linux/include/linux/fs.h
<<
>>
Prefs
   1#ifndef _LINUX_FS_H
   2#define _LINUX_FS_H
   3
   4
   5#include <linux/linkage.h>
   6#include <linux/wait.h>
   7#include <linux/kdev_t.h>
   8#include <linux/dcache.h>
   9#include <linux/path.h>
  10#include <linux/stat.h>
  11#include <linux/cache.h>
  12#include <linux/list.h>
  13#include <linux/radix-tree.h>
  14#include <linux/rbtree.h>
  15#include <linux/init.h>
  16#include <linux/pid.h>
  17#include <linux/bug.h>
  18#include <linux/mutex.h>
  19#include <linux/capability.h>
  20#include <linux/semaphore.h>
  21#include <linux/fiemap.h>
  22#include <linux/rculist_bl.h>
  23#include <linux/atomic.h>
  24#include <linux/shrinker.h>
  25#include <linux/migrate_mode.h>
  26#include <linux/uidgid.h>
  27#include <linux/lockdep.h>
  28#include <linux/percpu-rwsem.h>
  29#include <linux/blk_types.h>
  30
  31#include <asm/byteorder.h>
  32#include <uapi/linux/fs.h>
  33
  34struct export_operations;
  35struct hd_geometry;
  36struct iovec;
  37struct nameidata;
  38struct kiocb;
  39struct kobject;
  40struct pipe_inode_info;
  41struct poll_table_struct;
  42struct kstatfs;
  43struct vm_area_struct;
  44struct vfsmount;
  45struct cred;
  46struct swap_info_struct;
  47struct seq_file;
  48
  49extern void __init inode_init(void);
  50extern void __init inode_init_early(void);
  51extern void __init files_init(unsigned long);
  52
  53extern struct files_stat_struct files_stat;
  54extern unsigned long get_max_files(void);
  55extern int sysctl_nr_open;
  56extern struct inodes_stat_t inodes_stat;
  57extern int leases_enable, lease_break_time;
  58extern int sysctl_protected_symlinks;
  59extern int sysctl_protected_hardlinks;
  60
  61struct buffer_head;
  62typedef int (get_block_t)(struct inode *inode, sector_t iblock,
  63                        struct buffer_head *bh_result, int create);
  64typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
  65                        ssize_t bytes, void *private, int ret,
  66                        bool is_async);
  67
  68#define MAY_EXEC                0x00000001
  69#define MAY_WRITE               0x00000002
  70#define MAY_READ                0x00000004
  71#define MAY_APPEND              0x00000008
  72#define MAY_ACCESS              0x00000010
  73#define MAY_OPEN                0x00000020
  74#define MAY_CHDIR               0x00000040
  75/* called from RCU mode, don't block */
  76#define MAY_NOT_BLOCK           0x00000080
  77
  78/*
  79 * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
  80 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
  81 */
  82
  83/* file is open for reading */
  84#define FMODE_READ              ((__force fmode_t)0x1)
  85/* file is open for writing */
  86#define FMODE_WRITE             ((__force fmode_t)0x2)
  87/* file is seekable */
  88#define FMODE_LSEEK             ((__force fmode_t)0x4)
  89/* file can be accessed using pread */
  90#define FMODE_PREAD             ((__force fmode_t)0x8)
  91/* file can be accessed using pwrite */
  92#define FMODE_PWRITE            ((__force fmode_t)0x10)
  93/* File is opened for execution with sys_execve / sys_uselib */
  94#define FMODE_EXEC              ((__force fmode_t)0x20)
  95/* File is opened with O_NDELAY (only set for block devices) */
  96#define FMODE_NDELAY            ((__force fmode_t)0x40)
  97/* File is opened with O_EXCL (only set for block devices) */
  98#define FMODE_EXCL              ((__force fmode_t)0x80)
  99/* File is opened using open(.., 3, ..) and is writeable only for ioctls
 100   (specialy hack for floppy.c) */
 101#define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
 102/* 32bit hashes as llseek() offset (for directories) */
 103#define FMODE_32BITHASH         ((__force fmode_t)0x200)
 104/* 64bit hashes as llseek() offset (for directories) */
 105#define FMODE_64BITHASH         ((__force fmode_t)0x400)
 106
 107/*
 108 * Don't update ctime and mtime.
 109 *
 110 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
 111 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
 112 */
 113#define FMODE_NOCMTIME          ((__force fmode_t)0x800)
 114
 115/* Expect random access pattern */
 116#define FMODE_RANDOM            ((__force fmode_t)0x1000)
 117
 118/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
 119#define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
 120
 121/* File is opened with O_PATH; almost nothing can be done with it */
 122#define FMODE_PATH              ((__force fmode_t)0x4000)
 123
 124/* File was opened by fanotify and shouldn't generate fanotify events */
 125#define FMODE_NONOTIFY          ((__force fmode_t)0x1000000)
 126
 127/*
 128 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
 129 * that indicates that they should check the contents of the iovec are
 130 * valid, but not check the memory that the iovec elements
 131 * points too.
 132 */
 133#define CHECK_IOVEC_ONLY -1
 134
 135/*
 136 * The below are the various read and write types that we support. Some of
 137 * them include behavioral modifiers that send information down to the
 138 * block layer and IO scheduler. Terminology:
 139 *
 140 *      The block layer uses device plugging to defer IO a little bit, in
 141 *      the hope that we will see more IO very shortly. This increases
 142 *      coalescing of adjacent IO and thus reduces the number of IOs we
 143 *      have to send to the device. It also allows for better queuing,
 144 *      if the IO isn't mergeable. If the caller is going to be waiting
 145 *      for the IO, then he must ensure that the device is unplugged so
 146 *      that the IO is dispatched to the driver.
 147 *
 148 *      All IO is handled async in Linux. This is fine for background
 149 *      writes, but for reads or writes that someone waits for completion
 150 *      on, we want to notify the block layer and IO scheduler so that they
 151 *      know about it. That allows them to make better scheduling
 152 *      decisions. So when the below references 'sync' and 'async', it
 153 *      is referencing this priority hint.
 154 *
 155 * With that in mind, the available types are:
 156 *
 157 * READ                 A normal read operation. Device will be plugged.
 158 * READ_SYNC            A synchronous read. Device is not plugged, caller can
 159 *                      immediately wait on this read without caring about
 160 *                      unplugging.
 161 * READA                Used for read-ahead operations. Lower priority, and the
 162 *                      block layer could (in theory) choose to ignore this
 163 *                      request if it runs into resource problems.
 164 * WRITE                A normal async write. Device will be plugged.
 165 * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
 166 *                      the hint that someone will be waiting on this IO
 167 *                      shortly. The write equivalent of READ_SYNC.
 168 * WRITE_ODIRECT        Special case write for O_DIRECT only.
 169 * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
 170 * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
 171 *                      non-volatile media on completion.
 172 * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
 173 *                      by a cache flush and data is guaranteed to be on
 174 *                      non-volatile media on completion.
 175 *
 176 */
 177#define RW_MASK                 REQ_WRITE
 178#define RWA_MASK                REQ_RAHEAD
 179
 180#define READ                    0
 181#define WRITE                   RW_MASK
 182#define READA                   RWA_MASK
 183#define KERNEL_READ             (READ|REQ_KERNEL)
 184#define KERNEL_WRITE            (WRITE|REQ_KERNEL)
 185
 186#define READ_SYNC               (READ | REQ_SYNC)
 187#define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
 188#define WRITE_ODIRECT           (WRITE | REQ_SYNC)
 189#define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
 190#define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
 191#define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
 192
 193/*
 194 * Attribute flags.  These should be or-ed together to figure out what
 195 * has been changed!
 196 */
 197#define ATTR_MODE       (1 << 0)
 198#define ATTR_UID        (1 << 1)
 199#define ATTR_GID        (1 << 2)
 200#define ATTR_SIZE       (1 << 3)
 201#define ATTR_ATIME      (1 << 4)
 202#define ATTR_MTIME      (1 << 5)
 203#define ATTR_CTIME      (1 << 6)
 204#define ATTR_ATIME_SET  (1 << 7)
 205#define ATTR_MTIME_SET  (1 << 8)
 206#define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
 207#define ATTR_ATTR_FLAG  (1 << 10)
 208#define ATTR_KILL_SUID  (1 << 11)
 209#define ATTR_KILL_SGID  (1 << 12)
 210#define ATTR_FILE       (1 << 13)
 211#define ATTR_KILL_PRIV  (1 << 14)
 212#define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
 213#define ATTR_TIMES_SET  (1 << 16)
 214
 215/*
 216 * This is the Inode Attributes structure, used for notify_change().  It
 217 * uses the above definitions as flags, to know which values have changed.
 218 * Also, in this manner, a Filesystem can look at only the values it cares
 219 * about.  Basically, these are the attributes that the VFS layer can
 220 * request to change from the FS layer.
 221 *
 222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
 223 */
 224struct iattr {
 225        unsigned int    ia_valid;
 226        umode_t         ia_mode;
 227        kuid_t          ia_uid;
 228        kgid_t          ia_gid;
 229        loff_t          ia_size;
 230        struct timespec ia_atime;
 231        struct timespec ia_mtime;
 232        struct timespec ia_ctime;
 233
 234        /*
 235         * Not an attribute, but an auxiliary info for filesystems wanting to
 236         * implement an ftruncate() like method.  NOTE: filesystem should
 237         * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
 238         */
 239        struct file     *ia_file;
 240};
 241
 242/*
 243 * Includes for diskquotas.
 244 */
 245#include <linux/quota.h>
 246
 247/** 
 248 * enum positive_aop_returns - aop return codes with specific semantics
 249 *
 250 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
 251 *                          completed, that the page is still locked, and
 252 *                          should be considered active.  The VM uses this hint
 253 *                          to return the page to the active list -- it won't
 254 *                          be a candidate for writeback again in the near
 255 *                          future.  Other callers must be careful to unlock
 256 *                          the page if they get this return.  Returned by
 257 *                          writepage(); 
 258 *
 259 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
 260 *                      unlocked it and the page might have been truncated.
 261 *                      The caller should back up to acquiring a new page and
 262 *                      trying again.  The aop will be taking reasonable
 263 *                      precautions not to livelock.  If the caller held a page
 264 *                      reference, it should drop it before retrying.  Returned
 265 *                      by readpage().
 266 *
 267 * address_space_operation functions return these large constants to indicate
 268 * special semantics to the caller.  These are much larger than the bytes in a
 269 * page to allow for functions that return the number of bytes operated on in a
 270 * given page.
 271 */
 272
 273enum positive_aop_returns {
 274        AOP_WRITEPAGE_ACTIVATE  = 0x80000,
 275        AOP_TRUNCATED_PAGE      = 0x80001,
 276};
 277
 278#define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
 279#define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
 280#define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
 281                                                * helper code (eg buffer layer)
 282                                                * to clear GFP_FS from alloc */
 283
 284/*
 285 * oh the beauties of C type declarations.
 286 */
 287struct page;
 288struct address_space;
 289struct writeback_control;
 290
 291struct iov_iter {
 292        const struct iovec *iov;
 293        unsigned long nr_segs;
 294        size_t iov_offset;
 295        size_t count;
 296};
 297
 298size_t iov_iter_copy_from_user_atomic(struct page *page,
 299                struct iov_iter *i, unsigned long offset, size_t bytes);
 300size_t iov_iter_copy_from_user(struct page *page,
 301                struct iov_iter *i, unsigned long offset, size_t bytes);
 302void iov_iter_advance(struct iov_iter *i, size_t bytes);
 303int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
 304size_t iov_iter_single_seg_count(const struct iov_iter *i);
 305
 306static inline void iov_iter_init(struct iov_iter *i,
 307                        const struct iovec *iov, unsigned long nr_segs,
 308                        size_t count, size_t written)
 309{
 310        i->iov = iov;
 311        i->nr_segs = nr_segs;
 312        i->iov_offset = 0;
 313        i->count = count + written;
 314
 315        iov_iter_advance(i, written);
 316}
 317
 318static inline size_t iov_iter_count(struct iov_iter *i)
 319{
 320        return i->count;
 321}
 322
 323/*
 324 * "descriptor" for what we're up to with a read.
 325 * This allows us to use the same read code yet
 326 * have multiple different users of the data that
 327 * we read from a file.
 328 *
 329 * The simplest case just copies the data to user
 330 * mode.
 331 */
 332typedef struct {
 333        size_t written;
 334        size_t count;
 335        union {
 336                char __user *buf;
 337                void *data;
 338        } arg;
 339        int error;
 340} read_descriptor_t;
 341
 342typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
 343                unsigned long, unsigned long);
 344
 345struct address_space_operations {
 346        int (*writepage)(struct page *page, struct writeback_control *wbc);
 347        int (*readpage)(struct file *, struct page *);
 348
 349        /* Write back some dirty pages from this mapping. */
 350        int (*writepages)(struct address_space *, struct writeback_control *);
 351
 352        /* Set a page dirty.  Return true if this dirtied it */
 353        int (*set_page_dirty)(struct page *page);
 354
 355        int (*readpages)(struct file *filp, struct address_space *mapping,
 356                        struct list_head *pages, unsigned nr_pages);
 357
 358        int (*write_begin)(struct file *, struct address_space *mapping,
 359                                loff_t pos, unsigned len, unsigned flags,
 360                                struct page **pagep, void **fsdata);
 361        int (*write_end)(struct file *, struct address_space *mapping,
 362                                loff_t pos, unsigned len, unsigned copied,
 363                                struct page *page, void *fsdata);
 364
 365        /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
 366        sector_t (*bmap)(struct address_space *, sector_t);
 367        void (*invalidatepage) (struct page *, unsigned long);
 368        int (*releasepage) (struct page *, gfp_t);
 369        void (*freepage)(struct page *);
 370        ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
 371                        loff_t offset, unsigned long nr_segs);
 372        int (*get_xip_mem)(struct address_space *, pgoff_t, int,
 373                                                void **, unsigned long *);
 374        /*
 375         * migrate the contents of a page to the specified target. If sync
 376         * is false, it must not block.
 377         */
 378        int (*migratepage) (struct address_space *,
 379                        struct page *, struct page *, enum migrate_mode);
 380        int (*launder_page) (struct page *);
 381        int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
 382                                        unsigned long);
 383        int (*error_remove_page)(struct address_space *, struct page *);
 384
 385        /* swapfile support */
 386        int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
 387                                sector_t *span);
 388        void (*swap_deactivate)(struct file *file);
 389};
 390
 391extern const struct address_space_operations empty_aops;
 392
 393/*
 394 * pagecache_write_begin/pagecache_write_end must be used by general code
 395 * to write into the pagecache.
 396 */
 397int pagecache_write_begin(struct file *, struct address_space *mapping,
 398                                loff_t pos, unsigned len, unsigned flags,
 399                                struct page **pagep, void **fsdata);
 400
 401int pagecache_write_end(struct file *, struct address_space *mapping,
 402                                loff_t pos, unsigned len, unsigned copied,
 403                                struct page *page, void *fsdata);
 404
 405struct backing_dev_info;
 406struct address_space {
 407        struct inode            *host;          /* owner: inode, block_device */
 408        struct radix_tree_root  page_tree;      /* radix tree of all pages */
 409        spinlock_t              tree_lock;      /* and lock protecting it */
 410        unsigned int            i_mmap_writable;/* count VM_SHARED mappings */
 411        struct rb_root          i_mmap;         /* tree of private and shared mappings */
 412        struct list_head        i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
 413        struct mutex            i_mmap_mutex;   /* protect tree, count, list */
 414        /* Protected by tree_lock together with the radix tree */
 415        unsigned long           nrpages;        /* number of total pages */
 416        pgoff_t                 writeback_index;/* writeback starts here */
 417        const struct address_space_operations *a_ops;   /* methods */
 418        unsigned long           flags;          /* error bits/gfp mask */
 419        struct backing_dev_info *backing_dev_info; /* device readahead, etc */
 420        spinlock_t              private_lock;   /* for use by the address_space */
 421        struct list_head        private_list;   /* ditto */
 422        void                    *private_data;  /* ditto */
 423} __attribute__((aligned(sizeof(long))));
 424        /*
 425         * On most architectures that alignment is already the case; but
 426         * must be enforced here for CRIS, to let the least significant bit
 427         * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
 428         */
 429struct request_queue;
 430
 431struct block_device {
 432        dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
 433        int                     bd_openers;
 434        struct inode *          bd_inode;       /* will die */
 435        struct super_block *    bd_super;
 436        struct mutex            bd_mutex;       /* open/close mutex */
 437        struct list_head        bd_inodes;
 438        void *                  bd_claiming;
 439        void *                  bd_holder;
 440        int                     bd_holders;
 441        bool                    bd_write_holder;
 442#ifdef CONFIG_SYSFS
 443        struct list_head        bd_holder_disks;
 444#endif
 445        struct block_device *   bd_contains;
 446        unsigned                bd_block_size;
 447        struct hd_struct *      bd_part;
 448        /* number of times partitions within this device have been opened. */
 449        unsigned                bd_part_count;
 450        int                     bd_invalidated;
 451        struct gendisk *        bd_disk;
 452        struct request_queue *  bd_queue;
 453        struct list_head        bd_list;
 454        /*
 455         * Private data.  You must have bd_claim'ed the block_device
 456         * to use this.  NOTE:  bd_claim allows an owner to claim
 457         * the same device multiple times, the owner must take special
 458         * care to not mess up bd_private for that case.
 459         */
 460        unsigned long           bd_private;
 461
 462        /* The counter of freeze processes */
 463        int                     bd_fsfreeze_count;
 464        /* Mutex for freeze */
 465        struct mutex            bd_fsfreeze_mutex;
 466};
 467
 468/*
 469 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
 470 * radix trees
 471 */
 472#define PAGECACHE_TAG_DIRTY     0
 473#define PAGECACHE_TAG_WRITEBACK 1
 474#define PAGECACHE_TAG_TOWRITE   2
 475
 476int mapping_tagged(struct address_space *mapping, int tag);
 477
 478/*
 479 * Might pages of this file be mapped into userspace?
 480 */
 481static inline int mapping_mapped(struct address_space *mapping)
 482{
 483        return  !RB_EMPTY_ROOT(&mapping->i_mmap) ||
 484                !list_empty(&mapping->i_mmap_nonlinear);
 485}
 486
 487/*
 488 * Might pages of this file have been modified in userspace?
 489 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
 490 * marks vma as VM_SHARED if it is shared, and the file was opened for
 491 * writing i.e. vma may be mprotected writable even if now readonly.
 492 */
 493static inline int mapping_writably_mapped(struct address_space *mapping)
 494{
 495        return mapping->i_mmap_writable != 0;
 496}
 497
 498/*
 499 * Use sequence counter to get consistent i_size on 32-bit processors.
 500 */
 501#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 502#include <linux/seqlock.h>
 503#define __NEED_I_SIZE_ORDERED
 504#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
 505#else
 506#define i_size_ordered_init(inode) do { } while (0)
 507#endif
 508
 509struct posix_acl;
 510#define ACL_NOT_CACHED ((void *)(-1))
 511
 512#define IOP_FASTPERM    0x0001
 513#define IOP_LOOKUP      0x0002
 514#define IOP_NOFOLLOW    0x0004
 515
 516/*
 517 * Keep mostly read-only and often accessed (especially for
 518 * the RCU path lookup and 'stat' data) fields at the beginning
 519 * of the 'struct inode'
 520 */
 521struct inode {
 522        umode_t                 i_mode;
 523        unsigned short          i_opflags;
 524        kuid_t                  i_uid;
 525        kgid_t                  i_gid;
 526        unsigned int            i_flags;
 527
 528#ifdef CONFIG_FS_POSIX_ACL
 529        struct posix_acl        *i_acl;
 530        struct posix_acl        *i_default_acl;
 531#endif
 532
 533        const struct inode_operations   *i_op;
 534        struct super_block      *i_sb;
 535        struct address_space    *i_mapping;
 536
 537#ifdef CONFIG_SECURITY
 538        void                    *i_security;
 539#endif
 540
 541        /* Stat data, not accessed from path walking */
 542        unsigned long           i_ino;
 543        /*
 544         * Filesystems may only read i_nlink directly.  They shall use the
 545         * following functions for modification:
 546         *
 547         *    (set|clear|inc|drop)_nlink
 548         *    inode_(inc|dec)_link_count
 549         */
 550        union {
 551                const unsigned int i_nlink;
 552                unsigned int __i_nlink;
 553        };
 554        dev_t                   i_rdev;
 555        loff_t                  i_size;
 556        struct timespec         i_atime;
 557        struct timespec         i_mtime;
 558        struct timespec         i_ctime;
 559        spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
 560        unsigned short          i_bytes;
 561        unsigned int            i_blkbits;
 562        blkcnt_t                i_blocks;
 563
 564#ifdef __NEED_I_SIZE_ORDERED
 565        seqcount_t              i_size_seqcount;
 566#endif
 567
 568        /* Misc */
 569        unsigned long           i_state;
 570        struct mutex            i_mutex;
 571
 572        unsigned long           dirtied_when;   /* jiffies of first dirtying */
 573
 574        struct hlist_node       i_hash;
 575        struct list_head        i_wb_list;      /* backing dev IO list */
 576        struct list_head        i_lru;          /* inode LRU list */
 577        struct list_head        i_sb_list;
 578        union {
 579                struct hlist_head       i_dentry;
 580                struct rcu_head         i_rcu;
 581        };
 582        u64                     i_version;
 583        atomic_t                i_count;
 584        atomic_t                i_dio_count;
 585        atomic_t                i_writecount;
 586        const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
 587        struct file_lock        *i_flock;
 588        struct address_space    i_data;
 589#ifdef CONFIG_QUOTA
 590        struct dquot            *i_dquot[MAXQUOTAS];
 591#endif
 592        struct list_head        i_devices;
 593        union {
 594                struct pipe_inode_info  *i_pipe;
 595                struct block_device     *i_bdev;
 596                struct cdev             *i_cdev;
 597        };
 598
 599        __u32                   i_generation;
 600
 601#ifdef CONFIG_FSNOTIFY
 602        __u32                   i_fsnotify_mask; /* all events this inode cares about */
 603        struct hlist_head       i_fsnotify_marks;
 604#endif
 605
 606#ifdef CONFIG_IMA
 607        atomic_t                i_readcount; /* struct files open RO */
 608#endif
 609        void                    *i_private; /* fs or device private pointer */
 610};
 611
 612static inline int inode_unhashed(struct inode *inode)
 613{
 614        return hlist_unhashed(&inode->i_hash);
 615}
 616
 617/*
 618 * inode->i_mutex nesting subclasses for the lock validator:
 619 *
 620 * 0: the object of the current VFS operation
 621 * 1: parent
 622 * 2: child/target
 623 * 3: quota file
 624 *
 625 * The locking order between these classes is
 626 * parent -> child -> normal -> xattr -> quota
 627 */
 628enum inode_i_mutex_lock_class
 629{
 630        I_MUTEX_NORMAL,
 631        I_MUTEX_PARENT,
 632        I_MUTEX_CHILD,
 633        I_MUTEX_XATTR,
 634        I_MUTEX_QUOTA
 635};
 636
 637/*
 638 * NOTE: in a 32bit arch with a preemptable kernel and
 639 * an UP compile the i_size_read/write must be atomic
 640 * with respect to the local cpu (unlike with preempt disabled),
 641 * but they don't need to be atomic with respect to other cpus like in
 642 * true SMP (so they need either to either locally disable irq around
 643 * the read or for example on x86 they can be still implemented as a
 644 * cmpxchg8b without the need of the lock prefix). For SMP compiles
 645 * and 64bit archs it makes no difference if preempt is enabled or not.
 646 */
 647static inline loff_t i_size_read(const struct inode *inode)
 648{
 649#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 650        loff_t i_size;
 651        unsigned int seq;
 652
 653        do {
 654                seq = read_seqcount_begin(&inode->i_size_seqcount);
 655                i_size = inode->i_size;
 656        } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
 657        return i_size;
 658#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 659        loff_t i_size;
 660
 661        preempt_disable();
 662        i_size = inode->i_size;
 663        preempt_enable();
 664        return i_size;
 665#else
 666        return inode->i_size;
 667#endif
 668}
 669
 670/*
 671 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
 672 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
 673 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
 674 */
 675static inline void i_size_write(struct inode *inode, loff_t i_size)
 676{
 677#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 678        write_seqcount_begin(&inode->i_size_seqcount);
 679        inode->i_size = i_size;
 680        write_seqcount_end(&inode->i_size_seqcount);
 681#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 682        preempt_disable();
 683        inode->i_size = i_size;
 684        preempt_enable();
 685#else
 686        inode->i_size = i_size;
 687#endif
 688}
 689
 690/* Helper functions so that in most cases filesystems will
 691 * not need to deal directly with kuid_t and kgid_t and can
 692 * instead deal with the raw numeric values that are stored
 693 * in the filesystem.
 694 */
 695static inline uid_t i_uid_read(const struct inode *inode)
 696{
 697        return from_kuid(&init_user_ns, inode->i_uid);
 698}
 699
 700static inline gid_t i_gid_read(const struct inode *inode)
 701{
 702        return from_kgid(&init_user_ns, inode->i_gid);
 703}
 704
 705static inline void i_uid_write(struct inode *inode, uid_t uid)
 706{
 707        inode->i_uid = make_kuid(&init_user_ns, uid);
 708}
 709
 710static inline void i_gid_write(struct inode *inode, gid_t gid)
 711{
 712        inode->i_gid = make_kgid(&init_user_ns, gid);
 713}
 714
 715static inline unsigned iminor(const struct inode *inode)
 716{
 717        return MINOR(inode->i_rdev);
 718}
 719
 720static inline unsigned imajor(const struct inode *inode)
 721{
 722        return MAJOR(inode->i_rdev);
 723}
 724
 725extern struct block_device *I_BDEV(struct inode *inode);
 726
 727struct fown_struct {
 728        rwlock_t lock;          /* protects pid, uid, euid fields */
 729        struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
 730        enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
 731        kuid_t uid, euid;       /* uid/euid of process setting the owner */
 732        int signum;             /* posix.1b rt signal to be delivered on IO */
 733};
 734
 735/*
 736 * Track a single file's readahead state
 737 */
 738struct file_ra_state {
 739        pgoff_t start;                  /* where readahead started */
 740        unsigned int size;              /* # of readahead pages */
 741        unsigned int async_size;        /* do asynchronous readahead when
 742                                           there are only # of pages ahead */
 743
 744        unsigned int ra_pages;          /* Maximum readahead window */
 745        unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
 746        loff_t prev_pos;                /* Cache last read() position */
 747};
 748
 749/*
 750 * Check if @index falls in the readahead windows.
 751 */
 752static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
 753{
 754        return (index >= ra->start &&
 755                index <  ra->start + ra->size);
 756}
 757
 758#define FILE_MNT_WRITE_TAKEN    1
 759#define FILE_MNT_WRITE_RELEASED 2
 760
 761struct file {
 762        /*
 763         * fu_list becomes invalid after file_free is called and queued via
 764         * fu_rcuhead for RCU freeing
 765         */
 766        union {
 767                struct list_head        fu_list;
 768                struct rcu_head         fu_rcuhead;
 769        } f_u;
 770        struct path             f_path;
 771#define f_dentry        f_path.dentry
 772        struct inode            *f_inode;       /* cached value */
 773        const struct file_operations    *f_op;
 774
 775        /*
 776         * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
 777         * Must not be taken from IRQ context.
 778         */
 779        spinlock_t              f_lock;
 780#ifdef CONFIG_SMP
 781        int                     f_sb_list_cpu;
 782#endif
 783        atomic_long_t           f_count;
 784        unsigned int            f_flags;
 785        fmode_t                 f_mode;
 786        loff_t                  f_pos;
 787        struct fown_struct      f_owner;
 788        const struct cred       *f_cred;
 789        struct file_ra_state    f_ra;
 790
 791        u64                     f_version;
 792#ifdef CONFIG_SECURITY
 793        void                    *f_security;
 794#endif
 795        /* needed for tty driver, and maybe others */
 796        void                    *private_data;
 797
 798#ifdef CONFIG_EPOLL
 799        /* Used by fs/eventpoll.c to link all the hooks to this file */
 800        struct list_head        f_ep_links;
 801        struct list_head        f_tfile_llink;
 802#endif /* #ifdef CONFIG_EPOLL */
 803        struct address_space    *f_mapping;
 804#ifdef CONFIG_DEBUG_WRITECOUNT
 805        unsigned long f_mnt_write_state;
 806#endif
 807};
 808
 809struct file_handle {
 810        __u32 handle_bytes;
 811        int handle_type;
 812        /* file identifier */
 813        unsigned char f_handle[0];
 814};
 815
 816static inline struct file *get_file(struct file *f)
 817{
 818        atomic_long_inc(&f->f_count);
 819        return f;
 820}
 821#define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
 822#define file_count(x)   atomic_long_read(&(x)->f_count)
 823
 824#ifdef CONFIG_DEBUG_WRITECOUNT
 825static inline void file_take_write(struct file *f)
 826{
 827        WARN_ON(f->f_mnt_write_state != 0);
 828        f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
 829}
 830static inline void file_release_write(struct file *f)
 831{
 832        f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
 833}
 834static inline void file_reset_write(struct file *f)
 835{
 836        f->f_mnt_write_state = 0;
 837}
 838static inline void file_check_state(struct file *f)
 839{
 840        /*
 841         * At this point, either both or neither of these bits
 842         * should be set.
 843         */
 844        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
 845        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
 846}
 847static inline int file_check_writeable(struct file *f)
 848{
 849        if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
 850                return 0;
 851        printk(KERN_WARNING "writeable file with no "
 852                            "mnt_want_write()\n");
 853        WARN_ON(1);
 854        return -EINVAL;
 855}
 856#else /* !CONFIG_DEBUG_WRITECOUNT */
 857static inline void file_take_write(struct file *filp) {}
 858static inline void file_release_write(struct file *filp) {}
 859static inline void file_reset_write(struct file *filp) {}
 860static inline void file_check_state(struct file *filp) {}
 861static inline int file_check_writeable(struct file *filp)
 862{
 863        return 0;
 864}
 865#endif /* CONFIG_DEBUG_WRITECOUNT */
 866
 867#define MAX_NON_LFS     ((1UL<<31) - 1)
 868
 869/* Page cache limit. The filesystems should put that into their s_maxbytes 
 870   limits, otherwise bad things can happen in VM. */ 
 871#if BITS_PER_LONG==32
 872#define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
 873#elif BITS_PER_LONG==64
 874#define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
 875#endif
 876
 877#define FL_POSIX        1
 878#define FL_FLOCK        2
 879#define FL_ACCESS       8       /* not trying to lock, just looking */
 880#define FL_EXISTS       16      /* when unlocking, test for existence */
 881#define FL_LEASE        32      /* lease held on this file */
 882#define FL_CLOSE        64      /* unlock on close */
 883#define FL_SLEEP        128     /* A blocking lock */
 884#define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
 885#define FL_UNLOCK_PENDING       512 /* Lease is being broken */
 886
 887/*
 888 * Special return value from posix_lock_file() and vfs_lock_file() for
 889 * asynchronous locking.
 890 */
 891#define FILE_LOCK_DEFERRED 1
 892
 893/*
 894 * The POSIX file lock owner is determined by
 895 * the "struct files_struct" in the thread group
 896 * (or NULL for no owner - BSD locks).
 897 *
 898 * Lockd stuffs a "host" pointer into this.
 899 */
 900typedef struct files_struct *fl_owner_t;
 901
 902struct file_lock_operations {
 903        void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
 904        void (*fl_release_private)(struct file_lock *);
 905};
 906
 907struct lock_manager_operations {
 908        int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
 909        void (*lm_notify)(struct file_lock *);  /* unblock callback */
 910        int (*lm_grant)(struct file_lock *, struct file_lock *, int);
 911        void (*lm_break)(struct file_lock *);
 912        int (*lm_change)(struct file_lock **, int);
 913};
 914
 915struct lock_manager {
 916        struct list_head list;
 917};
 918
 919struct net;
 920void locks_start_grace(struct net *, struct lock_manager *);
 921void locks_end_grace(struct lock_manager *);
 922int locks_in_grace(struct net *);
 923
 924/* that will die - we need it for nfs_lock_info */
 925#include <linux/nfs_fs_i.h>
 926
 927struct file_lock {
 928        struct file_lock *fl_next;      /* singly linked list for this inode  */
 929        struct list_head fl_link;       /* doubly linked list of all locks */
 930        struct list_head fl_block;      /* circular list of blocked processes */
 931        fl_owner_t fl_owner;
 932        unsigned int fl_flags;
 933        unsigned char fl_type;
 934        unsigned int fl_pid;
 935        struct pid *fl_nspid;
 936        wait_queue_head_t fl_wait;
 937        struct file *fl_file;
 938        loff_t fl_start;
 939        loff_t fl_end;
 940
 941        struct fasync_struct *  fl_fasync; /* for lease break notifications */
 942        /* for lease breaks: */
 943        unsigned long fl_break_time;
 944        unsigned long fl_downgrade_time;
 945
 946        const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
 947        const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
 948        union {
 949                struct nfs_lock_info    nfs_fl;
 950                struct nfs4_lock_info   nfs4_fl;
 951                struct {
 952                        struct list_head link;  /* link in AFS vnode's pending_locks list */
 953                        int state;              /* state of grant or error if -ve */
 954                } afs;
 955        } fl_u;
 956};
 957
 958/* The following constant reflects the upper bound of the file/locking space */
 959#ifndef OFFSET_MAX
 960#define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
 961#define OFFSET_MAX      INT_LIMIT(loff_t)
 962#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
 963#endif
 964
 965#include <linux/fcntl.h>
 966
 967extern void send_sigio(struct fown_struct *fown, int fd, int band);
 968
 969#ifdef CONFIG_FILE_LOCKING
 970extern int fcntl_getlk(struct file *, struct flock __user *);
 971extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
 972                        struct flock __user *);
 973
 974#if BITS_PER_LONG == 32
 975extern int fcntl_getlk64(struct file *, struct flock64 __user *);
 976extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
 977                        struct flock64 __user *);
 978#endif
 979
 980extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
 981extern int fcntl_getlease(struct file *filp);
 982
 983/* fs/locks.c */
 984void locks_free_lock(struct file_lock *fl);
 985extern void locks_init_lock(struct file_lock *);
 986extern struct file_lock * locks_alloc_lock(void);
 987extern void locks_copy_lock(struct file_lock *, struct file_lock *);
 988extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
 989extern void locks_remove_posix(struct file *, fl_owner_t);
 990extern void locks_remove_flock(struct file *);
 991extern void locks_release_private(struct file_lock *);
 992extern void posix_test_lock(struct file *, struct file_lock *);
 993extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
 994extern int posix_lock_file_wait(struct file *, struct file_lock *);
 995extern int posix_unblock_lock(struct file *, struct file_lock *);
 996extern int vfs_test_lock(struct file *, struct file_lock *);
 997extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
 998extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
 999extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1000extern int __break_lease(struct inode *inode, unsigned int flags);
1001extern void lease_get_mtime(struct inode *, struct timespec *time);
1002extern int generic_setlease(struct file *, long, struct file_lock **);
1003extern int vfs_setlease(struct file *, long, struct file_lock **);
1004extern int lease_modify(struct file_lock **, int);
1005extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1006extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1007extern void locks_delete_block(struct file_lock *waiter);
1008extern void lock_flocks(void);
1009extern void unlock_flocks(void);
1010#else /* !CONFIG_FILE_LOCKING */
1011static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1012{
1013        return -EINVAL;
1014}
1015
1016static inline int fcntl_setlk(unsigned int fd, struct file *file,
1017                              unsigned int cmd, struct flock __user *user)
1018{
1019        return -EACCES;
1020}
1021
1022#if BITS_PER_LONG == 32
1023static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1024{
1025        return -EINVAL;
1026}
1027
1028static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1029                                unsigned int cmd, struct flock64 __user *user)
1030{
1031        return -EACCES;
1032}
1033#endif
1034static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1035{
1036        return 0;
1037}
1038
1039static inline int fcntl_getlease(struct file *filp)
1040{
1041        return 0;
1042}
1043
1044static inline void locks_init_lock(struct file_lock *fl)
1045{
1046        return;
1047}
1048
1049static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1050{
1051        return;
1052}
1053
1054static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1055{
1056        return;
1057}
1058
1059static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1060{
1061        return;
1062}
1063
1064static inline void locks_remove_flock(struct file *filp)
1065{
1066        return;
1067}
1068
1069static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1070{
1071        return;
1072}
1073
1074static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1075                                  struct file_lock *conflock)
1076{
1077        return -ENOLCK;
1078}
1079
1080static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1081{
1082        return -ENOLCK;
1083}
1084
1085static inline int posix_unblock_lock(struct file *filp,
1086                                     struct file_lock *waiter)
1087{
1088        return -ENOENT;
1089}
1090
1091static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1092{
1093        return 0;
1094}
1095
1096static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1097                                struct file_lock *fl, struct file_lock *conf)
1098{
1099        return -ENOLCK;
1100}
1101
1102static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1103{
1104        return 0;
1105}
1106
1107static inline int flock_lock_file_wait(struct file *filp,
1108                                       struct file_lock *request)
1109{
1110        return -ENOLCK;
1111}
1112
1113static inline int __break_lease(struct inode *inode, unsigned int mode)
1114{
1115        return 0;
1116}
1117
1118static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1119{
1120        return;
1121}
1122
1123static inline int generic_setlease(struct file *filp, long arg,
1124                                    struct file_lock **flp)
1125{
1126        return -EINVAL;
1127}
1128
1129static inline int vfs_setlease(struct file *filp, long arg,
1130                               struct file_lock **lease)
1131{
1132        return -EINVAL;
1133}
1134
1135static inline int lease_modify(struct file_lock **before, int arg)
1136{
1137        return -EINVAL;
1138}
1139
1140static inline int lock_may_read(struct inode *inode, loff_t start,
1141                                unsigned long len)
1142{
1143        return 1;
1144}
1145
1146static inline int lock_may_write(struct inode *inode, loff_t start,
1147                                 unsigned long len)
1148{
1149        return 1;
1150}
1151
1152static inline void locks_delete_block(struct file_lock *waiter)
1153{
1154}
1155
1156static inline void lock_flocks(void)
1157{
1158}
1159
1160static inline void unlock_flocks(void)
1161{
1162}
1163
1164#endif /* !CONFIG_FILE_LOCKING */
1165
1166
1167struct fasync_struct {
1168        spinlock_t              fa_lock;
1169        int                     magic;
1170        int                     fa_fd;
1171        struct fasync_struct    *fa_next; /* singly linked list */
1172        struct file             *fa_file;
1173        struct rcu_head         fa_rcu;
1174};
1175
1176#define FASYNC_MAGIC 0x4601
1177
1178/* SMP safe fasync helpers: */
1179extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1180extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1181extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1182extern struct fasync_struct *fasync_alloc(void);
1183extern void fasync_free(struct fasync_struct *);
1184
1185/* can be called from interrupts */
1186extern void kill_fasync(struct fasync_struct **, int, int);
1187
1188extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1189extern int f_setown(struct file *filp, unsigned long arg, int force);
1190extern void f_delown(struct file *filp);
1191extern pid_t f_getown(struct file *filp);
1192extern int send_sigurg(struct fown_struct *fown);
1193
1194struct mm_struct;
1195
1196/*
1197 *      Umount options
1198 */
1199
1200#define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1201#define MNT_DETACH      0x00000002      /* Just detach from the tree */
1202#define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1203#define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1204#define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1205
1206extern struct list_head super_blocks;
1207extern spinlock_t sb_lock;
1208
1209/* Possible states of 'frozen' field */
1210enum {
1211        SB_UNFROZEN = 0,                /* FS is unfrozen */
1212        SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1213        SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1214        SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1215                                         * internal threads if needed) */
1216        SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1217};
1218
1219#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1220
1221struct sb_writers {
1222        /* Counters for counting writers at each level */
1223        struct percpu_counter   counter[SB_FREEZE_LEVELS];
1224        wait_queue_head_t       wait;           /* queue for waiting for
1225                                                   writers / faults to finish */
1226        int                     frozen;         /* Is sb frozen? */
1227        wait_queue_head_t       wait_unfrozen;  /* queue for waiting for
1228                                                   sb to be thawed */
1229#ifdef CONFIG_DEBUG_LOCK_ALLOC
1230        struct lockdep_map      lock_map[SB_FREEZE_LEVELS];
1231#endif
1232};
1233
1234struct super_block {
1235        struct list_head        s_list;         /* Keep this first */
1236        dev_t                   s_dev;          /* search index; _not_ kdev_t */
1237        unsigned char           s_blocksize_bits;
1238        unsigned long           s_blocksize;
1239        loff_t                  s_maxbytes;     /* Max file size */
1240        struct file_system_type *s_type;
1241        const struct super_operations   *s_op;
1242        const struct dquot_operations   *dq_op;
1243        const struct quotactl_ops       *s_qcop;
1244        const struct export_operations *s_export_op;
1245        unsigned long           s_flags;
1246        unsigned long           s_magic;
1247        struct dentry           *s_root;
1248        struct rw_semaphore     s_umount;
1249        int                     s_count;
1250        atomic_t                s_active;
1251#ifdef CONFIG_SECURITY
1252        void                    *s_security;
1253#endif
1254        const struct xattr_handler **s_xattr;
1255
1256        struct list_head        s_inodes;       /* all inodes */
1257        struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1258#ifdef CONFIG_SMP
1259        struct list_head __percpu *s_files;
1260#else
1261        struct list_head        s_files;
1262#endif
1263        struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1264        /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1265        struct list_head        s_dentry_lru;   /* unused dentry lru */
1266        int                     s_nr_dentry_unused;     /* # of dentry on lru */
1267
1268        /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1269        spinlock_t              s_inode_lru_lock ____cacheline_aligned_in_smp;
1270        struct list_head        s_inode_lru;            /* unused inode lru */
1271        int                     s_nr_inodes_unused;     /* # of inodes on lru */
1272
1273        struct block_device     *s_bdev;
1274        struct backing_dev_info *s_bdi;
1275        struct mtd_info         *s_mtd;
1276        struct hlist_node       s_instances;
1277        struct quota_info       s_dquot;        /* Diskquota specific options */
1278
1279        struct sb_writers       s_writers;
1280
1281        char s_id[32];                          /* Informational name */
1282        u8 s_uuid[16];                          /* UUID */
1283
1284        void                    *s_fs_info;     /* Filesystem private info */
1285        unsigned int            s_max_links;
1286        fmode_t                 s_mode;
1287
1288        /* Granularity of c/m/atime in ns.
1289           Cannot be worse than a second */
1290        u32                s_time_gran;
1291
1292        /*
1293         * The next field is for VFS *only*. No filesystems have any business
1294         * even looking at it. You had been warned.
1295         */
1296        struct mutex s_vfs_rename_mutex;        /* Kludge */
1297
1298        /*
1299         * Filesystem subtype.  If non-empty the filesystem type field
1300         * in /proc/mounts will be "type.subtype"
1301         */
1302        char *s_subtype;
1303
1304        /*
1305         * Saved mount options for lazy filesystems using
1306         * generic_show_options()
1307         */
1308        char __rcu *s_options;
1309        const struct dentry_operations *s_d_op; /* default d_op for dentries */
1310
1311        /*
1312         * Saved pool identifier for cleancache (-1 means none)
1313         */
1314        int cleancache_poolid;
1315
1316        struct shrinker s_shrink;       /* per-sb shrinker handle */
1317
1318        /* Number of inodes with nlink == 0 but still referenced */
1319        atomic_long_t s_remove_count;
1320
1321        /* Being remounted read-only */
1322        int s_readonly_remount;
1323};
1324
1325/* superblock cache pruning functions */
1326extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1327extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1328
1329extern struct timespec current_fs_time(struct super_block *sb);
1330
1331/*
1332 * Snapshotting support.
1333 */
1334
1335void __sb_end_write(struct super_block *sb, int level);
1336int __sb_start_write(struct super_block *sb, int level, bool wait);
1337
1338/**
1339 * sb_end_write - drop write access to a superblock
1340 * @sb: the super we wrote to
1341 *
1342 * Decrement number of writers to the filesystem. Wake up possible waiters
1343 * wanting to freeze the filesystem.
1344 */
1345static inline void sb_end_write(struct super_block *sb)
1346{
1347        __sb_end_write(sb, SB_FREEZE_WRITE);
1348}
1349
1350/**
1351 * sb_end_pagefault - drop write access to a superblock from a page fault
1352 * @sb: the super we wrote to
1353 *
1354 * Decrement number of processes handling write page fault to the filesystem.
1355 * Wake up possible waiters wanting to freeze the filesystem.
1356 */
1357static inline void sb_end_pagefault(struct super_block *sb)
1358{
1359        __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1360}
1361
1362/**
1363 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1364 * @sb: the super we wrote to
1365 *
1366 * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1367 * waiters wanting to freeze the filesystem.
1368 */
1369static inline void sb_end_intwrite(struct super_block *sb)
1370{
1371        __sb_end_write(sb, SB_FREEZE_FS);
1372}
1373
1374/**
1375 * sb_start_write - get write access to a superblock
1376 * @sb: the super we write to
1377 *
1378 * When a process wants to write data or metadata to a file system (i.e. dirty
1379 * a page or an inode), it should embed the operation in a sb_start_write() -
1380 * sb_end_write() pair to get exclusion against file system freezing. This
1381 * function increments number of writers preventing freezing. If the file
1382 * system is already frozen, the function waits until the file system is
1383 * thawed.
1384 *
1385 * Since freeze protection behaves as a lock, users have to preserve
1386 * ordering of freeze protection and other filesystem locks. Generally,
1387 * freeze protection should be the outermost lock. In particular, we have:
1388 *
1389 * sb_start_write
1390 *   -> i_mutex                 (write path, truncate, directory ops, ...)
1391 *   -> s_umount                (freeze_super, thaw_super)
1392 */
1393static inline void sb_start_write(struct super_block *sb)
1394{
1395        __sb_start_write(sb, SB_FREEZE_WRITE, true);
1396}
1397
1398static inline int sb_start_write_trylock(struct super_block *sb)
1399{
1400        return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1401}
1402
1403/**
1404 * sb_start_pagefault - get write access to a superblock from a page fault
1405 * @sb: the super we write to
1406 *
1407 * When a process starts handling write page fault, it should embed the
1408 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1409 * exclusion against file system freezing. This is needed since the page fault
1410 * is going to dirty a page. This function increments number of running page
1411 * faults preventing freezing. If the file system is already frozen, the
1412 * function waits until the file system is thawed.
1413 *
1414 * Since page fault freeze protection behaves as a lock, users have to preserve
1415 * ordering of freeze protection and other filesystem locks. It is advised to
1416 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1417 * handling code implies lock dependency:
1418 *
1419 * mmap_sem
1420 *   -> sb_start_pagefault
1421 */
1422static inline void sb_start_pagefault(struct super_block *sb)
1423{
1424        __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1425}
1426
1427/*
1428 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1429 * @sb: the super we write to
1430 *
1431 * This is the third level of protection against filesystem freezing. It is
1432 * free for use by a filesystem. The only requirement is that it must rank
1433 * below sb_start_pagefault.
1434 *
1435 * For example filesystem can call sb_start_intwrite() when starting a
1436 * transaction which somewhat eases handling of freezing for internal sources
1437 * of filesystem changes (internal fs threads, discarding preallocation on file
1438 * close, etc.).
1439 */
1440static inline void sb_start_intwrite(struct super_block *sb)
1441{
1442        __sb_start_write(sb, SB_FREEZE_FS, true);
1443}
1444
1445
1446extern bool inode_owner_or_capable(const struct inode *inode);
1447
1448/*
1449 * VFS helper functions..
1450 */
1451extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1452extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1453extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1454extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1455extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1456extern int vfs_rmdir(struct inode *, struct dentry *);
1457extern int vfs_unlink(struct inode *, struct dentry *);
1458extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1459
1460/*
1461 * VFS dentry helper functions.
1462 */
1463extern void dentry_unhash(struct dentry *dentry);
1464
1465/*
1466 * VFS file helper functions.
1467 */
1468extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1469                        umode_t mode);
1470/*
1471 * VFS FS_IOC_FIEMAP helper definitions.
1472 */
1473struct fiemap_extent_info {
1474        unsigned int fi_flags;          /* Flags as passed from user */
1475        unsigned int fi_extents_mapped; /* Number of mapped extents */
1476        unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1477        struct fiemap_extent __user *fi_extents_start; /* Start of
1478                                                        fiemap_extent array */
1479};
1480int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1481                            u64 phys, u64 len, u32 flags);
1482int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1483
1484/*
1485 * File types
1486 *
1487 * NOTE! These match bits 12..15 of stat.st_mode
1488 * (ie "(i_mode >> 12) & 15").
1489 */
1490#define DT_UNKNOWN      0
1491#define DT_FIFO         1
1492#define DT_CHR          2
1493#define DT_DIR          4
1494#define DT_BLK          6
1495#define DT_REG          8
1496#define DT_LNK          10
1497#define DT_SOCK         12
1498#define DT_WHT          14
1499
1500/*
1501 * This is the "filldir" function type, used by readdir() to let
1502 * the kernel specify what kind of dirent layout it wants to have.
1503 * This allows the kernel to read directories into kernel space or
1504 * to have different dirent layouts depending on the binary type.
1505 */
1506typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1507struct block_device_operations;
1508
1509/* These macros are for out of kernel modules to test that
1510 * the kernel supports the unlocked_ioctl and compat_ioctl
1511 * fields in struct file_operations. */
1512#define HAVE_COMPAT_IOCTL 1
1513#define HAVE_UNLOCKED_IOCTL 1
1514
1515struct file_operations {
1516        struct module *owner;
1517        loff_t (*llseek) (struct file *, loff_t, int);
1518        ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1519        ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1520        ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1521        ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1522        int (*readdir) (struct file *, void *, filldir_t);
1523        unsigned int (*poll) (struct file *, struct poll_table_struct *);
1524        long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1525        long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1526        int (*mmap) (struct file *, struct vm_area_struct *);
1527        int (*open) (struct inode *, struct file *);
1528        int (*flush) (struct file *, fl_owner_t id);
1529        int (*release) (struct inode *, struct file *);
1530        int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1531        int (*aio_fsync) (struct kiocb *, int datasync);
1532        int (*fasync) (int, struct file *, int);
1533        int (*lock) (struct file *, int, struct file_lock *);
1534        ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1535        unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1536        int (*check_flags)(int);
1537        int (*flock) (struct file *, int, struct file_lock *);
1538        ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1539        ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1540        int (*setlease)(struct file *, long, struct file_lock **);
1541        long (*fallocate)(struct file *file, int mode, loff_t offset,
1542                          loff_t len);
1543        int (*show_fdinfo)(struct seq_file *m, struct file *f);
1544};
1545
1546struct inode_operations {
1547        struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1548        void * (*follow_link) (struct dentry *, struct nameidata *);
1549        int (*permission) (struct inode *, int);
1550        struct posix_acl * (*get_acl)(struct inode *, int);
1551
1552        int (*readlink) (struct dentry *, char __user *,int);
1553        void (*put_link) (struct dentry *, struct nameidata *, void *);
1554
1555        int (*create) (struct inode *,struct dentry *, umode_t, bool);
1556        int (*link) (struct dentry *,struct inode *,struct dentry *);
1557        int (*unlink) (struct inode *,struct dentry *);
1558        int (*symlink) (struct inode *,struct dentry *,const char *);
1559        int (*mkdir) (struct inode *,struct dentry *,umode_t);
1560        int (*rmdir) (struct inode *,struct dentry *);
1561        int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1562        int (*rename) (struct inode *, struct dentry *,
1563                        struct inode *, struct dentry *);
1564        int (*setattr) (struct dentry *, struct iattr *);
1565        int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1566        int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1567        ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1568        ssize_t (*listxattr) (struct dentry *, char *, size_t);
1569        int (*removexattr) (struct dentry *, const char *);
1570        int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1571                      u64 len);
1572        int (*update_time)(struct inode *, struct timespec *, int);
1573        int (*atomic_open)(struct inode *, struct dentry *,
1574                           struct file *, unsigned open_flag,
1575                           umode_t create_mode, int *opened);
1576} ____cacheline_aligned;
1577
1578ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1579                              unsigned long nr_segs, unsigned long fast_segs,
1580                              struct iovec *fast_pointer,
1581                              struct iovec **ret_pointer);
1582
1583extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1584extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1585extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1586                unsigned long, loff_t *);
1587extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1588                unsigned long, loff_t *);
1589
1590struct super_operations {
1591        struct inode *(*alloc_inode)(struct super_block *sb);
1592        void (*destroy_inode)(struct inode *);
1593
1594        void (*dirty_inode) (struct inode *, int flags);
1595        int (*write_inode) (struct inode *, struct writeback_control *wbc);
1596        int (*drop_inode) (struct inode *);
1597        void (*evict_inode) (struct inode *);
1598        void (*put_super) (struct super_block *);
1599        int (*sync_fs)(struct super_block *sb, int wait);
1600        int (*freeze_fs) (struct super_block *);
1601        int (*unfreeze_fs) (struct super_block *);
1602        int (*statfs) (struct dentry *, struct kstatfs *);
1603        int (*remount_fs) (struct super_block *, int *, char *);
1604        void (*umount_begin) (struct super_block *);
1605
1606        int (*show_options)(struct seq_file *, struct dentry *);
1607        int (*show_devname)(struct seq_file *, struct dentry *);
1608        int (*show_path)(struct seq_file *, struct dentry *);
1609        int (*show_stats)(struct seq_file *, struct dentry *);
1610#ifdef CONFIG_QUOTA
1611        ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1612        ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1613#endif
1614        int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1615        int (*nr_cached_objects)(struct super_block *);
1616        void (*free_cached_objects)(struct super_block *, int);
1617};
1618
1619/*
1620 * Inode flags - they have no relation to superblock flags now
1621 */
1622#define S_SYNC          1       /* Writes are synced at once */
1623#define S_NOATIME       2       /* Do not update access times */
1624#define S_APPEND        4       /* Append-only file */
1625#define S_IMMUTABLE     8       /* Immutable file */
1626#define S_DEAD          16      /* removed, but still open directory */
1627#define S_NOQUOTA       32      /* Inode is not counted to quota */
1628#define S_DIRSYNC       64      /* Directory modifications are synchronous */
1629#define S_NOCMTIME      128     /* Do not update file c/mtime */
1630#define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1631#define S_PRIVATE       512     /* Inode is fs-internal */
1632#define S_IMA           1024    /* Inode has an associated IMA struct */
1633#define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1634#define S_NOSEC         4096    /* no suid or xattr security attributes */
1635
1636/*
1637 * Note that nosuid etc flags are inode-specific: setting some file-system
1638 * flags just means all the inodes inherit those flags by default. It might be
1639 * possible to override it selectively if you really wanted to with some
1640 * ioctl() that is not currently implemented.
1641 *
1642 * Exception: MS_RDONLY is always applied to the entire file system.
1643 *
1644 * Unfortunately, it is possible to change a filesystems flags with it mounted
1645 * with files in use.  This means that all of the inodes will not have their
1646 * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1647 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1648 */
1649#define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1650
1651#define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1652#define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1653                                        ((inode)->i_flags & S_SYNC))
1654#define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1655                                        ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1656#define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1657#define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1658#define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1659
1660#define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1661#define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1662#define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1663#define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1664
1665#define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1666#define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1667#define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1668#define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1669#define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1670#define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1671#define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1672
1673/*
1674 * Inode state bits.  Protected by inode->i_lock
1675 *
1676 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1677 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1678 *
1679 * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1680 * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1681 * various stages of removing an inode.
1682 *
1683 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1684 *
1685 * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1686 *                      fdatasync().  i_atime is the usual cause.
1687 * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1688 *                      these changes separately from I_DIRTY_SYNC so that we
1689 *                      don't have to write inode on fdatasync() when only
1690 *                      mtime has changed in it.
1691 * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1692 * I_NEW                Serves as both a mutex and completion notification.
1693 *                      New inodes set I_NEW.  If two processes both create
1694 *                      the same inode, one of them will release its inode and
1695 *                      wait for I_NEW to be released before returning.
1696 *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1697 *                      also cause waiting on I_NEW, without I_NEW actually
1698 *                      being set.  find_inode() uses this to prevent returning
1699 *                      nearly-dead inodes.
1700 * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1701 *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1702 *                      cleared.
1703 * I_FREEING            Set when inode is about to be freed but still has dirty
1704 *                      pages or buffers attached or the inode itself is still
1705 *                      dirty.
1706 * I_CLEAR              Added by clear_inode().  In this state the inode is
1707 *                      clean and can be destroyed.  Inode keeps I_FREEING.
1708 *
1709 *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1710 *                      prohibited for many purposes.  iget() must wait for
1711 *                      the inode to be completely released, then create it
1712 *                      anew.  Other functions will just ignore such inodes,
1713 *                      if appropriate.  I_NEW is used for waiting.
1714 *
1715 * I_SYNC               Writeback of inode is running. The bit is set during
1716 *                      data writeback, and cleared with a wakeup on the bit
1717 *                      address once it is done. The bit is also used to pin
1718 *                      the inode in memory for flusher thread.
1719 *
1720 * I_REFERENCED         Marks the inode as recently references on the LRU list.
1721 *
1722 * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1723 *
1724 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1725 */
1726#define I_DIRTY_SYNC            (1 << 0)
1727#define I_DIRTY_DATASYNC        (1 << 1)
1728#define I_DIRTY_PAGES           (1 << 2)
1729#define __I_NEW                 3
1730#define I_NEW                   (1 << __I_NEW)
1731#define I_WILL_FREE             (1 << 4)
1732#define I_FREEING               (1 << 5)
1733#define I_CLEAR                 (1 << 6)
1734#define __I_SYNC                7
1735#define I_SYNC                  (1 << __I_SYNC)
1736#define I_REFERENCED            (1 << 8)
1737#define __I_DIO_WAKEUP          9
1738#define I_DIO_WAKEUP            (1 << I_DIO_WAKEUP)
1739
1740#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1741
1742extern void __mark_inode_dirty(struct inode *, int);
1743static inline void mark_inode_dirty(struct inode *inode)
1744{
1745        __mark_inode_dirty(inode, I_DIRTY);
1746}
1747
1748static inline void mark_inode_dirty_sync(struct inode *inode)
1749{
1750        __mark_inode_dirty(inode, I_DIRTY_SYNC);
1751}
1752
1753extern void inc_nlink(struct inode *inode);
1754extern void drop_nlink(struct inode *inode);
1755extern void clear_nlink(struct inode *inode);
1756extern void set_nlink(struct inode *inode, unsigned int nlink);
1757
1758static inline void inode_inc_link_count(struct inode *inode)
1759{
1760        inc_nlink(inode);
1761        mark_inode_dirty(inode);
1762}
1763
1764static inline void inode_dec_link_count(struct inode *inode)
1765{
1766        drop_nlink(inode);
1767        mark_inode_dirty(inode);
1768}
1769
1770/**
1771 * inode_inc_iversion - increments i_version
1772 * @inode: inode that need to be updated
1773 *
1774 * Every time the inode is modified, the i_version field will be incremented.
1775 * The filesystem has to be mounted with i_version flag
1776 */
1777
1778static inline void inode_inc_iversion(struct inode *inode)
1779{
1780       spin_lock(&inode->i_lock);
1781       inode->i_version++;
1782       spin_unlock(&inode->i_lock);
1783}
1784
1785enum file_time_flags {
1786        S_ATIME = 1,
1787        S_MTIME = 2,
1788        S_CTIME = 4,
1789        S_VERSION = 8,
1790};
1791
1792extern void touch_atime(struct path *);
1793static inline void file_accessed(struct file *file)
1794{
1795        if (!(file->f_flags & O_NOATIME))
1796                touch_atime(&file->f_path);
1797}
1798
1799int sync_inode(struct inode *inode, struct writeback_control *wbc);
1800int sync_inode_metadata(struct inode *inode, int wait);
1801
1802struct file_system_type {
1803        const char *name;
1804        int fs_flags;
1805#define FS_REQUIRES_DEV         1 
1806#define FS_BINARY_MOUNTDATA     2
1807#define FS_HAS_SUBTYPE          4
1808#define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
1809#define FS_USERNS_DEV_MOUNT     16 /* A userns mount does not imply MNT_NODEV */
1810#define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1811        struct dentry *(*mount) (struct file_system_type *, int,
1812                       const char *, void *);
1813        void (*kill_sb) (struct super_block *);
1814        struct module *owner;
1815        struct file_system_type * next;
1816        struct hlist_head fs_supers;
1817
1818        struct lock_class_key s_lock_key;
1819        struct lock_class_key s_umount_key;
1820        struct lock_class_key s_vfs_rename_key;
1821        struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1822
1823        struct lock_class_key i_lock_key;
1824        struct lock_class_key i_mutex_key;
1825        struct lock_class_key i_mutex_dir_key;
1826};
1827
1828#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1829
1830extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1831        void *data, int (*fill_super)(struct super_block *, void *, int));
1832extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1833        int flags, const char *dev_name, void *data,
1834        int (*fill_super)(struct super_block *, void *, int));
1835extern struct dentry *mount_single(struct file_system_type *fs_type,
1836        int flags, void *data,
1837        int (*fill_super)(struct super_block *, void *, int));
1838extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1839        int flags, void *data,
1840        int (*fill_super)(struct super_block *, void *, int));
1841extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1842void generic_shutdown_super(struct super_block *sb);
1843void kill_block_super(struct super_block *sb);
1844void kill_anon_super(struct super_block *sb);
1845void kill_litter_super(struct super_block *sb);
1846void deactivate_super(struct super_block *sb);
1847void deactivate_locked_super(struct super_block *sb);
1848int set_anon_super(struct super_block *s, void *data);
1849int get_anon_bdev(dev_t *);
1850void free_anon_bdev(dev_t);
1851struct super_block *sget(struct file_system_type *type,
1852                        int (*test)(struct super_block *,void *),
1853                        int (*set)(struct super_block *,void *),
1854                        int flags, void *data);
1855extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1856        const struct super_operations *ops,
1857        const struct dentry_operations *dops,
1858        unsigned long);
1859
1860/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1861#define fops_get(fops) \
1862        (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1863#define fops_put(fops) \
1864        do { if (fops) module_put((fops)->owner); } while(0)
1865
1866extern int register_filesystem(struct file_system_type *);
1867extern int unregister_filesystem(struct file_system_type *);
1868extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1869#define kern_mount(type) kern_mount_data(type, NULL)
1870extern void kern_unmount(struct vfsmount *mnt);
1871extern int may_umount_tree(struct vfsmount *);
1872extern int may_umount(struct vfsmount *);
1873extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1874extern struct vfsmount *collect_mounts(struct path *);
1875extern void drop_collected_mounts(struct vfsmount *);
1876extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1877                          struct vfsmount *);
1878extern int vfs_statfs(struct path *, struct kstatfs *);
1879extern int user_statfs(const char __user *, struct kstatfs *);
1880extern int fd_statfs(int, struct kstatfs *);
1881extern int vfs_ustat(dev_t, struct kstatfs *);
1882extern int freeze_super(struct super_block *super);
1883extern int thaw_super(struct super_block *super);
1884extern bool our_mnt(struct vfsmount *mnt);
1885
1886extern int current_umask(void);
1887
1888/* /sys/fs */
1889extern struct kobject *fs_kobj;
1890
1891#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1892extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1893
1894#define FLOCK_VERIFY_READ  1
1895#define FLOCK_VERIFY_WRITE 2
1896
1897#ifdef CONFIG_FILE_LOCKING
1898extern int locks_mandatory_locked(struct inode *);
1899extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1900
1901/*
1902 * Candidates for mandatory locking have the setgid bit set
1903 * but no group execute bit -  an otherwise meaningless combination.
1904 */
1905
1906static inline int __mandatory_lock(struct inode *ino)
1907{
1908        return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1909}
1910
1911/*
1912 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1913 * otherwise these will be advisory locks
1914 */
1915
1916static inline int mandatory_lock(struct inode *ino)
1917{
1918        return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1919}
1920
1921static inline int locks_verify_locked(struct inode *inode)
1922{
1923        if (mandatory_lock(inode))
1924                return locks_mandatory_locked(inode);
1925        return 0;
1926}
1927
1928static inline int locks_verify_truncate(struct inode *inode,
1929                                    struct file *filp,
1930                                    loff_t size)
1931{
1932        if (inode->i_flock && mandatory_lock(inode))
1933                return locks_mandatory_area(
1934                        FLOCK_VERIFY_WRITE, inode, filp,
1935                        size < inode->i_size ? size : inode->i_size,
1936                        (size < inode->i_size ? inode->i_size - size
1937                         : size - inode->i_size)
1938                );
1939        return 0;
1940}
1941
1942static inline int break_lease(struct inode *inode, unsigned int mode)
1943{
1944        if (inode->i_flock)
1945                return __break_lease(inode, mode);
1946        return 0;
1947}
1948#else /* !CONFIG_FILE_LOCKING */
1949static inline int locks_mandatory_locked(struct inode *inode)
1950{
1951        return 0;
1952}
1953
1954static inline int locks_mandatory_area(int rw, struct inode *inode,
1955                                       struct file *filp, loff_t offset,
1956                                       size_t count)
1957{
1958        return 0;
1959}
1960
1961static inline int __mandatory_lock(struct inode *inode)
1962{
1963        return 0;
1964}
1965
1966static inline int mandatory_lock(struct inode *inode)
1967{
1968        return 0;
1969}
1970
1971static inline int locks_verify_locked(struct inode *inode)
1972{
1973        return 0;
1974}
1975
1976static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1977                                        size_t size)
1978{
1979        return 0;
1980}
1981
1982static inline int break_lease(struct inode *inode, unsigned int mode)
1983{
1984        return 0;
1985}
1986
1987#endif /* CONFIG_FILE_LOCKING */
1988
1989/* fs/open.c */
1990struct audit_names;
1991struct filename {
1992        const char              *name;  /* pointer to actual string */
1993        const __user char       *uptr;  /* original userland pointer */
1994        struct audit_names      *aname;
1995        bool                    separate; /* should "name" be freed? */
1996};
1997
1998extern long vfs_truncate(struct path *, loff_t);
1999extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2000                       struct file *filp);
2001extern int do_fallocate(struct file *file, int mode, loff_t offset,
2002                        loff_t len);
2003extern long do_sys_open(int dfd, const char __user *filename, int flags,
2004                        umode_t mode);
2005extern struct file *file_open_name(struct filename *, int, umode_t);
2006extern struct file *filp_open(const char *, int, umode_t);
2007extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2008                                   const char *, int);
2009extern struct file * dentry_open(const struct path *, int, const struct cred *);
2010extern int filp_close(struct file *, fl_owner_t id);
2011
2012extern struct filename *getname(const char __user *);
2013
2014enum {
2015        FILE_CREATED = 1,
2016        FILE_OPENED = 2
2017};
2018extern int finish_open(struct file *file, struct dentry *dentry,
2019                        int (*open)(struct inode *, struct file *),
2020                        int *opened);
2021extern int finish_no_open(struct file *file, struct dentry *dentry);
2022
2023/* fs/ioctl.c */
2024
2025extern int ioctl_preallocate(struct file *filp, void __user *argp);
2026
2027/* fs/dcache.c */
2028extern void __init vfs_caches_init_early(void);
2029extern void __init vfs_caches_init(unsigned long);
2030
2031extern struct kmem_cache *names_cachep;
2032
2033extern void final_putname(struct filename *name);
2034
2035#define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2036#define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2037#ifndef CONFIG_AUDITSYSCALL
2038#define putname(name)           final_putname(name)
2039#else
2040extern void putname(struct filename *name);
2041#endif
2042
2043#ifdef CONFIG_BLOCK
2044extern int register_blkdev(unsigned int, const char *);
2045extern void unregister_blkdev(unsigned int, const char *);
2046extern struct block_device *bdget(dev_t);
2047extern struct block_device *bdgrab(struct block_device *bdev);
2048extern void bd_set_size(struct block_device *, loff_t size);
2049extern void bd_forget(struct inode *inode);
2050extern void bdput(struct block_device *);
2051extern void invalidate_bdev(struct block_device *);
2052extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2053extern int sync_blockdev(struct block_device *bdev);
2054extern void kill_bdev(struct block_device *);
2055extern struct super_block *freeze_bdev(struct block_device *);
2056extern void emergency_thaw_all(void);
2057extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2058extern int fsync_bdev(struct block_device *);
2059#else
2060static inline void bd_forget(struct inode *inode) {}
2061static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2062static inline void kill_bdev(struct block_device *bdev) {}
2063static inline void invalidate_bdev(struct block_device *bdev) {}
2064
2065static inline struct super_block *freeze_bdev(struct block_device *sb)
2066{
2067        return NULL;
2068}
2069
2070static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2071{
2072        return 0;
2073}
2074
2075static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2076{
2077}
2078#endif
2079extern int sync_filesystem(struct super_block *);
2080extern const struct file_operations def_blk_fops;
2081extern const struct file_operations def_chr_fops;
2082extern const struct file_operations bad_sock_fops;
2083extern const struct file_operations def_fifo_fops;
2084#ifdef CONFIG_BLOCK
2085extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2086extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2087extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2088extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2089extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2090                                               void *holder);
2091extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2092                                              void *holder);
2093extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2094#ifdef CONFIG_SYSFS
2095extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2096extern void bd_unlink_disk_holder(struct block_device *bdev,
2097                                  struct gendisk *disk);
2098#else
2099static inline int bd_link_disk_holder(struct block_device *bdev,
2100                                      struct gendisk *disk)
2101{
2102        return 0;
2103}
2104static inline void bd_unlink_disk_holder(struct block_device *bdev,
2105                                         struct gendisk *disk)
2106{
2107}
2108#endif
2109#endif
2110
2111/* fs/char_dev.c */
2112#define CHRDEV_MAJOR_HASH_SIZE  255
2113extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2114extern int register_chrdev_region(dev_t, unsigned, const char *);
2115extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2116                             unsigned int count, const char *name,
2117                             const struct file_operations *fops);
2118extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2119                                unsigned int count, const char *name);
2120extern void unregister_chrdev_region(dev_t, unsigned);
2121extern void chrdev_show(struct seq_file *,off_t);
2122
2123static inline int register_chrdev(unsigned int major, const char *name,
2124                                  const struct file_operations *fops)
2125{
2126        return __register_chrdev(major, 0, 256, name, fops);
2127}
2128
2129static inline void unregister_chrdev(unsigned int major, const char *name)
2130{
2131        __unregister_chrdev(major, 0, 256, name);
2132}
2133
2134/* fs/block_dev.c */
2135#define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2136#define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2137
2138#ifdef CONFIG_BLOCK
2139#define BLKDEV_MAJOR_HASH_SIZE  255
2140extern const char *__bdevname(dev_t, char *buffer);
2141extern const char *bdevname(struct block_device *bdev, char *buffer);
2142extern struct block_device *lookup_bdev(const char *);
2143extern void blkdev_show(struct seq_file *,off_t);
2144
2145#else
2146#define BLKDEV_MAJOR_HASH_SIZE  0
2147#endif
2148
2149extern void init_special_inode(struct inode *, umode_t, dev_t);
2150
2151/* Invalid inode operations -- fs/bad_inode.c */
2152extern void make_bad_inode(struct inode *);
2153extern int is_bad_inode(struct inode *);
2154
2155extern const struct file_operations read_pipefifo_fops;
2156extern const struct file_operations write_pipefifo_fops;
2157extern const struct file_operations rdwr_pipefifo_fops;
2158
2159#ifdef CONFIG_BLOCK
2160/*
2161 * return READ, READA, or WRITE
2162 */
2163#define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2164
2165/*
2166 * return data direction, READ or WRITE
2167 */
2168#define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2169
2170extern void check_disk_size_change(struct gendisk *disk,
2171                                   struct block_device *bdev);
2172extern int revalidate_disk(struct gendisk *);
2173extern int check_disk_change(struct block_device *);
2174extern int __invalidate_device(struct block_device *, bool);
2175extern int invalidate_partition(struct gendisk *, int);
2176#endif
2177unsigned long invalidate_mapping_pages(struct address_space *mapping,
2178                                        pgoff_t start, pgoff_t end);
2179
2180static inline void invalidate_remote_inode(struct inode *inode)
2181{
2182        if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2183            S_ISLNK(inode->i_mode))
2184                invalidate_mapping_pages(inode->i_mapping, 0, -1);
2185}
2186extern int invalidate_inode_pages2(struct address_space *mapping);
2187extern int invalidate_inode_pages2_range(struct address_space *mapping,
2188                                         pgoff_t start, pgoff_t end);
2189extern int write_inode_now(struct inode *, int);
2190extern int filemap_fdatawrite(struct address_space *);
2191extern int filemap_flush(struct address_space *);
2192extern int filemap_fdatawait(struct address_space *);
2193extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2194                                   loff_t lend);
2195extern int filemap_write_and_wait(struct address_space *mapping);
2196extern int filemap_write_and_wait_range(struct address_space *mapping,
2197                                        loff_t lstart, loff_t lend);
2198extern int __filemap_fdatawrite_range(struct address_space *mapping,
2199                                loff_t start, loff_t end, int sync_mode);
2200extern int filemap_fdatawrite_range(struct address_space *mapping,
2201                                loff_t start, loff_t end);
2202
2203extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2204                           int datasync);
2205extern int vfs_fsync(struct file *file, int datasync);
2206extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2207extern void emergency_sync(void);
2208extern void emergency_remount(void);
2209#ifdef CONFIG_BLOCK
2210extern sector_t bmap(struct inode *, sector_t);
2211#endif
2212extern int notify_change(struct dentry *, struct iattr *);
2213extern int inode_permission(struct inode *, int);
2214extern int generic_permission(struct inode *, int);
2215
2216static inline bool execute_ok(struct inode *inode)
2217{
2218        return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2219}
2220
2221static inline struct inode *file_inode(struct file *f)
2222{
2223        return f->f_inode;
2224}
2225
2226/*
2227 * get_write_access() gets write permission for a file.
2228 * put_write_access() releases this write permission.
2229 * This is used for regular files.
2230 * We cannot support write (and maybe mmap read-write shared) accesses and
2231 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2232 * can have the following values:
2233 * 0: no writers, no VM_DENYWRITE mappings
2234 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2235 * > 0: (i_writecount) users are writing to the file.
2236 *
2237 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2238 * except for the cases where we don't hold i_writecount yet. Then we need to
2239 * use {get,deny}_write_access() - these functions check the sign and refuse
2240 * to do the change if sign is wrong.
2241 */
2242static inline int get_write_access(struct inode *inode)
2243{
2244        return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2245}
2246static inline int deny_write_access(struct file *file)
2247{
2248        struct inode *inode = file_inode(file);
2249        return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2250}
2251static inline void put_write_access(struct inode * inode)
2252{
2253        atomic_dec(&inode->i_writecount);
2254}
2255static inline void allow_write_access(struct file *file)
2256{
2257        if (file)
2258                atomic_inc(&file_inode(file)->i_writecount);
2259}
2260#ifdef CONFIG_IMA
2261static inline void i_readcount_dec(struct inode *inode)
2262{
2263        BUG_ON(!atomic_read(&inode->i_readcount));
2264        atomic_dec(&inode->i_readcount);
2265}
2266static inline void i_readcount_inc(struct inode *inode)
2267{
2268        atomic_inc(&inode->i_readcount);
2269}
2270#else
2271static inline void i_readcount_dec(struct inode *inode)
2272{
2273        return;
2274}
2275static inline void i_readcount_inc(struct inode *inode)
2276{
2277        return;
2278}
2279#endif
2280extern int do_pipe_flags(int *, int);
2281
2282extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2283extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2284extern struct file * open_exec(const char *);
2285 
2286/* fs/dcache.c -- generic fs support functions */
2287extern int is_subdir(struct dentry *, struct dentry *);
2288extern int path_is_under(struct path *, struct path *);
2289extern ino_t find_inode_number(struct dentry *, struct qstr *);
2290
2291#include <linux/err.h>
2292
2293/* needed for stackable file system support */
2294extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2295
2296extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2297
2298extern int inode_init_always(struct super_block *, struct inode *);
2299extern void inode_init_once(struct inode *);
2300extern void address_space_init_once(struct address_space *mapping);
2301extern void ihold(struct inode * inode);
2302extern void iput(struct inode *);
2303extern struct inode * igrab(struct inode *);
2304extern ino_t iunique(struct super_block *, ino_t);
2305extern int inode_needs_sync(struct inode *inode);
2306extern int generic_delete_inode(struct inode *inode);
2307static inline int generic_drop_inode(struct inode *inode)
2308{
2309        return !inode->i_nlink || inode_unhashed(inode);
2310}
2311
2312extern struct inode *ilookup5_nowait(struct super_block *sb,
2313                unsigned long hashval, int (*test)(struct inode *, void *),
2314                void *data);
2315extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2316                int (*test)(struct inode *, void *), void *data);
2317extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2318
2319extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2320extern struct inode * iget_locked(struct super_block *, unsigned long);
2321extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2322extern int insert_inode_locked(struct inode *);
2323#ifdef CONFIG_DEBUG_LOCK_ALLOC
2324extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2325#else
2326static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2327#endif
2328extern void unlock_new_inode(struct inode *);
2329extern unsigned int get_next_ino(void);
2330
2331extern void __iget(struct inode * inode);
2332extern void iget_failed(struct inode *);
2333extern void clear_inode(struct inode *);
2334extern void __destroy_inode(struct inode *);
2335extern struct inode *new_inode_pseudo(struct super_block *sb);
2336extern struct inode *new_inode(struct super_block *sb);
2337extern void free_inode_nonrcu(struct inode *inode);
2338extern int should_remove_suid(struct dentry *);
2339extern int file_remove_suid(struct file *);
2340
2341extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2342static inline void insert_inode_hash(struct inode *inode)
2343{
2344        __insert_inode_hash(inode, inode->i_ino);
2345}
2346
2347extern void __remove_inode_hash(struct inode *);
2348static inline void remove_inode_hash(struct inode *inode)
2349{
2350        if (!inode_unhashed(inode))
2351                __remove_inode_hash(inode);
2352}
2353
2354extern void inode_sb_list_add(struct inode *inode);
2355
2356#ifdef CONFIG_BLOCK
2357extern void submit_bio(int, struct bio *);
2358extern int bdev_read_only(struct block_device *);
2359#endif
2360extern int set_blocksize(struct block_device *, int);
2361extern int sb_set_blocksize(struct super_block *, int);
2362extern int sb_min_blocksize(struct super_block *, int);
2363
2364extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2365extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2366extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2367                unsigned long size, pgoff_t pgoff);
2368extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2369int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2370extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2371extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2372                loff_t *);
2373extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2374extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2375                unsigned long *, loff_t, loff_t *, size_t, size_t);
2376extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2377                unsigned long, loff_t, loff_t *, size_t, ssize_t);
2378extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2379extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2380extern int generic_segment_checks(const struct iovec *iov,
2381                unsigned long *nr_segs, size_t *count, int access_flags);
2382
2383/* fs/block_dev.c */
2384extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2385                                unsigned long nr_segs, loff_t pos);
2386extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2387                        int datasync);
2388extern void block_sync_page(struct page *page);
2389
2390/* fs/splice.c */
2391extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2392                struct pipe_inode_info *, size_t, unsigned int);
2393extern ssize_t default_file_splice_read(struct file *, loff_t *,
2394                struct pipe_inode_info *, size_t, unsigned int);
2395extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2396                struct file *, loff_t *, size_t, unsigned int);
2397extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2398                struct file *out, loff_t *, size_t len, unsigned int flags);
2399extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2400                size_t len, unsigned int flags);
2401
2402extern void
2403file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2404extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2405extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2406extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2407extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2408                int whence, loff_t maxsize, loff_t eof);
2409extern int generic_file_open(struct inode * inode, struct file * filp);
2410extern int nonseekable_open(struct inode * inode, struct file * filp);
2411
2412#ifdef CONFIG_FS_XIP
2413extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2414                             loff_t *ppos);
2415extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2416extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2417                              size_t len, loff_t *ppos);
2418extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2419#else
2420static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2421{
2422        return 0;
2423}
2424#endif
2425
2426#ifdef CONFIG_BLOCK
2427typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2428                            loff_t file_offset);
2429
2430enum {
2431        /* need locking between buffered and direct access */
2432        DIO_LOCKING     = 0x01,
2433
2434        /* filesystem does not support filling holes */
2435        DIO_SKIP_HOLES  = 0x02,
2436};
2437
2438void dio_end_io(struct bio *bio, int error);
2439
2440ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2441        struct block_device *bdev, const struct iovec *iov, loff_t offset,
2442        unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2443        dio_submit_t submit_io, int flags);
2444
2445static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2446                struct inode *inode, const struct iovec *iov, loff_t offset,
2447                unsigned long nr_segs, get_block_t get_block)
2448{
2449        return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2450                                    offset, nr_segs, get_block, NULL, NULL,
2451                                    DIO_LOCKING | DIO_SKIP_HOLES);
2452}
2453#endif
2454
2455void inode_dio_wait(struct inode *inode);
2456void inode_dio_done(struct inode *inode);
2457
2458extern const struct file_operations generic_ro_fops;
2459
2460#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2461
2462extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2463extern int vfs_follow_link(struct nameidata *, const char *);
2464extern int page_readlink(struct dentry *, char __user *, int);
2465extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2466extern void page_put_link(struct dentry *, struct nameidata *, void *);
2467extern int __page_symlink(struct inode *inode, const char *symname, int len,
2468                int nofs);
2469extern int page_symlink(struct inode *inode, const char *symname, int len);
2470extern const struct inode_operations page_symlink_inode_operations;
2471extern int generic_readlink(struct dentry *, char __user *, int);
2472extern void generic_fillattr(struct inode *, struct kstat *);
2473extern int vfs_getattr(struct path *, struct kstat *);
2474void __inode_add_bytes(struct inode *inode, loff_t bytes);
2475void inode_add_bytes(struct inode *inode, loff_t bytes);
2476void inode_sub_bytes(struct inode *inode, loff_t bytes);
2477loff_t inode_get_bytes(struct inode *inode);
2478void inode_set_bytes(struct inode *inode, loff_t bytes);
2479
2480extern int vfs_readdir(struct file *, filldir_t, void *);
2481
2482extern int vfs_stat(const char __user *, struct kstat *);
2483extern int vfs_lstat(const char __user *, struct kstat *);
2484extern int vfs_fstat(unsigned int, struct kstat *);
2485extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2486
2487extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2488                    unsigned long arg);
2489extern int __generic_block_fiemap(struct inode *inode,
2490                                  struct fiemap_extent_info *fieinfo,
2491                                  loff_t start, loff_t len,
2492                                  get_block_t *get_block);
2493extern int generic_block_fiemap(struct inode *inode,
2494                                struct fiemap_extent_info *fieinfo, u64 start,
2495                                u64 len, get_block_t *get_block);
2496
2497extern void get_filesystem(struct file_system_type *fs);
2498extern void put_filesystem(struct file_system_type *fs);
2499extern struct file_system_type *get_fs_type(const char *name);
2500extern struct super_block *get_super(struct block_device *);
2501extern struct super_block *get_super_thawed(struct block_device *);
2502extern struct super_block *get_active_super(struct block_device *bdev);
2503extern void drop_super(struct super_block *sb);
2504extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2505extern void iterate_supers_type(struct file_system_type *,
2506                                void (*)(struct super_block *, void *), void *);
2507
2508extern int dcache_dir_open(struct inode *, struct file *);
2509extern int dcache_dir_close(struct inode *, struct file *);
2510extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2511extern int dcache_readdir(struct file *, void *, filldir_t);
2512extern int simple_setattr(struct dentry *, struct iattr *);
2513extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2514extern int simple_statfs(struct dentry *, struct kstatfs *);
2515extern int simple_open(struct inode *inode, struct file *file);
2516extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2517extern int simple_unlink(struct inode *, struct dentry *);
2518extern int simple_rmdir(struct inode *, struct dentry *);
2519extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2520extern int noop_fsync(struct file *, loff_t, loff_t, int);
2521extern int simple_empty(struct dentry *);
2522extern int simple_readpage(struct file *file, struct page *page);
2523extern int simple_write_begin(struct file *file, struct address_space *mapping,
2524                        loff_t pos, unsigned len, unsigned flags,
2525                        struct page **pagep, void **fsdata);
2526extern int simple_write_end(struct file *file, struct address_space *mapping,
2527                        loff_t pos, unsigned len, unsigned copied,
2528                        struct page *page, void *fsdata);
2529
2530extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2531extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2532extern const struct file_operations simple_dir_operations;
2533extern const struct inode_operations simple_dir_inode_operations;
2534struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2535struct dentry *d_alloc_name(struct dentry *, const char *);
2536extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2537extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2538extern void simple_release_fs(struct vfsmount **mount, int *count);
2539
2540extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2541                        loff_t *ppos, const void *from, size_t available);
2542extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2543                const void __user *from, size_t count);
2544
2545extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2546
2547extern int generic_check_addressable(unsigned, u64);
2548
2549#ifdef CONFIG_MIGRATION
2550extern int buffer_migrate_page(struct address_space *,
2551                                struct page *, struct page *,
2552                                enum migrate_mode);
2553#else
2554#define buffer_migrate_page NULL
2555#endif
2556
2557extern int inode_change_ok(const struct inode *, struct iattr *);
2558extern int inode_newsize_ok(const struct inode *, loff_t offset);
2559extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2560
2561extern int file_update_time(struct file *file);
2562
2563extern int generic_show_options(struct seq_file *m, struct dentry *root);
2564extern void save_mount_options(struct super_block *sb, char *options);
2565extern void replace_mount_options(struct super_block *sb, char *options);
2566
2567static inline ino_t parent_ino(struct dentry *dentry)
2568{
2569        ino_t res;
2570
2571        /*
2572         * Don't strictly need d_lock here? If the parent ino could change
2573         * then surely we'd have a deeper race in the caller?
2574         */
2575        spin_lock(&dentry->d_lock);
2576        res = dentry->d_parent->d_inode->i_ino;
2577        spin_unlock(&dentry->d_lock);
2578        return res;
2579}
2580
2581/* Transaction based IO helpers */
2582
2583/*
2584 * An argresp is stored in an allocated page and holds the
2585 * size of the argument or response, along with its content
2586 */
2587struct simple_transaction_argresp {
2588        ssize_t size;
2589        char data[0];
2590};
2591
2592#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2593
2594char *simple_transaction_get(struct file *file, const char __user *buf,
2595                                size_t size);
2596ssize_t simple_transaction_read(struct file *file, char __user *buf,
2597                                size_t size, loff_t *pos);
2598int simple_transaction_release(struct inode *inode, struct file *file);
2599
2600void simple_transaction_set(struct file *file, size_t n);
2601
2602/*
2603 * simple attribute files
2604 *
2605 * These attributes behave similar to those in sysfs:
2606 *
2607 * Writing to an attribute immediately sets a value, an open file can be
2608 * written to multiple times.
2609 *
2610 * Reading from an attribute creates a buffer from the value that might get
2611 * read with multiple read calls. When the attribute has been read
2612 * completely, no further read calls are possible until the file is opened
2613 * again.
2614 *
2615 * All attributes contain a text representation of a numeric value
2616 * that are accessed with the get() and set() functions.
2617 */
2618#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2619static int __fops ## _open(struct inode *inode, struct file *file)      \
2620{                                                                       \
2621        __simple_attr_check_format(__fmt, 0ull);                        \
2622        return simple_attr_open(inode, file, __get, __set, __fmt);      \
2623}                                                                       \
2624static const struct file_operations __fops = {                          \
2625        .owner   = THIS_MODULE,                                         \
2626        .open    = __fops ## _open,                                     \
2627        .release = simple_attr_release,                                 \
2628        .read    = simple_attr_read,                                    \
2629        .write   = simple_attr_write,                                   \
2630        .llseek  = generic_file_llseek,                                 \
2631};
2632
2633static inline __printf(1, 2)
2634void __simple_attr_check_format(const char *fmt, ...)
2635{
2636        /* don't do anything, just let the compiler check the arguments; */
2637}
2638
2639int simple_attr_open(struct inode *inode, struct file *file,
2640                     int (*get)(void *, u64 *), int (*set)(void *, u64),
2641                     const char *fmt);
2642int simple_attr_release(struct inode *inode, struct file *file);
2643ssize_t simple_attr_read(struct file *file, char __user *buf,
2644                         size_t len, loff_t *ppos);
2645ssize_t simple_attr_write(struct file *file, const char __user *buf,
2646                          size_t len, loff_t *ppos);
2647
2648struct ctl_table;
2649int proc_nr_files(struct ctl_table *table, int write,
2650                  void __user *buffer, size_t *lenp, loff_t *ppos);
2651int proc_nr_dentry(struct ctl_table *table, int write,
2652                  void __user *buffer, size_t *lenp, loff_t *ppos);
2653int proc_nr_inodes(struct ctl_table *table, int write,
2654                   void __user *buffer, size_t *lenp, loff_t *ppos);
2655int __init get_filesystem_list(char *buf);
2656
2657#define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2658#define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2659
2660#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2661#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2662                                            (flag & __FMODE_NONOTIFY)))
2663
2664static inline int is_sxid(umode_t mode)
2665{
2666        return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2667}
2668
2669static inline void inode_has_no_xattr(struct inode *inode)
2670{
2671        if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2672                inode->i_flags |= S_NOSEC;
2673}
2674
2675#endif /* _LINUX_FS_H */
2676