linux/include/linux/fs.h
<<
>>
Prefs
   1#ifndef _LINUX_FS_H
   2#define _LINUX_FS_H
   3
   4
   5#include <linux/linkage.h>
   6#include <linux/wait.h>
   7#include <linux/kdev_t.h>
   8#include <linux/dcache.h>
   9#include <linux/path.h>
  10#include <linux/stat.h>
  11#include <linux/cache.h>
  12#include <linux/list.h>
  13#include <linux/radix-tree.h>
  14#include <linux/rbtree.h>
  15#include <linux/init.h>
  16#include <linux/pid.h>
  17#include <linux/bug.h>
  18#include <linux/mutex.h>
  19#include <linux/capability.h>
  20#include <linux/semaphore.h>
  21#include <linux/fiemap.h>
  22#include <linux/rculist_bl.h>
  23#include <linux/atomic.h>
  24#include <linux/shrinker.h>
  25#include <linux/migrate_mode.h>
  26#include <linux/uidgid.h>
  27#include <linux/lockdep.h>
  28#include <linux/percpu-rwsem.h>
  29#include <linux/blk_types.h>
  30
  31#include <asm/byteorder.h>
  32#include <uapi/linux/fs.h>
  33
  34struct export_operations;
  35struct hd_geometry;
  36struct iovec;
  37struct nameidata;
  38struct kiocb;
  39struct kobject;
  40struct pipe_inode_info;
  41struct poll_table_struct;
  42struct kstatfs;
  43struct vm_area_struct;
  44struct vfsmount;
  45struct cred;
  46struct swap_info_struct;
  47
  48extern void __init inode_init(void);
  49extern void __init inode_init_early(void);
  50extern void __init files_init(unsigned long);
  51
  52extern struct files_stat_struct files_stat;
  53extern unsigned long get_max_files(void);
  54extern int sysctl_nr_open;
  55extern struct inodes_stat_t inodes_stat;
  56extern int leases_enable, lease_break_time;
  57extern int sysctl_protected_symlinks;
  58extern int sysctl_protected_hardlinks;
  59
  60struct buffer_head;
  61typedef int (get_block_t)(struct inode *inode, sector_t iblock,
  62                        struct buffer_head *bh_result, int create);
  63typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
  64                        ssize_t bytes, void *private, int ret,
  65                        bool is_async);
  66
  67#define MAY_EXEC                0x00000001
  68#define MAY_WRITE               0x00000002
  69#define MAY_READ                0x00000004
  70#define MAY_APPEND              0x00000008
  71#define MAY_ACCESS              0x00000010
  72#define MAY_OPEN                0x00000020
  73#define MAY_CHDIR               0x00000040
  74/* called from RCU mode, don't block */
  75#define MAY_NOT_BLOCK           0x00000080
  76
  77/*
  78 * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
  79 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
  80 */
  81
  82/* file is open for reading */
  83#define FMODE_READ              ((__force fmode_t)0x1)
  84/* file is open for writing */
  85#define FMODE_WRITE             ((__force fmode_t)0x2)
  86/* file is seekable */
  87#define FMODE_LSEEK             ((__force fmode_t)0x4)
  88/* file can be accessed using pread */
  89#define FMODE_PREAD             ((__force fmode_t)0x8)
  90/* file can be accessed using pwrite */
  91#define FMODE_PWRITE            ((__force fmode_t)0x10)
  92/* File is opened for execution with sys_execve / sys_uselib */
  93#define FMODE_EXEC              ((__force fmode_t)0x20)
  94/* File is opened with O_NDELAY (only set for block devices) */
  95#define FMODE_NDELAY            ((__force fmode_t)0x40)
  96/* File is opened with O_EXCL (only set for block devices) */
  97#define FMODE_EXCL              ((__force fmode_t)0x80)
  98/* File is opened using open(.., 3, ..) and is writeable only for ioctls
  99   (specialy hack for floppy.c) */
 100#define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
 101/* 32bit hashes as llseek() offset (for directories) */
 102#define FMODE_32BITHASH         ((__force fmode_t)0x200)
 103/* 64bit hashes as llseek() offset (for directories) */
 104#define FMODE_64BITHASH         ((__force fmode_t)0x400)
 105
 106/*
 107 * Don't update ctime and mtime.
 108 *
 109 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
 110 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
 111 */
 112#define FMODE_NOCMTIME          ((__force fmode_t)0x800)
 113
 114/* Expect random access pattern */
 115#define FMODE_RANDOM            ((__force fmode_t)0x1000)
 116
 117/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
 118#define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
 119
 120/* File is opened with O_PATH; almost nothing can be done with it */
 121#define FMODE_PATH              ((__force fmode_t)0x4000)
 122
 123/* File was opened by fanotify and shouldn't generate fanotify events */
 124#define FMODE_NONOTIFY          ((__force fmode_t)0x1000000)
 125
 126/*
 127 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
 128 * that indicates that they should check the contents of the iovec are
 129 * valid, but not check the memory that the iovec elements
 130 * points too.
 131 */
 132#define CHECK_IOVEC_ONLY -1
 133
 134/*
 135 * The below are the various read and write types that we support. Some of
 136 * them include behavioral modifiers that send information down to the
 137 * block layer and IO scheduler. Terminology:
 138 *
 139 *      The block layer uses device plugging to defer IO a little bit, in
 140 *      the hope that we will see more IO very shortly. This increases
 141 *      coalescing of adjacent IO and thus reduces the number of IOs we
 142 *      have to send to the device. It also allows for better queuing,
 143 *      if the IO isn't mergeable. If the caller is going to be waiting
 144 *      for the IO, then he must ensure that the device is unplugged so
 145 *      that the IO is dispatched to the driver.
 146 *
 147 *      All IO is handled async in Linux. This is fine for background
 148 *      writes, but for reads or writes that someone waits for completion
 149 *      on, we want to notify the block layer and IO scheduler so that they
 150 *      know about it. That allows them to make better scheduling
 151 *      decisions. So when the below references 'sync' and 'async', it
 152 *      is referencing this priority hint.
 153 *
 154 * With that in mind, the available types are:
 155 *
 156 * READ                 A normal read operation. Device will be plugged.
 157 * READ_SYNC            A synchronous read. Device is not plugged, caller can
 158 *                      immediately wait on this read without caring about
 159 *                      unplugging.
 160 * READA                Used for read-ahead operations. Lower priority, and the
 161 *                      block layer could (in theory) choose to ignore this
 162 *                      request if it runs into resource problems.
 163 * WRITE                A normal async write. Device will be plugged.
 164 * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
 165 *                      the hint that someone will be waiting on this IO
 166 *                      shortly. The write equivalent of READ_SYNC.
 167 * WRITE_ODIRECT        Special case write for O_DIRECT only.
 168 * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
 169 * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
 170 *                      non-volatile media on completion.
 171 * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
 172 *                      by a cache flush and data is guaranteed to be on
 173 *                      non-volatile media on completion.
 174 *
 175 */
 176#define RW_MASK                 REQ_WRITE
 177#define RWA_MASK                REQ_RAHEAD
 178
 179#define READ                    0
 180#define WRITE                   RW_MASK
 181#define READA                   RWA_MASK
 182#define KERNEL_READ             (READ|REQ_KERNEL)
 183#define KERNEL_WRITE            (WRITE|REQ_KERNEL)
 184
 185#define READ_SYNC               (READ | REQ_SYNC)
 186#define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
 187#define WRITE_ODIRECT           (WRITE | REQ_SYNC)
 188#define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
 189#define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
 190#define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
 191
 192/*
 193 * Attribute flags.  These should be or-ed together to figure out what
 194 * has been changed!
 195 */
 196#define ATTR_MODE       (1 << 0)
 197#define ATTR_UID        (1 << 1)
 198#define ATTR_GID        (1 << 2)
 199#define ATTR_SIZE       (1 << 3)
 200#define ATTR_ATIME      (1 << 4)
 201#define ATTR_MTIME      (1 << 5)
 202#define ATTR_CTIME      (1 << 6)
 203#define ATTR_ATIME_SET  (1 << 7)
 204#define ATTR_MTIME_SET  (1 << 8)
 205#define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
 206#define ATTR_ATTR_FLAG  (1 << 10)
 207#define ATTR_KILL_SUID  (1 << 11)
 208#define ATTR_KILL_SGID  (1 << 12)
 209#define ATTR_FILE       (1 << 13)
 210#define ATTR_KILL_PRIV  (1 << 14)
 211#define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
 212#define ATTR_TIMES_SET  (1 << 16)
 213
 214/*
 215 * This is the Inode Attributes structure, used for notify_change().  It
 216 * uses the above definitions as flags, to know which values have changed.
 217 * Also, in this manner, a Filesystem can look at only the values it cares
 218 * about.  Basically, these are the attributes that the VFS layer can
 219 * request to change from the FS layer.
 220 *
 221 * Derek Atkins <warlord@MIT.EDU> 94-10-20
 222 */
 223struct iattr {
 224        unsigned int    ia_valid;
 225        umode_t         ia_mode;
 226        kuid_t          ia_uid;
 227        kgid_t          ia_gid;
 228        loff_t          ia_size;
 229        struct timespec ia_atime;
 230        struct timespec ia_mtime;
 231        struct timespec ia_ctime;
 232
 233        /*
 234         * Not an attribute, but an auxiliary info for filesystems wanting to
 235         * implement an ftruncate() like method.  NOTE: filesystem should
 236         * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
 237         */
 238        struct file     *ia_file;
 239};
 240
 241/*
 242 * Includes for diskquotas.
 243 */
 244#include <linux/quota.h>
 245
 246/** 
 247 * enum positive_aop_returns - aop return codes with specific semantics
 248 *
 249 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
 250 *                          completed, that the page is still locked, and
 251 *                          should be considered active.  The VM uses this hint
 252 *                          to return the page to the active list -- it won't
 253 *                          be a candidate for writeback again in the near
 254 *                          future.  Other callers must be careful to unlock
 255 *                          the page if they get this return.  Returned by
 256 *                          writepage(); 
 257 *
 258 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
 259 *                      unlocked it and the page might have been truncated.
 260 *                      The caller should back up to acquiring a new page and
 261 *                      trying again.  The aop will be taking reasonable
 262 *                      precautions not to livelock.  If the caller held a page
 263 *                      reference, it should drop it before retrying.  Returned
 264 *                      by readpage().
 265 *
 266 * address_space_operation functions return these large constants to indicate
 267 * special semantics to the caller.  These are much larger than the bytes in a
 268 * page to allow for functions that return the number of bytes operated on in a
 269 * given page.
 270 */
 271
 272enum positive_aop_returns {
 273        AOP_WRITEPAGE_ACTIVATE  = 0x80000,
 274        AOP_TRUNCATED_PAGE      = 0x80001,
 275};
 276
 277#define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
 278#define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
 279#define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
 280                                                * helper code (eg buffer layer)
 281                                                * to clear GFP_FS from alloc */
 282
 283/*
 284 * oh the beauties of C type declarations.
 285 */
 286struct page;
 287struct address_space;
 288struct writeback_control;
 289
 290struct iov_iter {
 291        const struct iovec *iov;
 292        unsigned long nr_segs;
 293        size_t iov_offset;
 294        size_t count;
 295};
 296
 297size_t iov_iter_copy_from_user_atomic(struct page *page,
 298                struct iov_iter *i, unsigned long offset, size_t bytes);
 299size_t iov_iter_copy_from_user(struct page *page,
 300                struct iov_iter *i, unsigned long offset, size_t bytes);
 301void iov_iter_advance(struct iov_iter *i, size_t bytes);
 302int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
 303size_t iov_iter_single_seg_count(struct iov_iter *i);
 304
 305static inline void iov_iter_init(struct iov_iter *i,
 306                        const struct iovec *iov, unsigned long nr_segs,
 307                        size_t count, size_t written)
 308{
 309        i->iov = iov;
 310        i->nr_segs = nr_segs;
 311        i->iov_offset = 0;
 312        i->count = count + written;
 313
 314        iov_iter_advance(i, written);
 315}
 316
 317static inline size_t iov_iter_count(struct iov_iter *i)
 318{
 319        return i->count;
 320}
 321
 322/*
 323 * "descriptor" for what we're up to with a read.
 324 * This allows us to use the same read code yet
 325 * have multiple different users of the data that
 326 * we read from a file.
 327 *
 328 * The simplest case just copies the data to user
 329 * mode.
 330 */
 331typedef struct {
 332        size_t written;
 333        size_t count;
 334        union {
 335                char __user *buf;
 336                void *data;
 337        } arg;
 338        int error;
 339} read_descriptor_t;
 340
 341typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
 342                unsigned long, unsigned long);
 343
 344struct address_space_operations {
 345        int (*writepage)(struct page *page, struct writeback_control *wbc);
 346        int (*readpage)(struct file *, struct page *);
 347
 348        /* Write back some dirty pages from this mapping. */
 349        int (*writepages)(struct address_space *, struct writeback_control *);
 350
 351        /* Set a page dirty.  Return true if this dirtied it */
 352        int (*set_page_dirty)(struct page *page);
 353
 354        int (*readpages)(struct file *filp, struct address_space *mapping,
 355                        struct list_head *pages, unsigned nr_pages);
 356
 357        int (*write_begin)(struct file *, struct address_space *mapping,
 358                                loff_t pos, unsigned len, unsigned flags,
 359                                struct page **pagep, void **fsdata);
 360        int (*write_end)(struct file *, struct address_space *mapping,
 361                                loff_t pos, unsigned len, unsigned copied,
 362                                struct page *page, void *fsdata);
 363
 364        /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
 365        sector_t (*bmap)(struct address_space *, sector_t);
 366        void (*invalidatepage) (struct page *, unsigned long);
 367        int (*releasepage) (struct page *, gfp_t);
 368        void (*freepage)(struct page *);
 369        ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
 370                        loff_t offset, unsigned long nr_segs);
 371        int (*get_xip_mem)(struct address_space *, pgoff_t, int,
 372                                                void **, unsigned long *);
 373        /*
 374         * migrate the contents of a page to the specified target. If sync
 375         * is false, it must not block.
 376         */
 377        int (*migratepage) (struct address_space *,
 378                        struct page *, struct page *, enum migrate_mode);
 379        int (*launder_page) (struct page *);
 380        int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
 381                                        unsigned long);
 382        int (*error_remove_page)(struct address_space *, struct page *);
 383
 384        /* swapfile support */
 385        int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
 386                                sector_t *span);
 387        void (*swap_deactivate)(struct file *file);
 388};
 389
 390extern const struct address_space_operations empty_aops;
 391
 392/*
 393 * pagecache_write_begin/pagecache_write_end must be used by general code
 394 * to write into the pagecache.
 395 */
 396int pagecache_write_begin(struct file *, struct address_space *mapping,
 397                                loff_t pos, unsigned len, unsigned flags,
 398                                struct page **pagep, void **fsdata);
 399
 400int pagecache_write_end(struct file *, struct address_space *mapping,
 401                                loff_t pos, unsigned len, unsigned copied,
 402                                struct page *page, void *fsdata);
 403
 404struct backing_dev_info;
 405struct address_space {
 406        struct inode            *host;          /* owner: inode, block_device */
 407        struct radix_tree_root  page_tree;      /* radix tree of all pages */
 408        spinlock_t              tree_lock;      /* and lock protecting it */
 409        unsigned int            i_mmap_writable;/* count VM_SHARED mappings */
 410        struct rb_root          i_mmap;         /* tree of private and shared mappings */
 411        struct list_head        i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
 412        struct mutex            i_mmap_mutex;   /* protect tree, count, list */
 413        /* Protected by tree_lock together with the radix tree */
 414        unsigned long           nrpages;        /* number of total pages */
 415        pgoff_t                 writeback_index;/* writeback starts here */
 416        const struct address_space_operations *a_ops;   /* methods */
 417        unsigned long           flags;          /* error bits/gfp mask */
 418        struct backing_dev_info *backing_dev_info; /* device readahead, etc */
 419        spinlock_t              private_lock;   /* for use by the address_space */
 420        struct list_head        private_list;   /* ditto */
 421        struct address_space    *assoc_mapping; /* ditto */
 422} __attribute__((aligned(sizeof(long))));
 423        /*
 424         * On most architectures that alignment is already the case; but
 425         * must be enforced here for CRIS, to let the least significant bit
 426         * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
 427         */
 428struct request_queue;
 429
 430struct block_device {
 431        dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
 432        int                     bd_openers;
 433        struct inode *          bd_inode;       /* will die */
 434        struct super_block *    bd_super;
 435        struct mutex            bd_mutex;       /* open/close mutex */
 436        struct list_head        bd_inodes;
 437        void *                  bd_claiming;
 438        void *                  bd_holder;
 439        int                     bd_holders;
 440        bool                    bd_write_holder;
 441#ifdef CONFIG_SYSFS
 442        struct list_head        bd_holder_disks;
 443#endif
 444        struct block_device *   bd_contains;
 445        unsigned                bd_block_size;
 446        struct hd_struct *      bd_part;
 447        /* number of times partitions within this device have been opened. */
 448        unsigned                bd_part_count;
 449        int                     bd_invalidated;
 450        struct gendisk *        bd_disk;
 451        struct request_queue *  bd_queue;
 452        struct list_head        bd_list;
 453        /*
 454         * Private data.  You must have bd_claim'ed the block_device
 455         * to use this.  NOTE:  bd_claim allows an owner to claim
 456         * the same device multiple times, the owner must take special
 457         * care to not mess up bd_private for that case.
 458         */
 459        unsigned long           bd_private;
 460
 461        /* The counter of freeze processes */
 462        int                     bd_fsfreeze_count;
 463        /* Mutex for freeze */
 464        struct mutex            bd_fsfreeze_mutex;
 465};
 466
 467/*
 468 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
 469 * radix trees
 470 */
 471#define PAGECACHE_TAG_DIRTY     0
 472#define PAGECACHE_TAG_WRITEBACK 1
 473#define PAGECACHE_TAG_TOWRITE   2
 474
 475int mapping_tagged(struct address_space *mapping, int tag);
 476
 477/*
 478 * Might pages of this file be mapped into userspace?
 479 */
 480static inline int mapping_mapped(struct address_space *mapping)
 481{
 482        return  !RB_EMPTY_ROOT(&mapping->i_mmap) ||
 483                !list_empty(&mapping->i_mmap_nonlinear);
 484}
 485
 486/*
 487 * Might pages of this file have been modified in userspace?
 488 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
 489 * marks vma as VM_SHARED if it is shared, and the file was opened for
 490 * writing i.e. vma may be mprotected writable even if now readonly.
 491 */
 492static inline int mapping_writably_mapped(struct address_space *mapping)
 493{
 494        return mapping->i_mmap_writable != 0;
 495}
 496
 497/*
 498 * Use sequence counter to get consistent i_size on 32-bit processors.
 499 */
 500#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 501#include <linux/seqlock.h>
 502#define __NEED_I_SIZE_ORDERED
 503#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
 504#else
 505#define i_size_ordered_init(inode) do { } while (0)
 506#endif
 507
 508struct posix_acl;
 509#define ACL_NOT_CACHED ((void *)(-1))
 510
 511#define IOP_FASTPERM    0x0001
 512#define IOP_LOOKUP      0x0002
 513#define IOP_NOFOLLOW    0x0004
 514
 515/*
 516 * Keep mostly read-only and often accessed (especially for
 517 * the RCU path lookup and 'stat' data) fields at the beginning
 518 * of the 'struct inode'
 519 */
 520struct inode {
 521        umode_t                 i_mode;
 522        unsigned short          i_opflags;
 523        kuid_t                  i_uid;
 524        kgid_t                  i_gid;
 525        unsigned int            i_flags;
 526
 527#ifdef CONFIG_FS_POSIX_ACL
 528        struct posix_acl        *i_acl;
 529        struct posix_acl        *i_default_acl;
 530#endif
 531
 532        const struct inode_operations   *i_op;
 533        struct super_block      *i_sb;
 534        struct address_space    *i_mapping;
 535
 536#ifdef CONFIG_SECURITY
 537        void                    *i_security;
 538#endif
 539
 540        /* Stat data, not accessed from path walking */
 541        unsigned long           i_ino;
 542        /*
 543         * Filesystems may only read i_nlink directly.  They shall use the
 544         * following functions for modification:
 545         *
 546         *    (set|clear|inc|drop)_nlink
 547         *    inode_(inc|dec)_link_count
 548         */
 549        union {
 550                const unsigned int i_nlink;
 551                unsigned int __i_nlink;
 552        };
 553        dev_t                   i_rdev;
 554        loff_t                  i_size;
 555        struct timespec         i_atime;
 556        struct timespec         i_mtime;
 557        struct timespec         i_ctime;
 558        spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
 559        unsigned short          i_bytes;
 560        unsigned int            i_blkbits;
 561        blkcnt_t                i_blocks;
 562
 563#ifdef __NEED_I_SIZE_ORDERED
 564        seqcount_t              i_size_seqcount;
 565#endif
 566
 567        /* Misc */
 568        unsigned long           i_state;
 569        struct mutex            i_mutex;
 570
 571        unsigned long           dirtied_when;   /* jiffies of first dirtying */
 572
 573        struct hlist_node       i_hash;
 574        struct list_head        i_wb_list;      /* backing dev IO list */
 575        struct list_head        i_lru;          /* inode LRU list */
 576        struct list_head        i_sb_list;
 577        union {
 578                struct hlist_head       i_dentry;
 579                struct rcu_head         i_rcu;
 580        };
 581        u64                     i_version;
 582        atomic_t                i_count;
 583        atomic_t                i_dio_count;
 584        atomic_t                i_writecount;
 585        const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
 586        struct file_lock        *i_flock;
 587        struct address_space    i_data;
 588#ifdef CONFIG_QUOTA
 589        struct dquot            *i_dquot[MAXQUOTAS];
 590#endif
 591        struct list_head        i_devices;
 592        union {
 593                struct pipe_inode_info  *i_pipe;
 594                struct block_device     *i_bdev;
 595                struct cdev             *i_cdev;
 596        };
 597
 598        __u32                   i_generation;
 599
 600#ifdef CONFIG_FSNOTIFY
 601        __u32                   i_fsnotify_mask; /* all events this inode cares about */
 602        struct hlist_head       i_fsnotify_marks;
 603#endif
 604
 605#ifdef CONFIG_IMA
 606        atomic_t                i_readcount; /* struct files open RO */
 607#endif
 608        void                    *i_private; /* fs or device private pointer */
 609};
 610
 611static inline int inode_unhashed(struct inode *inode)
 612{
 613        return hlist_unhashed(&inode->i_hash);
 614}
 615
 616/*
 617 * inode->i_mutex nesting subclasses for the lock validator:
 618 *
 619 * 0: the object of the current VFS operation
 620 * 1: parent
 621 * 2: child/target
 622 * 3: quota file
 623 *
 624 * The locking order between these classes is
 625 * parent -> child -> normal -> xattr -> quota
 626 */
 627enum inode_i_mutex_lock_class
 628{
 629        I_MUTEX_NORMAL,
 630        I_MUTEX_PARENT,
 631        I_MUTEX_CHILD,
 632        I_MUTEX_XATTR,
 633        I_MUTEX_QUOTA
 634};
 635
 636/*
 637 * NOTE: in a 32bit arch with a preemptable kernel and
 638 * an UP compile the i_size_read/write must be atomic
 639 * with respect to the local cpu (unlike with preempt disabled),
 640 * but they don't need to be atomic with respect to other cpus like in
 641 * true SMP (so they need either to either locally disable irq around
 642 * the read or for example on x86 they can be still implemented as a
 643 * cmpxchg8b without the need of the lock prefix). For SMP compiles
 644 * and 64bit archs it makes no difference if preempt is enabled or not.
 645 */
 646static inline loff_t i_size_read(const struct inode *inode)
 647{
 648#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 649        loff_t i_size;
 650        unsigned int seq;
 651
 652        do {
 653                seq = read_seqcount_begin(&inode->i_size_seqcount);
 654                i_size = inode->i_size;
 655        } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
 656        return i_size;
 657#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 658        loff_t i_size;
 659
 660        preempt_disable();
 661        i_size = inode->i_size;
 662        preempt_enable();
 663        return i_size;
 664#else
 665        return inode->i_size;
 666#endif
 667}
 668
 669/*
 670 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
 671 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
 672 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
 673 */
 674static inline void i_size_write(struct inode *inode, loff_t i_size)
 675{
 676#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 677        write_seqcount_begin(&inode->i_size_seqcount);
 678        inode->i_size = i_size;
 679        write_seqcount_end(&inode->i_size_seqcount);
 680#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 681        preempt_disable();
 682        inode->i_size = i_size;
 683        preempt_enable();
 684#else
 685        inode->i_size = i_size;
 686#endif
 687}
 688
 689/* Helper functions so that in most cases filesystems will
 690 * not need to deal directly with kuid_t and kgid_t and can
 691 * instead deal with the raw numeric values that are stored
 692 * in the filesystem.
 693 */
 694static inline uid_t i_uid_read(const struct inode *inode)
 695{
 696        return from_kuid(&init_user_ns, inode->i_uid);
 697}
 698
 699static inline gid_t i_gid_read(const struct inode *inode)
 700{
 701        return from_kgid(&init_user_ns, inode->i_gid);
 702}
 703
 704static inline void i_uid_write(struct inode *inode, uid_t uid)
 705{
 706        inode->i_uid = make_kuid(&init_user_ns, uid);
 707}
 708
 709static inline void i_gid_write(struct inode *inode, gid_t gid)
 710{
 711        inode->i_gid = make_kgid(&init_user_ns, gid);
 712}
 713
 714static inline unsigned iminor(const struct inode *inode)
 715{
 716        return MINOR(inode->i_rdev);
 717}
 718
 719static inline unsigned imajor(const struct inode *inode)
 720{
 721        return MAJOR(inode->i_rdev);
 722}
 723
 724extern struct block_device *I_BDEV(struct inode *inode);
 725
 726struct fown_struct {
 727        rwlock_t lock;          /* protects pid, uid, euid fields */
 728        struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
 729        enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
 730        kuid_t uid, euid;       /* uid/euid of process setting the owner */
 731        int signum;             /* posix.1b rt signal to be delivered on IO */
 732};
 733
 734/*
 735 * Track a single file's readahead state
 736 */
 737struct file_ra_state {
 738        pgoff_t start;                  /* where readahead started */
 739        unsigned int size;              /* # of readahead pages */
 740        unsigned int async_size;        /* do asynchronous readahead when
 741                                           there are only # of pages ahead */
 742
 743        unsigned int ra_pages;          /* Maximum readahead window */
 744        unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
 745        loff_t prev_pos;                /* Cache last read() position */
 746};
 747
 748/*
 749 * Check if @index falls in the readahead windows.
 750 */
 751static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
 752{
 753        return (index >= ra->start &&
 754                index <  ra->start + ra->size);
 755}
 756
 757#define FILE_MNT_WRITE_TAKEN    1
 758#define FILE_MNT_WRITE_RELEASED 2
 759
 760struct file {
 761        /*
 762         * fu_list becomes invalid after file_free is called and queued via
 763         * fu_rcuhead for RCU freeing
 764         */
 765        union {
 766                struct list_head        fu_list;
 767                struct rcu_head         fu_rcuhead;
 768        } f_u;
 769        struct path             f_path;
 770#define f_dentry        f_path.dentry
 771#define f_vfsmnt        f_path.mnt
 772        const struct file_operations    *f_op;
 773
 774        /*
 775         * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
 776         * Must not be taken from IRQ context.
 777         */
 778        spinlock_t              f_lock;
 779#ifdef CONFIG_SMP
 780        int                     f_sb_list_cpu;
 781#endif
 782        atomic_long_t           f_count;
 783        unsigned int            f_flags;
 784        fmode_t                 f_mode;
 785        loff_t                  f_pos;
 786        struct fown_struct      f_owner;
 787        const struct cred       *f_cred;
 788        struct file_ra_state    f_ra;
 789
 790        u64                     f_version;
 791#ifdef CONFIG_SECURITY
 792        void                    *f_security;
 793#endif
 794        /* needed for tty driver, and maybe others */
 795        void                    *private_data;
 796
 797#ifdef CONFIG_EPOLL
 798        /* Used by fs/eventpoll.c to link all the hooks to this file */
 799        struct list_head        f_ep_links;
 800        struct list_head        f_tfile_llink;
 801#endif /* #ifdef CONFIG_EPOLL */
 802        struct address_space    *f_mapping;
 803#ifdef CONFIG_DEBUG_WRITECOUNT
 804        unsigned long f_mnt_write_state;
 805#endif
 806};
 807
 808struct file_handle {
 809        __u32 handle_bytes;
 810        int handle_type;
 811        /* file identifier */
 812        unsigned char f_handle[0];
 813};
 814
 815static inline struct file *get_file(struct file *f)
 816{
 817        atomic_long_inc(&f->f_count);
 818        return f;
 819}
 820#define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
 821#define file_count(x)   atomic_long_read(&(x)->f_count)
 822
 823#ifdef CONFIG_DEBUG_WRITECOUNT
 824static inline void file_take_write(struct file *f)
 825{
 826        WARN_ON(f->f_mnt_write_state != 0);
 827        f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
 828}
 829static inline void file_release_write(struct file *f)
 830{
 831        f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
 832}
 833static inline void file_reset_write(struct file *f)
 834{
 835        f->f_mnt_write_state = 0;
 836}
 837static inline void file_check_state(struct file *f)
 838{
 839        /*
 840         * At this point, either both or neither of these bits
 841         * should be set.
 842         */
 843        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
 844        WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
 845}
 846static inline int file_check_writeable(struct file *f)
 847{
 848        if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
 849                return 0;
 850        printk(KERN_WARNING "writeable file with no "
 851                            "mnt_want_write()\n");
 852        WARN_ON(1);
 853        return -EINVAL;
 854}
 855#else /* !CONFIG_DEBUG_WRITECOUNT */
 856static inline void file_take_write(struct file *filp) {}
 857static inline void file_release_write(struct file *filp) {}
 858static inline void file_reset_write(struct file *filp) {}
 859static inline void file_check_state(struct file *filp) {}
 860static inline int file_check_writeable(struct file *filp)
 861{
 862        return 0;
 863}
 864#endif /* CONFIG_DEBUG_WRITECOUNT */
 865
 866#define MAX_NON_LFS     ((1UL<<31) - 1)
 867
 868/* Page cache limit. The filesystems should put that into their s_maxbytes 
 869   limits, otherwise bad things can happen in VM. */ 
 870#if BITS_PER_LONG==32
 871#define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
 872#elif BITS_PER_LONG==64
 873#define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
 874#endif
 875
 876#define FL_POSIX        1
 877#define FL_FLOCK        2
 878#define FL_ACCESS       8       /* not trying to lock, just looking */
 879#define FL_EXISTS       16      /* when unlocking, test for existence */
 880#define FL_LEASE        32      /* lease held on this file */
 881#define FL_CLOSE        64      /* unlock on close */
 882#define FL_SLEEP        128     /* A blocking lock */
 883#define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
 884#define FL_UNLOCK_PENDING       512 /* Lease is being broken */
 885
 886/*
 887 * Special return value from posix_lock_file() and vfs_lock_file() for
 888 * asynchronous locking.
 889 */
 890#define FILE_LOCK_DEFERRED 1
 891
 892/*
 893 * The POSIX file lock owner is determined by
 894 * the "struct files_struct" in the thread group
 895 * (or NULL for no owner - BSD locks).
 896 *
 897 * Lockd stuffs a "host" pointer into this.
 898 */
 899typedef struct files_struct *fl_owner_t;
 900
 901struct file_lock_operations {
 902        void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
 903        void (*fl_release_private)(struct file_lock *);
 904};
 905
 906struct lock_manager_operations {
 907        int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
 908        void (*lm_notify)(struct file_lock *);  /* unblock callback */
 909        int (*lm_grant)(struct file_lock *, struct file_lock *, int);
 910        void (*lm_break)(struct file_lock *);
 911        int (*lm_change)(struct file_lock **, int);
 912};
 913
 914struct lock_manager {
 915        struct list_head list;
 916};
 917
 918struct net;
 919void locks_start_grace(struct net *, struct lock_manager *);
 920void locks_end_grace(struct lock_manager *);
 921int locks_in_grace(struct net *);
 922
 923/* that will die - we need it for nfs_lock_info */
 924#include <linux/nfs_fs_i.h>
 925
 926struct file_lock {
 927        struct file_lock *fl_next;      /* singly linked list for this inode  */
 928        struct list_head fl_link;       /* doubly linked list of all locks */
 929        struct list_head fl_block;      /* circular list of blocked processes */
 930        fl_owner_t fl_owner;
 931        unsigned int fl_flags;
 932        unsigned char fl_type;
 933        unsigned int fl_pid;
 934        struct pid *fl_nspid;
 935        wait_queue_head_t fl_wait;
 936        struct file *fl_file;
 937        loff_t fl_start;
 938        loff_t fl_end;
 939
 940        struct fasync_struct *  fl_fasync; /* for lease break notifications */
 941        /* for lease breaks: */
 942        unsigned long fl_break_time;
 943        unsigned long fl_downgrade_time;
 944
 945        const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
 946        const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
 947        union {
 948                struct nfs_lock_info    nfs_fl;
 949                struct nfs4_lock_info   nfs4_fl;
 950                struct {
 951                        struct list_head link;  /* link in AFS vnode's pending_locks list */
 952                        int state;              /* state of grant or error if -ve */
 953                } afs;
 954        } fl_u;
 955};
 956
 957/* The following constant reflects the upper bound of the file/locking space */
 958#ifndef OFFSET_MAX
 959#define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
 960#define OFFSET_MAX      INT_LIMIT(loff_t)
 961#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
 962#endif
 963
 964#include <linux/fcntl.h>
 965
 966extern void send_sigio(struct fown_struct *fown, int fd, int band);
 967
 968#ifdef CONFIG_FILE_LOCKING
 969extern int fcntl_getlk(struct file *, struct flock __user *);
 970extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
 971                        struct flock __user *);
 972
 973#if BITS_PER_LONG == 32
 974extern int fcntl_getlk64(struct file *, struct flock64 __user *);
 975extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
 976                        struct flock64 __user *);
 977#endif
 978
 979extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
 980extern int fcntl_getlease(struct file *filp);
 981
 982/* fs/locks.c */
 983void locks_free_lock(struct file_lock *fl);
 984extern void locks_init_lock(struct file_lock *);
 985extern struct file_lock * locks_alloc_lock(void);
 986extern void locks_copy_lock(struct file_lock *, struct file_lock *);
 987extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
 988extern void locks_remove_posix(struct file *, fl_owner_t);
 989extern void locks_remove_flock(struct file *);
 990extern void locks_release_private(struct file_lock *);
 991extern void posix_test_lock(struct file *, struct file_lock *);
 992extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
 993extern int posix_lock_file_wait(struct file *, struct file_lock *);
 994extern int posix_unblock_lock(struct file *, struct file_lock *);
 995extern int vfs_test_lock(struct file *, struct file_lock *);
 996extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
 997extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
 998extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
 999extern int __break_lease(struct inode *inode, unsigned int flags);
1000extern void lease_get_mtime(struct inode *, struct timespec *time);
1001extern int generic_setlease(struct file *, long, struct file_lock **);
1002extern int vfs_setlease(struct file *, long, struct file_lock **);
1003extern int lease_modify(struct file_lock **, int);
1004extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1005extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1006extern void locks_delete_block(struct file_lock *waiter);
1007extern void lock_flocks(void);
1008extern void unlock_flocks(void);
1009#else /* !CONFIG_FILE_LOCKING */
1010static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1011{
1012        return -EINVAL;
1013}
1014
1015static inline int fcntl_setlk(unsigned int fd, struct file *file,
1016                              unsigned int cmd, struct flock __user *user)
1017{
1018        return -EACCES;
1019}
1020
1021#if BITS_PER_LONG == 32
1022static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1023{
1024        return -EINVAL;
1025}
1026
1027static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1028                                unsigned int cmd, struct flock64 __user *user)
1029{
1030        return -EACCES;
1031}
1032#endif
1033static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1034{
1035        return 0;
1036}
1037
1038static inline int fcntl_getlease(struct file *filp)
1039{
1040        return 0;
1041}
1042
1043static inline void locks_init_lock(struct file_lock *fl)
1044{
1045        return;
1046}
1047
1048static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1049{
1050        return;
1051}
1052
1053static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1054{
1055        return;
1056}
1057
1058static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1059{
1060        return;
1061}
1062
1063static inline void locks_remove_flock(struct file *filp)
1064{
1065        return;
1066}
1067
1068static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1069{
1070        return;
1071}
1072
1073static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1074                                  struct file_lock *conflock)
1075{
1076        return -ENOLCK;
1077}
1078
1079static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1080{
1081        return -ENOLCK;
1082}
1083
1084static inline int posix_unblock_lock(struct file *filp,
1085                                     struct file_lock *waiter)
1086{
1087        return -ENOENT;
1088}
1089
1090static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1091{
1092        return 0;
1093}
1094
1095static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1096                                struct file_lock *fl, struct file_lock *conf)
1097{
1098        return -ENOLCK;
1099}
1100
1101static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1102{
1103        return 0;
1104}
1105
1106static inline int flock_lock_file_wait(struct file *filp,
1107                                       struct file_lock *request)
1108{
1109        return -ENOLCK;
1110}
1111
1112static inline int __break_lease(struct inode *inode, unsigned int mode)
1113{
1114        return 0;
1115}
1116
1117static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1118{
1119        return;
1120}
1121
1122static inline int generic_setlease(struct file *filp, long arg,
1123                                    struct file_lock **flp)
1124{
1125        return -EINVAL;
1126}
1127
1128static inline int vfs_setlease(struct file *filp, long arg,
1129                               struct file_lock **lease)
1130{
1131        return -EINVAL;
1132}
1133
1134static inline int lease_modify(struct file_lock **before, int arg)
1135{
1136        return -EINVAL;
1137}
1138
1139static inline int lock_may_read(struct inode *inode, loff_t start,
1140                                unsigned long len)
1141{
1142        return 1;
1143}
1144
1145static inline int lock_may_write(struct inode *inode, loff_t start,
1146                                 unsigned long len)
1147{
1148        return 1;
1149}
1150
1151static inline void locks_delete_block(struct file_lock *waiter)
1152{
1153}
1154
1155static inline void lock_flocks(void)
1156{
1157}
1158
1159static inline void unlock_flocks(void)
1160{
1161}
1162
1163#endif /* !CONFIG_FILE_LOCKING */
1164
1165
1166struct fasync_struct {
1167        spinlock_t              fa_lock;
1168        int                     magic;
1169        int                     fa_fd;
1170        struct fasync_struct    *fa_next; /* singly linked list */
1171        struct file             *fa_file;
1172        struct rcu_head         fa_rcu;
1173};
1174
1175#define FASYNC_MAGIC 0x4601
1176
1177/* SMP safe fasync helpers: */
1178extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1179extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1180extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1181extern struct fasync_struct *fasync_alloc(void);
1182extern void fasync_free(struct fasync_struct *);
1183
1184/* can be called from interrupts */
1185extern void kill_fasync(struct fasync_struct **, int, int);
1186
1187extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1188extern int f_setown(struct file *filp, unsigned long arg, int force);
1189extern void f_delown(struct file *filp);
1190extern pid_t f_getown(struct file *filp);
1191extern int send_sigurg(struct fown_struct *fown);
1192
1193struct mm_struct;
1194
1195/*
1196 *      Umount options
1197 */
1198
1199#define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1200#define MNT_DETACH      0x00000002      /* Just detach from the tree */
1201#define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1202#define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1203#define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1204
1205extern struct list_head super_blocks;
1206extern spinlock_t sb_lock;
1207
1208/* Possible states of 'frozen' field */
1209enum {
1210        SB_UNFROZEN = 0,                /* FS is unfrozen */
1211        SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1212        SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1213        SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1214                                         * internal threads if needed) */
1215        SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1216};
1217
1218#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1219
1220struct sb_writers {
1221        /* Counters for counting writers at each level */
1222        struct percpu_counter   counter[SB_FREEZE_LEVELS];
1223        wait_queue_head_t       wait;           /* queue for waiting for
1224                                                   writers / faults to finish */
1225        int                     frozen;         /* Is sb frozen? */
1226        wait_queue_head_t       wait_unfrozen;  /* queue for waiting for
1227                                                   sb to be thawed */
1228#ifdef CONFIG_DEBUG_LOCK_ALLOC
1229        struct lockdep_map      lock_map[SB_FREEZE_LEVELS];
1230#endif
1231};
1232
1233struct super_block {
1234        struct list_head        s_list;         /* Keep this first */
1235        dev_t                   s_dev;          /* search index; _not_ kdev_t */
1236        unsigned char           s_blocksize_bits;
1237        unsigned long           s_blocksize;
1238        loff_t                  s_maxbytes;     /* Max file size */
1239        struct file_system_type *s_type;
1240        const struct super_operations   *s_op;
1241        const struct dquot_operations   *dq_op;
1242        const struct quotactl_ops       *s_qcop;
1243        const struct export_operations *s_export_op;
1244        unsigned long           s_flags;
1245        unsigned long           s_magic;
1246        struct dentry           *s_root;
1247        struct rw_semaphore     s_umount;
1248        int                     s_count;
1249        atomic_t                s_active;
1250#ifdef CONFIG_SECURITY
1251        void                    *s_security;
1252#endif
1253        const struct xattr_handler **s_xattr;
1254
1255        struct list_head        s_inodes;       /* all inodes */
1256        struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1257#ifdef CONFIG_SMP
1258        struct list_head __percpu *s_files;
1259#else
1260        struct list_head        s_files;
1261#endif
1262        struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1263        /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1264        struct list_head        s_dentry_lru;   /* unused dentry lru */
1265        int                     s_nr_dentry_unused;     /* # of dentry on lru */
1266
1267        /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1268        spinlock_t              s_inode_lru_lock ____cacheline_aligned_in_smp;
1269        struct list_head        s_inode_lru;            /* unused inode lru */
1270        int                     s_nr_inodes_unused;     /* # of inodes on lru */
1271
1272        struct block_device     *s_bdev;
1273        struct backing_dev_info *s_bdi;
1274        struct mtd_info         *s_mtd;
1275        struct hlist_node       s_instances;
1276        struct quota_info       s_dquot;        /* Diskquota specific options */
1277
1278        struct sb_writers       s_writers;
1279
1280        char s_id[32];                          /* Informational name */
1281        u8 s_uuid[16];                          /* UUID */
1282
1283        void                    *s_fs_info;     /* Filesystem private info */
1284        unsigned int            s_max_links;
1285        fmode_t                 s_mode;
1286
1287        /* Granularity of c/m/atime in ns.
1288           Cannot be worse than a second */
1289        u32                s_time_gran;
1290
1291        /*
1292         * The next field is for VFS *only*. No filesystems have any business
1293         * even looking at it. You had been warned.
1294         */
1295        struct mutex s_vfs_rename_mutex;        /* Kludge */
1296
1297        /*
1298         * Filesystem subtype.  If non-empty the filesystem type field
1299         * in /proc/mounts will be "type.subtype"
1300         */
1301        char *s_subtype;
1302
1303        /*
1304         * Saved mount options for lazy filesystems using
1305         * generic_show_options()
1306         */
1307        char __rcu *s_options;
1308        const struct dentry_operations *s_d_op; /* default d_op for dentries */
1309
1310        /*
1311         * Saved pool identifier for cleancache (-1 means none)
1312         */
1313        int cleancache_poolid;
1314
1315        struct shrinker s_shrink;       /* per-sb shrinker handle */
1316
1317        /* Number of inodes with nlink == 0 but still referenced */
1318        atomic_long_t s_remove_count;
1319
1320        /* Being remounted read-only */
1321        int s_readonly_remount;
1322};
1323
1324/* superblock cache pruning functions */
1325extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1326extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1327
1328extern struct timespec current_fs_time(struct super_block *sb);
1329
1330/*
1331 * Snapshotting support.
1332 */
1333
1334void __sb_end_write(struct super_block *sb, int level);
1335int __sb_start_write(struct super_block *sb, int level, bool wait);
1336
1337/**
1338 * sb_end_write - drop write access to a superblock
1339 * @sb: the super we wrote to
1340 *
1341 * Decrement number of writers to the filesystem. Wake up possible waiters
1342 * wanting to freeze the filesystem.
1343 */
1344static inline void sb_end_write(struct super_block *sb)
1345{
1346        __sb_end_write(sb, SB_FREEZE_WRITE);
1347}
1348
1349/**
1350 * sb_end_pagefault - drop write access to a superblock from a page fault
1351 * @sb: the super we wrote to
1352 *
1353 * Decrement number of processes handling write page fault to the filesystem.
1354 * Wake up possible waiters wanting to freeze the filesystem.
1355 */
1356static inline void sb_end_pagefault(struct super_block *sb)
1357{
1358        __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1359}
1360
1361/**
1362 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1363 * @sb: the super we wrote to
1364 *
1365 * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1366 * waiters wanting to freeze the filesystem.
1367 */
1368static inline void sb_end_intwrite(struct super_block *sb)
1369{
1370        __sb_end_write(sb, SB_FREEZE_FS);
1371}
1372
1373/**
1374 * sb_start_write - get write access to a superblock
1375 * @sb: the super we write to
1376 *
1377 * When a process wants to write data or metadata to a file system (i.e. dirty
1378 * a page or an inode), it should embed the operation in a sb_start_write() -
1379 * sb_end_write() pair to get exclusion against file system freezing. This
1380 * function increments number of writers preventing freezing. If the file
1381 * system is already frozen, the function waits until the file system is
1382 * thawed.
1383 *
1384 * Since freeze protection behaves as a lock, users have to preserve
1385 * ordering of freeze protection and other filesystem locks. Generally,
1386 * freeze protection should be the outermost lock. In particular, we have:
1387 *
1388 * sb_start_write
1389 *   -> i_mutex                 (write path, truncate, directory ops, ...)
1390 *   -> s_umount                (freeze_super, thaw_super)
1391 */
1392static inline void sb_start_write(struct super_block *sb)
1393{
1394        __sb_start_write(sb, SB_FREEZE_WRITE, true);
1395}
1396
1397static inline int sb_start_write_trylock(struct super_block *sb)
1398{
1399        return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1400}
1401
1402/**
1403 * sb_start_pagefault - get write access to a superblock from a page fault
1404 * @sb: the super we write to
1405 *
1406 * When a process starts handling write page fault, it should embed the
1407 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1408 * exclusion against file system freezing. This is needed since the page fault
1409 * is going to dirty a page. This function increments number of running page
1410 * faults preventing freezing. If the file system is already frozen, the
1411 * function waits until the file system is thawed.
1412 *
1413 * Since page fault freeze protection behaves as a lock, users have to preserve
1414 * ordering of freeze protection and other filesystem locks. It is advised to
1415 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1416 * handling code implies lock dependency:
1417 *
1418 * mmap_sem
1419 *   -> sb_start_pagefault
1420 */
1421static inline void sb_start_pagefault(struct super_block *sb)
1422{
1423        __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1424}
1425
1426/*
1427 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1428 * @sb: the super we write to
1429 *
1430 * This is the third level of protection against filesystem freezing. It is
1431 * free for use by a filesystem. The only requirement is that it must rank
1432 * below sb_start_pagefault.
1433 *
1434 * For example filesystem can call sb_start_intwrite() when starting a
1435 * transaction which somewhat eases handling of freezing for internal sources
1436 * of filesystem changes (internal fs threads, discarding preallocation on file
1437 * close, etc.).
1438 */
1439static inline void sb_start_intwrite(struct super_block *sb)
1440{
1441        __sb_start_write(sb, SB_FREEZE_FS, true);
1442}
1443
1444
1445extern bool inode_owner_or_capable(const struct inode *inode);
1446
1447/* not quite ready to be deprecated, but... */
1448extern void lock_super(struct super_block *);
1449extern void unlock_super(struct super_block *);
1450
1451/*
1452 * VFS helper functions..
1453 */
1454extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1455extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1456extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1457extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1458extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1459extern int vfs_rmdir(struct inode *, struct dentry *);
1460extern int vfs_unlink(struct inode *, struct dentry *);
1461extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1462
1463/*
1464 * VFS dentry helper functions.
1465 */
1466extern void dentry_unhash(struct dentry *dentry);
1467
1468/*
1469 * VFS file helper functions.
1470 */
1471extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1472                        umode_t mode);
1473/*
1474 * VFS FS_IOC_FIEMAP helper definitions.
1475 */
1476struct fiemap_extent_info {
1477        unsigned int fi_flags;          /* Flags as passed from user */
1478        unsigned int fi_extents_mapped; /* Number of mapped extents */
1479        unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1480        struct fiemap_extent __user *fi_extents_start; /* Start of
1481                                                        fiemap_extent array */
1482};
1483int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1484                            u64 phys, u64 len, u32 flags);
1485int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1486
1487/*
1488 * File types
1489 *
1490 * NOTE! These match bits 12..15 of stat.st_mode
1491 * (ie "(i_mode >> 12) & 15").
1492 */
1493#define DT_UNKNOWN      0
1494#define DT_FIFO         1
1495#define DT_CHR          2
1496#define DT_DIR          4
1497#define DT_BLK          6
1498#define DT_REG          8
1499#define DT_LNK          10
1500#define DT_SOCK         12
1501#define DT_WHT          14
1502
1503/*
1504 * This is the "filldir" function type, used by readdir() to let
1505 * the kernel specify what kind of dirent layout it wants to have.
1506 * This allows the kernel to read directories into kernel space or
1507 * to have different dirent layouts depending on the binary type.
1508 */
1509typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1510struct block_device_operations;
1511
1512/* These macros are for out of kernel modules to test that
1513 * the kernel supports the unlocked_ioctl and compat_ioctl
1514 * fields in struct file_operations. */
1515#define HAVE_COMPAT_IOCTL 1
1516#define HAVE_UNLOCKED_IOCTL 1
1517
1518struct file_operations {
1519        struct module *owner;
1520        loff_t (*llseek) (struct file *, loff_t, int);
1521        ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1522        ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1523        ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1524        ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1525        int (*readdir) (struct file *, void *, filldir_t);
1526        unsigned int (*poll) (struct file *, struct poll_table_struct *);
1527        long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1528        long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1529        int (*mmap) (struct file *, struct vm_area_struct *);
1530        int (*open) (struct inode *, struct file *);
1531        int (*flush) (struct file *, fl_owner_t id);
1532        int (*release) (struct inode *, struct file *);
1533        int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1534        int (*aio_fsync) (struct kiocb *, int datasync);
1535        int (*fasync) (int, struct file *, int);
1536        int (*lock) (struct file *, int, struct file_lock *);
1537        ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1538        unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1539        int (*check_flags)(int);
1540        int (*flock) (struct file *, int, struct file_lock *);
1541        ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1542        ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1543        int (*setlease)(struct file *, long, struct file_lock **);
1544        long (*fallocate)(struct file *file, int mode, loff_t offset,
1545                          loff_t len);
1546};
1547
1548struct inode_operations {
1549        struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1550        void * (*follow_link) (struct dentry *, struct nameidata *);
1551        int (*permission) (struct inode *, int);
1552        struct posix_acl * (*get_acl)(struct inode *, int);
1553
1554        int (*readlink) (struct dentry *, char __user *,int);
1555        void (*put_link) (struct dentry *, struct nameidata *, void *);
1556
1557        int (*create) (struct inode *,struct dentry *, umode_t, bool);
1558        int (*link) (struct dentry *,struct inode *,struct dentry *);
1559        int (*unlink) (struct inode *,struct dentry *);
1560        int (*symlink) (struct inode *,struct dentry *,const char *);
1561        int (*mkdir) (struct inode *,struct dentry *,umode_t);
1562        int (*rmdir) (struct inode *,struct dentry *);
1563        int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1564        int (*rename) (struct inode *, struct dentry *,
1565                        struct inode *, struct dentry *);
1566        void (*truncate) (struct inode *);
1567        int (*setattr) (struct dentry *, struct iattr *);
1568        int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1569        int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1570        ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1571        ssize_t (*listxattr) (struct dentry *, char *, size_t);
1572        int (*removexattr) (struct dentry *, const char *);
1573        int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1574                      u64 len);
1575        int (*update_time)(struct inode *, struct timespec *, int);
1576        int (*atomic_open)(struct inode *, struct dentry *,
1577                           struct file *, unsigned open_flag,
1578                           umode_t create_mode, int *opened);
1579} ____cacheline_aligned;
1580
1581struct seq_file;
1582
1583ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1584                              unsigned long nr_segs, unsigned long fast_segs,
1585                              struct iovec *fast_pointer,
1586                              struct iovec **ret_pointer);
1587
1588extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1589extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1590extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1591                unsigned long, loff_t *);
1592extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1593                unsigned long, loff_t *);
1594
1595struct super_operations {
1596        struct inode *(*alloc_inode)(struct super_block *sb);
1597        void (*destroy_inode)(struct inode *);
1598
1599        void (*dirty_inode) (struct inode *, int flags);
1600        int (*write_inode) (struct inode *, struct writeback_control *wbc);
1601        int (*drop_inode) (struct inode *);
1602        void (*evict_inode) (struct inode *);
1603        void (*put_super) (struct super_block *);
1604        int (*sync_fs)(struct super_block *sb, int wait);
1605        int (*freeze_fs) (struct super_block *);
1606        int (*unfreeze_fs) (struct super_block *);
1607        int (*statfs) (struct dentry *, struct kstatfs *);
1608        int (*remount_fs) (struct super_block *, int *, char *);
1609        void (*umount_begin) (struct super_block *);
1610
1611        int (*show_options)(struct seq_file *, struct dentry *);
1612        int (*show_devname)(struct seq_file *, struct dentry *);
1613        int (*show_path)(struct seq_file *, struct dentry *);
1614        int (*show_stats)(struct seq_file *, struct dentry *);
1615#ifdef CONFIG_QUOTA
1616        ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1617        ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1618#endif
1619        int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1620        int (*nr_cached_objects)(struct super_block *);
1621        void (*free_cached_objects)(struct super_block *, int);
1622};
1623
1624/*
1625 * Inode flags - they have no relation to superblock flags now
1626 */
1627#define S_SYNC          1       /* Writes are synced at once */
1628#define S_NOATIME       2       /* Do not update access times */
1629#define S_APPEND        4       /* Append-only file */
1630#define S_IMMUTABLE     8       /* Immutable file */
1631#define S_DEAD          16      /* removed, but still open directory */
1632#define S_NOQUOTA       32      /* Inode is not counted to quota */
1633#define S_DIRSYNC       64      /* Directory modifications are synchronous */
1634#define S_NOCMTIME      128     /* Do not update file c/mtime */
1635#define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1636#define S_PRIVATE       512     /* Inode is fs-internal */
1637#define S_IMA           1024    /* Inode has an associated IMA struct */
1638#define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1639#define S_NOSEC         4096    /* no suid or xattr security attributes */
1640
1641/*
1642 * Note that nosuid etc flags are inode-specific: setting some file-system
1643 * flags just means all the inodes inherit those flags by default. It might be
1644 * possible to override it selectively if you really wanted to with some
1645 * ioctl() that is not currently implemented.
1646 *
1647 * Exception: MS_RDONLY is always applied to the entire file system.
1648 *
1649 * Unfortunately, it is possible to change a filesystems flags with it mounted
1650 * with files in use.  This means that all of the inodes will not have their
1651 * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1652 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1653 */
1654#define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1655
1656#define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1657#define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1658                                        ((inode)->i_flags & S_SYNC))
1659#define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1660                                        ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1661#define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1662#define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1663#define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1664
1665#define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1666#define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1667#define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1668#define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1669
1670#define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1671#define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1672#define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1673#define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1674#define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1675#define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1676#define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1677
1678/*
1679 * Inode state bits.  Protected by inode->i_lock
1680 *
1681 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1682 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1683 *
1684 * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1685 * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1686 * various stages of removing an inode.
1687 *
1688 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1689 *
1690 * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1691 *                      fdatasync().  i_atime is the usual cause.
1692 * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1693 *                      these changes separately from I_DIRTY_SYNC so that we
1694 *                      don't have to write inode on fdatasync() when only
1695 *                      mtime has changed in it.
1696 * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1697 * I_NEW                Serves as both a mutex and completion notification.
1698 *                      New inodes set I_NEW.  If two processes both create
1699 *                      the same inode, one of them will release its inode and
1700 *                      wait for I_NEW to be released before returning.
1701 *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1702 *                      also cause waiting on I_NEW, without I_NEW actually
1703 *                      being set.  find_inode() uses this to prevent returning
1704 *                      nearly-dead inodes.
1705 * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1706 *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1707 *                      cleared.
1708 * I_FREEING            Set when inode is about to be freed but still has dirty
1709 *                      pages or buffers attached or the inode itself is still
1710 *                      dirty.
1711 * I_CLEAR              Added by clear_inode().  In this state the inode is
1712 *                      clean and can be destroyed.  Inode keeps I_FREEING.
1713 *
1714 *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1715 *                      prohibited for many purposes.  iget() must wait for
1716 *                      the inode to be completely released, then create it
1717 *                      anew.  Other functions will just ignore such inodes,
1718 *                      if appropriate.  I_NEW is used for waiting.
1719 *
1720 * I_SYNC               Writeback of inode is running. The bit is set during
1721 *                      data writeback, and cleared with a wakeup on the bit
1722 *                      address once it is done. The bit is also used to pin
1723 *                      the inode in memory for flusher thread.
1724 *
1725 * I_REFERENCED         Marks the inode as recently references on the LRU list.
1726 *
1727 * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1728 *
1729 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1730 */
1731#define I_DIRTY_SYNC            (1 << 0)
1732#define I_DIRTY_DATASYNC        (1 << 1)
1733#define I_DIRTY_PAGES           (1 << 2)
1734#define __I_NEW                 3
1735#define I_NEW                   (1 << __I_NEW)
1736#define I_WILL_FREE             (1 << 4)
1737#define I_FREEING               (1 << 5)
1738#define I_CLEAR                 (1 << 6)
1739#define __I_SYNC                7
1740#define I_SYNC                  (1 << __I_SYNC)
1741#define I_REFERENCED            (1 << 8)
1742#define __I_DIO_WAKEUP          9
1743#define I_DIO_WAKEUP            (1 << I_DIO_WAKEUP)
1744
1745#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1746
1747extern void __mark_inode_dirty(struct inode *, int);
1748static inline void mark_inode_dirty(struct inode *inode)
1749{
1750        __mark_inode_dirty(inode, I_DIRTY);
1751}
1752
1753static inline void mark_inode_dirty_sync(struct inode *inode)
1754{
1755        __mark_inode_dirty(inode, I_DIRTY_SYNC);
1756}
1757
1758extern void inc_nlink(struct inode *inode);
1759extern void drop_nlink(struct inode *inode);
1760extern void clear_nlink(struct inode *inode);
1761extern void set_nlink(struct inode *inode, unsigned int nlink);
1762
1763static inline void inode_inc_link_count(struct inode *inode)
1764{
1765        inc_nlink(inode);
1766        mark_inode_dirty(inode);
1767}
1768
1769static inline void inode_dec_link_count(struct inode *inode)
1770{
1771        drop_nlink(inode);
1772        mark_inode_dirty(inode);
1773}
1774
1775/**
1776 * inode_inc_iversion - increments i_version
1777 * @inode: inode that need to be updated
1778 *
1779 * Every time the inode is modified, the i_version field will be incremented.
1780 * The filesystem has to be mounted with i_version flag
1781 */
1782
1783static inline void inode_inc_iversion(struct inode *inode)
1784{
1785       spin_lock(&inode->i_lock);
1786       inode->i_version++;
1787       spin_unlock(&inode->i_lock);
1788}
1789
1790enum file_time_flags {
1791        S_ATIME = 1,
1792        S_MTIME = 2,
1793        S_CTIME = 4,
1794        S_VERSION = 8,
1795};
1796
1797extern void touch_atime(struct path *);
1798static inline void file_accessed(struct file *file)
1799{
1800        if (!(file->f_flags & O_NOATIME))
1801                touch_atime(&file->f_path);
1802}
1803
1804int sync_inode(struct inode *inode, struct writeback_control *wbc);
1805int sync_inode_metadata(struct inode *inode, int wait);
1806
1807struct file_system_type {
1808        const char *name;
1809        int fs_flags;
1810#define FS_REQUIRES_DEV         1 
1811#define FS_BINARY_MOUNTDATA     2
1812#define FS_HAS_SUBTYPE          4
1813#define FS_REVAL_DOT            16384   /* Check the paths ".", ".." for staleness */
1814#define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1815        struct dentry *(*mount) (struct file_system_type *, int,
1816                       const char *, void *);
1817        void (*kill_sb) (struct super_block *);
1818        struct module *owner;
1819        struct file_system_type * next;
1820        struct hlist_head fs_supers;
1821
1822        struct lock_class_key s_lock_key;
1823        struct lock_class_key s_umount_key;
1824        struct lock_class_key s_vfs_rename_key;
1825        struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1826
1827        struct lock_class_key i_lock_key;
1828        struct lock_class_key i_mutex_key;
1829        struct lock_class_key i_mutex_dir_key;
1830};
1831
1832extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1833        void *data, int (*fill_super)(struct super_block *, void *, int));
1834extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1835        int flags, const char *dev_name, void *data,
1836        int (*fill_super)(struct super_block *, void *, int));
1837extern struct dentry *mount_single(struct file_system_type *fs_type,
1838        int flags, void *data,
1839        int (*fill_super)(struct super_block *, void *, int));
1840extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1841        int flags, void *data,
1842        int (*fill_super)(struct super_block *, void *, int));
1843extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1844void generic_shutdown_super(struct super_block *sb);
1845void kill_block_super(struct super_block *sb);
1846void kill_anon_super(struct super_block *sb);
1847void kill_litter_super(struct super_block *sb);
1848void deactivate_super(struct super_block *sb);
1849void deactivate_locked_super(struct super_block *sb);
1850int set_anon_super(struct super_block *s, void *data);
1851int get_anon_bdev(dev_t *);
1852void free_anon_bdev(dev_t);
1853struct super_block *sget(struct file_system_type *type,
1854                        int (*test)(struct super_block *,void *),
1855                        int (*set)(struct super_block *,void *),
1856                        int flags, void *data);
1857extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1858        const struct super_operations *ops,
1859        const struct dentry_operations *dops,
1860        unsigned long);
1861
1862/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1863#define fops_get(fops) \
1864        (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1865#define fops_put(fops) \
1866        do { if (fops) module_put((fops)->owner); } while(0)
1867
1868extern int register_filesystem(struct file_system_type *);
1869extern int unregister_filesystem(struct file_system_type *);
1870extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1871#define kern_mount(type) kern_mount_data(type, NULL)
1872extern void kern_unmount(struct vfsmount *mnt);
1873extern int may_umount_tree(struct vfsmount *);
1874extern int may_umount(struct vfsmount *);
1875extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1876extern struct vfsmount *collect_mounts(struct path *);
1877extern void drop_collected_mounts(struct vfsmount *);
1878extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1879                          struct vfsmount *);
1880extern int vfs_statfs(struct path *, struct kstatfs *);
1881extern int user_statfs(const char __user *, struct kstatfs *);
1882extern int fd_statfs(int, struct kstatfs *);
1883extern int vfs_ustat(dev_t, struct kstatfs *);
1884extern int freeze_super(struct super_block *super);
1885extern int thaw_super(struct super_block *super);
1886extern bool our_mnt(struct vfsmount *mnt);
1887
1888extern int current_umask(void);
1889
1890/* /sys/fs */
1891extern struct kobject *fs_kobj;
1892
1893#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1894extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1895
1896#define FLOCK_VERIFY_READ  1
1897#define FLOCK_VERIFY_WRITE 2
1898
1899#ifdef CONFIG_FILE_LOCKING
1900extern int locks_mandatory_locked(struct inode *);
1901extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1902
1903/*
1904 * Candidates for mandatory locking have the setgid bit set
1905 * but no group execute bit -  an otherwise meaningless combination.
1906 */
1907
1908static inline int __mandatory_lock(struct inode *ino)
1909{
1910        return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1911}
1912
1913/*
1914 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1915 * otherwise these will be advisory locks
1916 */
1917
1918static inline int mandatory_lock(struct inode *ino)
1919{
1920        return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1921}
1922
1923static inline int locks_verify_locked(struct inode *inode)
1924{
1925        if (mandatory_lock(inode))
1926                return locks_mandatory_locked(inode);
1927        return 0;
1928}
1929
1930static inline int locks_verify_truncate(struct inode *inode,
1931                                    struct file *filp,
1932                                    loff_t size)
1933{
1934        if (inode->i_flock && mandatory_lock(inode))
1935                return locks_mandatory_area(
1936                        FLOCK_VERIFY_WRITE, inode, filp,
1937                        size < inode->i_size ? size : inode->i_size,
1938                        (size < inode->i_size ? inode->i_size - size
1939                         : size - inode->i_size)
1940                );
1941        return 0;
1942}
1943
1944static inline int break_lease(struct inode *inode, unsigned int mode)
1945{
1946        if (inode->i_flock)
1947                return __break_lease(inode, mode);
1948        return 0;
1949}
1950#else /* !CONFIG_FILE_LOCKING */
1951static inline int locks_mandatory_locked(struct inode *inode)
1952{
1953        return 0;
1954}
1955
1956static inline int locks_mandatory_area(int rw, struct inode *inode,
1957                                       struct file *filp, loff_t offset,
1958                                       size_t count)
1959{
1960        return 0;
1961}
1962
1963static inline int __mandatory_lock(struct inode *inode)
1964{
1965        return 0;
1966}
1967
1968static inline int mandatory_lock(struct inode *inode)
1969{
1970        return 0;
1971}
1972
1973static inline int locks_verify_locked(struct inode *inode)
1974{
1975        return 0;
1976}
1977
1978static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1979                                        size_t size)
1980{
1981        return 0;
1982}
1983
1984static inline int break_lease(struct inode *inode, unsigned int mode)
1985{
1986        return 0;
1987}
1988
1989#endif /* CONFIG_FILE_LOCKING */
1990
1991/* fs/open.c */
1992struct audit_names;
1993struct filename {
1994        const char              *name;  /* pointer to actual string */
1995        const __user char       *uptr;  /* original userland pointer */
1996        struct audit_names      *aname;
1997        bool                    separate; /* should "name" be freed? */
1998};
1999
2000extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2001                       struct file *filp);
2002extern int do_fallocate(struct file *file, int mode, loff_t offset,
2003                        loff_t len);
2004extern long do_sys_open(int dfd, const char __user *filename, int flags,
2005                        umode_t mode);
2006extern struct file *file_open_name(struct filename *, int, umode_t);
2007extern struct file *filp_open(const char *, int, umode_t);
2008extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2009                                   const char *, int);
2010extern struct file * dentry_open(const struct path *, int, const struct cred *);
2011extern int filp_close(struct file *, fl_owner_t id);
2012
2013extern struct filename *getname(const char __user *);
2014
2015enum {
2016        FILE_CREATED = 1,
2017        FILE_OPENED = 2
2018};
2019extern int finish_open(struct file *file, struct dentry *dentry,
2020                        int (*open)(struct inode *, struct file *),
2021                        int *opened);
2022extern int finish_no_open(struct file *file, struct dentry *dentry);
2023
2024/* fs/ioctl.c */
2025
2026extern int ioctl_preallocate(struct file *filp, void __user *argp);
2027
2028/* fs/dcache.c */
2029extern void __init vfs_caches_init_early(void);
2030extern void __init vfs_caches_init(unsigned long);
2031
2032extern struct kmem_cache *names_cachep;
2033
2034extern void final_putname(struct filename *name);
2035
2036#define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2037#define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2038#ifndef CONFIG_AUDITSYSCALL
2039#define putname(name)           final_putname(name)
2040#else
2041extern void putname(struct filename *name);
2042#endif
2043
2044#ifdef CONFIG_BLOCK
2045extern int register_blkdev(unsigned int, const char *);
2046extern void unregister_blkdev(unsigned int, const char *);
2047extern struct block_device *bdget(dev_t);
2048extern struct block_device *bdgrab(struct block_device *bdev);
2049extern void bd_set_size(struct block_device *, loff_t size);
2050extern void bd_forget(struct inode *inode);
2051extern void bdput(struct block_device *);
2052extern void invalidate_bdev(struct block_device *);
2053extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2054extern int sync_blockdev(struct block_device *bdev);
2055extern void kill_bdev(struct block_device *);
2056extern struct super_block *freeze_bdev(struct block_device *);
2057extern void emergency_thaw_all(void);
2058extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2059extern int fsync_bdev(struct block_device *);
2060#else
2061static inline void bd_forget(struct inode *inode) {}
2062static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2063static inline void kill_bdev(struct block_device *bdev) {}
2064static inline void invalidate_bdev(struct block_device *bdev) {}
2065
2066static inline struct super_block *freeze_bdev(struct block_device *sb)
2067{
2068        return NULL;
2069}
2070
2071static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2072{
2073        return 0;
2074}
2075
2076static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2077{
2078}
2079#endif
2080extern int sync_filesystem(struct super_block *);
2081extern const struct file_operations def_blk_fops;
2082extern const struct file_operations def_chr_fops;
2083extern const struct file_operations bad_sock_fops;
2084extern const struct file_operations def_fifo_fops;
2085#ifdef CONFIG_BLOCK
2086extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2087extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2088extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2089extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2090extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2091                                               void *holder);
2092extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2093                                              void *holder);
2094extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2095#ifdef CONFIG_SYSFS
2096extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2097extern void bd_unlink_disk_holder(struct block_device *bdev,
2098                                  struct gendisk *disk);
2099#else
2100static inline int bd_link_disk_holder(struct block_device *bdev,
2101                                      struct gendisk *disk)
2102{
2103        return 0;
2104}
2105static inline void bd_unlink_disk_holder(struct block_device *bdev,
2106                                         struct gendisk *disk)
2107{
2108}
2109#endif
2110#endif
2111
2112/* fs/char_dev.c */
2113#define CHRDEV_MAJOR_HASH_SIZE  255
2114extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2115extern int register_chrdev_region(dev_t, unsigned, const char *);
2116extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2117                             unsigned int count, const char *name,
2118                             const struct file_operations *fops);
2119extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2120                                unsigned int count, const char *name);
2121extern void unregister_chrdev_region(dev_t, unsigned);
2122extern void chrdev_show(struct seq_file *,off_t);
2123
2124static inline int register_chrdev(unsigned int major, const char *name,
2125                                  const struct file_operations *fops)
2126{
2127        return __register_chrdev(major, 0, 256, name, fops);
2128}
2129
2130static inline void unregister_chrdev(unsigned int major, const char *name)
2131{
2132        __unregister_chrdev(major, 0, 256, name);
2133}
2134
2135/* fs/block_dev.c */
2136#define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2137#define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2138
2139#ifdef CONFIG_BLOCK
2140#define BLKDEV_MAJOR_HASH_SIZE  255
2141extern const char *__bdevname(dev_t, char *buffer);
2142extern const char *bdevname(struct block_device *bdev, char *buffer);
2143extern struct block_device *lookup_bdev(const char *);
2144extern void blkdev_show(struct seq_file *,off_t);
2145
2146#else
2147#define BLKDEV_MAJOR_HASH_SIZE  0
2148#endif
2149
2150extern void init_special_inode(struct inode *, umode_t, dev_t);
2151
2152/* Invalid inode operations -- fs/bad_inode.c */
2153extern void make_bad_inode(struct inode *);
2154extern int is_bad_inode(struct inode *);
2155
2156extern const struct file_operations read_pipefifo_fops;
2157extern const struct file_operations write_pipefifo_fops;
2158extern const struct file_operations rdwr_pipefifo_fops;
2159
2160#ifdef CONFIG_BLOCK
2161/*
2162 * return READ, READA, or WRITE
2163 */
2164#define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2165
2166/*
2167 * return data direction, READ or WRITE
2168 */
2169#define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2170
2171extern void check_disk_size_change(struct gendisk *disk,
2172                                   struct block_device *bdev);
2173extern int revalidate_disk(struct gendisk *);
2174extern int check_disk_change(struct block_device *);
2175extern int __invalidate_device(struct block_device *, bool);
2176extern int invalidate_partition(struct gendisk *, int);
2177#endif
2178unsigned long invalidate_mapping_pages(struct address_space *mapping,
2179                                        pgoff_t start, pgoff_t end);
2180
2181static inline void invalidate_remote_inode(struct inode *inode)
2182{
2183        if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2184            S_ISLNK(inode->i_mode))
2185                invalidate_mapping_pages(inode->i_mapping, 0, -1);
2186}
2187extern int invalidate_inode_pages2(struct address_space *mapping);
2188extern int invalidate_inode_pages2_range(struct address_space *mapping,
2189                                         pgoff_t start, pgoff_t end);
2190extern int write_inode_now(struct inode *, int);
2191extern int filemap_fdatawrite(struct address_space *);
2192extern int filemap_flush(struct address_space *);
2193extern int filemap_fdatawait(struct address_space *);
2194extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2195                                   loff_t lend);
2196extern int filemap_write_and_wait(struct address_space *mapping);
2197extern int filemap_write_and_wait_range(struct address_space *mapping,
2198                                        loff_t lstart, loff_t lend);
2199extern int __filemap_fdatawrite_range(struct address_space *mapping,
2200                                loff_t start, loff_t end, int sync_mode);
2201extern int filemap_fdatawrite_range(struct address_space *mapping,
2202                                loff_t start, loff_t end);
2203
2204extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2205                           int datasync);
2206extern int vfs_fsync(struct file *file, int datasync);
2207extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2208extern void emergency_sync(void);
2209extern void emergency_remount(void);
2210#ifdef CONFIG_BLOCK
2211extern sector_t bmap(struct inode *, sector_t);
2212#endif
2213extern int notify_change(struct dentry *, struct iattr *);
2214extern int inode_permission(struct inode *, int);
2215extern int generic_permission(struct inode *, int);
2216
2217static inline bool execute_ok(struct inode *inode)
2218{
2219        return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2220}
2221
2222/*
2223 * get_write_access() gets write permission for a file.
2224 * put_write_access() releases this write permission.
2225 * This is used for regular files.
2226 * We cannot support write (and maybe mmap read-write shared) accesses and
2227 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2228 * can have the following values:
2229 * 0: no writers, no VM_DENYWRITE mappings
2230 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2231 * > 0: (i_writecount) users are writing to the file.
2232 *
2233 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2234 * except for the cases where we don't hold i_writecount yet. Then we need to
2235 * use {get,deny}_write_access() - these functions check the sign and refuse
2236 * to do the change if sign is wrong.
2237 */
2238static inline int get_write_access(struct inode *inode)
2239{
2240        return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2241}
2242static inline int deny_write_access(struct file *file)
2243{
2244        struct inode *inode = file->f_path.dentry->d_inode;
2245        return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2246}
2247static inline void put_write_access(struct inode * inode)
2248{
2249        atomic_dec(&inode->i_writecount);
2250}
2251static inline void allow_write_access(struct file *file)
2252{
2253        if (file)
2254                atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2255}
2256#ifdef CONFIG_IMA
2257static inline void i_readcount_dec(struct inode *inode)
2258{
2259        BUG_ON(!atomic_read(&inode->i_readcount));
2260        atomic_dec(&inode->i_readcount);
2261}
2262static inline void i_readcount_inc(struct inode *inode)
2263{
2264        atomic_inc(&inode->i_readcount);
2265}
2266#else
2267static inline void i_readcount_dec(struct inode *inode)
2268{
2269        return;
2270}
2271static inline void i_readcount_inc(struct inode *inode)
2272{
2273        return;
2274}
2275#endif
2276extern int do_pipe_flags(int *, int);
2277
2278extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2279extern struct file * open_exec(const char *);
2280 
2281/* fs/dcache.c -- generic fs support functions */
2282extern int is_subdir(struct dentry *, struct dentry *);
2283extern int path_is_under(struct path *, struct path *);
2284extern ino_t find_inode_number(struct dentry *, struct qstr *);
2285
2286#include <linux/err.h>
2287
2288/* needed for stackable file system support */
2289extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2290
2291extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2292
2293extern int inode_init_always(struct super_block *, struct inode *);
2294extern void inode_init_once(struct inode *);
2295extern void address_space_init_once(struct address_space *mapping);
2296extern void ihold(struct inode * inode);
2297extern void iput(struct inode *);
2298extern struct inode * igrab(struct inode *);
2299extern ino_t iunique(struct super_block *, ino_t);
2300extern int inode_needs_sync(struct inode *inode);
2301extern int generic_delete_inode(struct inode *inode);
2302static inline int generic_drop_inode(struct inode *inode)
2303{
2304        return !inode->i_nlink || inode_unhashed(inode);
2305}
2306
2307extern struct inode *ilookup5_nowait(struct super_block *sb,
2308                unsigned long hashval, int (*test)(struct inode *, void *),
2309                void *data);
2310extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2311                int (*test)(struct inode *, void *), void *data);
2312extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2313
2314extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2315extern struct inode * iget_locked(struct super_block *, unsigned long);
2316extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2317extern int insert_inode_locked(struct inode *);
2318#ifdef CONFIG_DEBUG_LOCK_ALLOC
2319extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2320#else
2321static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2322#endif
2323extern void unlock_new_inode(struct inode *);
2324extern unsigned int get_next_ino(void);
2325
2326extern void __iget(struct inode * inode);
2327extern void iget_failed(struct inode *);
2328extern void clear_inode(struct inode *);
2329extern void __destroy_inode(struct inode *);
2330extern struct inode *new_inode_pseudo(struct super_block *sb);
2331extern struct inode *new_inode(struct super_block *sb);
2332extern void free_inode_nonrcu(struct inode *inode);
2333extern int should_remove_suid(struct dentry *);
2334extern int file_remove_suid(struct file *);
2335
2336extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2337static inline void insert_inode_hash(struct inode *inode)
2338{
2339        __insert_inode_hash(inode, inode->i_ino);
2340}
2341
2342extern void __remove_inode_hash(struct inode *);
2343static inline void remove_inode_hash(struct inode *inode)
2344{
2345        if (!inode_unhashed(inode))
2346                __remove_inode_hash(inode);
2347}
2348
2349extern void inode_sb_list_add(struct inode *inode);
2350
2351#ifdef CONFIG_BLOCK
2352extern void submit_bio(int, struct bio *);
2353extern int bdev_read_only(struct block_device *);
2354#endif
2355extern int set_blocksize(struct block_device *, int);
2356extern int sb_set_blocksize(struct super_block *, int);
2357extern int sb_min_blocksize(struct super_block *, int);
2358
2359extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2360extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2361extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2362                unsigned long size, pgoff_t pgoff);
2363extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2364int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2365extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2366extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2367                loff_t *);
2368extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2369extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2370                unsigned long *, loff_t, loff_t *, size_t, size_t);
2371extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2372                unsigned long, loff_t, loff_t *, size_t, ssize_t);
2373extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2374extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2375extern int generic_segment_checks(const struct iovec *iov,
2376                unsigned long *nr_segs, size_t *count, int access_flags);
2377
2378/* fs/block_dev.c */
2379extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2380                                unsigned long nr_segs, loff_t pos);
2381extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2382                        int datasync);
2383extern void block_sync_page(struct page *page);
2384
2385/* fs/splice.c */
2386extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2387                struct pipe_inode_info *, size_t, unsigned int);
2388extern ssize_t default_file_splice_read(struct file *, loff_t *,
2389                struct pipe_inode_info *, size_t, unsigned int);
2390extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2391                struct file *, loff_t *, size_t, unsigned int);
2392extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2393                struct file *out, loff_t *, size_t len, unsigned int flags);
2394extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2395                size_t len, unsigned int flags);
2396
2397extern void
2398file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2399extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2400extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2401extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2402extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2403                int origin, loff_t maxsize, loff_t eof);
2404extern int generic_file_open(struct inode * inode, struct file * filp);
2405extern int nonseekable_open(struct inode * inode, struct file * filp);
2406
2407#ifdef CONFIG_FS_XIP
2408extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2409                             loff_t *ppos);
2410extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2411extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2412                              size_t len, loff_t *ppos);
2413extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2414#else
2415static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2416{
2417        return 0;
2418}
2419#endif
2420
2421#ifdef CONFIG_BLOCK
2422typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2423                            loff_t file_offset);
2424
2425enum {
2426        /* need locking between buffered and direct access */
2427        DIO_LOCKING     = 0x01,
2428
2429        /* filesystem does not support filling holes */
2430        DIO_SKIP_HOLES  = 0x02,
2431};
2432
2433void dio_end_io(struct bio *bio, int error);
2434
2435ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2436        struct block_device *bdev, const struct iovec *iov, loff_t offset,
2437        unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2438        dio_submit_t submit_io, int flags);
2439
2440static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2441                struct inode *inode, const struct iovec *iov, loff_t offset,
2442                unsigned long nr_segs, get_block_t get_block)
2443{
2444        return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2445                                    offset, nr_segs, get_block, NULL, NULL,
2446                                    DIO_LOCKING | DIO_SKIP_HOLES);
2447}
2448#endif
2449
2450void inode_dio_wait(struct inode *inode);
2451void inode_dio_done(struct inode *inode);
2452
2453extern const struct file_operations generic_ro_fops;
2454
2455#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2456
2457extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2458extern int vfs_follow_link(struct nameidata *, const char *);
2459extern int page_readlink(struct dentry *, char __user *, int);
2460extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2461extern void page_put_link(struct dentry *, struct nameidata *, void *);
2462extern int __page_symlink(struct inode *inode, const char *symname, int len,
2463                int nofs);
2464extern int page_symlink(struct inode *inode, const char *symname, int len);
2465extern const struct inode_operations page_symlink_inode_operations;
2466extern int generic_readlink(struct dentry *, char __user *, int);
2467extern void generic_fillattr(struct inode *, struct kstat *);
2468extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2469void __inode_add_bytes(struct inode *inode, loff_t bytes);
2470void inode_add_bytes(struct inode *inode, loff_t bytes);
2471void inode_sub_bytes(struct inode *inode, loff_t bytes);
2472loff_t inode_get_bytes(struct inode *inode);
2473void inode_set_bytes(struct inode *inode, loff_t bytes);
2474
2475extern int vfs_readdir(struct file *, filldir_t, void *);
2476
2477extern int vfs_stat(const char __user *, struct kstat *);
2478extern int vfs_lstat(const char __user *, struct kstat *);
2479extern int vfs_fstat(unsigned int, struct kstat *);
2480extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2481
2482extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2483                    unsigned long arg);
2484extern int __generic_block_fiemap(struct inode *inode,
2485                                  struct fiemap_extent_info *fieinfo,
2486                                  loff_t start, loff_t len,
2487                                  get_block_t *get_block);
2488extern int generic_block_fiemap(struct inode *inode,
2489                                struct fiemap_extent_info *fieinfo, u64 start,
2490                                u64 len, get_block_t *get_block);
2491
2492extern void get_filesystem(struct file_system_type *fs);
2493extern void put_filesystem(struct file_system_type *fs);
2494extern struct file_system_type *get_fs_type(const char *name);
2495extern struct super_block *get_super(struct block_device *);
2496extern struct super_block *get_super_thawed(struct block_device *);
2497extern struct super_block *get_active_super(struct block_device *bdev);
2498extern void drop_super(struct super_block *sb);
2499extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2500extern void iterate_supers_type(struct file_system_type *,
2501                                void (*)(struct super_block *, void *), void *);
2502
2503extern int dcache_dir_open(struct inode *, struct file *);
2504extern int dcache_dir_close(struct inode *, struct file *);
2505extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2506extern int dcache_readdir(struct file *, void *, filldir_t);
2507extern int simple_setattr(struct dentry *, struct iattr *);
2508extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2509extern int simple_statfs(struct dentry *, struct kstatfs *);
2510extern int simple_open(struct inode *inode, struct file *file);
2511extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2512extern int simple_unlink(struct inode *, struct dentry *);
2513extern int simple_rmdir(struct inode *, struct dentry *);
2514extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2515extern int noop_fsync(struct file *, loff_t, loff_t, int);
2516extern int simple_empty(struct dentry *);
2517extern int simple_readpage(struct file *file, struct page *page);
2518extern int simple_write_begin(struct file *file, struct address_space *mapping,
2519                        loff_t pos, unsigned len, unsigned flags,
2520                        struct page **pagep, void **fsdata);
2521extern int simple_write_end(struct file *file, struct address_space *mapping,
2522                        loff_t pos, unsigned len, unsigned copied,
2523                        struct page *page, void *fsdata);
2524
2525extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2526extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2527extern const struct file_operations simple_dir_operations;
2528extern const struct inode_operations simple_dir_inode_operations;
2529struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2530struct dentry *d_alloc_name(struct dentry *, const char *);
2531extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2532extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2533extern void simple_release_fs(struct vfsmount **mount, int *count);
2534
2535extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2536                        loff_t *ppos, const void *from, size_t available);
2537extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2538                const void __user *from, size_t count);
2539
2540extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2541
2542extern int generic_check_addressable(unsigned, u64);
2543
2544#ifdef CONFIG_MIGRATION
2545extern int buffer_migrate_page(struct address_space *,
2546                                struct page *, struct page *,
2547                                enum migrate_mode);
2548#else
2549#define buffer_migrate_page NULL
2550#endif
2551
2552extern int inode_change_ok(const struct inode *, struct iattr *);
2553extern int inode_newsize_ok(const struct inode *, loff_t offset);
2554extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2555
2556extern int file_update_time(struct file *file);
2557
2558extern int generic_show_options(struct seq_file *m, struct dentry *root);
2559extern void save_mount_options(struct super_block *sb, char *options);
2560extern void replace_mount_options(struct super_block *sb, char *options);
2561
2562static inline ino_t parent_ino(struct dentry *dentry)
2563{
2564        ino_t res;
2565
2566        /*
2567         * Don't strictly need d_lock here? If the parent ino could change
2568         * then surely we'd have a deeper race in the caller?
2569         */
2570        spin_lock(&dentry->d_lock);
2571        res = dentry->d_parent->d_inode->i_ino;
2572        spin_unlock(&dentry->d_lock);
2573        return res;
2574}
2575
2576/* Transaction based IO helpers */
2577
2578/*
2579 * An argresp is stored in an allocated page and holds the
2580 * size of the argument or response, along with its content
2581 */
2582struct simple_transaction_argresp {
2583        ssize_t size;
2584        char data[0];
2585};
2586
2587#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2588
2589char *simple_transaction_get(struct file *file, const char __user *buf,
2590                                size_t size);
2591ssize_t simple_transaction_read(struct file *file, char __user *buf,
2592                                size_t size, loff_t *pos);
2593int simple_transaction_release(struct inode *inode, struct file *file);
2594
2595void simple_transaction_set(struct file *file, size_t n);
2596
2597/*
2598 * simple attribute files
2599 *
2600 * These attributes behave similar to those in sysfs:
2601 *
2602 * Writing to an attribute immediately sets a value, an open file can be
2603 * written to multiple times.
2604 *
2605 * Reading from an attribute creates a buffer from the value that might get
2606 * read with multiple read calls. When the attribute has been read
2607 * completely, no further read calls are possible until the file is opened
2608 * again.
2609 *
2610 * All attributes contain a text representation of a numeric value
2611 * that are accessed with the get() and set() functions.
2612 */
2613#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2614static int __fops ## _open(struct inode *inode, struct file *file)      \
2615{                                                                       \
2616        __simple_attr_check_format(__fmt, 0ull);                        \
2617        return simple_attr_open(inode, file, __get, __set, __fmt);      \
2618}                                                                       \
2619static const struct file_operations __fops = {                          \
2620        .owner   = THIS_MODULE,                                         \
2621        .open    = __fops ## _open,                                     \
2622        .release = simple_attr_release,                                 \
2623        .read    = simple_attr_read,                                    \
2624        .write   = simple_attr_write,                                   \
2625        .llseek  = generic_file_llseek,                                 \
2626};
2627
2628static inline __printf(1, 2)
2629void __simple_attr_check_format(const char *fmt, ...)
2630{
2631        /* don't do anything, just let the compiler check the arguments; */
2632}
2633
2634int simple_attr_open(struct inode *inode, struct file *file,
2635                     int (*get)(void *, u64 *), int (*set)(void *, u64),
2636                     const char *fmt);
2637int simple_attr_release(struct inode *inode, struct file *file);
2638ssize_t simple_attr_read(struct file *file, char __user *buf,
2639                         size_t len, loff_t *ppos);
2640ssize_t simple_attr_write(struct file *file, const char __user *buf,
2641                          size_t len, loff_t *ppos);
2642
2643struct ctl_table;
2644int proc_nr_files(struct ctl_table *table, int write,
2645                  void __user *buffer, size_t *lenp, loff_t *ppos);
2646int proc_nr_dentry(struct ctl_table *table, int write,
2647                  void __user *buffer, size_t *lenp, loff_t *ppos);
2648int proc_nr_inodes(struct ctl_table *table, int write,
2649                   void __user *buffer, size_t *lenp, loff_t *ppos);
2650int __init get_filesystem_list(char *buf);
2651
2652#define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2653#define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2654
2655#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2656#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2657                                            (flag & __FMODE_NONOTIFY)))
2658
2659static inline int is_sxid(umode_t mode)
2660{
2661        return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2662}
2663
2664static inline void inode_has_no_xattr(struct inode *inode)
2665{
2666        if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2667                inode->i_flags |= S_NOSEC;
2668}
2669
2670#endif /* _LINUX_FS_H */
2671
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.