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/list_lru.h>
  14#include <linux/llist.h>
  15#include <linux/radix-tree.h>
  16#include <linux/rbtree.h>
  17#include <linux/init.h>
  18#include <linux/pid.h>
  19#include <linux/bug.h>
  20#include <linux/mutex.h>
  21#include <linux/capability.h>
  22#include <linux/semaphore.h>
  23#include <linux/fiemap.h>
  24#include <linux/rculist_bl.h>
  25#include <linux/atomic.h>
  26#include <linux/shrinker.h>
  27#include <linux/migrate_mode.h>
  28#include <linux/uidgid.h>
  29#include <linux/lockdep.h>
  30#include <linux/percpu-rwsem.h>
  31#include <linux/blk_types.h>
  32
  33#include <asm/byteorder.h>
  34#include <uapi/linux/fs.h>
  35
  36struct export_operations;
  37struct hd_geometry;
  38struct iovec;
  39struct nameidata;
  40struct kiocb;
  41struct kobject;
  42struct pipe_inode_info;
  43struct poll_table_struct;
  44struct kstatfs;
  45struct vm_area_struct;
  46struct vfsmount;
  47struct cred;
  48struct swap_info_struct;
  49struct seq_file;
  50struct workqueue_struct;
  51
  52extern void __init inode_init(void);
  53extern void __init inode_init_early(void);
  54extern void __init files_init(unsigned long);
  55
  56extern struct files_stat_struct files_stat;
  57extern unsigned long get_max_files(void);
  58extern int sysctl_nr_open;
  59extern struct inodes_stat_t inodes_stat;
  60extern int leases_enable, lease_break_time;
  61extern int sysctl_protected_symlinks;
  62extern int sysctl_protected_hardlinks;
  63
  64struct buffer_head;
  65typedef int (get_block_t)(struct inode *inode, sector_t iblock,
  66                        struct buffer_head *bh_result, int create);
  67typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
  68                        ssize_t bytes, void *private);
  69
  70#define MAY_EXEC                0x00000001
  71#define MAY_WRITE               0x00000002
  72#define MAY_READ                0x00000004
  73#define MAY_APPEND              0x00000008
  74#define MAY_ACCESS              0x00000010
  75#define MAY_OPEN                0x00000020
  76#define MAY_CHDIR               0x00000040
  77/* called from RCU mode, don't block */
  78#define MAY_NOT_BLOCK           0x00000080
  79
  80/*
  81 * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
  82 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
  83 */
  84
  85/* file is open for reading */
  86#define FMODE_READ              ((__force fmode_t)0x1)
  87/* file is open for writing */
  88#define FMODE_WRITE             ((__force fmode_t)0x2)
  89/* file is seekable */
  90#define FMODE_LSEEK             ((__force fmode_t)0x4)
  91/* file can be accessed using pread */
  92#define FMODE_PREAD             ((__force fmode_t)0x8)
  93/* file can be accessed using pwrite */
  94#define FMODE_PWRITE            ((__force fmode_t)0x10)
  95/* File is opened for execution with sys_execve / sys_uselib */
  96#define FMODE_EXEC              ((__force fmode_t)0x20)
  97/* File is opened with O_NDELAY (only set for block devices) */
  98#define FMODE_NDELAY            ((__force fmode_t)0x40)
  99/* File is opened with O_EXCL (only set for block devices) */
 100#define FMODE_EXCL              ((__force fmode_t)0x80)
 101/* File is opened using open(.., 3, ..) and is writeable only for ioctls
 102   (specialy hack for floppy.c) */
 103#define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
 104/* 32bit hashes as llseek() offset (for directories) */
 105#define FMODE_32BITHASH         ((__force fmode_t)0x200)
 106/* 64bit hashes as llseek() offset (for directories) */
 107#define FMODE_64BITHASH         ((__force fmode_t)0x400)
 108
 109/*
 110 * Don't update ctime and mtime.
 111 *
 112 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
 113 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
 114 */
 115#define FMODE_NOCMTIME          ((__force fmode_t)0x800)
 116
 117/* Expect random access pattern */
 118#define FMODE_RANDOM            ((__force fmode_t)0x1000)
 119
 120/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
 121#define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
 122
 123/* File is opened with O_PATH; almost nothing can be done with it */
 124#define FMODE_PATH              ((__force fmode_t)0x4000)
 125
 126/* File needs atomic accesses to f_pos */
 127#define FMODE_ATOMIC_POS        ((__force fmode_t)0x8000)
 128
 129/* File was opened by fanotify and shouldn't generate fanotify events */
 130#define FMODE_NONOTIFY          ((__force fmode_t)0x1000000)
 131
 132/*
 133 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
 134 * that indicates that they should check the contents of the iovec are
 135 * valid, but not check the memory that the iovec elements
 136 * points too.
 137 */
 138#define CHECK_IOVEC_ONLY -1
 139
 140/*
 141 * The below are the various read and write types that we support. Some of
 142 * them include behavioral modifiers that send information down to the
 143 * block layer and IO scheduler. Terminology:
 144 *
 145 *      The block layer uses device plugging to defer IO a little bit, in
 146 *      the hope that we will see more IO very shortly. This increases
 147 *      coalescing of adjacent IO and thus reduces the number of IOs we
 148 *      have to send to the device. It also allows for better queuing,
 149 *      if the IO isn't mergeable. If the caller is going to be waiting
 150 *      for the IO, then he must ensure that the device is unplugged so
 151 *      that the IO is dispatched to the driver.
 152 *
 153 *      All IO is handled async in Linux. This is fine for background
 154 *      writes, but for reads or writes that someone waits for completion
 155 *      on, we want to notify the block layer and IO scheduler so that they
 156 *      know about it. That allows them to make better scheduling
 157 *      decisions. So when the below references 'sync' and 'async', it
 158 *      is referencing this priority hint.
 159 *
 160 * With that in mind, the available types are:
 161 *
 162 * READ                 A normal read operation. Device will be plugged.
 163 * READ_SYNC            A synchronous read. Device is not plugged, caller can
 164 *                      immediately wait on this read without caring about
 165 *                      unplugging.
 166 * READA                Used for read-ahead operations. Lower priority, and the
 167 *                      block layer could (in theory) choose to ignore this
 168 *                      request if it runs into resource problems.
 169 * WRITE                A normal async write. Device will be plugged.
 170 * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
 171 *                      the hint that someone will be waiting on this IO
 172 *                      shortly. The write equivalent of READ_SYNC.
 173 * WRITE_ODIRECT        Special case write for O_DIRECT only.
 174 * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
 175 * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
 176 *                      non-volatile media on completion.
 177 * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
 178 *                      by a cache flush and data is guaranteed to be on
 179 *                      non-volatile media on completion.
 180 *
 181 */
 182#define RW_MASK                 REQ_WRITE
 183#define RWA_MASK                REQ_RAHEAD
 184
 185#define READ                    0
 186#define WRITE                   RW_MASK
 187#define READA                   RWA_MASK
 188#define KERNEL_READ             (READ|REQ_KERNEL)
 189#define KERNEL_WRITE            (WRITE|REQ_KERNEL)
 190
 191#define READ_SYNC               (READ | REQ_SYNC)
 192#define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
 193#define WRITE_ODIRECT           (WRITE | REQ_SYNC)
 194#define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
 195#define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
 196#define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
 197
 198/*
 199 * Attribute flags.  These should be or-ed together to figure out what
 200 * has been changed!
 201 */
 202#define ATTR_MODE       (1 << 0)
 203#define ATTR_UID        (1 << 1)
 204#define ATTR_GID        (1 << 2)
 205#define ATTR_SIZE       (1 << 3)
 206#define ATTR_ATIME      (1 << 4)
 207#define ATTR_MTIME      (1 << 5)
 208#define ATTR_CTIME      (1 << 6)
 209#define ATTR_ATIME_SET  (1 << 7)
 210#define ATTR_MTIME_SET  (1 << 8)
 211#define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
 212#define ATTR_ATTR_FLAG  (1 << 10)
 213#define ATTR_KILL_SUID  (1 << 11)
 214#define ATTR_KILL_SGID  (1 << 12)
 215#define ATTR_FILE       (1 << 13)
 216#define ATTR_KILL_PRIV  (1 << 14)
 217#define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
 218#define ATTR_TIMES_SET  (1 << 16)
 219
 220/*
 221 * This is the Inode Attributes structure, used for notify_change().  It
 222 * uses the above definitions as flags, to know which values have changed.
 223 * Also, in this manner, a Filesystem can look at only the values it cares
 224 * about.  Basically, these are the attributes that the VFS layer can
 225 * request to change from the FS layer.
 226 *
 227 * Derek Atkins <warlord@MIT.EDU> 94-10-20
 228 */
 229struct iattr {
 230        unsigned int    ia_valid;
 231        umode_t         ia_mode;
 232        kuid_t          ia_uid;
 233        kgid_t          ia_gid;
 234        loff_t          ia_size;
 235        struct timespec ia_atime;
 236        struct timespec ia_mtime;
 237        struct timespec ia_ctime;
 238
 239        /*
 240         * Not an attribute, but an auxiliary info for filesystems wanting to
 241         * implement an ftruncate() like method.  NOTE: filesystem should
 242         * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
 243         */
 244        struct file     *ia_file;
 245};
 246
 247/*
 248 * Includes for diskquotas.
 249 */
 250#include <linux/quota.h>
 251
 252/** 
 253 * enum positive_aop_returns - aop return codes with specific semantics
 254 *
 255 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
 256 *                          completed, that the page is still locked, and
 257 *                          should be considered active.  The VM uses this hint
 258 *                          to return the page to the active list -- it won't
 259 *                          be a candidate for writeback again in the near
 260 *                          future.  Other callers must be careful to unlock
 261 *                          the page if they get this return.  Returned by
 262 *                          writepage(); 
 263 *
 264 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
 265 *                      unlocked it and the page might have been truncated.
 266 *                      The caller should back up to acquiring a new page and
 267 *                      trying again.  The aop will be taking reasonable
 268 *                      precautions not to livelock.  If the caller held a page
 269 *                      reference, it should drop it before retrying.  Returned
 270 *                      by readpage().
 271 *
 272 * address_space_operation functions return these large constants to indicate
 273 * special semantics to the caller.  These are much larger than the bytes in a
 274 * page to allow for functions that return the number of bytes operated on in a
 275 * given page.
 276 */
 277
 278enum positive_aop_returns {
 279        AOP_WRITEPAGE_ACTIVATE  = 0x80000,
 280        AOP_TRUNCATED_PAGE      = 0x80001,
 281};
 282
 283#define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
 284#define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
 285#define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
 286                                                * helper code (eg buffer layer)
 287                                                * to clear GFP_FS from alloc */
 288
 289/*
 290 * oh the beauties of C type declarations.
 291 */
 292struct page;
 293struct address_space;
 294struct writeback_control;
 295
 296struct iov_iter {
 297        const struct iovec *iov;
 298        unsigned long nr_segs;
 299        size_t iov_offset;
 300        size_t count;
 301};
 302
 303size_t iov_iter_copy_from_user_atomic(struct page *page,
 304                struct iov_iter *i, unsigned long offset, size_t bytes);
 305size_t iov_iter_copy_from_user(struct page *page,
 306                struct iov_iter *i, unsigned long offset, size_t bytes);
 307void iov_iter_advance(struct iov_iter *i, size_t bytes);
 308int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
 309size_t iov_iter_single_seg_count(const struct iov_iter *i);
 310
 311static inline void iov_iter_init(struct iov_iter *i,
 312                        const struct iovec *iov, unsigned long nr_segs,
 313                        size_t count, size_t written)
 314{
 315        i->iov = iov;
 316        i->nr_segs = nr_segs;
 317        i->iov_offset = 0;
 318        i->count = count + written;
 319
 320        iov_iter_advance(i, written);
 321}
 322
 323static inline size_t iov_iter_count(struct iov_iter *i)
 324{
 325        return i->count;
 326}
 327
 328/*
 329 * "descriptor" for what we're up to with a read.
 330 * This allows us to use the same read code yet
 331 * have multiple different users of the data that
 332 * we read from a file.
 333 *
 334 * The simplest case just copies the data to user
 335 * mode.
 336 */
 337typedef struct {
 338        size_t written;
 339        size_t count;
 340        union {
 341                char __user *buf;
 342                void *data;
 343        } arg;
 344        int error;
 345} read_descriptor_t;
 346
 347typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
 348                unsigned long, unsigned long);
 349
 350struct address_space_operations {
 351        int (*writepage)(struct page *page, struct writeback_control *wbc);
 352        int (*readpage)(struct file *, struct page *);
 353
 354        /* Write back some dirty pages from this mapping. */
 355        int (*writepages)(struct address_space *, struct writeback_control *);
 356
 357        /* Set a page dirty.  Return true if this dirtied it */
 358        int (*set_page_dirty)(struct page *page);
 359
 360        int (*readpages)(struct file *filp, struct address_space *mapping,
 361                        struct list_head *pages, unsigned nr_pages);
 362
 363        int (*write_begin)(struct file *, struct address_space *mapping,
 364                                loff_t pos, unsigned len, unsigned flags,
 365                                struct page **pagep, void **fsdata);
 366        int (*write_end)(struct file *, struct address_space *mapping,
 367                                loff_t pos, unsigned len, unsigned copied,
 368                                struct page *page, void *fsdata);
 369
 370        /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
 371        sector_t (*bmap)(struct address_space *, sector_t);
 372        void (*invalidatepage) (struct page *, unsigned int, unsigned int);
 373        int (*releasepage) (struct page *, gfp_t);
 374        void (*freepage)(struct page *);
 375        ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
 376                        loff_t offset, unsigned long nr_segs);
 377        int (*get_xip_mem)(struct address_space *, pgoff_t, int,
 378                                                void **, unsigned long *);
 379        /*
 380         * migrate the contents of a page to the specified target. If
 381         * migrate_mode is MIGRATE_ASYNC, it must not block.
 382         */
 383        int (*migratepage) (struct address_space *,
 384                        struct page *, struct page *, enum migrate_mode);
 385        int (*launder_page) (struct page *);
 386        int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
 387                                        unsigned long);
 388        void (*is_dirty_writeback) (struct page *, bool *, bool *);
 389        int (*error_remove_page)(struct address_space *, struct page *);
 390
 391        /* swapfile support */
 392        int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
 393                                sector_t *span);
 394        void (*swap_deactivate)(struct file *file);
 395};
 396
 397extern const struct address_space_operations empty_aops;
 398
 399/*
 400 * pagecache_write_begin/pagecache_write_end must be used by general code
 401 * to write into the pagecache.
 402 */
 403int pagecache_write_begin(struct file *, struct address_space *mapping,
 404                                loff_t pos, unsigned len, unsigned flags,
 405                                struct page **pagep, void **fsdata);
 406
 407int pagecache_write_end(struct file *, struct address_space *mapping,
 408                                loff_t pos, unsigned len, unsigned copied,
 409                                struct page *page, void *fsdata);
 410
 411struct backing_dev_info;
 412struct address_space {
 413        struct inode            *host;          /* owner: inode, block_device */
 414        struct radix_tree_root  page_tree;      /* radix tree of all pages */
 415        spinlock_t              tree_lock;      /* and lock protecting it */
 416        unsigned int            i_mmap_writable;/* count VM_SHARED mappings */
 417        struct rb_root          i_mmap;         /* tree of private and shared mappings */
 418        struct list_head        i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
 419        struct mutex            i_mmap_mutex;   /* protect tree, count, list */
 420        /* Protected by tree_lock together with the radix tree */
 421        unsigned long           nrpages;        /* number of total pages */
 422        pgoff_t                 writeback_index;/* writeback starts here */
 423        const struct address_space_operations *a_ops;   /* methods */
 424        unsigned long           flags;          /* error bits/gfp mask */
 425        struct backing_dev_info *backing_dev_info; /* device readahead, etc */
 426        spinlock_t              private_lock;   /* for use by the address_space */
 427        struct list_head        private_list;   /* ditto */
 428        void                    *private_data;  /* ditto */
 429} __attribute__((aligned(sizeof(long))));
 430        /*
 431         * On most architectures that alignment is already the case; but
 432         * must be enforced here for CRIS, to let the least significant bit
 433         * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
 434         */
 435struct request_queue;
 436
 437struct block_device {
 438        dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
 439        int                     bd_openers;
 440        struct inode *          bd_inode;       /* will die */
 441        struct super_block *    bd_super;
 442        struct mutex            bd_mutex;       /* open/close mutex */
 443        struct list_head        bd_inodes;
 444        void *                  bd_claiming;
 445        void *                  bd_holder;
 446        int                     bd_holders;
 447        bool                    bd_write_holder;
 448#ifdef CONFIG_SYSFS
 449        struct list_head        bd_holder_disks;
 450#endif
 451        struct block_device *   bd_contains;
 452        unsigned                bd_block_size;
 453        struct hd_struct *      bd_part;
 454        /* number of times partitions within this device have been opened. */
 455        unsigned                bd_part_count;
 456        int                     bd_invalidated;
 457        struct gendisk *        bd_disk;
 458        struct request_queue *  bd_queue;
 459        struct list_head        bd_list;
 460        /*
 461         * Private data.  You must have bd_claim'ed the block_device
 462         * to use this.  NOTE:  bd_claim allows an owner to claim
 463         * the same device multiple times, the owner must take special
 464         * care to not mess up bd_private for that case.
 465         */
 466        unsigned long           bd_private;
 467
 468        /* The counter of freeze processes */
 469        int                     bd_fsfreeze_count;
 470        /* Mutex for freeze */
 471        struct mutex            bd_fsfreeze_mutex;
 472};
 473
 474/*
 475 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
 476 * radix trees
 477 */
 478#define PAGECACHE_TAG_DIRTY     0
 479#define PAGECACHE_TAG_WRITEBACK 1
 480#define PAGECACHE_TAG_TOWRITE   2
 481
 482int mapping_tagged(struct address_space *mapping, int tag);
 483
 484/*
 485 * Might pages of this file be mapped into userspace?
 486 */
 487static inline int mapping_mapped(struct address_space *mapping)
 488{
 489        return  !RB_EMPTY_ROOT(&mapping->i_mmap) ||
 490                !list_empty(&mapping->i_mmap_nonlinear);
 491}
 492
 493/*
 494 * Might pages of this file have been modified in userspace?
 495 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
 496 * marks vma as VM_SHARED if it is shared, and the file was opened for
 497 * writing i.e. vma may be mprotected writable even if now readonly.
 498 */
 499static inline int mapping_writably_mapped(struct address_space *mapping)
 500{
 501        return mapping->i_mmap_writable != 0;
 502}
 503
 504/*
 505 * Use sequence counter to get consistent i_size on 32-bit processors.
 506 */
 507#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 508#include <linux/seqlock.h>
 509#define __NEED_I_SIZE_ORDERED
 510#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
 511#else
 512#define i_size_ordered_init(inode) do { } while (0)
 513#endif
 514
 515struct posix_acl;
 516#define ACL_NOT_CACHED ((void *)(-1))
 517
 518#define IOP_FASTPERM    0x0001
 519#define IOP_LOOKUP      0x0002
 520#define IOP_NOFOLLOW    0x0004
 521
 522/*
 523 * Keep mostly read-only and often accessed (especially for
 524 * the RCU path lookup and 'stat' data) fields at the beginning
 525 * of the 'struct inode'
 526 */
 527struct inode {
 528        umode_t                 i_mode;
 529        unsigned short          i_opflags;
 530        kuid_t                  i_uid;
 531        kgid_t                  i_gid;
 532        unsigned int            i_flags;
 533
 534#ifdef CONFIG_FS_POSIX_ACL
 535        struct posix_acl        *i_acl;
 536        struct posix_acl        *i_default_acl;
 537#endif
 538
 539        const struct inode_operations   *i_op;
 540        struct super_block      *i_sb;
 541        struct address_space    *i_mapping;
 542
 543#ifdef CONFIG_SECURITY
 544        void                    *i_security;
 545#endif
 546
 547        /* Stat data, not accessed from path walking */
 548        unsigned long           i_ino;
 549        /*
 550         * Filesystems may only read i_nlink directly.  They shall use the
 551         * following functions for modification:
 552         *
 553         *    (set|clear|inc|drop)_nlink
 554         *    inode_(inc|dec)_link_count
 555         */
 556        union {
 557                const unsigned int i_nlink;
 558                unsigned int __i_nlink;
 559        };
 560        dev_t                   i_rdev;
 561        loff_t                  i_size;
 562        struct timespec         i_atime;
 563        struct timespec         i_mtime;
 564        struct timespec         i_ctime;
 565        spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
 566        unsigned short          i_bytes;
 567        unsigned int            i_blkbits;
 568        blkcnt_t                i_blocks;
 569
 570#ifdef __NEED_I_SIZE_ORDERED
 571        seqcount_t              i_size_seqcount;
 572#endif
 573
 574        /* Misc */
 575        unsigned long           i_state;
 576        struct mutex            i_mutex;
 577
 578        unsigned long           dirtied_when;   /* jiffies of first dirtying */
 579
 580        struct hlist_node       i_hash;
 581        struct list_head        i_wb_list;      /* backing dev IO list */
 582        struct list_head        i_lru;          /* inode LRU list */
 583        struct list_head        i_sb_list;
 584        union {
 585                struct hlist_head       i_dentry;
 586                struct rcu_head         i_rcu;
 587        };
 588        u64                     i_version;
 589        atomic_t                i_count;
 590        atomic_t                i_dio_count;
 591        atomic_t                i_writecount;
 592        const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
 593        struct file_lock        *i_flock;
 594        struct address_space    i_data;
 595#ifdef CONFIG_QUOTA
 596        struct dquot            *i_dquot[MAXQUOTAS];
 597#endif
 598        struct list_head        i_devices;
 599        union {
 600                struct pipe_inode_info  *i_pipe;
 601                struct block_device     *i_bdev;
 602                struct cdev             *i_cdev;
 603        };
 604
 605        __u32                   i_generation;
 606
 607#ifdef CONFIG_FSNOTIFY
 608        __u32                   i_fsnotify_mask; /* all events this inode cares about */
 609        struct hlist_head       i_fsnotify_marks;
 610#endif
 611
 612#ifdef CONFIG_IMA
 613        atomic_t                i_readcount; /* struct files open RO */
 614#endif
 615        void                    *i_private; /* fs or device private pointer */
 616};
 617
 618static inline int inode_unhashed(struct inode *inode)
 619{
 620        return hlist_unhashed(&inode->i_hash);
 621}
 622
 623/*
 624 * inode->i_mutex nesting subclasses for the lock validator:
 625 *
 626 * 0: the object of the current VFS operation
 627 * 1: parent
 628 * 2: child/target
 629 * 3: xattr
 630 * 4: second non-directory
 631 * The last is for certain operations (such as rename) which lock two
 632 * non-directories at once.
 633 *
 634 * The locking order between these classes is
 635 * parent -> child -> normal -> xattr -> second non-directory
 636 */
 637enum inode_i_mutex_lock_class
 638{
 639        I_MUTEX_NORMAL,
 640        I_MUTEX_PARENT,
 641        I_MUTEX_CHILD,
 642        I_MUTEX_XATTR,
 643        I_MUTEX_NONDIR2
 644};
 645
 646void lock_two_nondirectories(struct inode *, struct inode*);
 647void unlock_two_nondirectories(struct inode *, struct inode*);
 648
 649/*
 650 * NOTE: in a 32bit arch with a preemptable kernel and
 651 * an UP compile the i_size_read/write must be atomic
 652 * with respect to the local cpu (unlike with preempt disabled),
 653 * but they don't need to be atomic with respect to other cpus like in
 654 * true SMP (so they need either to either locally disable irq around
 655 * the read or for example on x86 they can be still implemented as a
 656 * cmpxchg8b without the need of the lock prefix). For SMP compiles
 657 * and 64bit archs it makes no difference if preempt is enabled or not.
 658 */
 659static inline loff_t i_size_read(const struct inode *inode)
 660{
 661#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 662        loff_t i_size;
 663        unsigned int seq;
 664
 665        do {
 666                seq = read_seqcount_begin(&inode->i_size_seqcount);
 667                i_size = inode->i_size;
 668        } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
 669        return i_size;
 670#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 671        loff_t i_size;
 672
 673        preempt_disable();
 674        i_size = inode->i_size;
 675        preempt_enable();
 676        return i_size;
 677#else
 678        return inode->i_size;
 679#endif
 680}
 681
 682/*
 683 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
 684 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
 685 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
 686 */
 687static inline void i_size_write(struct inode *inode, loff_t i_size)
 688{
 689#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 690        preempt_disable();
 691        write_seqcount_begin(&inode->i_size_seqcount);
 692        inode->i_size = i_size;
 693        write_seqcount_end(&inode->i_size_seqcount);
 694        preempt_enable();
 695#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 696        preempt_disable();
 697        inode->i_size = i_size;
 698        preempt_enable();
 699#else
 700        inode->i_size = i_size;
 701#endif
 702}
 703
 704/* Helper functions so that in most cases filesystems will
 705 * not need to deal directly with kuid_t and kgid_t and can
 706 * instead deal with the raw numeric values that are stored
 707 * in the filesystem.
 708 */
 709static inline uid_t i_uid_read(const struct inode *inode)
 710{
 711        return from_kuid(&init_user_ns, inode->i_uid);
 712}
 713
 714static inline gid_t i_gid_read(const struct inode *inode)
 715{
 716        return from_kgid(&init_user_ns, inode->i_gid);
 717}
 718
 719static inline void i_uid_write(struct inode *inode, uid_t uid)
 720{
 721        inode->i_uid = make_kuid(&init_user_ns, uid);
 722}
 723
 724static inline void i_gid_write(struct inode *inode, gid_t gid)
 725{
 726        inode->i_gid = make_kgid(&init_user_ns, gid);
 727}
 728
 729static inline unsigned iminor(const struct inode *inode)
 730{
 731        return MINOR(inode->i_rdev);
 732}
 733
 734static inline unsigned imajor(const struct inode *inode)
 735{
 736        return MAJOR(inode->i_rdev);
 737}
 738
 739extern struct block_device *I_BDEV(struct inode *inode);
 740
 741struct fown_struct {
 742        rwlock_t lock;          /* protects pid, uid, euid fields */
 743        struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
 744        enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
 745        kuid_t uid, euid;       /* uid/euid of process setting the owner */
 746        int signum;             /* posix.1b rt signal to be delivered on IO */
 747};
 748
 749/*
 750 * Track a single file's readahead state
 751 */
 752struct file_ra_state {
 753        pgoff_t start;                  /* where readahead started */
 754        unsigned int size;              /* # of readahead pages */
 755        unsigned int async_size;        /* do asynchronous readahead when
 756                                           there are only # of pages ahead */
 757
 758        unsigned int ra_pages;          /* Maximum readahead window */
 759        unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
 760        loff_t prev_pos;                /* Cache last read() position */
 761};
 762
 763/*
 764 * Check if @index falls in the readahead windows.
 765 */
 766static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
 767{
 768        return (index >= ra->start &&
 769                index <  ra->start + ra->size);
 770}
 771
 772#define FILE_MNT_WRITE_TAKEN    1
 773#define FILE_MNT_WRITE_RELEASED 2
 774
 775struct file {
 776        union {
 777                struct llist_node       fu_llist;
 778                struct rcu_head         fu_rcuhead;
 779        } f_u;
 780        struct path             f_path;
 781#define f_dentry        f_path.dentry
 782        struct inode            *f_inode;       /* cached value */
 783        const struct file_operations    *f_op;
 784
 785        /*
 786         * Protects f_ep_links, f_flags.
 787         * Must not be taken from IRQ context.
 788         */
 789        spinlock_t              f_lock;
 790        atomic_long_t           f_count;
 791        unsigned int            f_flags;
 792        fmode_t                 f_mode;
 793        struct mutex            f_pos_lock;
 794        loff_t                  f_pos;
 795        struct fown_struct      f_owner;
 796        const struct cred       *f_cred;
 797        struct file_ra_state    f_ra;
 798
 799        u64                     f_version;
 800#ifdef CONFIG_SECURITY
 801        void                    *f_security;
 802#endif
 803        /* needed for tty driver, and maybe others */
 804        void                    *private_data;
 805
 806#ifdef CONFIG_EPOLL
 807        /* Used by fs/eventpoll.c to link all the hooks to this file */
 808        struct list_head        f_ep_links;
 809        struct list_head        f_tfile_llink;
 810#endif /* #ifdef CONFIG_EPOLL */
 811        struct address_space    *f_mapping;
 812#ifdef CONFIG_DEBUG_WRITECOUNT
 813        unsigned long f_mnt_write_state;
 814#endif
 815} __attribute__((aligned(4)));  /* lest something weird decides that 2 is OK */
 816
 817struct file_handle {
 818        __u32 handle_bytes;
 819        int handle_type;
 820        /* file identifier */
 821        unsigned char f_handle[0];
 822};
 823
 824static inline struct file *get_file(struct file *f)
 825{
 826        atomic_long_inc(&f->f_count);
 827        return f;
 828}
 829#define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
 830#define file_count(x)   atomic_long_read(&(x)->f_count)
 831
 832#ifdef CONFIG_DEBUG_WRITECOUNT
 833static inline void file_take_write(struct file *f)
 834{
 835        WARN_ON(f->f_mnt_write_state != 0);
 836        f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
 837}
 838static inline void file_release_write(struct file *f)
 839{
 840        f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
 841}
 842static inline void file_reset_write(struct file *f)
 843{
 844        f->f_mnt_write_state = 0;
 845}
 846static inline void file_check_state(struct file *f)
 847{
 848        /*
 849         * At this point, either both or neither of these bits
 850         * should be set.
 851         */
 852        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
 853        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
 854}
 855static inline int file_check_writeable(struct file *f)
 856{
 857        if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
 858                return 0;
 859        printk(KERN_WARNING "writeable file with no "
 860                            "mnt_want_write()\n");
 861        WARN_ON(1);
 862        return -EINVAL;
 863}
 864#else /* !CONFIG_DEBUG_WRITECOUNT */
 865static inline void file_take_write(struct file *filp) {}
 866static inline void file_release_write(struct file *filp) {}
 867static inline void file_reset_write(struct file *filp) {}
 868static inline void file_check_state(struct file *filp) {}
 869static inline int file_check_writeable(struct file *filp)
 870{
 871        return 0;
 872}
 873#endif /* CONFIG_DEBUG_WRITECOUNT */
 874
 875#define MAX_NON_LFS     ((1UL<<31) - 1)
 876
 877/* Page cache limit. The filesystems should put that into their s_maxbytes 
 878   limits, otherwise bad things can happen in VM. */ 
 879#if BITS_PER_LONG==32
 880#define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
 881#elif BITS_PER_LONG==64
 882#define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
 883#endif
 884
 885#define FL_POSIX        1
 886#define FL_FLOCK        2
 887#define FL_DELEG        4       /* NFSv4 delegation */
 888#define FL_ACCESS       8       /* not trying to lock, just looking */
 889#define FL_EXISTS       16      /* when unlocking, test for existence */
 890#define FL_LEASE        32      /* lease held on this file */
 891#define FL_CLOSE        64      /* unlock on close */
 892#define FL_SLEEP        128     /* A blocking lock */
 893#define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
 894#define FL_UNLOCK_PENDING       512 /* Lease is being broken */
 895
 896/*
 897 * Special return value from posix_lock_file() and vfs_lock_file() for
 898 * asynchronous locking.
 899 */
 900#define FILE_LOCK_DEFERRED 1
 901
 902/*
 903 * The POSIX file lock owner is determined by
 904 * the "struct files_struct" in the thread group
 905 * (or NULL for no owner - BSD locks).
 906 *
 907 * Lockd stuffs a "host" pointer into this.
 908 */
 909typedef struct files_struct *fl_owner_t;
 910
 911struct file_lock_operations {
 912        void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
 913        void (*fl_release_private)(struct file_lock *);
 914};
 915
 916struct lock_manager_operations {
 917        int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
 918        unsigned long (*lm_owner_key)(struct file_lock *);
 919        void (*lm_notify)(struct file_lock *);  /* unblock callback */
 920        int (*lm_grant)(struct file_lock *, struct file_lock *, int);
 921        void (*lm_break)(struct file_lock *);
 922        int (*lm_change)(struct file_lock **, int);
 923};
 924
 925struct lock_manager {
 926        struct list_head list;
 927};
 928
 929struct net;
 930void locks_start_grace(struct net *, struct lock_manager *);
 931void locks_end_grace(struct lock_manager *);
 932int locks_in_grace(struct net *);
 933
 934/* that will die - we need it for nfs_lock_info */
 935#include <linux/nfs_fs_i.h>
 936
 937/*
 938 * struct file_lock represents a generic "file lock". It's used to represent
 939 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
 940 * note that the same struct is used to represent both a request for a lock and
 941 * the lock itself, but the same object is never used for both.
 942 *
 943 * FIXME: should we create a separate "struct lock_request" to help distinguish
 944 * these two uses?
 945 *
 946 * The i_flock list is ordered by:
 947 *
 948 * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
 949 * 2) lock owner
 950 * 3) lock range start
 951 * 4) lock range end
 952 *
 953 * Obviously, the last two criteria only matter for POSIX locks.
 954 */
 955struct file_lock {
 956        struct file_lock *fl_next;      /* singly linked list for this inode  */
 957        struct hlist_node fl_link;      /* node in global lists */
 958        struct list_head fl_block;      /* circular list of blocked processes */
 959        fl_owner_t fl_owner;
 960        unsigned int fl_flags;
 961        unsigned char fl_type;
 962        unsigned int fl_pid;
 963        int fl_link_cpu;                /* what cpu's list is this on? */
 964        struct pid *fl_nspid;
 965        wait_queue_head_t fl_wait;
 966        struct file *fl_file;
 967        loff_t fl_start;
 968        loff_t fl_end;
 969
 970        struct fasync_struct *  fl_fasync; /* for lease break notifications */
 971        /* for lease breaks: */
 972        unsigned long fl_break_time;
 973        unsigned long fl_downgrade_time;
 974
 975        const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
 976        const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
 977        union {
 978                struct nfs_lock_info    nfs_fl;
 979                struct nfs4_lock_info   nfs4_fl;
 980                struct {
 981                        struct list_head link;  /* link in AFS vnode's pending_locks list */
 982                        int state;              /* state of grant or error if -ve */
 983                } afs;
 984        } fl_u;
 985};
 986
 987/* The following constant reflects the upper bound of the file/locking space */
 988#ifndef OFFSET_MAX
 989#define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
 990#define OFFSET_MAX      INT_LIMIT(loff_t)
 991#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
 992#endif
 993
 994#include <linux/fcntl.h>
 995
 996extern void send_sigio(struct fown_struct *fown, int fd, int band);
 997
 998#ifdef CONFIG_FILE_LOCKING
 999extern int fcntl_getlk(struct file *, struct flock __user *);
1000extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1001                        struct flock __user *);
1002
1003#if BITS_PER_LONG == 32
1004extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1005extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1006                        struct flock64 __user *);
1007#endif
1008
1009extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1010extern int fcntl_getlease(struct file *filp);
1011
1012/* fs/locks.c */
1013void locks_free_lock(struct file_lock *fl);
1014extern void locks_init_lock(struct file_lock *);
1015extern struct file_lock * locks_alloc_lock(void);
1016extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1017extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1018extern void locks_remove_posix(struct file *, fl_owner_t);
1019extern void locks_remove_flock(struct file *);
1020extern void locks_release_private(struct file_lock *);
1021extern void posix_test_lock(struct file *, struct file_lock *);
1022extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1023extern int posix_lock_file_wait(struct file *, struct file_lock *);
1024extern int posix_unblock_lock(struct file_lock *);
1025extern int vfs_test_lock(struct file *, struct file_lock *);
1026extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1027extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1028extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1029extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1030extern void lease_get_mtime(struct inode *, struct timespec *time);
1031extern int generic_setlease(struct file *, long, struct file_lock **);
1032extern int vfs_setlease(struct file *, long, struct file_lock **);
1033extern int lease_modify(struct file_lock **, int);
1034extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1035extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1036#else /* !CONFIG_FILE_LOCKING */
1037static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1038{
1039        return -EINVAL;
1040}
1041
1042static inline int fcntl_setlk(unsigned int fd, struct file *file,
1043                              unsigned int cmd, struct flock __user *user)
1044{
1045        return -EACCES;
1046}
1047
1048#if BITS_PER_LONG == 32
1049static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1050{
1051        return -EINVAL;
1052}
1053
1054static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1055                                unsigned int cmd, struct flock64 __user *user)
1056{
1057        return -EACCES;
1058}
1059#endif
1060static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1061{
1062        return 0;
1063}
1064
1065static inline int fcntl_getlease(struct file *filp)
1066{
1067        return 0;
1068}
1069
1070static inline void locks_init_lock(struct file_lock *fl)
1071{
1072        return;
1073}
1074
1075static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1076{
1077        return;
1078}
1079
1080static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1081{
1082        return;
1083}
1084
1085static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1086{
1087        return;
1088}
1089
1090static inline void locks_remove_flock(struct file *filp)
1091{
1092        return;
1093}
1094
1095static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1096{
1097        return;
1098}
1099
1100static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1101                                  struct file_lock *conflock)
1102{
1103        return -ENOLCK;
1104}
1105
1106static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1107{
1108        return -ENOLCK;
1109}
1110
1111static inline int posix_unblock_lock(struct file_lock *waiter)
1112{
1113        return -ENOENT;
1114}
1115
1116static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1117{
1118        return 0;
1119}
1120
1121static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1122                                struct file_lock *fl, struct file_lock *conf)
1123{
1124        return -ENOLCK;
1125}
1126
1127static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1128{
1129        return 0;
1130}
1131
1132static inline int flock_lock_file_wait(struct file *filp,
1133                                       struct file_lock *request)
1134{
1135        return -ENOLCK;
1136}
1137
1138static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1139{
1140        return 0;
1141}
1142
1143static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1144{
1145        return;
1146}
1147
1148static inline int generic_setlease(struct file *filp, long arg,
1149                                    struct file_lock **flp)
1150{
1151        return -EINVAL;
1152}
1153
1154static inline int vfs_setlease(struct file *filp, long arg,
1155                               struct file_lock **lease)
1156{
1157        return -EINVAL;
1158}
1159
1160static inline int lease_modify(struct file_lock **before, int arg)
1161{
1162        return -EINVAL;
1163}
1164
1165static inline int lock_may_read(struct inode *inode, loff_t start,
1166                                unsigned long len)
1167{
1168        return 1;
1169}
1170
1171static inline int lock_may_write(struct inode *inode, loff_t start,
1172                                 unsigned long len)
1173{
1174        return 1;
1175}
1176#endif /* !CONFIG_FILE_LOCKING */
1177
1178
1179struct fasync_struct {
1180        spinlock_t              fa_lock;
1181        int                     magic;
1182        int                     fa_fd;
1183        struct fasync_struct    *fa_next; /* singly linked list */
1184        struct file             *fa_file;
1185        struct rcu_head         fa_rcu;
1186};
1187
1188#define FASYNC_MAGIC 0x4601
1189
1190/* SMP safe fasync helpers: */
1191extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1192extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1193extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1194extern struct fasync_struct *fasync_alloc(void);
1195extern void fasync_free(struct fasync_struct *);
1196
1197/* can be called from interrupts */
1198extern void kill_fasync(struct fasync_struct **, int, int);
1199
1200extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1201extern int f_setown(struct file *filp, unsigned long arg, int force);
1202extern void f_delown(struct file *filp);
1203extern pid_t f_getown(struct file *filp);
1204extern int send_sigurg(struct fown_struct *fown);
1205
1206struct mm_struct;
1207
1208/*
1209 *      Umount options
1210 */
1211
1212#define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1213#define MNT_DETACH      0x00000002      /* Just detach from the tree */
1214#define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1215#define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1216#define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1217
1218extern struct list_head super_blocks;
1219extern spinlock_t sb_lock;
1220
1221/* Possible states of 'frozen' field */
1222enum {
1223        SB_UNFROZEN = 0,                /* FS is unfrozen */
1224        SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1225        SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1226        SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1227                                         * internal threads if needed) */
1228        SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1229};
1230
1231#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1232
1233struct sb_writers {
1234        /* Counters for counting writers at each level */
1235        struct percpu_counter   counter[SB_FREEZE_LEVELS];
1236        wait_queue_head_t       wait;           /* queue for waiting for
1237                                                   writers / faults to finish */
1238        int                     frozen;         /* Is sb frozen? */
1239        wait_queue_head_t       wait_unfrozen;  /* queue for waiting for
1240                                                   sb to be thawed */
1241#ifdef CONFIG_DEBUG_LOCK_ALLOC
1242        struct lockdep_map      lock_map[SB_FREEZE_LEVELS];
1243#endif
1244};
1245
1246struct super_block {
1247        struct list_head        s_list;         /* Keep this first */
1248        dev_t                   s_dev;          /* search index; _not_ kdev_t */
1249        unsigned char           s_blocksize_bits;
1250        unsigned long           s_blocksize;
1251        loff_t                  s_maxbytes;     /* Max file size */
1252        struct file_system_type *s_type;
1253        const struct super_operations   *s_op;
1254        const struct dquot_operations   *dq_op;
1255        const struct quotactl_ops       *s_qcop;
1256        const struct export_operations *s_export_op;
1257        unsigned long           s_flags;
1258        unsigned long           s_magic;
1259        struct dentry           *s_root;
1260        struct rw_semaphore     s_umount;
1261        int                     s_count;
1262        atomic_t                s_active;
1263#ifdef CONFIG_SECURITY
1264        void                    *s_security;
1265#endif
1266        const struct xattr_handler **s_xattr;
1267
1268        struct list_head        s_inodes;       /* all inodes */
1269        struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1270        struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1271        struct block_device     *s_bdev;
1272        struct backing_dev_info *s_bdi;
1273        struct mtd_info         *s_mtd;
1274        struct hlist_node       s_instances;
1275        struct quota_info       s_dquot;        /* Diskquota specific options */
1276
1277        struct sb_writers       s_writers;
1278
1279        char s_id[32];                          /* Informational name */
1280        u8 s_uuid[16];                          /* UUID */
1281
1282        void                    *s_fs_info;     /* Filesystem private info */
1283        unsigned int            s_max_links;
1284        fmode_t                 s_mode;
1285
1286        /* Granularity of c/m/atime in ns.
1287           Cannot be worse than a second */
1288        u32                s_time_gran;
1289
1290        /*
1291         * The next field is for VFS *only*. No filesystems have any business
1292         * even looking at it. You had been warned.
1293         */
1294        struct mutex s_vfs_rename_mutex;        /* Kludge */
1295
1296        /*
1297         * Filesystem subtype.  If non-empty the filesystem type field
1298         * in /proc/mounts will be "type.subtype"
1299         */
1300        char *s_subtype;
1301
1302        /*
1303         * Saved mount options for lazy filesystems using
1304         * generic_show_options()
1305         */
1306        char __rcu *s_options;
1307        const struct dentry_operations *s_d_op; /* default d_op for dentries */
1308
1309        /*
1310         * Saved pool identifier for cleancache (-1 means none)
1311         */
1312        int cleancache_poolid;
1313
1314        struct shrinker s_shrink;       /* per-sb shrinker handle */
1315
1316        /* Number of inodes with nlink == 0 but still referenced */
1317        atomic_long_t s_remove_count;
1318
1319        /* Being remounted read-only */
1320        int s_readonly_remount;
1321
1322        /* AIO completions deferred from interrupt context */
1323        struct workqueue_struct *s_dio_done_wq;
1324
1325        /*
1326         * Keep the lru lists last in the structure so they always sit on their
1327         * own individual cachelines.
1328         */
1329        struct list_lru         s_dentry_lru ____cacheline_aligned_in_smp;
1330        struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
1331        struct rcu_head         rcu;
1332};
1333
1334extern struct timespec current_fs_time(struct super_block *sb);
1335
1336/*
1337 * Snapshotting support.
1338 */
1339
1340void __sb_end_write(struct super_block *sb, int level);
1341int __sb_start_write(struct super_block *sb, int level, bool wait);
1342
1343/**
1344 * sb_end_write - drop write access to a superblock
1345 * @sb: the super we wrote to
1346 *
1347 * Decrement number of writers to the filesystem. Wake up possible waiters
1348 * wanting to freeze the filesystem.
1349 */
1350static inline void sb_end_write(struct super_block *sb)
1351{
1352        __sb_end_write(sb, SB_FREEZE_WRITE);
1353}
1354
1355/**
1356 * sb_end_pagefault - drop write access to a superblock from a page fault
1357 * @sb: the super we wrote to
1358 *
1359 * Decrement number of processes handling write page fault to the filesystem.
1360 * Wake up possible waiters wanting to freeze the filesystem.
1361 */
1362static inline void sb_end_pagefault(struct super_block *sb)
1363{
1364        __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1365}
1366
1367/**
1368 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1369 * @sb: the super we wrote to
1370 *
1371 * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1372 * waiters wanting to freeze the filesystem.
1373 */
1374static inline void sb_end_intwrite(struct super_block *sb)
1375{
1376        __sb_end_write(sb, SB_FREEZE_FS);
1377}
1378
1379/**
1380 * sb_start_write - get write access to a superblock
1381 * @sb: the super we write to
1382 *
1383 * When a process wants to write data or metadata to a file system (i.e. dirty
1384 * a page or an inode), it should embed the operation in a sb_start_write() -
1385 * sb_end_write() pair to get exclusion against file system freezing. This
1386 * function increments number of writers preventing freezing. If the file
1387 * system is already frozen, the function waits until the file system is
1388 * thawed.
1389 *
1390 * Since freeze protection behaves as a lock, users have to preserve
1391 * ordering of freeze protection and other filesystem locks. Generally,
1392 * freeze protection should be the outermost lock. In particular, we have:
1393 *
1394 * sb_start_write
1395 *   -> i_mutex                 (write path, truncate, directory ops, ...)
1396 *   -> s_umount                (freeze_super, thaw_super)
1397 */
1398static inline void sb_start_write(struct super_block *sb)
1399{
1400        __sb_start_write(sb, SB_FREEZE_WRITE, true);
1401}
1402
1403static inline int sb_start_write_trylock(struct super_block *sb)
1404{
1405        return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1406}
1407
1408/**
1409 * sb_start_pagefault - get write access to a superblock from a page fault
1410 * @sb: the super we write to
1411 *
1412 * When a process starts handling write page fault, it should embed the
1413 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1414 * exclusion against file system freezing. This is needed since the page fault
1415 * is going to dirty a page. This function increments number of running page
1416 * faults preventing freezing. If the file system is already frozen, the
1417 * function waits until the file system is thawed.
1418 *
1419 * Since page fault freeze protection behaves as a lock, users have to preserve
1420 * ordering of freeze protection and other filesystem locks. It is advised to
1421 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1422 * handling code implies lock dependency:
1423 *
1424 * mmap_sem
1425 *   -> sb_start_pagefault
1426 */
1427static inline void sb_start_pagefault(struct super_block *sb)
1428{
1429        __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1430}
1431
1432/*
1433 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1434 * @sb: the super we write to
1435 *
1436 * This is the third level of protection against filesystem freezing. It is
1437 * free for use by a filesystem. The only requirement is that it must rank
1438 * below sb_start_pagefault.
1439 *
1440 * For example filesystem can call sb_start_intwrite() when starting a
1441 * transaction which somewhat eases handling of freezing for internal sources
1442 * of filesystem changes (internal fs threads, discarding preallocation on file
1443 * close, etc.).
1444 */
1445static inline void sb_start_intwrite(struct super_block *sb)
1446{
1447        __sb_start_write(sb, SB_FREEZE_FS, true);
1448}
1449
1450
1451extern bool inode_owner_or_capable(const struct inode *inode);
1452
1453/*
1454 * VFS helper functions..
1455 */
1456extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1457extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1458extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1459extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1460extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1461extern int vfs_rmdir(struct inode *, struct dentry *);
1462extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1463extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **);
1464
1465/*
1466 * VFS dentry helper functions.
1467 */
1468extern void dentry_unhash(struct dentry *dentry);
1469
1470/*
1471 * VFS file helper functions.
1472 */
1473extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1474                        umode_t mode);
1475/*
1476 * VFS FS_IOC_FIEMAP helper definitions.
1477 */
1478struct fiemap_extent_info {
1479        unsigned int fi_flags;          /* Flags as passed from user */
1480        unsigned int fi_extents_mapped; /* Number of mapped extents */
1481        unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1482        struct fiemap_extent __user *fi_extents_start; /* Start of
1483                                                        fiemap_extent array */
1484};
1485int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1486                            u64 phys, u64 len, u32 flags);
1487int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1488
1489/*
1490 * File types
1491 *
1492 * NOTE! These match bits 12..15 of stat.st_mode
1493 * (ie "(i_mode >> 12) & 15").
1494 */
1495#define DT_UNKNOWN      0
1496#define DT_FIFO         1
1497#define DT_CHR          2
1498#define DT_DIR          4
1499#define DT_BLK          6
1500#define DT_REG          8
1501#define DT_LNK          10
1502#define DT_SOCK         12
1503#define DT_WHT          14
1504
1505/*
1506 * This is the "filldir" function type, used by readdir() to let
1507 * the kernel specify what kind of dirent layout it wants to have.
1508 * This allows the kernel to read directories into kernel space or
1509 * to have different dirent layouts depending on the binary type.
1510 */
1511typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1512struct dir_context {
1513        const filldir_t actor;
1514        loff_t pos;
1515};
1516
1517struct block_device_operations;
1518
1519/* These macros are for out of kernel modules to test that
1520 * the kernel supports the unlocked_ioctl and compat_ioctl
1521 * fields in struct file_operations. */
1522#define HAVE_COMPAT_IOCTL 1
1523#define HAVE_UNLOCKED_IOCTL 1
1524
1525struct file_operations {
1526        struct module *owner;
1527        loff_t (*llseek) (struct file *, loff_t, int);
1528        ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1529        ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1530        ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1531        ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1532        int (*iterate) (struct file *, struct dir_context *);
1533        unsigned int (*poll) (struct file *, struct poll_table_struct *);
1534        long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1535        long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1536        int (*mmap) (struct file *, struct vm_area_struct *);
1537        int (*open) (struct inode *, struct file *);
1538        int (*flush) (struct file *, fl_owner_t id);
1539        int (*release) (struct inode *, struct file *);
1540        int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1541        int (*aio_fsync) (struct kiocb *, int datasync);
1542        int (*fasync) (int, struct file *, int);
1543        int (*lock) (struct file *, int, struct file_lock *);
1544        ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1545        unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1546        int (*check_flags)(int);
1547        int (*flock) (struct file *, int, struct file_lock *);
1548        ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1549        ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1550        int (*setlease)(struct file *, long, struct file_lock **);
1551        long (*fallocate)(struct file *file, int mode, loff_t offset,
1552                          loff_t len);
1553        int (*show_fdinfo)(struct seq_file *m, struct file *f);
1554};
1555
1556struct inode_operations {
1557        struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1558        void * (*follow_link) (struct dentry *, struct nameidata *);
1559        int (*permission) (struct inode *, int);
1560        struct posix_acl * (*get_acl)(struct inode *, int);
1561
1562        int (*readlink) (struct dentry *, char __user *,int);
1563        void (*put_link) (struct dentry *, struct nameidata *, void *);
1564
1565        int (*create) (struct inode *,struct dentry *, umode_t, bool);
1566        int (*link) (struct dentry *,struct inode *,struct dentry *);
1567        int (*unlink) (struct inode *,struct dentry *);
1568        int (*symlink) (struct inode *,struct dentry *,const char *);
1569        int (*mkdir) (struct inode *,struct dentry *,umode_t);
1570        int (*rmdir) (struct inode *,struct dentry *);
1571        int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1572        int (*rename) (struct inode *, struct dentry *,
1573                        struct inode *, struct dentry *);
1574        int (*setattr) (struct dentry *, struct iattr *);
1575        int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1576        int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1577        ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1578        ssize_t (*listxattr) (struct dentry *, char *, size_t);
1579        int (*removexattr) (struct dentry *, const char *);
1580        int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1581                      u64 len);
1582        int (*update_time)(struct inode *, struct timespec *, int);
1583        int (*atomic_open)(struct inode *, struct dentry *,
1584                           struct file *, unsigned open_flag,
1585                           umode_t create_mode, int *opened);
1586        int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1587        int (*set_acl)(struct inode *, struct posix_acl *, int);
1588} ____cacheline_aligned;
1589
1590ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1591                              unsigned long nr_segs, unsigned long fast_segs,
1592                              struct iovec *fast_pointer,
1593                              struct iovec **ret_pointer);
1594
1595extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1596extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1597extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1598                unsigned long, loff_t *);
1599extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1600                unsigned long, loff_t *);
1601
1602struct super_operations {
1603        struct inode *(*alloc_inode)(struct super_block *sb);
1604        void (*destroy_inode)(struct inode *);
1605
1606        void (*dirty_inode) (struct inode *, int flags);
1607        int (*write_inode) (struct inode *, struct writeback_control *wbc);
1608        int (*drop_inode) (struct inode *);
1609        void (*evict_inode) (struct inode *);
1610        void (*put_super) (struct super_block *);
1611        int (*sync_fs)(struct super_block *sb, int wait);
1612        int (*freeze_fs) (struct super_block *);
1613        int (*unfreeze_fs) (struct super_block *);
1614        int (*statfs) (struct dentry *, struct kstatfs *);
1615        int (*remount_fs) (struct super_block *, int *, char *);
1616        void (*umount_begin) (struct super_block *);
1617
1618        int (*show_options)(struct seq_file *, struct dentry *);
1619        int (*show_devname)(struct seq_file *, struct dentry *);
1620        int (*show_path)(struct seq_file *, struct dentry *);
1621        int (*show_stats)(struct seq_file *, struct dentry *);
1622#ifdef CONFIG_QUOTA
1623        ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1624        ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1625#endif
1626        int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1627        long (*nr_cached_objects)(struct super_block *, int);
1628        long (*free_cached_objects)(struct super_block *, long, int);
1629};
1630
1631/*
1632 * Inode flags - they have no relation to superblock flags now
1633 */
1634#define S_SYNC          1       /* Writes are synced at once */
1635#define S_NOATIME       2       /* Do not update access times */
1636#define S_APPEND        4       /* Append-only file */
1637#define S_IMMUTABLE     8       /* Immutable file */
1638#define S_DEAD          16      /* removed, but still open directory */
1639#define S_NOQUOTA       32      /* Inode is not counted to quota */
1640#define S_DIRSYNC       64      /* Directory modifications are synchronous */
1641#define S_NOCMTIME      128     /* Do not update file c/mtime */
1642#define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1643#define S_PRIVATE       512     /* Inode is fs-internal */
1644#define S_IMA           1024    /* Inode has an associated IMA struct */
1645#define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1646#define S_NOSEC         4096    /* no suid or xattr security attributes */
1647
1648/*
1649 * Note that nosuid etc flags are inode-specific: setting some file-system
1650 * flags just means all the inodes inherit those flags by default. It might be
1651 * possible to override it selectively if you really wanted to with some
1652 * ioctl() that is not currently implemented.
1653 *
1654 * Exception: MS_RDONLY is always applied to the entire file system.
1655 *
1656 * Unfortunately, it is possible to change a filesystems flags with it mounted
1657 * with files in use.  This means that all of the inodes will not have their
1658 * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1659 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1660 */
1661#define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1662
1663#define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1664#define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1665                                        ((inode)->i_flags & S_SYNC))
1666#define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1667                                        ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1668#define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1669#define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1670#define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1671
1672#define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1673#define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1674#define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1675#define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1676
1677#define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1678#define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1679#define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1680#define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1681#define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1682#define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1683#define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1684
1685/*
1686 * Inode state bits.  Protected by inode->i_lock
1687 *
1688 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1689 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1690 *
1691 * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1692 * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1693 * various stages of removing an inode.
1694 *
1695 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1696 *
1697 * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1698 *                      fdatasync().  i_atime is the usual cause.
1699 * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1700 *                      these changes separately from I_DIRTY_SYNC so that we
1701 *                      don't have to write inode on fdatasync() when only
1702 *                      mtime has changed in it.
1703 * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1704 * I_NEW                Serves as both a mutex and completion notification.
1705 *                      New inodes set I_NEW.  If two processes both create
1706 *                      the same inode, one of them will release its inode and
1707 *                      wait for I_NEW to be released before returning.
1708 *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1709 *                      also cause waiting on I_NEW, without I_NEW actually
1710 *                      being set.  find_inode() uses this to prevent returning
1711 *                      nearly-dead inodes.
1712 * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1713 *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1714 *                      cleared.
1715 * I_FREEING            Set when inode is about to be freed but still has dirty
1716 *                      pages or buffers attached or the inode itself is still
1717 *                      dirty.
1718 * I_CLEAR              Added by clear_inode().  In this state the inode is
1719 *                      clean and can be destroyed.  Inode keeps I_FREEING.
1720 *
1721 *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1722 *                      prohibited for many purposes.  iget() must wait for
1723 *                      the inode to be completely released, then create it
1724 *                      anew.  Other functions will just ignore such inodes,
1725 *                      if appropriate.  I_NEW is used for waiting.
1726 *
1727 * I_SYNC               Writeback of inode is running. The bit is set during
1728 *                      data writeback, and cleared with a wakeup on the bit
1729 *                      address once it is done. The bit is also used to pin
1730 *                      the inode in memory for flusher thread.
1731 *
1732 * I_REFERENCED         Marks the inode as recently references on the LRU list.
1733 *
1734 * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1735 *
1736 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1737 */
1738#define I_DIRTY_SYNC            (1 << 0)
1739#define I_DIRTY_DATASYNC        (1 << 1)
1740#define I_DIRTY_PAGES           (1 << 2)
1741#define __I_NEW                 3
1742#define I_NEW                   (1 << __I_NEW)
1743#define I_WILL_FREE             (1 << 4)
1744#define I_FREEING               (1 << 5)
1745#define I_CLEAR                 (1 << 6)
1746#define __I_SYNC                7
1747#define I_SYNC                  (1 << __I_SYNC)
1748#define I_REFERENCED            (1 << 8)
1749#define __I_DIO_WAKEUP          9
1750#define I_DIO_WAKEUP            (1 << I_DIO_WAKEUP)
1751#define I_LINKABLE              (1 << 10)
1752
1753#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1754
1755extern void __mark_inode_dirty(struct inode *, int);
1756static inline void mark_inode_dirty(struct inode *inode)
1757{
1758        __mark_inode_dirty(inode, I_DIRTY);
1759}
1760
1761static inline void mark_inode_dirty_sync(struct inode *inode)
1762{
1763        __mark_inode_dirty(inode, I_DIRTY_SYNC);
1764}
1765
1766extern void inc_nlink(struct inode *inode);
1767extern void drop_nlink(struct inode *inode);
1768extern void clear_nlink(struct inode *inode);
1769extern void set_nlink(struct inode *inode, unsigned int nlink);
1770
1771static inline void inode_inc_link_count(struct inode *inode)
1772{
1773        inc_nlink(inode);
1774        mark_inode_dirty(inode);
1775}
1776
1777static inline void inode_dec_link_count(struct inode *inode)
1778{
1779        drop_nlink(inode);
1780        mark_inode_dirty(inode);
1781}
1782
1783/**
1784 * inode_inc_iversion - increments i_version
1785 * @inode: inode that need to be updated
1786 *
1787 * Every time the inode is modified, the i_version field will be incremented.
1788 * The filesystem has to be mounted with i_version flag
1789 */
1790
1791static inline void inode_inc_iversion(struct inode *inode)
1792{
1793       spin_lock(&inode->i_lock);
1794       inode->i_version++;
1795       spin_unlock(&inode->i_lock);
1796}
1797
1798enum file_time_flags {
1799        S_ATIME = 1,
1800        S_MTIME = 2,
1801        S_CTIME = 4,
1802        S_VERSION = 8,
1803};
1804
1805extern void touch_atime(const struct path *);
1806static inline void file_accessed(struct file *file)
1807{
1808        if (!(file->f_flags & O_NOATIME))
1809                touch_atime(&file->f_path);
1810}
1811
1812int sync_inode(struct inode *inode, struct writeback_control *wbc);
1813int sync_inode_metadata(struct inode *inode, int wait);
1814
1815struct file_system_type {
1816        const char *name;
1817        int fs_flags;
1818#define FS_REQUIRES_DEV         1 
1819#define FS_BINARY_MOUNTDATA     2
1820#define FS_HAS_SUBTYPE          4
1821#define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
1822#define FS_USERNS_DEV_MOUNT     16 /* A userns mount does not imply MNT_NODEV */
1823#define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1824        struct dentry *(*mount) (struct file_system_type *, int,
1825                       const char *, void *);
1826        void (*kill_sb) (struct super_block *);
1827        struct module *owner;
1828        struct file_system_type * next;
1829        struct hlist_head fs_supers;
1830
1831        struct lock_class_key s_lock_key;
1832        struct lock_class_key s_umount_key;
1833        struct lock_class_key s_vfs_rename_key;
1834        struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1835
1836        struct lock_class_key i_lock_key;
1837        struct lock_class_key i_mutex_key;
1838        struct lock_class_key i_mutex_dir_key;
1839};
1840
1841#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1842
1843extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1844        void *data, int (*fill_super)(struct super_block *, void *, int));
1845extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1846        int flags, const char *dev_name, void *data,
1847        int (*fill_super)(struct super_block *, void *, int));
1848extern struct dentry *mount_single(struct file_system_type *fs_type,
1849        int flags, void *data,
1850        int (*fill_super)(struct super_block *, void *, int));
1851extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1852        int flags, void *data,
1853        int (*fill_super)(struct super_block *, void *, int));
1854extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1855void generic_shutdown_super(struct super_block *sb);
1856void kill_block_super(struct super_block *sb);
1857void kill_anon_super(struct super_block *sb);
1858void kill_litter_super(struct super_block *sb);
1859void deactivate_super(struct super_block *sb);
1860void deactivate_locked_super(struct super_block *sb);
1861int set_anon_super(struct super_block *s, void *data);
1862int get_anon_bdev(dev_t *);
1863void free_anon_bdev(dev_t);
1864struct super_block *sget(struct file_system_type *type,
1865                        int (*test)(struct super_block *,void *),
1866                        int (*set)(struct super_block *,void *),
1867                        int flags, void *data);
1868extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1869        const struct super_operations *ops,
1870        const struct dentry_operations *dops,
1871        unsigned long);
1872
1873/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1874#define fops_get(fops) \
1875        (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1876#define fops_put(fops) \
1877        do { if (fops) module_put((fops)->owner); } while(0)
1878/*
1879 * This one is to be used *ONLY* from ->open() instances.
1880 * fops must be non-NULL, pinned down *and* module dependencies
1881 * should be sufficient to pin the caller down as well.
1882 */
1883#define replace_fops(f, fops) \
1884        do {    \
1885                struct file *__file = (f); \
1886                fops_put(__file->f_op); \
1887                BUG_ON(!(__file->f_op = (fops))); \
1888        } while(0)
1889
1890extern int register_filesystem(struct file_system_type *);
1891extern int unregister_filesystem(struct file_system_type *);
1892extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1893#define kern_mount(type) kern_mount_data(type, NULL)
1894extern void kern_unmount(struct vfsmount *mnt);
1895extern int may_umount_tree(struct vfsmount *);
1896extern int may_umount(struct vfsmount *);
1897extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1898extern struct vfsmount *collect_mounts(struct path *);
1899extern void drop_collected_mounts(struct vfsmount *);
1900extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1901                          struct vfsmount *);
1902extern int vfs_statfs(struct path *, struct kstatfs *);
1903extern int user_statfs(const char __user *, struct kstatfs *);
1904extern int fd_statfs(int, struct kstatfs *);
1905extern int vfs_ustat(dev_t, struct kstatfs *);
1906extern int freeze_super(struct super_block *super);
1907extern int thaw_super(struct super_block *super);
1908extern bool our_mnt(struct vfsmount *mnt);
1909extern bool fs_fully_visible(struct file_system_type *);
1910
1911extern int current_umask(void);
1912
1913extern void ihold(struct inode * inode);
1914extern void iput(struct inode *);
1915
1916/* /sys/fs */
1917extern struct kobject *fs_kobj;
1918
1919#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1920
1921#define FLOCK_VERIFY_READ  1
1922#define FLOCK_VERIFY_WRITE 2
1923
1924#ifdef CONFIG_FILE_LOCKING
1925extern int locks_mandatory_locked(struct inode *);
1926extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1927
1928/*
1929 * Candidates for mandatory locking have the setgid bit set
1930 * but no group execute bit -  an otherwise meaningless combination.
1931 */
1932
1933static inline int __mandatory_lock(struct inode *ino)
1934{
1935        return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1936}
1937
1938/*
1939 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1940 * otherwise these will be advisory locks
1941 */
1942
1943static inline int mandatory_lock(struct inode *ino)
1944{
1945        return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1946}
1947
1948static inline int locks_verify_locked(struct inode *inode)
1949{
1950        if (mandatory_lock(inode))
1951                return locks_mandatory_locked(inode);
1952        return 0;
1953}
1954
1955static inline int locks_verify_truncate(struct inode *inode,
1956                                    struct file *filp,
1957                                    loff_t size)
1958{
1959        if (inode->i_flock && mandatory_lock(inode))
1960                return locks_mandatory_area(
1961                        FLOCK_VERIFY_WRITE, inode, filp,
1962                        size < inode->i_size ? size : inode->i_size,
1963                        (size < inode->i_size ? inode->i_size - size
1964                         : size - inode->i_size)
1965                );
1966        return 0;
1967}
1968
1969static inline int break_lease(struct inode *inode, unsigned int mode)
1970{
1971        if (inode->i_flock)
1972                return __break_lease(inode, mode, FL_LEASE);
1973        return 0;
1974}
1975
1976static inline int break_deleg(struct inode *inode, unsigned int mode)
1977{
1978        if (inode->i_flock)
1979                return __break_lease(inode, mode, FL_DELEG);
1980        return 0;
1981}
1982
1983static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
1984{
1985        int ret;
1986
1987        ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
1988        if (ret == -EWOULDBLOCK && delegated_inode) {
1989                *delegated_inode = inode;
1990                ihold(inode);
1991        }
1992        return ret;
1993}
1994
1995static inline int break_deleg_wait(struct inode **delegated_inode)
1996{
1997        int ret;
1998
1999        ret = break_deleg(*delegated_inode, O_WRONLY);
2000        iput(*delegated_inode);
2001        *delegated_inode = NULL;
2002        return ret;
2003}
2004
2005#else /* !CONFIG_FILE_LOCKING */
2006static inline int locks_mandatory_locked(struct inode *inode)
2007{
2008        return 0;
2009}
2010
2011static inline int locks_mandatory_area(int rw, struct inode *inode,
2012                                       struct file *filp, loff_t offset,
2013                                       size_t count)
2014{
2015        return 0;
2016}
2017
2018static inline int __mandatory_lock(struct inode *inode)
2019{
2020        return 0;
2021}
2022
2023static inline int mandatory_lock(struct inode *inode)
2024{
2025        return 0;
2026}
2027
2028static inline int locks_verify_locked(struct inode *inode)
2029{
2030        return 0;
2031}
2032
2033static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2034                                        size_t size)
2035{
2036        return 0;
2037}
2038
2039static inline int break_lease(struct inode *inode, unsigned int mode)
2040{
2041        return 0;
2042}
2043
2044static inline int break_deleg(struct inode *inode, unsigned int mode)
2045{
2046        return 0;
2047}
2048
2049static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2050{
2051        return 0;
2052}
2053
2054static inline int break_deleg_wait(struct inode **delegated_inode)
2055{
2056        BUG();
2057        return 0;
2058}
2059
2060#endif /* CONFIG_FILE_LOCKING */
2061
2062/* fs/open.c */
2063struct audit_names;
2064struct filename {
2065        const char              *name;  /* pointer to actual string */
2066        const __user char       *uptr;  /* original userland pointer */
2067        struct audit_names      *aname;
2068        bool                    separate; /* should "name" be freed? */
2069};
2070
2071extern long vfs_truncate(struct path *, loff_t);
2072extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2073                       struct file *filp);
2074extern int do_fallocate(struct file *file, int mode, loff_t offset,
2075                        loff_t len);
2076extern long do_sys_open(int dfd, const char __user *filename, int flags,
2077                        umode_t mode);
2078extern struct file *file_open_name(struct filename *, int, umode_t);
2079extern struct file *filp_open(const char *, int, umode_t);
2080extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2081                                   const char *, int);
2082extern struct file * dentry_open(const struct path *, int, const struct cred *);
2083extern int filp_close(struct file *, fl_owner_t id);
2084
2085extern struct filename *getname(const char __user *);
2086extern struct filename *getname_kernel(const char *);
2087
2088enum {
2089        FILE_CREATED = 1,
2090        FILE_OPENED = 2
2091};
2092extern int finish_open(struct file *file, struct dentry *dentry,
2093                        int (*open)(struct inode *, struct file *),
2094                        int *opened);
2095extern int finish_no_open(struct file *file, struct dentry *dentry);
2096
2097/* fs/ioctl.c */
2098
2099extern int ioctl_preallocate(struct file *filp, void __user *argp);
2100
2101/* fs/dcache.c */
2102extern void __init vfs_caches_init_early(void);
2103extern void __init vfs_caches_init(unsigned long);
2104
2105extern struct kmem_cache *names_cachep;
2106
2107extern void final_putname(struct filename *name);
2108
2109#define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2110#define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2111#ifndef CONFIG_AUDITSYSCALL
2112#define putname(name)           final_putname(name)
2113#else
2114extern void putname(struct filename *name);
2115#endif
2116
2117#ifdef CONFIG_BLOCK
2118extern int register_blkdev(unsigned int, const char *);
2119extern void unregister_blkdev(unsigned int, const char *);
2120extern struct block_device *bdget(dev_t);
2121extern struct block_device *bdgrab(struct block_device *bdev);
2122extern void bd_set_size(struct block_device *, loff_t size);
2123extern void bd_forget(struct inode *inode);
2124extern void bdput(struct block_device *);
2125extern void invalidate_bdev(struct block_device *);
2126extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2127extern int sync_blockdev(struct block_device *bdev);
2128extern void kill_bdev(struct block_device *);
2129extern struct super_block *freeze_bdev(struct block_device *);
2130extern void emergency_thaw_all(void);
2131extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2132extern int fsync_bdev(struct block_device *);
2133extern int sb_is_blkdev_sb(struct super_block *sb);
2134#else
2135static inline void bd_forget(struct inode *inode) {}
2136static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2137static inline void kill_bdev(struct block_device *bdev) {}
2138static inline void invalidate_bdev(struct block_device *bdev) {}
2139
2140static inline struct super_block *freeze_bdev(struct block_device *sb)
2141{
2142        return NULL;
2143}
2144
2145static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2146{
2147        return 0;
2148}
2149
2150static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2151{
2152}
2153
2154static inline int sb_is_blkdev_sb(struct super_block *sb)
2155{
2156        return 0;
2157}
2158#endif
2159extern int sync_filesystem(struct super_block *);
2160extern const struct file_operations def_blk_fops;
2161extern const struct file_operations def_chr_fops;
2162extern const struct file_operations bad_sock_fops;
2163#ifdef CONFIG_BLOCK
2164extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2165extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2166extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2167extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2168extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2169                                               void *holder);
2170extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2171                                              void *holder);
2172extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2173#ifdef CONFIG_SYSFS
2174extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2175extern void bd_unlink_disk_holder(struct block_device *bdev,
2176                                  struct gendisk *disk);
2177#else
2178static inline int bd_link_disk_holder(struct block_device *bdev,
2179                                      struct gendisk *disk)
2180{
2181        return 0;
2182}
2183static inline void bd_unlink_disk_holder(struct block_device *bdev,
2184                                         struct gendisk *disk)
2185{
2186}
2187#endif
2188#endif
2189
2190/* fs/char_dev.c */
2191#define CHRDEV_MAJOR_HASH_SIZE  255
2192extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2193extern int register_chrdev_region(dev_t, unsigned, const char *);
2194extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2195                             unsigned int count, const char *name,
2196                             const struct file_operations *fops);
2197extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2198                                unsigned int count, const char *name);
2199extern void unregister_chrdev_region(dev_t, unsigned);
2200extern void chrdev_show(struct seq_file *,off_t);
2201
2202static inline int register_chrdev(unsigned int major, const char *name,
2203                                  const struct file_operations *fops)
2204{
2205        return __register_chrdev(major, 0, 256, name, fops);
2206}
2207
2208static inline void unregister_chrdev(unsigned int major, const char *name)
2209{
2210        __unregister_chrdev(major, 0, 256, name);
2211}
2212
2213/* fs/block_dev.c */
2214#define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2215#define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2216
2217#ifdef CONFIG_BLOCK
2218#define BLKDEV_MAJOR_HASH_SIZE  255
2219extern const char *__bdevname(dev_t, char *buffer);
2220extern const char *bdevname(struct block_device *bdev, char *buffer);
2221extern struct block_device *lookup_bdev(const char *);
2222extern void blkdev_show(struct seq_file *,off_t);
2223
2224#else
2225#define BLKDEV_MAJOR_HASH_SIZE  0
2226#endif
2227
2228extern void init_special_inode(struct inode *, umode_t, dev_t);
2229
2230/* Invalid inode operations -- fs/bad_inode.c */
2231extern void make_bad_inode(struct inode *);
2232extern int is_bad_inode(struct inode *);
2233
2234#ifdef CONFIG_BLOCK
2235/*
2236 * return READ, READA, or WRITE
2237 */
2238#define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2239
2240/*
2241 * return data direction, READ or WRITE
2242 */
2243#define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2244
2245extern void check_disk_size_change(struct gendisk *disk,
2246                                   struct block_device *bdev);
2247extern int revalidate_disk(struct gendisk *);
2248extern int check_disk_change(struct block_device *);
2249extern int __invalidate_device(struct block_device *, bool);
2250extern int invalidate_partition(struct gendisk *, int);
2251#endif
2252unsigned long invalidate_mapping_pages(struct address_space *mapping,
2253                                        pgoff_t start, pgoff_t end);
2254
2255static inline void invalidate_remote_inode(struct inode *inode)
2256{
2257        if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2258            S_ISLNK(inode->i_mode))
2259                invalidate_mapping_pages(inode->i_mapping, 0, -1);
2260}
2261extern int invalidate_inode_pages2(struct address_space *mapping);
2262extern int invalidate_inode_pages2_range(struct address_space *mapping,
2263                                         pgoff_t start, pgoff_t end);
2264extern int write_inode_now(struct inode *, int);
2265extern int filemap_fdatawrite(struct address_space *);
2266extern int filemap_flush(struct address_space *);
2267extern int filemap_fdatawait(struct address_space *);
2268extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2269                                   loff_t lend);
2270extern int filemap_write_and_wait(struct address_space *mapping);
2271extern int filemap_write_and_wait_range(struct address_space *mapping,
2272                                        loff_t lstart, loff_t lend);
2273extern int __filemap_fdatawrite_range(struct address_space *mapping,
2274                                loff_t start, loff_t end, int sync_mode);
2275extern int filemap_fdatawrite_range(struct address_space *mapping,
2276                                loff_t start, loff_t end);
2277
2278extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2279                           int datasync);
2280extern int vfs_fsync(struct file *file, int datasync);
2281static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2282{
2283        if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2284                return 0;
2285        return vfs_fsync_range(file, pos, pos + count - 1,
2286                               (file->f_flags & __O_SYNC) ? 0 : 1);
2287}
2288extern void emergency_sync(void);
2289extern void emergency_remount(void);
2290#ifdef CONFIG_BLOCK
2291extern sector_t bmap(struct inode *, sector_t);
2292#endif
2293extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2294extern int inode_permission(struct inode *, int);
2295extern int generic_permission(struct inode *, int);
2296
2297static inline bool execute_ok(struct inode *inode)
2298{
2299        return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2300}
2301
2302static inline struct inode *file_inode(struct file *f)
2303{
2304        return f->f_inode;
2305}
2306
2307static inline void file_start_write(struct file *file)
2308{
2309        if (!S_ISREG(file_inode(file)->i_mode))
2310                return;
2311        __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2312}
2313
2314static inline bool file_start_write_trylock(struct file *file)
2315{
2316        if (!S_ISREG(file_inode(file)->i_mode))
2317                return true;
2318        return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2319}
2320
2321static inline void file_end_write(struct file *file)
2322{
2323        if (!S_ISREG(file_inode(file)->i_mode))
2324                return;
2325        __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2326}
2327
2328/*
2329 * get_write_access() gets write permission for a file.
2330 * put_write_access() releases this write permission.
2331 * This is used for regular files.
2332 * We cannot support write (and maybe mmap read-write shared) accesses and
2333 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2334 * can have the following values:
2335 * 0: no writers, no VM_DENYWRITE mappings
2336 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2337 * > 0: (i_writecount) users are writing to the file.
2338 *
2339 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2340 * except for the cases where we don't hold i_writecount yet. Then we need to
2341 * use {get,deny}_write_access() - these functions check the sign and refuse
2342 * to do the change if sign is wrong.
2343 */
2344static inline int get_write_access(struct inode *inode)
2345{
2346        return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2347}
2348static inline int deny_write_access(struct file *file)
2349{
2350        struct inode *inode = file_inode(file);
2351        return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2352}
2353static inline void put_write_access(struct inode * inode)
2354{
2355        atomic_dec(&inode->i_writecount);
2356}
2357static inline void allow_write_access(struct file *file)
2358{
2359        if (file)
2360                atomic_inc(&file_inode(file)->i_writecount);
2361}
2362static inline bool inode_is_open_for_write(const struct inode *inode)
2363{
2364        return atomic_read(&inode->i_writecount) > 0;
2365}
2366
2367#ifdef CONFIG_IMA
2368static inline void i_readcount_dec(struct inode *inode)
2369{
2370        BUG_ON(!atomic_read(&inode->i_readcount));
2371        atomic_dec(&inode->i_readcount);
2372}
2373static inline void i_readcount_inc(struct inode *inode)
2374{
2375        atomic_inc(&inode->i_readcount);
2376}
2377#else
2378static inline void i_readcount_dec(struct inode *inode)
2379{
2380        return;
2381}
2382static inline void i_readcount_inc(struct inode *inode)
2383{
2384        return;
2385}
2386#endif
2387extern int do_pipe_flags(int *, int);
2388
2389extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2390extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2391extern struct file * open_exec(const char *);
2392 
2393/* fs/dcache.c -- generic fs support functions */
2394extern int is_subdir(struct dentry *, struct dentry *);
2395extern int path_is_under(struct path *, struct path *);
2396
2397#include <linux/err.h>
2398
2399/* needed for stackable file system support */
2400extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2401
2402extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2403
2404extern int inode_init_always(struct super_block *, struct inode *);
2405extern void inode_init_once(struct inode *);
2406extern void address_space_init_once(struct address_space *mapping);
2407extern struct inode * igrab(struct inode *);
2408extern ino_t iunique(struct super_block *, ino_t);
2409extern int inode_needs_sync(struct inode *inode);
2410extern int generic_delete_inode(struct inode *inode);
2411static inline int generic_drop_inode(struct inode *inode)
2412{
2413        return !inode->i_nlink || inode_unhashed(inode);
2414}
2415
2416extern struct inode *ilookup5_nowait(struct super_block *sb,
2417                unsigned long hashval, int (*test)(struct inode *, void *),
2418                void *data);
2419extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2420                int (*test)(struct inode *, void *), void *data);
2421extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2422
2423extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2424extern struct inode * iget_locked(struct super_block *, unsigned long);
2425extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2426extern int insert_inode_locked(struct inode *);
2427#ifdef CONFIG_DEBUG_LOCK_ALLOC
2428extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2429#else
2430static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2431#endif
2432extern void unlock_new_inode(struct inode *);
2433extern unsigned int get_next_ino(void);
2434
2435extern void __iget(struct inode * inode);
2436extern void iget_failed(struct inode *);
2437extern void clear_inode(struct inode *);
2438extern void __destroy_inode(struct inode *);
2439extern struct inode *new_inode_pseudo(struct super_block *sb);
2440extern struct inode *new_inode(struct super_block *sb);
2441extern void free_inode_nonrcu(struct inode *inode);
2442extern int should_remove_suid(struct dentry *);
2443extern int file_remove_suid(struct file *);
2444
2445extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2446static inline void insert_inode_hash(struct inode *inode)
2447{
2448        __insert_inode_hash(inode, inode->i_ino);
2449}
2450
2451extern void __remove_inode_hash(struct inode *);
2452static inline void remove_inode_hash(struct inode *inode)
2453{
2454        if (!inode_unhashed(inode))
2455                __remove_inode_hash(inode);
2456}
2457
2458extern void inode_sb_list_add(struct inode *inode);
2459
2460#ifdef CONFIG_BLOCK
2461extern void submit_bio(int, struct bio *);
2462extern int bdev_read_only(struct block_device *);
2463#endif
2464extern int set_blocksize(struct block_device *, int);
2465extern int sb_set_blocksize(struct super_block *, int);
2466extern int sb_min_blocksize(struct super_block *, int);
2467
2468extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2469extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2470extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2471                unsigned long size, pgoff_t pgoff);
2472extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2473int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2474extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2475extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2476                loff_t *);
2477extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2478extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2479                unsigned long *, loff_t, loff_t *, size_t, size_t);
2480extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2481                unsigned long, loff_t, loff_t *, size_t, ssize_t);
2482extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2483extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2484extern int generic_segment_checks(const struct iovec *iov,
2485                unsigned long *nr_segs, size_t *count, int access_flags);
2486
2487/* fs/block_dev.c */
2488extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2489                                unsigned long nr_segs, loff_t pos);
2490extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2491                        int datasync);
2492extern void block_sync_page(struct page *page);
2493
2494/* fs/splice.c */
2495extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2496                struct pipe_inode_info *, size_t, unsigned int);
2497extern ssize_t default_file_splice_read(struct file *, loff_t *,
2498                struct pipe_inode_info *, size_t, unsigned int);
2499extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2500                struct file *, loff_t *, size_t, unsigned int);
2501extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2502                struct file *out, loff_t *, size_t len, unsigned int flags);
2503
2504extern void
2505file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2506extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2507extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2508extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2509extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2510extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2511                int whence, loff_t maxsize, loff_t eof);
2512extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2513                int whence, loff_t size);
2514extern int generic_file_open(struct inode * inode, struct file * filp);
2515extern int nonseekable_open(struct inode * inode, struct file * filp);
2516
2517#ifdef CONFIG_FS_XIP
2518extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2519                             loff_t *ppos);
2520extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2521extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2522                              size_t len, loff_t *ppos);
2523extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2524#else
2525static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2526{
2527        return 0;
2528}
2529#endif
2530
2531#ifdef CONFIG_BLOCK
2532typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2533                            loff_t file_offset);
2534
2535enum {
2536        /* need locking between buffered and direct access */
2537        DIO_LOCKING     = 0x01,
2538
2539        /* filesystem does not support filling holes */
2540        DIO_SKIP_HOLES  = 0x02,
2541};
2542
2543void dio_end_io(struct bio *bio, int error);
2544
2545ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2546        struct block_device *bdev, const struct iovec *iov, loff_t offset,
2547        unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2548        dio_submit_t submit_io, int flags);
2549
2550static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2551                struct inode *inode, const struct iovec *iov, loff_t offset,
2552                unsigned long nr_segs, get_block_t get_block)
2553{
2554        return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2555                                    offset, nr_segs, get_block, NULL, NULL,
2556                                    DIO_LOCKING | DIO_SKIP_HOLES);
2557}
2558#endif
2559
2560void inode_dio_wait(struct inode *inode);
2561void inode_dio_done(struct inode *inode);
2562
2563extern const struct file_operations generic_ro_fops;
2564
2565#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2566
2567extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2568extern int page_readlink(struct dentry *, char __user *, int);
2569extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2570extern void page_put_link(struct dentry *, struct nameidata *, void *);
2571extern int __page_symlink(struct inode *inode, const char *symname, int len,
2572                int nofs);
2573extern int page_symlink(struct inode *inode, const char *symname, int len);
2574extern const struct inode_operations page_symlink_inode_operations;
2575extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
2576extern int generic_readlink(struct dentry *, char __user *, int);
2577extern void generic_fillattr(struct inode *, struct kstat *);
2578int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2579extern int vfs_getattr(struct path *, struct kstat *);
2580void __inode_add_bytes(struct inode *inode, loff_t bytes);
2581void inode_add_bytes(struct inode *inode, loff_t bytes);
2582void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2583void inode_sub_bytes(struct inode *inode, loff_t bytes);
2584loff_t inode_get_bytes(struct inode *inode);
2585void inode_set_bytes(struct inode *inode, loff_t bytes);
2586
2587extern int vfs_readdir(struct file *, filldir_t, void *);
2588extern int iterate_dir(struct file *, struct dir_context *);
2589
2590extern int vfs_stat(const char __user *, struct kstat *);
2591extern int vfs_lstat(const char __user *, struct kstat *);
2592extern int vfs_fstat(unsigned int, struct kstat *);
2593extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2594
2595extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2596                    unsigned long arg);
2597extern int __generic_block_fiemap(struct inode *inode,
2598                                  struct fiemap_extent_info *fieinfo,
2599                                  loff_t start, loff_t len,
2600                                  get_block_t *get_block);
2601extern int generic_block_fiemap(struct inode *inode,
2602                                struct fiemap_extent_info *fieinfo, u64 start,
2603                                u64 len, get_block_t *get_block);
2604
2605extern void get_filesystem(struct file_system_type *fs);
2606extern void put_filesystem(struct file_system_type *fs);
2607extern struct file_system_type *get_fs_type(const char *name);
2608extern struct super_block *get_super(struct block_device *);
2609extern struct super_block *get_super_thawed(struct block_device *);
2610extern struct super_block *get_active_super(struct block_device *bdev);
2611extern void drop_super(struct super_block *sb);
2612extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2613extern void iterate_supers_type(struct file_system_type *,
2614                                void (*)(struct super_block *, void *), void *);
2615
2616extern int dcache_dir_open(struct inode *, struct file *);
2617extern int dcache_dir_close(struct inode *, struct file *);
2618extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2619extern int dcache_readdir(struct file *, struct dir_context *);
2620extern int simple_setattr(struct dentry *, struct iattr *);
2621extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2622extern int simple_statfs(struct dentry *, struct kstatfs *);
2623extern int simple_open(struct inode *inode, struct file *file);
2624extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2625extern int simple_unlink(struct inode *, struct dentry *);
2626extern int simple_rmdir(struct inode *, struct dentry *);
2627extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2628extern int noop_fsync(struct file *, loff_t, loff_t, int);
2629extern int simple_empty(struct dentry *);
2630extern int simple_readpage(struct file *file, struct page *page);
2631extern int simple_write_begin(struct file *file, struct address_space *mapping,
2632                        loff_t pos, unsigned len, unsigned flags,
2633                        struct page **pagep, void **fsdata);
2634extern int simple_write_end(struct file *file, struct address_space *mapping,
2635                        loff_t pos, unsigned len, unsigned copied,
2636                        struct page *page, void *fsdata);
2637extern int always_delete_dentry(const struct dentry *);
2638extern struct inode *alloc_anon_inode(struct super_block *);
2639extern const struct dentry_operations simple_dentry_operations;
2640
2641extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2642extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2643extern const struct file_operations simple_dir_operations;
2644extern const struct inode_operations simple_dir_inode_operations;
2645struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2646struct dentry *d_alloc_name(struct dentry *, const char *);
2647extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2648extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2649extern void simple_release_fs(struct vfsmount **mount, int *count);
2650
2651extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2652                        loff_t *ppos, const void *from, size_t available);
2653extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2654                const void __user *from, size_t count);
2655
2656extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2657
2658extern int generic_check_addressable(unsigned, u64);
2659
2660#ifdef CONFIG_MIGRATION
2661extern int buffer_migrate_page(struct address_space *,
2662                                struct page *, struct page *,
2663                                enum migrate_mode);
2664#else
2665#define buffer_migrate_page NULL
2666#endif
2667
2668extern int inode_change_ok(const struct inode *, struct iattr *);
2669extern int inode_newsize_ok(const struct inode *, loff_t offset);
2670extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2671
2672extern int file_update_time(struct file *file);
2673
2674extern int generic_show_options(struct seq_file *m, struct dentry *root);
2675extern void save_mount_options(struct super_block *sb, char *options);
2676extern void replace_mount_options(struct super_block *sb, char *options);
2677
2678static inline ino_t parent_ino(struct dentry *dentry)
2679{
2680        ino_t res;
2681
2682        /*
2683         * Don't strictly need d_lock here? If the parent ino could change
2684         * then surely we'd have a deeper race in the caller?
2685         */
2686        spin_lock(&dentry->d_lock);
2687        res = dentry->d_parent->d_inode->i_ino;
2688        spin_unlock(&dentry->d_lock);
2689        return res;
2690}
2691
2692/* Transaction based IO helpers */
2693
2694/*
2695 * An argresp is stored in an allocated page and holds the
2696 * size of the argument or response, along with its content
2697 */
2698struct simple_transaction_argresp {
2699        ssize_t size;
2700        char data[0];
2701};
2702
2703#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2704
2705char *simple_transaction_get(struct file *file, const char __user *buf,
2706                                size_t size);
2707ssize_t simple_transaction_read(struct file *file, char __user *buf,
2708                                size_t size, loff_t *pos);
2709int simple_transaction_release(struct inode *inode, struct file *file);
2710
2711void simple_transaction_set(struct file *file, size_t n);
2712
2713/*
2714 * simple attribute files
2715 *
2716 * These attributes behave similar to those in sysfs:
2717 *
2718 * Writing to an attribute immediately sets a value, an open file can be
2719 * written to multiple times.
2720 *
2721 * Reading from an attribute creates a buffer from the value that might get
2722 * read with multiple read calls. When the attribute has been read
2723 * completely, no further read calls are possible until the file is opened
2724 * again.
2725 *
2726 * All attributes contain a text representation of a numeric value
2727 * that are accessed with the get() and set() functions.
2728 */
2729#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2730static int __fops ## _open(struct inode *inode, struct file *file)      \
2731{                                                                       \
2732        __simple_attr_check_format(__fmt, 0ull);                        \
2733        return simple_attr_open(inode, file, __get, __set, __fmt);      \
2734}                                                                       \
2735static const struct file_operations __fops = {                          \
2736        .owner   = THIS_MODULE,                                         \
2737        .open    = __fops ## _open,                                     \
2738        .release = simple_attr_release,                                 \
2739        .read    = simple_attr_read,                                    \
2740        .write   = simple_attr_write,                                   \
2741        .llseek  = generic_file_llseek,                                 \
2742};
2743
2744static inline __printf(1, 2)
2745void __simple_attr_check_format(const char *fmt, ...)
2746{
2747        /* don't do anything, just let the compiler check the arguments; */
2748}
2749
2750int simple_attr_open(struct inode *inode, struct file *file,
2751                     int (*get)(void *, u64 *), int (*set)(void *, u64),
2752                     const char *fmt);
2753int simple_attr_release(struct inode *inode, struct file *file);
2754ssize_t simple_attr_read(struct file *file, char __user *buf,
2755                         size_t len, loff_t *ppos);
2756ssize_t simple_attr_write(struct file *file, const char __user *buf,
2757                          size_t len, loff_t *ppos);
2758
2759struct ctl_table;
2760int proc_nr_files(struct ctl_table *table, int write,
2761                  void __user *buffer, size_t *lenp, loff_t *ppos);
2762int proc_nr_dentry(struct ctl_table *table, int write,
2763                  void __user *buffer, size_t *lenp, loff_t *ppos);
2764int proc_nr_inodes(struct ctl_table *table, int write,
2765                   void __user *buffer, size_t *lenp, loff_t *ppos);
2766int __init get_filesystem_list(char *buf);
2767
2768#define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2769#define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2770
2771#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2772#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2773                                            (flag & __FMODE_NONOTIFY)))
2774
2775static inline int is_sxid(umode_t mode)
2776{
2777        return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2778}
2779
2780static inline void inode_has_no_xattr(struct inode *inode)
2781{
2782        if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2783                inode->i_flags |= S_NOSEC;
2784}
2785
2786static inline bool dir_emit(struct dir_context *ctx,
2787                            const char *name, int namelen,
2788                            u64 ino, unsigned type)
2789{
2790        return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2791}
2792static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2793{
2794        return ctx->actor(ctx, ".", 1, ctx->pos,
2795                          file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2796}
2797static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2798{
2799        return ctx->actor(ctx, "..", 2, ctx->pos,
2800                          parent_ino(file->f_path.dentry), DT_DIR) == 0;
2801}
2802static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2803{
2804        if (ctx->pos == 0) {
2805                if (!dir_emit_dot(file, ctx))
2806                        return false;
2807                ctx->pos = 1;
2808        }
2809        if (ctx->pos == 1) {
2810                if (!dir_emit_dotdot(file, ctx))
2811                        return false;
2812                ctx->pos = 2;
2813        }
2814        return true;
2815}
2816static inline bool dir_relax(struct inode *inode)
2817{
2818        mutex_unlock(&inode->i_mutex);
2819        mutex_lock(&inode->i_mutex);
2820        return !IS_DEADDIR(inode);
2821}
2822
2823#endif /* _LINUX_FS_H */
2824