The Design and Implementation of the FreeBSD Operating System, Second Edition
Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]

FreeBSD/Linux Kernel Cross Reference
sys/include/linux/fs.h

Version: -  FREEBSD  -  FREEBSD-13-STABLE  -  FREEBSD-13-0  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  l41  -  OPENBSD  -  linux-2.6  -  MK84  -  PLAN9  -  xnu-8792 
SearchContext: -  none  -  3  -  10 

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

Cache object: 7a64eb2f5daade311c9643f646923b0b


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]


This page is part of the FreeBSD/Linux Linux Kernel Cross-Reference, and was automatically generated using a modified version of the LXR engine.