block *sb);
1153void unnamed_dev_init(void);
1154
1155/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1156#define fops_get(fops) \
1157        (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1158#define fops_put(fops) \
1159        do { if (fops) module_put((fops)->owner); } while(0)
1160
1161extern int register_filesystem(struct file_system_type *);
1162extern int unregister_filesystem(struct file_system_type *);
1163extern struct vfsmount *kern_mount(struct file_system_type *);
1164extern int may_umount_tree(struct vfsmount *);
1165extern int may_umount(struct vfsmount *);
1166extern long do_mount(char *, char *, char *, unsigned long, void *);
1167
1168extern int vfs_statfs(struct super_block *, struct kstatfs *);
1169
1170/* Return value for VFS lock functions - tells locks.c to lock conventionally
1171 * REALLY kosha for root NFS and nfs_lock
1172 */ 
1173#define LOCK_USE_CLNT 1
1174
1175#define FLOCK_VERIFY_READ  1
1176#define FLOCK_VERIFY_WRITE 2
1177
1178extern int locks_mandatory_locked(struct inode *);
1179extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1180
1181/*
1182 * Candidates for mandatory locking have the setgid bit set
1183 * but no group execute bit -  an otherwise meaningless combination.
1184 */
1185#define MANDATORY_LOCK(inode) \
1186        (IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1187
1188static inline int locks_verify_locked(struct inode *inode)
1189{
1190        if (MANDATORY_LOCK(inode))
1191                return locks_mandatory_locked(inode);
1192        return 0;
1193}
1194
1195static inline int locks_verify_area(int read_write, struct inode *inode,
1196                                    struct file *filp, loff_t offset,
1197                                    size_t count)
1198{
1199        if (inode->i_flock && MANDATORY_LOCK(inode))
1200                return locks_mandatory_area(read_write, inode, filp, offset, count);
1201        return 0;
1202}
1203
1204static inline int locks_verify_truncate(struct inode *inode,
1205                                    struct file *filp,
1206                                    loff_t size)
1207{
1208        if (inode->i_flock && MANDATORY_LOCK(inode))
1209                return locks_mandatory_area(
1210                        FLOCK_VERIFY_WRITE, inode, filp,
1211                        size < inode->i_size ? size : inode->i_size,
1212                        (size < inode->i_size ? inode->i_size - size
1213                         : size - inode->i_size)
1214                );
1215        return 0;
1216}
1217
1218static inline int break_lease(struct inode *inode, unsigned int mode)
1219{
1220        if (inode->i_flock)
1221                return __break_lease(inode, mode);
1222        return 0;
1223}
1224
1225/* fs/open.c */
1226
1227extern int do_truncate(struct dentry *, loff_t start);
1228extern struct file *filp_open(const char *, int, int);
1229extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1230extern int filp_close(struct file *, fl_owner_t id);
1231extern char * getname(const char __user *);
1232
1233/* fs/dcache.c */
1234extern void __init vfs_caches_init_early(void);
1235extern void __init vfs_caches_init(unsigned long);
1236
1237#define __getname()     kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1238#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1239#ifndef CONFIG_AUDITSYSCALL
1240#define putname(name)   __putname(name)
1241#else
1242#define putname(name)                                                   \
1243        do {                                                            \
1244                if (unlikely(current->audit_context))                   \
1245                        audit_putname(name);                            \
1246                else                                                    \
1247                        __putname(name);                                \
1248        } while (0)
1249#endif
1250
1251extern int register_blkdev(unsigned int, const char *);
1252extern int unregister_blkdev(unsigned int, const char *);
1253extern struct block_device *bdget(dev_t);
1254extern void bd_set_size(struct block_device *, loff_t size);
1255extern void bd_forget(struct inode *inode);
1256extern void bdput(struct block_device *);
1257extern int blkdev_open(struct inode *, struct file *);
1258extern struct block_device *open_by_devnum(dev_t, unsigned);
1259extern struct file_operations def_blk_fops;
1260extern struct address_space_operations def_blk_aops;
1261extern struct file_operations def_chr_fops;
1262extern struct file_operations bad_sock_fops;
1263extern struct file_operations def_fifo_fops;
1264extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1265extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1266extern int blkdev_get(struct block_device *, mode_t, unsigned);
1267extern int blkdev_put(struct block_device *);
1268extern int bd_claim(struct block_device *, void *);
1269extern void bd_release(struct block_device *);
1270
1271/* fs/char_dev.c */
1272extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, char *);
1273extern int register_chrdev_region(dev_t, unsigned, char *);
1274extern int register_chrdev(unsigned int, const char *,
1275                           struct file_operations *);
1276extern int unregister_chrdev(unsigned int, const char *);
1277extern void unregister_chrdev_region(dev_t, unsigned);
1278extern int chrdev_open(struct inode *, struct file *);
1279
1280/* fs/block_dev.c */
1281#define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
1282extern const char *__bdevname(dev_t, char *buffer);
1283extern const char *bdevname(struct block_device *bdev, char *buffer);
1284extern struct block_device *lookup_bdev(const char *);
1285extern struct block_device *open_bdev_excl(const char *, int, void *);
1286extern void close_bdev_excl(struct block_device *);
1287
1288extern void init_special_inode(struct inode *, umode_t, dev_t);
1289
1290/* Invalid inode operations -- fs/bad_inode.c */
1291extern void make_bad_inode(struct inode *);
1292extern int is_bad_inode(struct inode *);
1293
1294extern struct file_operations read_fifo_fops;
1295extern struct file_operations write_fifo_fops;
1296extern struct file_operations rdwr_fifo_fops;
1297extern struct file_operations read_pipe_fops;
1298extern struct file_operations write_pipe_fops;
1299extern struct file_operations rdwr_pipe_fops;
1300
1301extern int fs_may_remount_ro(struct super_block *);
1302
1303/*
1304 * return READ, READA, or WRITE
1305 */
1306#define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1307
1308/*
1309 * return data direction, READ or WRITE
1310 */
1311#define bio_data_dir(bio)       ((bio)->bi_rw & 1)
1312
1313extern int check_disk_change(struct block_device *);
1314extern int invalidate_inodes(struct super_block *);
1315extern int __invalidate_device(struct block_device *, int);
1316extern int invalidate_partition(struct gendisk *, int);
1317unsigned long invalidate_mapping_pages(struct address_space *mapping,
1318                                        pgoff_t start, pgoff_t end);
1319unsigned long invalidate_inode_pages(struct address_space *mapping);
1320static inline void invalidate_remote_inode(struct inode *inode)
1321{
1322        if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1323            S_ISLNK(inode->i_mode))
1324                invalidate_inode_pages(inode->i_mapping);
1325}
1326extern void invalidate_inode_pages2(struct address_space *mapping);
1327extern void write_inode_now(struct inode *, int);
1328extern int filemap_fdatawrite(struct address_space *);
1329extern int filemap_flush(struct address_space *);
1330extern int filemap_fdatawait(struct address_space *);
1331extern int filemap_write_and_wait(struct address_space *mapping);
1332extern void sync_supers(void);
1333extern void sync_filesystems(int wait);
1334extern void emergency_sync(void);
1335extern void emergency_remount(void);
1336extern int do_remount_sb(struct super_block *sb, int flags,
1337                         void *data, int force);
1338extern sector_t bmap(struct inode *, sector_t);
1339extern int setattr_mask(unsigned int);
1340extern int notify_change(struct dentry *, struct iattr *);
1341extern int permission(struct inode *, int, struct nameidata *);
1342extern int vfs_permission(struct inode *, int);
1343extern int get_write_access(struct inode *);
1344extern int deny_write_access(struct file *);
1345static inline void put_write_access(struct inode * inode)
1346{
1347        atomic_dec(&inode->i_writecount);
1348}
1349static inline void allow_write_access(struct file *file)
1350{
1351        if (file)
1352                atomic_inc(&file->f_dentry->d_inode->i_writecount);
1353}
1354extern int do_pipe(int *);
1355
1356extern int open_namei(const char *, int, int, struct nameidata *);
1357extern int may_open(struct nameidata *, int, int);
1358
1359extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1360extern struct file * open_exec(const char *);
1361 
1362/* fs/dcache.c -- generic fs support functions */
1363extern int is_subdir(struct dentry *, struct dentry *);
1364extern ino_t find_inode_number(struct dentry *, struct qstr *);
1365
1366#include <linux/err.h>
1367
1368/* needed for stackable file system support */
1369extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1370
1371extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1372
1373extern void inode_init_once(struct inode *);
1374extern void iput(struct inode *);
1375extern struct inode * igrab(struct inode *);
1376extern ino_t iunique(struct super_block *, ino_t);
1377extern int inode_needs_sync(struct inode *inode);
1378extern void generic_delete_inode(struct inode *inode);
1379
1380extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1381                int (*test)(struct inode *, void *), void *data);
1382extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1383
1384extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1385extern struct inode * iget_locked(struct super_block *, unsigned long);
1386extern void unlock_new_inode(struct inode *);
1387
1388static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1389{
1390        struct inode *inode = iget_locked(sb, ino);
1391        
1392        if (inode && (inode->i_state & I_NEW)) {
1393                sb->s_op->read_inode(inode);
1394                unlock_new_inode(inode);
1395        }
1396
1397        return inode;
1398}
1399
1400extern void __iget(struct inode * inode);
1401extern void clear_inode(struct inode *);
1402extern void destroy_inode(struct inode *);
1403extern struct inode *new_inode(struct super_block *);
1404extern int remove_suid(struct dentry *);
1405extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1406extern struct semaphore iprune_sem;
1407
1408extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1409extern void remove_inode_hash(struct inode *);
1410static inline void insert_inode_hash(struct inode *inode) {
1411        __insert_inode_hash(inode, inode->i_ino);
1412}
1413
1414extern struct file * get_empty_filp(void);
1415extern void file_move(struct file *f, struct list_head *list);
1416extern void file_kill(struct file *f);
1417struct bio;
1418extern void submit_bio(int, struct bio *);
1419extern int bdev_read_only(struct block_device *);
1420extern int set_blocksize(struct block_device *, int);
1421extern int sb_set_blocksize(struct super_block *, int);
1422extern int sb_min_blocksize(struct super_block *, int);
1423
1424extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1425extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1426extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1427extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1428extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1429int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1430extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1431extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1432extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1433extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1434extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1435                unsigned long, loff_t *);
1436extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1437                unsigned long *, loff_t, loff_t *, size_t, size_t);
1438extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1439                unsigned long, loff_t, loff_t *, size_t, ssize_t);
1440extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1441extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1442ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1443                                unsigned long nr_segs, loff_t *ppos);
1444extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1445extern void do_generic_mapping_read(struct address_space *mapping,
1446                                    struct file_ra_state *, struct file *,
1447                                    loff_t *, read_descriptor_t *, read_actor_t);
1448extern void
1449file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1450extern ssize_t generic_file_direct_IO(int rw, struct kiocb *iocb,
1451        const struct iovec *iov, loff_t offset, unsigned long nr_segs);
1452extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov, 
1453        unsigned long nr_segs, loff_t *ppos);
1454ssize_t generic_file_writev(struct file *filp, const struct iovec *iov, 
1455                        unsigned long nr_segs, loff_t *ppos);
1456extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1457extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1458extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1459extern int generic_file_open(struct inode * inode, struct file * filp);
1460extern int nonseekable_open(struct inode * inode, struct file * filp);
1461
1462static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1463                                        read_descriptor_t * desc,
1464                                        read_actor_t actor)
1465{
1466        do_generic_mapping_read(filp->f_mapping,
1467                                &filp->f_ra,
1468                                filp,
1469                                ppos,
1470                                desc,
1471                                actor);
1472}
1473
1474ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1475        struct block_device *bdev, const struct iovec *iov, loff_t offset,
1476        unsigned long nr_segs, get_blocks_t get_blocks, dio_iodone_t end_io,
1477        int lock_type);
1478
1479enum {
1480        DIO_LOCKING = 1, /* need locking between buffered and direct access */
1481        DIO_NO_LOCKING,  /* bdev; no locking at all between buffered/direct */
1482        DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1483};
1484
1485static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1486        struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1487        loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1488        dio_iodone_t end_io)
1489{
1490        return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1491                                nr_segs, get_blocks, end_io, DIO_LOCKING);
1492}
1493
1494static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1495        struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1496        loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1497        dio_iodone_t end_io)
1498{
1499        return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1500                                nr_segs, get_blocks, end_io, DIO_NO_LOCKING);
1501}
1502
1503static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1504        struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1505        loff_t offset, unsigned long nr_segs, get_blocks_t get_blocks,
1506        dio_iodone_t end_io)
1507{
1508        return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1509                                nr_segs, get_blocks, end_io, DIO_OWN_LOCKING);
1510}
1511
1512extern struct file_operations generic_ro_fops;
1513
1514#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1515
1516extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1517extern int vfs_follow_link(struct nameidata *, const char *);
1518extern int page_readlink(struct dentry *, char __user *, int);
1519extern int page_follow_link(struct dentry *, struct nameidata *);
1520extern int page_follow_link_light(struct dentry *, struct nameidata *);
1521extern void page_put_link(struct dentry *, struct nameidata *);
1522extern int page_symlink(struct inode *inode, const char *symname, int len);
1523extern struct inode_operations page_symlink_inode_operations;
1524extern int generic_readlink(struct dentry *, char __user *, int);
1525extern void generic_fillattr(struct inode *, struct kstat *);
1526extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1527void inode_add_bytes(struct inode *inode, loff_t bytes);
1528void inode_sub_bytes(struct inode *inode, loff_t bytes);
1529loff_t inode_get_bytes(struct inode *inode);
1530void inode_set_bytes(struct inode *inode, loff_t bytes);
1531
1532extern int vfs_readdir(struct file *, filldir_t, void *);
1533
1534extern int vfs_stat(char __user *, struct kstat *);
1535extern int vfs_lstat(char __user *, struct kstat *);
1536extern int vfs_fstat(unsigned int, struct kstat *);
1537
1538extern struct file_system_type *get_fs_type(const char *name);
1539extern struct super_block *get_super(struct block_device *);
1540extern struct super_block *user_get_super(dev_t);
1541extern void drop_super(struct super_block *sb);
1542
1543extern int dcache_dir_open(struct inode *, struct file *);
1544extern int dcache_dir_close(struct inode *, struct file *);
1545extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1546extern int dcache_readdir(struct file *, void *, filldir_t);
1547extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1548extern int simple_statfs(struct super_block *, struct kstatfs *);
1549extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1550extern int simple_unlink(struct inode *, struct dentry *);
1551extern int simple_rmdir(struct inode *, struct dentry *);
1552extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1553extern int simple_sync_file(struct file *, struct dentry *, int);
1554extern int simple_empty(struct dentry *);
1555extern int simple_readpage(struct file *file, struct page *page);
1556extern int simple_prepare_write(struct file *file, struct page *page,
1557                        unsigned offset, unsigned to);
1558extern int simple_commit_write(struct file *file, struct page *page,
1559                                unsigned offset, unsigned to);
1560
1561extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1562extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1563extern struct file_operations simple_dir_operations;
1564extern struct inode_operations simple_dir_inode_operations;
1565struct tree_descr { char *name; struct file_operations *ops; int mode; };
1566extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1567extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1568extern void simple_release_fs(struct vfsmount **mount, int *count);
1569
1570extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1571
1572extern int inode_change_ok(struct inode *, struct iattr *);
1573extern int __must_check inode_setattr(struct inode *, struct iattr *);
1574
1575extern void inode_update_time(struct inode *inode, int ctime_too);
1576
1577static inline ino_t parent_ino(struct dentry *dentry)
1578{
1579        ino_t res;
1580
1581        spin_lock(&dentry->d_lock);
1582        res = dentry->d_parent->d_inode->i_ino;
1583        spin_unlock(&dentry->d_lock);
1584        return res;
1585}
1586
1587/* kernel/fork.c */
1588extern int unshare_files(void);
1589
1590/* Transaction based IO helpers */
1591
1592/*
1593 * An argresp is stored in an allocated page and holds the
1594 * size of the argument or response, along with its content
1595 */
1596struct simple_transaction_argresp {
1597        ssize_t size;
1598        char data[0];
1599};
1600
1601#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1602
1603char *simple_transaction_get(struct file *file, const char __user *buf,
1604                                size_t size);
1605ssize_t simple_transaction_read(struct file *file, char __user *buf,
1606                                size_t size, loff_t *pos);
1607int simple_transaction_release(struct inode *inode, struct file *file);
1608
1609static inline void simple_transaction_set(struct file *file, size_t n)
1610{
1611        struct simple_transaction_argresp *ar = file->private_data;
1612
1613        BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1614
1615        /*
1616         * The barrier ensures that ar->size will really remain zero until
1617         * ar->data is ready for reading.
1618         */
1619        smp_mb();
1620        ar->size = n;
1621}
1622
1623#ifdef CONFIG_SECURITY
1624static inline char *alloc_secdata(void)
1625{
1626        return (char *)get_zeroed_page(GFP_KERNEL);
1627}
1628
1629static inline void free_secdata(void *secdata)
1630{
1631        free_page((unsigned long)secdata);
1632}
1633#else
1634static inline char *alloc_secdata(void)
1635{
1636        return (char *)1;
1637}
1638
1639static inline void free_secdata(void *secdata)
1640{ }
1641#endif  /* CONFIG_SECURITY */
1642
1643#endif /* __KERNEL__ */
1644#endif /* _LINUX_FS_H */
1645
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.