Version:  2.0.40 2.2.26 2.4.37 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17

Linux/include/linux/fs.h

  1 #ifndef _LINUX_FS_H
  2 #define _LINUX_FS_H
  3 
  4 
  5 #include <linux/linkage.h>
  6 #include <linux/wait.h>
  7 #include <linux/kdev_t.h>
  8 #include <linux/dcache.h>
  9 #include <linux/path.h>
 10 #include <linux/stat.h>
 11 #include <linux/cache.h>
 12 #include <linux/list.h>
 13 #include <linux/list_lru.h>
 14 #include <linux/llist.h>
 15 #include <linux/radix-tree.h>
 16 #include <linux/rbtree.h>
 17 #include <linux/init.h>
 18 #include <linux/pid.h>
 19 #include <linux/bug.h>
 20 #include <linux/mutex.h>
 21 #include <linux/capability.h>
 22 #include <linux/semaphore.h>
 23 #include <linux/fiemap.h>
 24 #include <linux/rculist_bl.h>
 25 #include <linux/atomic.h>
 26 #include <linux/shrinker.h>
 27 #include <linux/migrate_mode.h>
 28 #include <linux/uidgid.h>
 29 #include <linux/lockdep.h>
 30 #include <linux/percpu-rwsem.h>
 31 #include <linux/blk_types.h>
 32 
 33 #include <asm/byteorder.h>
 34 #include <uapi/linux/fs.h>
 35 
 36 struct export_operations;
 37 struct hd_geometry;
 38 struct iovec;
 39 struct nameidata;
 40 struct kiocb;
 41 struct kobject;
 42 struct pipe_inode_info;
 43 struct poll_table_struct;
 44 struct kstatfs;
 45 struct vm_area_struct;
 46 struct vfsmount;
 47 struct cred;
 48 struct swap_info_struct;
 49 struct seq_file;
 50 struct workqueue_struct;
 51 struct iov_iter;
 52 
 53 extern void __init inode_init(void);
 54 extern void __init inode_init_early(void);
 55 extern void __init files_init(unsigned long);
 56 
 57 extern struct files_stat_struct files_stat;
 58 extern unsigned long get_max_files(void);
 59 extern int sysctl_nr_open;
 60 extern struct inodes_stat_t inodes_stat;
 61 extern int leases_enable, lease_break_time;
 62 extern int sysctl_protected_symlinks;
 63 extern int sysctl_protected_hardlinks;
 64 
 65 struct buffer_head;
 66 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 67                         struct buffer_head *bh_result, int create);
 68 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 69                         ssize_t bytes, void *private);
 70 
 71 #define MAY_EXEC                0x00000001
 72 #define MAY_WRITE               0x00000002
 73 #define MAY_READ                0x00000004
 74 #define MAY_APPEND              0x00000008
 75 #define MAY_ACCESS              0x00000010
 76 #define MAY_OPEN                0x00000020
 77 #define MAY_CHDIR               0x00000040
 78 /* called from RCU mode, don't block */
 79 #define MAY_NOT_BLOCK           0x00000080
 80 
 81 /*
 82  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
 83  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
 84  */
 85 
 86 /* file is open for reading */
 87 #define FMODE_READ              ((__force fmode_t)0x1)
 88 /* file is open for writing */
 89 #define FMODE_WRITE             ((__force fmode_t)0x2)
 90 /* file is seekable */
 91 #define FMODE_LSEEK             ((__force fmode_t)0x4)
 92 /* file can be accessed using pread */
 93 #define FMODE_PREAD             ((__force fmode_t)0x8)
 94 /* file can be accessed using pwrite */
 95 #define FMODE_PWRITE            ((__force fmode_t)0x10)
 96 /* File is opened for execution with sys_execve / sys_uselib */
 97 #define FMODE_EXEC              ((__force fmode_t)0x20)
 98 /* File is opened with O_NDELAY (only set for block devices) */
 99 #define FMODE_NDELAY            ((__force fmode_t)0x40)
100 /* File is opened with O_EXCL (only set for block devices) */
101 #define FMODE_EXCL              ((__force fmode_t)0x80)
102 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
103    (specialy hack for floppy.c) */
104 #define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
105 /* 32bit hashes as llseek() offset (for directories) */
106 #define FMODE_32BITHASH         ((__force fmode_t)0x200)
107 /* 64bit hashes as llseek() offset (for directories) */
108 #define FMODE_64BITHASH         ((__force fmode_t)0x400)
109 
110 /*
111  * Don't update ctime and mtime.
112  *
113  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
114  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
115  */
116 #define FMODE_NOCMTIME          ((__force fmode_t)0x800)
117 
118 /* Expect random access pattern */
119 #define FMODE_RANDOM            ((__force fmode_t)0x1000)
120 
121 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
122 #define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
123 
124 /* File is opened with O_PATH; almost nothing can be done with it */
125 #define FMODE_PATH              ((__force fmode_t)0x4000)
126 
127 /* File needs atomic accesses to f_pos */
128 #define FMODE_ATOMIC_POS        ((__force fmode_t)0x8000)
129 /* Write access to underlying fs */
130 #define FMODE_WRITER            ((__force fmode_t)0x10000)
131 /* Has read method(s) */
132 #define FMODE_CAN_READ          ((__force fmode_t)0x20000)
133 /* Has write method(s) */
134 #define FMODE_CAN_WRITE         ((__force fmode_t)0x40000)
135 
136 /* File was opened by fanotify and shouldn't generate fanotify events */
137 #define FMODE_NONOTIFY          ((__force fmode_t)0x1000000)
138 
139 /*
140  * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
141  * that indicates that they should check the contents of the iovec are
142  * valid, but not check the memory that the iovec elements
143  * points too.
144  */
145 #define CHECK_IOVEC_ONLY -1
146 
147 /*
148  * The below are the various read and write types that we support. Some of
149  * them include behavioral modifiers that send information down to the
150  * block layer and IO scheduler. Terminology:
151  *
152  *      The block layer uses device plugging to defer IO a little bit, in
153  *      the hope that we will see more IO very shortly. This increases
154  *      coalescing of adjacent IO and thus reduces the number of IOs we
155  *      have to send to the device. It also allows for better queuing,
156  *      if the IO isn't mergeable. If the caller is going to be waiting
157  *      for the IO, then he must ensure that the device is unplugged so
158  *      that the IO is dispatched to the driver.
159  *
160  *      All IO is handled async in Linux. This is fine for background
161  *      writes, but for reads or writes that someone waits for completion
162  *      on, we want to notify the block layer and IO scheduler so that they
163  *      know about it. That allows them to make better scheduling
164  *      decisions. So when the below references 'sync' and 'async', it
165  *      is referencing this priority hint.
166  *
167  * With that in mind, the available types are:
168  *
169  * READ                 A normal read operation. Device will be plugged.
170  * READ_SYNC            A synchronous read. Device is not plugged, caller can
171  *                      immediately wait on this read without caring about
172  *                      unplugging.
173  * READA                Used for read-ahead operations. Lower priority, and the
174  *                      block layer could (in theory) choose to ignore this
175  *                      request if it runs into resource problems.
176  * WRITE                A normal async write. Device will be plugged.
177  * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
178  *                      the hint that someone will be waiting on this IO
179  *                      shortly. The write equivalent of READ_SYNC.
180  * WRITE_ODIRECT        Special case write for O_DIRECT only.
181  * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
182  * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
183  *                      non-volatile media on completion.
184  * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
185  *                      by a cache flush and data is guaranteed to be on
186  *                      non-volatile media on completion.
187  *
188  */
189 #define RW_MASK                 REQ_WRITE
190 #define RWA_MASK                REQ_RAHEAD
191 
192 #define READ                    0
193 #define WRITE                   RW_MASK
194 #define READA                   RWA_MASK
195 #define KERNEL_READ             (READ|REQ_KERNEL)
196 #define KERNEL_WRITE            (WRITE|REQ_KERNEL)
197 
198 #define READ_SYNC               (READ | REQ_SYNC)
199 #define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
200 #define WRITE_ODIRECT           (WRITE | REQ_SYNC)
201 #define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
202 #define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
203 #define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
204 
205 /*
206  * Attribute flags.  These should be or-ed together to figure out what
207  * has been changed!
208  */
209 #define ATTR_MODE       (1 << 0)
210 #define ATTR_UID        (1 << 1)
211 #define ATTR_GID        (1 << 2)
212 #define ATTR_SIZE       (1 << 3)
213 #define ATTR_ATIME      (1 << 4)
214 #define ATTR_MTIME      (1 << 5)
215 #define ATTR_CTIME      (1 << 6)
216 #define ATTR_ATIME_SET  (1 << 7)
217 #define ATTR_MTIME_SET  (1 << 8)
218 #define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
219 #define ATTR_ATTR_FLAG  (1 << 10)
220 #define ATTR_KILL_SUID  (1 << 11)
221 #define ATTR_KILL_SGID  (1 << 12)
222 #define ATTR_FILE       (1 << 13)
223 #define ATTR_KILL_PRIV  (1 << 14)
224 #define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
225 #define ATTR_TIMES_SET  (1 << 16)
226 
227 /*
228  * This is the Inode Attributes structure, used for notify_change().  It
229  * uses the above definitions as flags, to know which values have changed.
230  * Also, in this manner, a Filesystem can look at only the values it cares
231  * about.  Basically, these are the attributes that the VFS layer can
232  * request to change from the FS layer.
233  *
234  * Derek Atkins <warlord@MIT.EDU> 94-10-20
235  */
236 struct iattr {
237         unsigned int    ia_valid;
238         umode_t         ia_mode;
239         kuid_t          ia_uid;
240         kgid_t          ia_gid;
241         loff_t          ia_size;
242         struct timespec ia_atime;
243         struct timespec ia_mtime;
244         struct timespec ia_ctime;
245 
246         /*
247          * Not an attribute, but an auxiliary info for filesystems wanting to
248          * implement an ftruncate() like method.  NOTE: filesystem should
249          * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
250          */
251         struct file     *ia_file;
252 };
253 
254 /*
255  * Includes for diskquotas.
256  */
257 #include <linux/quota.h>
258 
259 /** 
260  * enum positive_aop_returns - aop return codes with specific semantics
261  *
262  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
263  *                          completed, that the page is still locked, and
264  *                          should be considered active.  The VM uses this hint
265  *                          to return the page to the active list -- it won't
266  *                          be a candidate for writeback again in the near
267  *                          future.  Other callers must be careful to unlock
268  *                          the page if they get this return.  Returned by
269  *                          writepage(); 
270  *
271  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
272  *                      unlocked it and the page might have been truncated.
273  *                      The caller should back up to acquiring a new page and
274  *                      trying again.  The aop will be taking reasonable
275  *                      precautions not to livelock.  If the caller held a page
276  *                      reference, it should drop it before retrying.  Returned
277  *                      by readpage().
278  *
279  * address_space_operation functions return these large constants to indicate
280  * special semantics to the caller.  These are much larger than the bytes in a
281  * page to allow for functions that return the number of bytes operated on in a
282  * given page.
283  */
284 
285 enum positive_aop_returns {
286         AOP_WRITEPAGE_ACTIVATE  = 0x80000,
287         AOP_TRUNCATED_PAGE      = 0x80001,
288 };
289 
290 #define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
291 #define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
292 #define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
293                                                 * helper code (eg buffer layer)
294                                                 * to clear GFP_FS from alloc */
295 
296 /*
297  * oh the beauties of C type declarations.
298  */
299 struct page;
300 struct address_space;
301 struct writeback_control;
302 
303 /*
304  * "descriptor" for what we're up to with a read.
305  * This allows us to use the same read code yet
306  * have multiple different users of the data that
307  * we read from a file.
308  *
309  * The simplest case just copies the data to user
310  * mode.
311  */
312 typedef struct {
313         size_t written;
314         size_t count;
315         union {
316                 char __user *buf;
317                 void *data;
318         } arg;
319         int error;
320 } read_descriptor_t;
321 
322 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
323                 unsigned long, unsigned long);
324 
325 struct address_space_operations {
326         int (*writepage)(struct page *page, struct writeback_control *wbc);
327         int (*readpage)(struct file *, struct page *);
328 
329         /* Write back some dirty pages from this mapping. */
330         int (*writepages)(struct address_space *, struct writeback_control *);
331 
332         /* Set a page dirty.  Return true if this dirtied it */
333         int (*set_page_dirty)(struct page *page);
334 
335         int (*readpages)(struct file *filp, struct address_space *mapping,
336                         struct list_head *pages, unsigned nr_pages);
337 
338         int (*write_begin)(struct file *, struct address_space *mapping,
339                                 loff_t pos, unsigned len, unsigned flags,
340                                 struct page **pagep, void **fsdata);
341         int (*write_end)(struct file *, struct address_space *mapping,
342                                 loff_t pos, unsigned len, unsigned copied,
343                                 struct page *page, void *fsdata);
344 
345         /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
346         sector_t (*bmap)(struct address_space *, sector_t);
347         void (*invalidatepage) (struct page *, unsigned int, unsigned int);
348         int (*releasepage) (struct page *, gfp_t);
349         void (*freepage)(struct page *);
350         ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset);
351         int (*get_xip_mem)(struct address_space *, pgoff_t, int,
352                                                 void **, unsigned long *);
353         /*
354          * migrate the contents of a page to the specified target. If
355          * migrate_mode is MIGRATE_ASYNC, it must not block.
356          */
357         int (*migratepage) (struct address_space *,
358                         struct page *, struct page *, enum migrate_mode);
359         int (*launder_page) (struct page *);
360         int (*is_partially_uptodate) (struct page *, unsigned long,
361                                         unsigned long);
362         void (*is_dirty_writeback) (struct page *, bool *, bool *);
363         int (*error_remove_page)(struct address_space *, struct page *);
364 
365         /* swapfile support */
366         int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
367                                 sector_t *span);
368         void (*swap_deactivate)(struct file *file);
369 };
370 
371 extern const struct address_space_operations empty_aops;
372 
373 /*
374  * pagecache_write_begin/pagecache_write_end must be used by general code
375  * to write into the pagecache.
376  */
377 int pagecache_write_begin(struct file *, struct address_space *mapping,
378                                 loff_t pos, unsigned len, unsigned flags,
379                                 struct page **pagep, void **fsdata);
380 
381 int pagecache_write_end(struct file *, struct address_space *mapping,
382                                 loff_t pos, unsigned len, unsigned copied,
383                                 struct page *page, void *fsdata);
384 
385 struct backing_dev_info;
386 struct address_space {
387         struct inode            *host;          /* owner: inode, block_device */
388         struct radix_tree_root  page_tree;      /* radix tree of all pages */
389         spinlock_t              tree_lock;      /* and lock protecting it */
390         atomic_t                i_mmap_writable;/* count VM_SHARED mappings */
391         struct rb_root          i_mmap;         /* tree of private and shared mappings */
392         struct list_head        i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
393         struct mutex            i_mmap_mutex;   /* protect tree, count, list */
394         /* Protected by tree_lock together with the radix tree */
395         unsigned long           nrpages;        /* number of total pages */
396         unsigned long           nrshadows;      /* number of shadow entries */
397         pgoff_t                 writeback_index;/* writeback starts here */
398         const struct address_space_operations *a_ops;   /* methods */
399         unsigned long           flags;          /* error bits/gfp mask */
400         struct backing_dev_info *backing_dev_info; /* device readahead, etc */
401         spinlock_t              private_lock;   /* for use by the address_space */
402         struct list_head        private_list;   /* ditto */
403         void                    *private_data;  /* ditto */
404 } __attribute__((aligned(sizeof(long))));
405         /*
406          * On most architectures that alignment is already the case; but
407          * must be enforced here for CRIS, to let the least significant bit
408          * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
409          */
410 struct request_queue;
411 
412 struct block_device {
413         dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
414         int                     bd_openers;
415         struct inode *          bd_inode;       /* will die */
416         struct super_block *    bd_super;
417         struct mutex            bd_mutex;       /* open/close mutex */
418         struct list_head        bd_inodes;
419         void *                  bd_claiming;
420         void *                  bd_holder;
421         int                     bd_holders;
422         bool                    bd_write_holder;
423 #ifdef CONFIG_SYSFS
424         struct list_head        bd_holder_disks;
425 #endif
426         struct block_device *   bd_contains;
427         unsigned                bd_block_size;
428         struct hd_struct *      bd_part;
429         /* number of times partitions within this device have been opened. */
430         unsigned                bd_part_count;
431         int                     bd_invalidated;
432         struct gendisk *        bd_disk;
433         struct request_queue *  bd_queue;
434         struct list_head        bd_list;
435         /*
436          * Private data.  You must have bd_claim'ed the block_device
437          * to use this.  NOTE:  bd_claim allows an owner to claim
438          * the same device multiple times, the owner must take special
439          * care to not mess up bd_private for that case.
440          */
441         unsigned long           bd_private;
442 
443         /* The counter of freeze processes */
444         int                     bd_fsfreeze_count;
445         /* Mutex for freeze */
446         struct mutex            bd_fsfreeze_mutex;
447 };
448 
449 /*
450  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
451  * radix trees
452  */
453 #define PAGECACHE_TAG_DIRTY     0
454 #define PAGECACHE_TAG_WRITEBACK 1
455 #define PAGECACHE_TAG_TOWRITE   2
456 
457 int mapping_tagged(struct address_space *mapping, int tag);
458 
459 /*
460  * Might pages of this file be mapped into userspace?
461  */
462 static inline int mapping_mapped(struct address_space *mapping)
463 {
464         return  !RB_EMPTY_ROOT(&mapping->i_mmap) ||
465                 !list_empty(&mapping->i_mmap_nonlinear);
466 }
467 
468 /*
469  * Might pages of this file have been modified in userspace?
470  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
471  * marks vma as VM_SHARED if it is shared, and the file was opened for
472  * writing i.e. vma may be mprotected writable even if now readonly.
473  *
474  * If i_mmap_writable is negative, no new writable mappings are allowed. You
475  * can only deny writable mappings, if none exists right now.
476  */
477 static inline int mapping_writably_mapped(struct address_space *mapping)
478 {
479         return atomic_read(&mapping->i_mmap_writable) > 0;
480 }
481 
482 static inline int mapping_map_writable(struct address_space *mapping)
483 {
484         return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
485                 0 : -EPERM;
486 }
487 
488 static inline void mapping_unmap_writable(struct address_space *mapping)
489 {
490         atomic_dec(&mapping->i_mmap_writable);
491 }
492 
493 static inline int mapping_deny_writable(struct address_space *mapping)
494 {
495         return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
496                 0 : -EBUSY;
497 }
498 
499 static inline void mapping_allow_writable(struct address_space *mapping)
500 {
501         atomic_inc(&mapping->i_mmap_writable);
502 }
503 
504 /*
505  * Use sequence counter to get consistent i_size on 32-bit processors.
506  */
507 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
508 #include <linux/seqlock.h>
509 #define __NEED_I_SIZE_ORDERED
510 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
511 #else
512 #define i_size_ordered_init(inode) do { } while (0)
513 #endif
514 
515 struct posix_acl;
516 #define ACL_NOT_CACHED ((void *)(-1))
517 
518 #define IOP_FASTPERM    0x0001
519 #define IOP_LOOKUP      0x0002
520 #define IOP_NOFOLLOW    0x0004
521 
522 /*
523  * Keep mostly read-only and often accessed (especially for
524  * the RCU path lookup and 'stat' data) fields at the beginning
525  * of the 'struct inode'
526  */
527 struct inode {
528         umode_t                 i_mode;
529         unsigned short          i_opflags;
530         kuid_t                  i_uid;
531         kgid_t                  i_gid;
532         unsigned int            i_flags;
533 
534 #ifdef CONFIG_FS_POSIX_ACL
535         struct posix_acl        *i_acl;
536         struct posix_acl        *i_default_acl;
537 #endif
538 
539         const struct inode_operations   *i_op;
540         struct super_block      *i_sb;
541         struct address_space    *i_mapping;
542 
543 #ifdef CONFIG_SECURITY
544         void                    *i_security;
545 #endif
546 
547         /* Stat data, not accessed from path walking */
548         unsigned long           i_ino;
549         /*
550          * Filesystems may only read i_nlink directly.  They shall use the
551          * following functions for modification:
552          *
553          *    (set|clear|inc|drop)_nlink
554          *    inode_(inc|dec)_link_count
555          */
556         union {
557                 const unsigned int i_nlink;
558                 unsigned int __i_nlink;
559         };
560         dev_t                   i_rdev;
561         loff_t                  i_size;
562         struct timespec         i_atime;
563         struct timespec         i_mtime;
564         struct timespec         i_ctime;
565         spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
566         unsigned short          i_bytes;
567         unsigned int            i_blkbits;
568         blkcnt_t                i_blocks;
569 
570 #ifdef __NEED_I_SIZE_ORDERED
571         seqcount_t              i_size_seqcount;
572 #endif
573 
574         /* Misc */
575         unsigned long           i_state;
576         struct mutex            i_mutex;
577 
578         unsigned long           dirtied_when;   /* jiffies of first dirtying */
579 
580         struct hlist_node       i_hash;
581         struct list_head        i_wb_list;      /* backing dev IO list */
582         struct list_head        i_lru;          /* inode LRU list */
583         struct list_head        i_sb_list;
584         union {
585                 struct hlist_head       i_dentry;
586                 struct rcu_head         i_rcu;
587         };
588         u64                     i_version;
589         atomic_t                i_count;
590         atomic_t                i_dio_count;
591         atomic_t                i_writecount;
592 #ifdef CONFIG_IMA
593         atomic_t                i_readcount; /* struct files open RO */
594 #endif
595         const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
596         struct file_lock        *i_flock;
597         struct address_space    i_data;
598 #ifdef CONFIG_QUOTA
599         struct dquot            *i_dquot[MAXQUOTAS];
600 #endif
601         struct list_head        i_devices;
602         union {
603                 struct pipe_inode_info  *i_pipe;
604                 struct block_device     *i_bdev;
605                 struct cdev             *i_cdev;
606         };
607 
608         __u32                   i_generation;
609 
610 #ifdef CONFIG_FSNOTIFY
611         __u32                   i_fsnotify_mask; /* all events this inode cares about */
612         struct hlist_head       i_fsnotify_marks;
613 #endif
614 
615         void                    *i_private; /* fs or device private pointer */
616 };
617 
618 static inline int inode_unhashed(struct inode *inode)
619 {
620         return hlist_unhashed(&inode->i_hash);
621 }
622 
623 /*
624  * inode->i_mutex nesting subclasses for the lock validator:
625  *
626  * 0: the object of the current VFS operation
627  * 1: parent
628  * 2: child/target
629  * 3: xattr
630  * 4: second non-directory
631  * The last is for certain operations (such as rename) which lock two
632  * non-directories at once.
633  *
634  * The locking order between these classes is
635  * parent -> child -> normal -> xattr -> second non-directory
636  */
637 enum inode_i_mutex_lock_class
638 {
639         I_MUTEX_NORMAL,
640         I_MUTEX_PARENT,
641         I_MUTEX_CHILD,
642         I_MUTEX_XATTR,
643         I_MUTEX_NONDIR2
644 };
645 
646 void lock_two_nondirectories(struct inode *, struct inode*);
647 void unlock_two_nondirectories(struct inode *, struct inode*);
648 
649 /*
650  * NOTE: in a 32bit arch with a preemptable kernel and
651  * an UP compile the i_size_read/write must be atomic
652  * with respect to the local cpu (unlike with preempt disabled),
653  * but they don't need to be atomic with respect to other cpus like in
654  * true SMP (so they need either to either locally disable irq around
655  * the read or for example on x86 they can be still implemented as a
656  * cmpxchg8b without the need of the lock prefix). For SMP compiles
657  * and 64bit archs it makes no difference if preempt is enabled or not.
658  */
659 static inline loff_t i_size_read(const struct inode *inode)
660 {
661 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
662         loff_t i_size;
663         unsigned int seq;
664 
665         do {
666                 seq = read_seqcount_begin(&inode->i_size_seqcount);
667                 i_size = inode->i_size;
668         } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
669         return i_size;
670 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
671         loff_t i_size;
672 
673         preempt_disable();
674         i_size = inode->i_size;
675         preempt_enable();
676         return i_size;
677 #else
678         return inode->i_size;
679 #endif
680 }
681 
682 /*
683  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
684  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
685  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
686  */
687 static inline void i_size_write(struct inode *inode, loff_t i_size)
688 {
689 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
690         preempt_disable();
691         write_seqcount_begin(&inode->i_size_seqcount);
692         inode->i_size = i_size;
693         write_seqcount_end(&inode->i_size_seqcount);
694         preempt_enable();
695 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
696         preempt_disable();
697         inode->i_size = i_size;
698         preempt_enable();
699 #else
700         inode->i_size = i_size;
701 #endif
702 }
703 
704 /* Helper functions so that in most cases filesystems will
705  * not need to deal directly with kuid_t and kgid_t and can
706  * instead deal with the raw numeric values that are stored
707  * in the filesystem.
708  */
709 static inline uid_t i_uid_read(const struct inode *inode)
710 {
711         return from_kuid(&init_user_ns, inode->i_uid);
712 }
713 
714 static inline gid_t i_gid_read(const struct inode *inode)
715 {
716         return from_kgid(&init_user_ns, inode->i_gid);
717 }
718 
719 static inline void i_uid_write(struct inode *inode, uid_t uid)
720 {
721         inode->i_uid = make_kuid(&init_user_ns, uid);
722 }
723 
724 static inline void i_gid_write(struct inode *inode, gid_t gid)
725 {
726         inode->i_gid = make_kgid(&init_user_ns, gid);
727 }
728 
729 static inline unsigned iminor(const struct inode *inode)
730 {
731         return MINOR(inode->i_rdev);
732 }
733 
734 static inline unsigned imajor(const struct inode *inode)
735 {
736         return MAJOR(inode->i_rdev);
737 }
738 
739 extern struct block_device *I_BDEV(struct inode *inode);
740 
741 struct fown_struct {
742         rwlock_t lock;          /* protects pid, uid, euid fields */
743         struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
744         enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
745         kuid_t uid, euid;       /* uid/euid of process setting the owner */
746         int signum;             /* posix.1b rt signal to be delivered on IO */
747 };
748 
749 /*
750  * Track a single file's readahead state
751  */
752 struct file_ra_state {
753         pgoff_t start;                  /* where readahead started */
754         unsigned int size;              /* # of readahead pages */
755         unsigned int async_size;        /* do asynchronous readahead when
756                                            there are only # of pages ahead */
757 
758         unsigned int ra_pages;          /* Maximum readahead window */
759         unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
760         loff_t prev_pos;                /* Cache last read() position */
761 };
762 
763 /*
764  * Check if @index falls in the readahead windows.
765  */
766 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
767 {
768         return (index >= ra->start &&
769                 index <  ra->start + ra->size);
770 }
771 
772 struct file {
773         union {
774                 struct llist_node       fu_llist;
775                 struct rcu_head         fu_rcuhead;
776         } f_u;
777         struct path             f_path;
778 #define f_dentry        f_path.dentry
779         struct inode            *f_inode;       /* cached value */
780         const struct file_operations    *f_op;
781 
782         /*
783          * Protects f_ep_links, f_flags.
784          * Must not be taken from IRQ context.
785          */
786         spinlock_t              f_lock;
787         atomic_long_t           f_count;
788         unsigned int            f_flags;
789         fmode_t                 f_mode;
790         struct mutex            f_pos_lock;
791         loff_t                  f_pos;
792         struct fown_struct      f_owner;
793         const struct cred       *f_cred;
794         struct file_ra_state    f_ra;
795 
796         u64                     f_version;
797 #ifdef CONFIG_SECURITY
798         void                    *f_security;
799 #endif
800         /* needed for tty driver, and maybe others */
801         void                    *private_data;
802 
803 #ifdef CONFIG_EPOLL
804         /* Used by fs/eventpoll.c to link all the hooks to this file */
805         struct list_head        f_ep_links;
806         struct list_head        f_tfile_llink;
807 #endif /* #ifdef CONFIG_EPOLL */
808         struct address_space    *f_mapping;
809 } __attribute__((aligned(4)));  /* lest something weird decides that 2 is OK */
810 
811 struct file_handle {
812         __u32 handle_bytes;
813         int handle_type;
814         /* file identifier */
815         unsigned char f_handle[0];
816 };
817 
818 static inline struct file *get_file(struct file *f)
819 {
820         atomic_long_inc(&f->f_count);
821         return f;
822 }
823 #define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
824 #define file_count(x)   atomic_long_read(&(x)->f_count)
825 
826 #define MAX_NON_LFS     ((1UL<<31) - 1)
827 
828 /* Page cache limit. The filesystems should put that into their s_maxbytes 
829    limits, otherwise bad things can happen in VM. */ 
830 #if BITS_PER_LONG==32
831 #define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
832 #elif BITS_PER_LONG==64
833 #define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
834 #endif
835 
836 #define FL_POSIX        1
837 #define FL_FLOCK        2
838 #define FL_DELEG        4       /* NFSv4 delegation */
839 #define FL_ACCESS       8       /* not trying to lock, just looking */
840 #define FL_EXISTS       16      /* when unlocking, test for existence */
841 #define FL_LEASE        32      /* lease held on this file */
842 #define FL_CLOSE        64      /* unlock on close */
843 #define FL_SLEEP        128     /* A blocking lock */
844 #define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
845 #define FL_UNLOCK_PENDING       512 /* Lease is being broken */
846 #define FL_OFDLCK       1024    /* lock is "owned" by struct file */
847 
848 /*
849  * Special return value from posix_lock_file() and vfs_lock_file() for
850  * asynchronous locking.
851  */
852 #define FILE_LOCK_DEFERRED 1
853 
854 /*
855  * The POSIX file lock owner is determined by
856  * the "struct files_struct" in the thread group
857  * (or NULL for no owner - BSD locks).
858  *
859  * Lockd stuffs a "host" pointer into this.
860  */
861 typedef void *fl_owner_t;
862 
863 struct file_lock_operations {
864         void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
865         void (*fl_release_private)(struct file_lock *);
866 };
867 
868 struct lock_manager_operations {
869         int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
870         unsigned long (*lm_owner_key)(struct file_lock *);
871         void (*lm_notify)(struct file_lock *);  /* unblock callback */
872         int (*lm_grant)(struct file_lock *, struct file_lock *, int);
873         void (*lm_break)(struct file_lock *);
874         int (*lm_change)(struct file_lock **, int);
875 };
876 
877 struct lock_manager {
878         struct list_head list;
879 };
880 
881 struct net;
882 void locks_start_grace(struct net *, struct lock_manager *);
883 void locks_end_grace(struct lock_manager *);
884 int locks_in_grace(struct net *);
885 
886 /* that will die - we need it for nfs_lock_info */
887 #include <linux/nfs_fs_i.h>
888 
889 /*
890  * struct file_lock represents a generic "file lock". It's used to represent
891  * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
892  * note that the same struct is used to represent both a request for a lock and
893  * the lock itself, but the same object is never used for both.
894  *
895  * FIXME: should we create a separate "struct lock_request" to help distinguish
896  * these two uses?
897  *
898  * The i_flock list is ordered by:
899  *
900  * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
901  * 2) lock owner
902  * 3) lock range start
903  * 4) lock range end
904  *
905  * Obviously, the last two criteria only matter for POSIX locks.
906  */
907 struct file_lock {
908         struct file_lock *fl_next;      /* singly linked list for this inode  */
909         struct hlist_node fl_link;      /* node in global lists */
910         struct list_head fl_block;      /* circular list of blocked processes */
911         fl_owner_t fl_owner;
912         unsigned int fl_flags;
913         unsigned char fl_type;
914         unsigned int fl_pid;
915         int fl_link_cpu;                /* what cpu's list is this on? */
916         struct pid *fl_nspid;
917         wait_queue_head_t fl_wait;
918         struct file *fl_file;
919         loff_t fl_start;
920         loff_t fl_end;
921 
922         struct fasync_struct *  fl_fasync; /* for lease break notifications */
923         /* for lease breaks: */
924         unsigned long fl_break_time;
925         unsigned long fl_downgrade_time;
926 
927         const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
928         const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
929         union {
930                 struct nfs_lock_info    nfs_fl;
931                 struct nfs4_lock_info   nfs4_fl;
932                 struct {
933                         struct list_head link;  /* link in AFS vnode's pending_locks list */
934                         int state;              /* state of grant or error if -ve */
935                 } afs;
936         } fl_u;
937 };
938 
939 /* The following constant reflects the upper bound of the file/locking space */
940 #ifndef OFFSET_MAX
941 #define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
942 #define OFFSET_MAX      INT_LIMIT(loff_t)
943 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
944 #endif
945 
946 #include <linux/fcntl.h>
947 
948 extern void send_sigio(struct fown_struct *fown, int fd, int band);
949 
950 #ifdef CONFIG_FILE_LOCKING
951 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
952 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
953                         struct flock __user *);
954 
955 #if BITS_PER_LONG == 32
956 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
957 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
958                         struct flock64 __user *);
959 #endif
960 
961 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
962 extern int fcntl_getlease(struct file *filp);
963 
964 /* fs/locks.c */
965 void locks_free_lock(struct file_lock *fl);
966 extern void locks_init_lock(struct file_lock *);
967 extern struct file_lock * locks_alloc_lock(void);
968 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
969 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
970 extern void locks_remove_posix(struct file *, fl_owner_t);
971 extern void locks_remove_file(struct file *);
972 extern void locks_release_private(struct file_lock *);
973 extern void posix_test_lock(struct file *, struct file_lock *);
974 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
975 extern int posix_lock_file_wait(struct file *, struct file_lock *);
976 extern int posix_unblock_lock(struct file_lock *);
977 extern int vfs_test_lock(struct file *, struct file_lock *);
978 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
979 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
980 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
981 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
982 extern void lease_get_mtime(struct inode *, struct timespec *time);
983 extern int generic_setlease(struct file *, long, struct file_lock **);
984 extern int vfs_setlease(struct file *, long, struct file_lock **);
985 extern int lease_modify(struct file_lock **, int);
986 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
987 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
988 #else /* !CONFIG_FILE_LOCKING */
989 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
990                               struct flock __user *user)
991 {
992         return -EINVAL;
993 }
994 
995 static inline int fcntl_setlk(unsigned int fd, struct file *file,
996                               unsigned int cmd, struct flock __user *user)
997 {
998         return -EACCES;
999 }
1000 
1001 #if BITS_PER_LONG == 32
1002 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1003                                 struct flock64 __user *user)
1004 {
1005         return -EINVAL;
1006 }
1007 
1008 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1009                                 unsigned int cmd, struct flock64 __user *user)
1010 {
1011         return -EACCES;
1012 }
1013 #endif
1014 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1015 {
1016         return 0;
1017 }
1018 
1019 static inline int fcntl_getlease(struct file *filp)
1020 {
1021         return 0;
1022 }
1023 
1024 static inline void locks_init_lock(struct file_lock *fl)
1025 {
1026         return;
1027 }
1028 
1029 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1030 {
1031         return;
1032 }
1033 
1034 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1035 {
1036         return;
1037 }
1038 
1039 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1040 {
1041         return;
1042 }
1043 
1044 static inline void locks_remove_file(struct file *filp)
1045 {
1046         return;
1047 }
1048 
1049 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1050 {
1051         return;
1052 }
1053 
1054 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1055                                   struct file_lock *conflock)
1056 {
1057         return -ENOLCK;
1058 }
1059 
1060 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1061 {
1062         return -ENOLCK;
1063 }
1064 
1065 static inline int posix_unblock_lock(struct file_lock *waiter)
1066 {
1067         return -ENOENT;
1068 }
1069 
1070 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1071 {
1072         return 0;
1073 }
1074 
1075 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1076                                 struct file_lock *fl, struct file_lock *conf)
1077 {
1078         return -ENOLCK;
1079 }
1080 
1081 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1082 {
1083         return 0;
1084 }
1085 
1086 static inline int flock_lock_file_wait(struct file *filp,
1087                                        struct file_lock *request)
1088 {
1089         return -ENOLCK;
1090 }
1091 
1092 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1093 {
1094         return 0;
1095 }
1096 
1097 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1098 {
1099         return;
1100 }
1101 
1102 static inline int generic_setlease(struct file *filp, long arg,
1103                                     struct file_lock **flp)
1104 {
1105         return -EINVAL;
1106 }
1107 
1108 static inline int vfs_setlease(struct file *filp, long arg,
1109                                struct file_lock **lease)
1110 {
1111         return -EINVAL;
1112 }
1113 
1114 static inline int lease_modify(struct file_lock **before, int arg)
1115 {
1116         return -EINVAL;
1117 }
1118 
1119 static inline int lock_may_read(struct inode *inode, loff_t start,
1120                                 unsigned long len)
1121 {
1122         return 1;
1123 }
1124 
1125 static inline int lock_may_write(struct inode *inode, loff_t start,
1126                                  unsigned long len)
1127 {
1128         return 1;
1129 }
1130 #endif /* !CONFIG_FILE_LOCKING */
1131 
1132 
1133 struct fasync_struct {
1134         spinlock_t              fa_lock;
1135         int                     magic;
1136         int                     fa_fd;
1137         struct fasync_struct    *fa_next; /* singly linked list */
1138         struct file             *fa_file;
1139         struct rcu_head         fa_rcu;
1140 };
1141 
1142 #define FASYNC_MAGIC 0x4601
1143 
1144 /* SMP safe fasync helpers: */
1145 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1146 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1147 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1148 extern struct fasync_struct *fasync_alloc(void);
1149 extern void fasync_free(struct fasync_struct *);
1150 
1151 /* can be called from interrupts */
1152 extern void kill_fasync(struct fasync_struct **, int, int);
1153 
1154 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1155 extern int f_setown(struct file *filp, unsigned long arg, int force);
1156 extern void f_delown(struct file *filp);
1157 extern pid_t f_getown(struct file *filp);
1158 extern int send_sigurg(struct fown_struct *fown);
1159 
1160 struct mm_struct;
1161 
1162 /*
1163  *      Umount options
1164  */
1165 
1166 #define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1167 #define MNT_DETACH      0x00000002      /* Just detach from the tree */
1168 #define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1169 #define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1170 #define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1171 
1172 extern struct list_head super_blocks;
1173 extern spinlock_t sb_lock;
1174 
1175 /* Possible states of 'frozen' field */
1176 enum {
1177         SB_UNFROZEN = 0,                /* FS is unfrozen */
1178         SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1179         SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1180         SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1181                                          * internal threads if needed) */
1182         SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1183 };
1184 
1185 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1186 
1187 struct sb_writers {
1188         /* Counters for counting writers at each level */
1189         struct percpu_counter   counter[SB_FREEZE_LEVELS];
1190         wait_queue_head_t       wait;           /* queue for waiting for
1191                                                    writers / faults to finish */
1192         int                     frozen;         /* Is sb frozen? */
1193         wait_queue_head_t       wait_unfrozen;  /* queue for waiting for
1194                                                    sb to be thawed */
1195 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1196         struct lockdep_map      lock_map[SB_FREEZE_LEVELS];
1197 #endif
1198 };
1199 
1200 struct super_block {
1201         struct list_head        s_list;         /* Keep this first */
1202         dev_t                   s_dev;          /* search index; _not_ kdev_t */
1203         unsigned char           s_blocksize_bits;
1204         unsigned long           s_blocksize;
1205         loff_t                  s_maxbytes;     /* Max file size */
1206         struct file_system_type *s_type;
1207         const struct super_operations   *s_op;
1208         const struct dquot_operations   *dq_op;
1209         const struct quotactl_ops       *s_qcop;
1210         const struct export_operations *s_export_op;
1211         unsigned long           s_flags;
1212         unsigned long           s_magic;
1213         struct dentry           *s_root;
1214         struct rw_semaphore     s_umount;
1215         int                     s_count;
1216         atomic_t                s_active;
1217 #ifdef CONFIG_SECURITY
1218         void                    *s_security;
1219 #endif
1220         const struct xattr_handler **s_xattr;
1221 
1222         struct list_head        s_inodes;       /* all inodes */
1223         struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1224         struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1225         struct block_device     *s_bdev;
1226         struct backing_dev_info *s_bdi;
1227         struct mtd_info         *s_mtd;
1228         struct hlist_node       s_instances;
1229         struct quota_info       s_dquot;        /* Diskquota specific options */
1230 
1231         struct sb_writers       s_writers;
1232 
1233         char s_id[32];                          /* Informational name */
1234         u8 s_uuid[16];                          /* UUID */
1235 
1236         void                    *s_fs_info;     /* Filesystem private info */
1237         unsigned int            s_max_links;
1238         fmode_t                 s_mode;
1239 
1240         /* Granularity of c/m/atime in ns.
1241            Cannot be worse than a second */
1242         u32                s_time_gran;
1243 
1244         /*
1245          * The next field is for VFS *only*. No filesystems have any business
1246          * even looking at it. You had been warned.
1247          */
1248         struct mutex s_vfs_rename_mutex;        /* Kludge */
1249 
1250         /*
1251          * Filesystem subtype.  If non-empty the filesystem type field
1252          * in /proc/mounts will be "type.subtype"
1253          */
1254         char *s_subtype;
1255 
1256         /*
1257          * Saved mount options for lazy filesystems using
1258          * generic_show_options()
1259          */
1260         char __rcu *s_options;
1261         const struct dentry_operations *s_d_op; /* default d_op for dentries */
1262 
1263         /*
1264          * Saved pool identifier for cleancache (-1 means none)
1265          */
1266         int cleancache_poolid;
1267 
1268         struct shrinker s_shrink;       /* per-sb shrinker handle */
1269 
1270         /* Number of inodes with nlink == 0 but still referenced */
1271         atomic_long_t s_remove_count;
1272 
1273         /* Being remounted read-only */
1274         int s_readonly_remount;
1275 
1276         /* AIO completions deferred from interrupt context */
1277         struct workqueue_struct *s_dio_done_wq;
1278         struct hlist_head s_pins;
1279 
1280         /*
1281          * Keep the lru lists last in the structure so they always sit on their
1282          * own individual cachelines.
1283          */
1284         struct list_lru         s_dentry_lru ____cacheline_aligned_in_smp;
1285         struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
1286         struct rcu_head         rcu;
1287 };
1288 
1289 extern struct timespec current_fs_time(struct super_block *sb);
1290 
1291 /*
1292  * Snapshotting support.
1293  */
1294 
1295 void __sb_end_write(struct super_block *sb, int level);
1296 int __sb_start_write(struct super_block *sb, int level, bool wait);
1297 
1298 /**
1299  * sb_end_write - drop write access to a superblock
1300  * @sb: the super we wrote to
1301  *
1302  * Decrement number of writers to the filesystem. Wake up possible waiters
1303  * wanting to freeze the filesystem.
1304  */
1305 static inline void sb_end_write(struct super_block *sb)
1306 {
1307         __sb_end_write(sb, SB_FREEZE_WRITE);
1308 }
1309 
1310 /**
1311  * sb_end_pagefault - drop write access to a superblock from a page fault
1312  * @sb: the super we wrote to
1313  *
1314  * Decrement number of processes handling write page fault to the filesystem.
1315  * Wake up possible waiters wanting to freeze the filesystem.
1316  */
1317 static inline void sb_end_pagefault(struct super_block *sb)
1318 {
1319         __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1320 }
1321 
1322 /**
1323  * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1324  * @sb: the super we wrote to
1325  *
1326  * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1327  * waiters wanting to freeze the filesystem.
1328  */
1329 static inline void sb_end_intwrite(struct super_block *sb)
1330 {
1331         __sb_end_write(sb, SB_FREEZE_FS);
1332 }
1333 
1334 /**
1335  * sb_start_write - get write access to a superblock
1336  * @sb: the super we write to
1337  *
1338  * When a process wants to write data or metadata to a file system (i.e. dirty
1339  * a page or an inode), it should embed the operation in a sb_start_write() -
1340  * sb_end_write() pair to get exclusion against file system freezing. This
1341  * function increments number of writers preventing freezing. If the file
1342  * system is already frozen, the function waits until the file system is
1343  * thawed.
1344  *
1345  * Since freeze protection behaves as a lock, users have to preserve
1346  * ordering of freeze protection and other filesystem locks. Generally,
1347  * freeze protection should be the outermost lock. In particular, we have:
1348  *
1349  * sb_start_write
1350  *   -> i_mutex                 (write path, truncate, directory ops, ...)
1351  *   -> s_umount                (freeze_super, thaw_super)
1352  */
1353 static inline void sb_start_write(struct super_block *sb)
1354 {
1355         __sb_start_write(sb, SB_FREEZE_WRITE, true);
1356 }
1357 
1358 static inline int sb_start_write_trylock(struct super_block *sb)
1359 {
1360         return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1361 }
1362 
1363 /**
1364  * sb_start_pagefault - get write access to a superblock from a page fault
1365  * @sb: the super we write to
1366  *
1367  * When a process starts handling write page fault, it should embed the
1368  * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1369  * exclusion against file system freezing. This is needed since the page fault
1370  * is going to dirty a page. This function increments number of running page
1371  * faults preventing freezing. If the file system is already frozen, the
1372  * function waits until the file system is thawed.
1373  *
1374  * Since page fault freeze protection behaves as a lock, users have to preserve
1375  * ordering of freeze protection and other filesystem locks. It is advised to
1376  * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1377  * handling code implies lock dependency:
1378  *
1379  * mmap_sem
1380  *   -> sb_start_pagefault
1381  */
1382 static inline void sb_start_pagefault(struct super_block *sb)
1383 {
1384         __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1385 }
1386 
1387 /*
1388  * sb_start_intwrite - get write access to a superblock for internal fs purposes
1389  * @sb: the super we write to
1390  *
1391  * This is the third level of protection against filesystem freezing. It is
1392  * free for use by a filesystem. The only requirement is that it must rank
1393  * below sb_start_pagefault.
1394  *
1395  * For example filesystem can call sb_start_intwrite() when starting a
1396  * transaction which somewhat eases handling of freezing for internal sources
1397  * of filesystem changes (internal fs threads, discarding preallocation on file
1398  * close, etc.).
1399  */
1400 static inline void sb_start_intwrite(struct super_block *sb)
1401 {
1402         __sb_start_write(sb, SB_FREEZE_FS, true);
1403 }
1404 
1405 
1406 extern bool inode_owner_or_capable(const struct inode *inode);
1407 
1408 /*
1409  * VFS helper functions..
1410  */
1411 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1412 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1413 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1414 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1415 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1416 extern int vfs_rmdir(struct inode *, struct dentry *);
1417 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1418 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1419 
1420 /*
1421  * VFS dentry helper functions.
1422  */
1423 extern void dentry_unhash(struct dentry *dentry);
1424 
1425 /*
1426  * VFS file helper functions.
1427  */
1428 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1429                         umode_t mode);
1430 /*
1431  * VFS FS_IOC_FIEMAP helper definitions.
1432  */
1433 struct fiemap_extent_info {
1434         unsigned int fi_flags;          /* Flags as passed from user */
1435         unsigned int fi_extents_mapped; /* Number of mapped extents */
1436         unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1437         struct fiemap_extent __user *fi_extents_start; /* Start of
1438                                                         fiemap_extent array */
1439 };
1440 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1441                             u64 phys, u64 len, u32 flags);
1442 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1443 
1444 /*
1445  * File types
1446  *
1447  * NOTE! These match bits 12..15 of stat.st_mode
1448  * (ie "(i_mode >> 12) & 15").
1449  */
1450 #define DT_UNKNOWN      0
1451 #define DT_FIFO         1
1452 #define DT_CHR          2
1453 #define DT_DIR          4
1454 #define DT_BLK          6
1455 #define DT_REG          8
1456 #define DT_LNK          10
1457 #define DT_SOCK         12
1458 #define DT_WHT          14
1459 
1460 /*
1461  * This is the "filldir" function type, used by readdir() to let
1462  * the kernel specify what kind of dirent layout it wants to have.
1463  * This allows the kernel to read directories into kernel space or
1464  * to have different dirent layouts depending on the binary type.
1465  */
1466 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1467 struct dir_context {
1468         const filldir_t actor;
1469         loff_t pos;
1470 };
1471 
1472 struct block_device_operations;
1473 
1474 /* These macros are for out of kernel modules to test that
1475  * the kernel supports the unlocked_ioctl and compat_ioctl
1476  * fields in struct file_operations. */
1477 #define HAVE_COMPAT_IOCTL 1
1478 #define HAVE_UNLOCKED_IOCTL 1
1479 
1480 struct iov_iter;
1481 
1482 struct file_operations {
1483         struct module *owner;
1484         loff_t (*llseek) (struct file *, loff_t, int);
1485         ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1486         ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1487         ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1488         ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1489         ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1490         ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1491         int (*iterate) (struct file *, struct dir_context *);
1492         unsigned int (*poll) (struct file *, struct poll_table_struct *);
1493         long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1494         long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1495         int (*mmap) (struct file *, struct vm_area_struct *);
1496         int (*open) (struct inode *, struct file *);
1497         int (*flush) (struct file *, fl_owner_t id);
1498         int (*release) (struct inode *, struct file *);
1499         int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1500         int (*aio_fsync) (struct kiocb *, int datasync);
1501         int (*fasync) (int, struct file *, int);
1502         int (*lock) (struct file *, int, struct file_lock *);
1503         ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1504         unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1505         int (*check_flags)(int);
1506         int (*flock) (struct file *, int, struct file_lock *);
1507         ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1508         ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1509         int (*setlease)(struct file *, long, struct file_lock **);
1510         long (*fallocate)(struct file *file, int mode, loff_t offset,
1511                           loff_t len);
1512         int (*show_fdinfo)(struct seq_file *m, struct file *f);
1513 };
1514 
1515 struct inode_operations {
1516         struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1517         void * (*follow_link) (struct dentry *, struct nameidata *);
1518         int (*permission) (struct inode *, int);
1519         struct posix_acl * (*get_acl)(struct inode *, int);
1520 
1521         int (*readlink) (struct dentry *, char __user *,int);
1522         void (*put_link) (struct dentry *, struct nameidata *, void *);
1523 
1524         int (*create) (struct inode *,struct dentry *, umode_t, bool);
1525         int (*link) (struct dentry *,struct inode *,struct dentry *);
1526         int (*unlink) (struct inode *,struct dentry *);
1527         int (*symlink) (struct inode *,struct dentry *,const char *);
1528         int (*mkdir) (struct inode *,struct dentry *,umode_t);
1529         int (*rmdir) (struct inode *,struct dentry *);
1530         int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1531         int (*rename) (struct inode *, struct dentry *,
1532                         struct inode *, struct dentry *);
1533         int (*rename2) (struct inode *, struct dentry *,
1534                         struct inode *, struct dentry *, unsigned int);
1535         int (*setattr) (struct dentry *, struct iattr *);
1536         int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1537         int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1538         ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1539         ssize_t (*listxattr) (struct dentry *, char *, size_t);
1540         int (*removexattr) (struct dentry *, const char *);
1541         int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1542                       u64 len);
1543         int (*update_time)(struct inode *, struct timespec *, int);
1544         int (*atomic_open)(struct inode *, struct dentry *,
1545                            struct file *, unsigned open_flag,
1546                            umode_t create_mode, int *opened);
1547         int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1548         int (*set_acl)(struct inode *, struct posix_acl *, int);
1549 } ____cacheline_aligned;
1550 
1551 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1552                               unsigned long nr_segs, unsigned long fast_segs,
1553                               struct iovec *fast_pointer,
1554                               struct iovec **ret_pointer);
1555 
1556 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1557 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1558 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1559                 unsigned long, loff_t *);
1560 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1561                 unsigned long, loff_t *);
1562 
1563 struct super_operations {
1564         struct inode *(*alloc_inode)(struct super_block *sb);
1565         void (*destroy_inode)(struct inode *);
1566 
1567         void (*dirty_inode) (struct inode *, int flags);
1568         int (*write_inode) (struct inode *, struct writeback_control *wbc);
1569         int (*drop_inode) (struct inode *);
1570         void (*evict_inode) (struct inode *);
1571         void (*put_super) (struct super_block *);
1572         int (*sync_fs)(struct super_block *sb, int wait);
1573         int (*freeze_fs) (struct super_block *);
1574         int (*unfreeze_fs) (struct super_block *);
1575         int (*statfs) (struct dentry *, struct kstatfs *);
1576         int (*remount_fs) (struct super_block *, int *, char *);
1577         void (*umount_begin) (struct super_block *);
1578 
1579         int (*show_options)(struct seq_file *, struct dentry *);
1580         int (*show_devname)(struct seq_file *, struct dentry *);
1581         int (*show_path)(struct seq_file *, struct dentry *);
1582         int (*show_stats)(struct seq_file *, struct dentry *);
1583 #ifdef CONFIG_QUOTA
1584         ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1585         ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1586 #endif
1587         int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1588         long (*nr_cached_objects)(struct super_block *, int);
1589         long (*free_cached_objects)(struct super_block *, long, int);
1590 };
1591 
1592 /*
1593  * Inode flags - they have no relation to superblock flags now
1594  */
1595 #define S_SYNC          1       /* Writes are synced at once */
1596 #define S_NOATIME       2       /* Do not update access times */
1597 #define S_APPEND        4       /* Append-only file */
1598 #define S_IMMUTABLE     8       /* Immutable file */
1599 #define S_DEAD          16      /* removed, but still open directory */
1600 #define S_NOQUOTA       32      /* Inode is not counted to quota */
1601 #define S_DIRSYNC       64      /* Directory modifications are synchronous */
1602 #define S_NOCMTIME      128     /* Do not update file c/mtime */
1603 #define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1604 #define S_PRIVATE       512     /* Inode is fs-internal */
1605 #define S_IMA           1024    /* Inode has an associated IMA struct */
1606 #define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1607 #define S_NOSEC         4096    /* no suid or xattr security attributes */
1608 
1609 /*
1610  * Note that nosuid etc flags are inode-specific: setting some file-system
1611  * flags just means all the inodes inherit those flags by default. It might be
1612  * possible to override it selectively if you really wanted to with some
1613  * ioctl() that is not currently implemented.
1614  *
1615  * Exception: MS_RDONLY is always applied to the entire file system.
1616  *
1617  * Unfortunately, it is possible to change a filesystems flags with it mounted
1618  * with files in use.  This means that all of the inodes will not have their
1619  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1620  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1621  */
1622 #define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1623 
1624 #define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1625 #define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1626                                         ((inode)->i_flags & S_SYNC))
1627 #define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1628                                         ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1629 #define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1630 #define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1631 #define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1632 
1633 #define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1634 #define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1635 #define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1636 #define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1637 
1638 #define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1639 #define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1640 #define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1641 #define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1642 #define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1643 #define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1644 #define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1645 
1646 /*
1647  * Inode state bits.  Protected by inode->i_lock
1648  *
1649  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1650  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1651  *
1652  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1653  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1654  * various stages of removing an inode.
1655  *
1656  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1657  *
1658  * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1659  *                      fdatasync().  i_atime is the usual cause.
1660  * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1661  *                      these changes separately from I_DIRTY_SYNC so that we
1662  *                      don't have to write inode on fdatasync() when only
1663  *                      mtime has changed in it.
1664  * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1665  * I_NEW                Serves as both a mutex and completion notification.
1666  *                      New inodes set I_NEW.  If two processes both create
1667  *                      the same inode, one of them will release its inode and
1668  *                      wait for I_NEW to be released before returning.
1669  *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1670  *                      also cause waiting on I_NEW, without I_NEW actually
1671  *                      being set.  find_inode() uses this to prevent returning
1672  *                      nearly-dead inodes.
1673  * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1674  *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1675  *                      cleared.
1676  * I_FREEING            Set when inode is about to be freed but still has dirty
1677  *                      pages or buffers attached or the inode itself is still
1678  *                      dirty.
1679  * I_CLEAR              Added by clear_inode().  In this state the inode is
1680  *                      clean and can be destroyed.  Inode keeps I_FREEING.
1681  *
1682  *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1683  *                      prohibited for many purposes.  iget() must wait for
1684  *                      the inode to be completely released, then create it
1685  *                      anew.  Other functions will just ignore such inodes,
1686  *                      if appropriate.  I_NEW is used for waiting.
1687  *
1688  * I_SYNC               Writeback of inode is running. The bit is set during
1689  *                      data writeback, and cleared with a wakeup on the bit
1690  *                      address once it is done. The bit is also used to pin
1691  *                      the inode in memory for flusher thread.
1692  *
1693  * I_REFERENCED         Marks the inode as recently references on the LRU list.
1694  *
1695  * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1696  *
1697  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1698  */
1699 #define I_DIRTY_SYNC            (1 << 0)
1700 #define I_DIRTY_DATASYNC        (1 << 1)
1701 #define I_DIRTY_PAGES           (1 << 2)
1702 #define __I_NEW                 3
1703 #define I_NEW                   (1 << __I_NEW)
1704 #define I_WILL_FREE             (1 << 4)
1705 #define I_FREEING               (1 << 5)
1706 #define I_CLEAR                 (1 << 6)
1707 #define __I_SYNC                7
1708 #define I_SYNC                  (1 << __I_SYNC)
1709 #define I_REFERENCED            (1 << 8)
1710 #define __I_DIO_WAKEUP          9
1711 #define I_DIO_WAKEUP            (1 << I_DIO_WAKEUP)
1712 #define I_LINKABLE              (1 << 10)
1713 
1714 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1715 
1716 extern void __mark_inode_dirty(struct inode *, int);
1717 static inline void mark_inode_dirty(struct inode *inode)
1718 {
1719         __mark_inode_dirty(inode, I_DIRTY);
1720 }
1721 
1722 static inline void mark_inode_dirty_sync(struct inode *inode)
1723 {
1724         __mark_inode_dirty(inode, I_DIRTY_SYNC);
1725 }
1726 
1727 extern void inc_nlink(struct inode *inode);
1728 extern void drop_nlink(struct inode *inode);
1729 extern void clear_nlink(struct inode *inode);
1730 extern void set_nlink(struct inode *inode, unsigned int nlink);
1731 
1732 static inline void inode_inc_link_count(struct inode *inode)
1733 {
1734         inc_nlink(inode);
1735         mark_inode_dirty(inode);
1736 }
1737 
1738 static inline void inode_dec_link_count(struct inode *inode)
1739 {
1740         drop_nlink(inode);
1741         mark_inode_dirty(inode);
1742 }
1743 
1744 /**
1745  * inode_inc_iversion - increments i_version
1746  * @inode: inode that need to be updated
1747  *
1748  * Every time the inode is modified, the i_version field will be incremented.
1749  * The filesystem has to be mounted with i_version flag
1750  */
1751 
1752 static inline void inode_inc_iversion(struct inode *inode)
1753 {
1754        spin_lock(&inode->i_lock);
1755        inode->i_version++;
1756        spin_unlock(&inode->i_lock);
1757 }
1758 
1759 enum file_time_flags {
1760         S_ATIME = 1,
1761         S_MTIME = 2,
1762         S_CTIME = 4,
1763         S_VERSION = 8,
1764 };
1765 
1766 extern void touch_atime(const struct path *);
1767 static inline void file_accessed(struct file *file)
1768 {
1769         if (!(file->f_flags & O_NOATIME))
1770                 touch_atime(&file->f_path);
1771 }
1772 
1773 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1774 int sync_inode_metadata(struct inode *inode, int wait);
1775 
1776 struct file_system_type {
1777         const char *name;
1778         int fs_flags;
1779 #define FS_REQUIRES_DEV         1 
1780 #define FS_BINARY_MOUNTDATA     2
1781 #define FS_HAS_SUBTYPE          4
1782 #define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
1783 #define FS_USERNS_DEV_MOUNT     16 /* A userns mount does not imply MNT_NODEV */
1784 #define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1785         struct dentry *(*mount) (struct file_system_type *, int,
1786                        const char *, void *);
1787         void (*kill_sb) (struct super_block *);
1788         struct module *owner;
1789         struct file_system_type * next;
1790         struct hlist_head fs_supers;
1791 
1792         struct lock_class_key s_lock_key;
1793         struct lock_class_key s_umount_key;
1794         struct lock_class_key s_vfs_rename_key;
1795         struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1796 
1797         struct lock_class_key i_lock_key;
1798         struct lock_class_key i_mutex_key;
1799         struct lock_class_key i_mutex_dir_key;
1800 };
1801 
1802 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1803 
1804 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1805         void *data, int (*fill_super)(struct super_block *, void *, int));
1806 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1807         int flags, const char *dev_name, void *data,
1808         int (*fill_super)(struct super_block *, void *, int));
1809 extern struct dentry *mount_single(struct file_system_type *fs_type,
1810         int flags, void *data,
1811         int (*fill_super)(struct super_block *, void *, int));
1812 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1813         int flags, void *data,
1814         int (*fill_super)(struct super_block *, void *, int));
1815 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1816 void generic_shutdown_super(struct super_block *sb);
1817 void kill_block_super(struct super_block *sb);
1818 void kill_anon_super(struct super_block *sb);
1819 void kill_litter_super(struct super_block *sb);
1820 void deactivate_super(struct super_block *sb);
1821 void deactivate_locked_super(struct super_block *sb);
1822 int set_anon_super(struct super_block *s, void *data);
1823 int get_anon_bdev(dev_t *);
1824 void free_anon_bdev(dev_t);
1825 struct super_block *sget(struct file_system_type *type,
1826                         int (*test)(struct super_block *,void *),
1827                         int (*set)(struct super_block *,void *),
1828                         int flags, void *data);
1829 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1830         const struct super_operations *ops,
1831         const struct dentry_operations *dops,
1832         unsigned long);
1833 
1834 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1835 #define fops_get(fops) \
1836         (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1837 #define fops_put(fops) \
1838         do { if (fops) module_put((fops)->owner); } while(0)
1839 /*
1840  * This one is to be used *ONLY* from ->open() instances.
1841  * fops must be non-NULL, pinned down *and* module dependencies
1842  * should be sufficient to pin the caller down as well.
1843  */
1844 #define replace_fops(f, fops) \
1845         do {    \
1846                 struct file *__file = (f); \
1847                 fops_put(__file->f_op); \
1848                 BUG_ON(!(__file->f_op = (fops))); \
1849         } while(0)
1850 
1851 extern int register_filesystem(struct file_system_type *);
1852 extern int unregister_filesystem(struct file_system_type *);
1853 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1854 #define kern_mount(type) kern_mount_data(type, NULL)
1855 extern void kern_unmount(struct vfsmount *mnt);
1856 extern int may_umount_tree(struct vfsmount *);
1857 extern int may_umount(struct vfsmount *);
1858 extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1859 extern struct vfsmount *collect_mounts(struct path *);
1860 extern void drop_collected_mounts(struct vfsmount *);
1861 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1862                           struct vfsmount *);
1863 extern int vfs_statfs(struct path *, struct kstatfs *);
1864 extern int user_statfs(const char __user *, struct kstatfs *);
1865 extern int fd_statfs(int, struct kstatfs *);
1866 extern int vfs_ustat(dev_t, struct kstatfs *);
1867 extern int freeze_super(struct super_block *super);
1868 extern int thaw_super(struct super_block *super);
1869 extern bool our_mnt(struct vfsmount *mnt);
1870 extern bool fs_fully_visible(struct file_system_type *);
1871 
1872 extern int current_umask(void);
1873 
1874 extern void ihold(struct inode * inode);
1875 extern void iput(struct inode *);
1876 
1877 static inline struct inode *file_inode(struct file *f)
1878 {
1879         return f->f_inode;
1880 }
1881 
1882 /* /sys/fs */
1883 extern struct kobject *fs_kobj;
1884 
1885 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1886 
1887 #define FLOCK_VERIFY_READ  1
1888 #define FLOCK_VERIFY_WRITE 2
1889 
1890 #ifdef CONFIG_FILE_LOCKING
1891 extern int locks_mandatory_locked(struct file *);
1892 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1893 
1894 /*
1895  * Candidates for mandatory locking have the setgid bit set
1896  * but no group execute bit -  an otherwise meaningless combination.
1897  */
1898 
1899 static inline int __mandatory_lock(struct inode *ino)
1900 {
1901         return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1902 }
1903 
1904 /*
1905  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1906  * otherwise these will be advisory locks
1907  */
1908 
1909 static inline int mandatory_lock(struct inode *ino)
1910 {
1911         return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1912 }
1913 
1914 static inline int locks_verify_locked(struct file *file)
1915 {
1916         if (mandatory_lock(file_inode(file)))
1917                 return locks_mandatory_locked(file);
1918         return 0;
1919 }
1920 
1921 static inline int locks_verify_truncate(struct inode *inode,
1922                                     struct file *filp,
1923                                     loff_t size)
1924 {
1925         if (inode->i_flock && mandatory_lock(inode))
1926                 return locks_mandatory_area(
1927                         FLOCK_VERIFY_WRITE, inode, filp,
1928                         size < inode->i_size ? size : inode->i_size,
1929                         (size < inode->i_size ? inode->i_size - size
1930                          : size - inode->i_size)
1931                 );
1932         return 0;
1933 }
1934 
1935 static inline int break_lease(struct inode *inode, unsigned int mode)
1936 {
1937         /*
1938          * Since this check is lockless, we must ensure that any refcounts
1939          * taken are done before checking inode->i_flock. Otherwise, we could
1940          * end up racing with tasks trying to set a new lease on this file.
1941          */
1942         smp_mb();
1943         if (inode->i_flock)
1944                 return __break_lease(inode, mode, FL_LEASE);
1945         return 0;
1946 }
1947 
1948 static inline int break_deleg(struct inode *inode, unsigned int mode)
1949 {
1950         /*
1951          * Since this check is lockless, we must ensure that any refcounts
1952          * taken are done before checking inode->i_flock. Otherwise, we could
1953          * end up racing with tasks trying to set a new lease on this file.
1954          */
1955         smp_mb();
1956         if (inode->i_flock)
1957                 return __break_lease(inode, mode, FL_DELEG);
1958         return 0;
1959 }
1960 
1961 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
1962 {
1963         int ret;
1964 
1965         ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
1966         if (ret == -EWOULDBLOCK && delegated_inode) {
1967                 *delegated_inode = inode;
1968                 ihold(inode);
1969         }
1970         return ret;
1971 }
1972 
1973 static inline int break_deleg_wait(struct inode **delegated_inode)
1974 {
1975         int ret;
1976 
1977         ret = break_deleg(*delegated_inode, O_WRONLY);
1978         iput(*delegated_inode);
1979         *delegated_inode = NULL;
1980         return ret;
1981 }
1982 
1983 #else /* !CONFIG_FILE_LOCKING */
1984 static inline int locks_mandatory_locked(struct file *file)
1985 {
1986         return 0;
1987 }
1988 
1989 static inline int locks_mandatory_area(int rw, struct inode *inode,
1990                                        struct file *filp, loff_t offset,
1991                                        size_t count)
1992 {
1993         return 0;
1994 }
1995 
1996 static inline int __mandatory_lock(struct inode *inode)
1997 {
1998         return 0;
1999 }
2000 
2001 static inline int mandatory_lock(struct inode *inode)
2002 {
2003         return 0;
2004 }
2005 
2006 static inline int locks_verify_locked(struct file *file)
2007 {
2008         return 0;
2009 }
2010 
2011 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2012                                         size_t size)
2013 {
2014         return 0;
2015 }
2016 
2017 static inline int break_lease(struct inode *inode, unsigned int mode)
2018 {
2019         return 0;
2020 }
2021 
2022 static inline int break_deleg(struct inode *inode, unsigned int mode)
2023 {
2024         return 0;
2025 }
2026 
2027 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2028 {
2029         return 0;
2030 }
2031 
2032 static inline int break_deleg_wait(struct inode **delegated_inode)
2033 {
2034         BUG();
2035         return 0;
2036 }
2037 
2038 #endif /* CONFIG_FILE_LOCKING */
2039 
2040 /* fs/open.c */
2041 struct audit_names;
2042 struct filename {
2043         const char              *name;  /* pointer to actual string */
2044         const __user char       *uptr;  /* original userland pointer */
2045         struct audit_names      *aname;
2046         bool                    separate; /* should "name" be freed? */
2047 };
2048 
2049 extern long vfs_truncate(struct path *, loff_t);
2050 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2051                        struct file *filp);
2052 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2053                         loff_t len);
2054 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2055                         umode_t mode);
2056 extern struct file *file_open_name(struct filename *, int, umode_t);
2057 extern struct file *filp_open(const char *, int, umode_t);
2058 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2059                                    const char *, int);
2060 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2061 extern int filp_close(struct file *, fl_owner_t id);
2062 
2063 extern struct filename *getname(const char __user *);
2064 extern struct filename *getname_kernel(const char *);
2065 
2066 enum {
2067         FILE_CREATED = 1,
2068         FILE_OPENED = 2
2069 };
2070 extern int finish_open(struct file *file, struct dentry *dentry,
2071                         int (*open)(struct inode *, struct file *),
2072                         int *opened);
2073 extern int finish_no_open(struct file *file, struct dentry *dentry);
2074 
2075 /* fs/ioctl.c */
2076 
2077 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2078 
2079 /* fs/dcache.c */
2080 extern void __init vfs_caches_init_early(void);
2081 extern void __init vfs_caches_init(unsigned long);
2082 
2083 extern struct kmem_cache *names_cachep;
2084 
2085 extern void final_putname(struct filename *name);
2086 
2087 #define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2088 #define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2089 #ifndef CONFIG_AUDITSYSCALL
2090 #define putname(name)           final_putname(name)
2091 #else
2092 extern void putname(struct filename *name);
2093 #endif
2094 
2095 #ifdef CONFIG_BLOCK
2096 extern int register_blkdev(unsigned int, const char *);
2097 extern void unregister_blkdev(unsigned int, const char *);
2098 extern struct block_device *bdget(dev_t);
2099 extern struct block_device *bdgrab(struct block_device *bdev);
2100 extern void bd_set_size(struct block_device *, loff_t size);
2101 extern void bd_forget(struct inode *inode);
2102 extern void bdput(struct block_device *);
2103 extern void invalidate_bdev(struct block_device *);
2104 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2105 extern int sync_blockdev(struct block_device *bdev);
2106 extern void kill_bdev(struct block_device *);
2107 extern struct super_block *freeze_bdev(struct block_device *);
2108 extern void emergency_thaw_all(void);
2109 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2110 extern int fsync_bdev(struct block_device *);
2111 extern int sb_is_blkdev_sb(struct super_block *sb);
2112 #else
2113 static inline void bd_forget(struct inode *inode) {}
2114 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2115 static inline void kill_bdev(struct block_device *bdev) {}
2116 static inline void invalidate_bdev(struct block_device *bdev) {}
2117 
2118 static inline struct super_block *freeze_bdev(struct block_device *sb)
2119 {
2120         return NULL;
2121 }
2122 
2123 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2124 {
2125         return 0;
2126 }
2127 
2128 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2129 {
2130 }
2131 
2132 static inline int sb_is_blkdev_sb(struct super_block *sb)
2133 {
2134         return 0;
2135 }
2136 #endif
2137 extern int sync_filesystem(struct super_block *);
2138 extern const struct file_operations def_blk_fops;
2139 extern const struct file_operations def_chr_fops;
2140 extern const struct file_operations bad_sock_fops;
2141 #ifdef CONFIG_BLOCK
2142 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2143 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2144 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2145 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2146 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2147                                                void *holder);
2148 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2149                                               void *holder);
2150 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2151 #ifdef CONFIG_SYSFS
2152 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2153 extern void bd_unlink_disk_holder(struct block_device *bdev,
2154                                   struct gendisk *disk);
2155 #else
2156 static inline int bd_link_disk_holder(struct block_device *bdev,
2157                                       struct gendisk *disk)
2158 {
2159         return 0;
2160 }
2161 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2162                                          struct gendisk *disk)
2163 {
2164 }
2165 #endif
2166 #endif
2167 
2168 /* fs/char_dev.c */
2169 #define CHRDEV_MAJOR_HASH_SIZE  255
2170 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2171 extern int register_chrdev_region(dev_t, unsigned, const char *);
2172 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2173                              unsigned int count, const char *name,
2174                              const struct file_operations *fops);
2175 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2176                                 unsigned int count, const char *name);
2177 extern void unregister_chrdev_region(dev_t, unsigned);
2178 extern void chrdev_show(struct seq_file *,off_t);
2179 
2180 static inline int register_chrdev(unsigned int major, const char *name,
2181                                   const struct file_operations *fops)
2182 {
2183         return __register_chrdev(major, 0, 256, name, fops);
2184 }
2185 
2186 static inline void unregister_chrdev(unsigned int major, const char *name)
2187 {
2188         __unregister_chrdev(major, 0, 256, name);
2189 }
2190 
2191 /* fs/block_dev.c */
2192 #define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2193 #define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2194 
2195 #ifdef CONFIG_BLOCK
2196 #define BLKDEV_MAJOR_HASH_SIZE  255
2197 extern const char *__bdevname(dev_t, char *buffer);
2198 extern const char *bdevname(struct block_device *bdev, char *buffer);
2199 extern struct block_device *lookup_bdev(const char *);
2200 extern void blkdev_show(struct seq_file *,off_t);
2201 
2202 #else
2203 #define BLKDEV_MAJOR_HASH_SIZE  0
2204 #endif
2205 
2206 extern void init_special_inode(struct inode *, umode_t, dev_t);
2207 
2208 /* Invalid inode operations -- fs/bad_inode.c */
2209 extern void make_bad_inode(struct inode *);
2210 extern int is_bad_inode(struct inode *);
2211 
2212 #ifdef CONFIG_BLOCK
2213 /*
2214  * return READ, READA, or WRITE
2215  */
2216 #define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2217 
2218 /*
2219  * return data direction, READ or WRITE
2220  */
2221 #define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2222 
2223 extern void check_disk_size_change(struct gendisk *disk,
2224                                    struct block_device *bdev);
2225 extern int revalidate_disk(struct gendisk *);
2226 extern int check_disk_change(struct block_device *);
2227 extern int __invalidate_device(struct block_device *, bool);
2228 extern int invalidate_partition(struct gendisk *, int);
2229 #endif
2230 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2231                                         pgoff_t start, pgoff_t end);
2232 
2233 static inline void invalidate_remote_inode(struct inode *inode)
2234 {
2235         if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2236             S_ISLNK(inode->i_mode))
2237                 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2238 }
2239 extern int invalidate_inode_pages2(struct address_space *mapping);
2240 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2241                                          pgoff_t start, pgoff_t end);
2242 extern int write_inode_now(struct inode *, int);
2243 extern int filemap_fdatawrite(struct address_space *);
2244 extern int filemap_flush(struct address_space *);
2245 extern int filemap_fdatawait(struct address_space *);
2246 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2247                                    loff_t lend);
2248 extern int filemap_write_and_wait(struct address_space *mapping);
2249 extern int filemap_write_and_wait_range(struct address_space *mapping,
2250                                         loff_t lstart, loff_t lend);
2251 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2252                                 loff_t start, loff_t end, int sync_mode);
2253 extern int filemap_fdatawrite_range(struct address_space *mapping,
2254                                 loff_t start, loff_t end);
2255 
2256 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2257                            int datasync);
2258 extern int vfs_fsync(struct file *file, int datasync);
2259 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2260 {
2261         if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2262                 return 0;
2263         return vfs_fsync_range(file, pos, pos + count - 1,
2264                                (file->f_flags & __O_SYNC) ? 0 : 1);
2265 }
2266 extern void emergency_sync(void);
2267 extern void emergency_remount(void);
2268 #ifdef CONFIG_BLOCK
2269 extern sector_t bmap(struct inode *, sector_t);
2270 #endif
2271 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2272 extern int inode_permission(struct inode *, int);
2273 extern int generic_permission(struct inode *, int);
2274 
2275 static inline bool execute_ok(struct inode *inode)
2276 {
2277         return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2278 }
2279 
2280 static inline void file_start_write(struct file *file)
2281 {
2282         if (!S_ISREG(file_inode(file)->i_mode))
2283                 return;
2284         __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2285 }
2286 
2287 static inline bool file_start_write_trylock(struct file *file)
2288 {
2289         if (!S_ISREG(file_inode(file)->i_mode))
2290                 return true;
2291         return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2292 }
2293 
2294 static inline void file_end_write(struct file *file)
2295 {
2296         if (!S_ISREG(file_inode(file)->i_mode))
2297                 return;
2298         __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2299 }
2300 
2301 /*
2302  * get_write_access() gets write permission for a file.
2303  * put_write_access() releases this write permission.
2304  * This is used for regular files.
2305  * We cannot support write (and maybe mmap read-write shared) accesses and
2306  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2307  * can have the following values:
2308  * 0: no writers, no VM_DENYWRITE mappings
2309  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2310  * > 0: (i_writecount) users are writing to the file.
2311  *
2312  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2313  * except for the cases where we don't hold i_writecount yet. Then we need to
2314  * use {get,deny}_write_access() - these functions check the sign and refuse
2315  * to do the change if sign is wrong.
2316  */
2317 static inline int get_write_access(struct inode *inode)
2318 {
2319         return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2320 }
2321 static inline int deny_write_access(struct file *file)
2322 {
2323         struct inode *inode = file_inode(file);
2324         return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2325 }
2326 static inline void put_write_access(struct inode * inode)
2327 {
2328         atomic_dec(&inode->i_writecount);
2329 }
2330 static inline void allow_write_access(struct file *file)
2331 {
2332         if (file)
2333                 atomic_inc(&file_inode(file)->i_writecount);
2334 }
2335 static inline bool inode_is_open_for_write(const struct inode *inode)
2336 {
2337         return atomic_read(&inode->i_writecount) > 0;
2338 }
2339 
2340 #ifdef CONFIG_IMA
2341 static inline void i_readcount_dec(struct inode *inode)
2342 {
2343         BUG_ON(!atomic_read(&inode->i_readcount));
2344         atomic_dec(&inode->i_readcount);
2345 }
2346 static inline void i_readcount_inc(struct inode *inode)
2347 {
2348         atomic_inc(&inode->i_readcount);
2349 }
2350 #else
2351 static inline void i_readcount_dec(struct inode *inode)
2352 {
2353         return;
2354 }
2355 static inline void i_readcount_inc(struct inode *inode)
2356 {
2357         return;
2358 }
2359 #endif
2360 extern int do_pipe_flags(int *, int);
2361 
2362 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2363 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2364 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2365 extern struct file * open_exec(const char *);
2366  
2367 /* fs/dcache.c -- generic fs support functions */
2368 extern int is_subdir(struct dentry *, struct dentry *);
2369 extern int path_is_under(struct path *, struct path *);
2370 
2371 #include <linux/err.h>
2372 
2373 /* needed for stackable file system support */
2374 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2375 
2376 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2377 
2378 extern int inode_init_always(struct super_block *, struct inode *);
2379 extern void inode_init_once(struct inode *);
2380 extern void address_space_init_once(struct address_space *mapping);
2381 extern struct inode * igrab(struct inode *);
2382 extern ino_t iunique(struct super_block *, ino_t);
2383 extern int inode_needs_sync(struct inode *inode);
2384 extern int generic_delete_inode(struct inode *inode);
2385 static inline int generic_drop_inode(struct inode *inode)
2386 {
2387         return !inode->i_nlink || inode_unhashed(inode);
2388 }
2389 
2390 extern struct inode *ilookup5_nowait(struct super_block *sb,
2391                 unsigned long hashval, int (*test)(struct inode *, void *),
2392                 void *data);
2393 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2394                 int (*test)(struct inode *, void *), void *data);
2395 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2396 
2397 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2398 extern struct inode * iget_locked(struct super_block *, unsigned long);
2399 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2400 extern int insert_inode_locked(struct inode *);
2401 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2402 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2403 #else
2404 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2405 #endif
2406 extern void unlock_new_inode(struct inode *);
2407 extern unsigned int get_next_ino(void);
2408 
2409 extern void __iget(struct inode * inode);
2410 extern void iget_failed(struct inode *);
2411 extern void clear_inode(struct inode *);
2412 extern void __destroy_inode(struct inode *);
2413 extern struct inode *new_inode_pseudo(struct super_block *sb);
2414 extern struct inode *new_inode(struct super_block *sb);
2415 extern void free_inode_nonrcu(struct inode *inode);
2416 extern int should_remove_suid(struct dentry *);
2417 extern int file_remove_suid(struct file *);
2418 
2419 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2420 static inline void insert_inode_hash(struct inode *inode)
2421 {
2422         __insert_inode_hash(inode, inode->i_ino);
2423 }
2424 
2425 extern void __remove_inode_hash(struct inode *);
2426 static inline void remove_inode_hash(struct inode *inode)
2427 {
2428         if (!inode_unhashed(inode))
2429                 __remove_inode_hash(inode);
2430 }
2431 
2432 extern void inode_sb_list_add(struct inode *inode);
2433 
2434 #ifdef CONFIG_BLOCK
2435 extern void submit_bio(int, struct bio *);
2436 extern int bdev_read_only(struct block_device *);
2437 #endif
2438 extern int set_blocksize(struct block_device *, int);
2439 extern int sb_set_blocksize(struct super_block *, int);
2440 extern int sb_min_blocksize(struct super_block *, int);
2441 
2442 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2443 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2444 extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2445                 unsigned long size, pgoff_t pgoff);
2446 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2447 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2448 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2449 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2450 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2451 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2452 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2453 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2454 extern ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2455 extern ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2456 
2457 /* fs/block_dev.c */
2458 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2459 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2460                         int datasync);
2461 extern void block_sync_page(struct page *page);
2462 
2463 /* fs/splice.c */
2464 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2465                 struct pipe_inode_info *, size_t, unsigned int);
2466 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2467                 struct pipe_inode_info *, size_t, unsigned int);
2468 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2469                 struct file *, loff_t *, size_t, unsigned int);
2470 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2471                 struct file *out, loff_t *, size_t len, unsigned int flags);
2472 
2473 extern void
2474 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2475 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2476 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2477 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2478 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2479 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2480                 int whence, loff_t maxsize, loff_t eof);
2481 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2482                 int whence, loff_t size);
2483 extern int generic_file_open(struct inode * inode, struct file * filp);
2484 extern int nonseekable_open(struct inode * inode, struct file * filp);
2485 
2486 #ifdef CONFIG_FS_XIP
2487 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2488                              loff_t *ppos);
2489 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2490 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2491                               size_t len, loff_t *ppos);
2492 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2493 #else
2494 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2495 {
2496         return 0;
2497 }
2498 #endif
2499 
2500 #ifdef CONFIG_BLOCK
2501 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2502                             loff_t file_offset);
2503 
2504 enum {
2505         /* need locking between buffered and direct access */
2506         DIO_LOCKING     = 0x01,
2507 
2508         /* filesystem does not support filling holes */
2509         DIO_SKIP_HOLES  = 0x02,
2510 
2511         /* filesystem can handle aio writes beyond i_size */
2512         DIO_ASYNC_EXTEND = 0x04,
2513 };
2514 
2515 void dio_end_io(struct bio *bio, int error);
2516 
2517 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2518         struct block_device *bdev, struct iov_iter *iter, loff_t offset,
2519         get_block_t get_block, dio_iodone_t end_io,
2520         dio_submit_t submit_io, int flags);
2521 
2522 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2523                 struct inode *inode, struct iov_iter *iter, loff_t offset,
2524                 get_block_t get_block)
2525 {
2526         return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iter,
2527                                     offset, get_block, NULL, NULL,
2528                                     DIO_LOCKING | DIO_SKIP_HOLES);
2529 }
2530 #endif
2531 
2532 void inode_dio_wait(struct inode *inode);
2533 void inode_dio_done(struct inode *inode);
2534 
2535 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2536                             unsigned int mask);
2537 
2538 extern const struct file_operations generic_ro_fops;
2539 
2540 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2541 
2542 extern int readlink_copy(char __user *, int, const char *);
2543 extern int page_readlink(struct dentry *, char __user *, int);
2544 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2545 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2546 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2547                 int nofs);
2548 extern int page_symlink(struct inode *inode, const char *symname, int len);
2549 extern const struct inode_operations page_symlink_inode_operations;
2550 extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
2551 extern int generic_readlink(struct dentry *, char __user *, int);
2552 extern void generic_fillattr(struct inode *, struct kstat *);
2553 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2554 extern int vfs_getattr(struct path *, struct kstat *);
2555 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2556 void inode_add_bytes(struct inode *inode, loff_t bytes);
2557 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2558 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2559 loff_t inode_get_bytes(struct inode *inode);
2560 void inode_set_bytes(struct inode *inode, loff_t bytes);
2561 
2562 extern int vfs_readdir(struct file *, filldir_t, void *);
2563 extern int iterate_dir(struct file *, struct dir_context *);
2564 
2565 extern int vfs_stat(const char __user *, struct kstat *);
2566 extern int vfs_lstat(const char __user *, struct kstat *);
2567 extern int vfs_fstat(unsigned int, struct kstat *);
2568 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2569 
2570 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2571                     unsigned long arg);
2572 extern int __generic_block_fiemap(struct inode *inode,
2573                                   struct fiemap_extent_info *fieinfo,
2574                                   loff_t start, loff_t len,
2575                                   get_block_t *get_block);
2576 extern int generic_block_fiemap(struct inode *inode,
2577                                 struct fiemap_extent_info *fieinfo, u64 start,
2578                                 u64 len, get_block_t *get_block);
2579 
2580 extern void get_filesystem(struct file_system_type *fs);
2581 extern void put_filesystem(struct file_system_type *fs);
2582 extern struct file_system_type *get_fs_type(const char *name);
2583 extern struct super_block *get_super(struct block_device *);
2584 extern struct super_block *get_super_thawed(struct block_device *);
2585 extern struct super_block *get_active_super(struct block_device *bdev);
2586 extern void drop_super(struct super_block *sb);
2587 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2588 extern void iterate_supers_type(struct file_system_type *,
2589                                 void (*)(struct super_block *, void *), void *);
2590 
2591 extern int dcache_dir_open(struct inode *, struct file *);
2592 extern int dcache_dir_close(struct inode *, struct file *);
2593 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2594 extern int dcache_readdir(struct file *, struct dir_context *);
2595 extern int simple_setattr(struct dentry *, struct iattr *);
2596 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2597 extern int simple_statfs(struct dentry *, struct kstatfs *);
2598 extern int simple_open(struct inode *inode, struct file *file);
2599 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2600 extern int simple_unlink(struct inode *, struct dentry *);
2601 extern int simple_rmdir(struct inode *, struct dentry *);
2602 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2603 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2604 extern int simple_empty(struct dentry *);
2605 extern int simple_readpage(struct file *file, struct page *page);
2606 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2607                         loff_t pos, unsigned len, unsigned flags,
2608                         struct page **pagep, void **fsdata);
2609 extern int simple_write_end(struct file *file, struct address_space *mapping,
2610                         loff_t pos, unsigned len, unsigned copied,
2611                         struct page *page, void *fsdata);
2612 extern int always_delete_dentry(const struct dentry *);
2613 extern struct inode *alloc_anon_inode(struct super_block *);
2614 extern const struct dentry_operations simple_dentry_operations;
2615 
2616 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2617 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2618 extern const struct file_operations simple_dir_operations;
2619 extern const struct inode_operations simple_dir_inode_operations;
2620 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2621 struct dentry *d_alloc_name(struct dentry *, const char *);
2622 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2623 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2624 extern void simple_release_fs(struct vfsmount **mount, int *count);
2625 
2626 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2627                         loff_t *ppos, const void *from, size_t available);
2628 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2629                 const void __user *from, size_t count);
2630 
2631 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2632 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2633 
2634 extern int generic_check_addressable(unsigned, u64);
2635 
2636 #ifdef CONFIG_MIGRATION
2637 extern int buffer_migrate_page(struct address_space *,
2638                                 struct page *, struct page *,
2639                                 enum migrate_mode);
2640 #else
2641 #define buffer_migrate_page NULL
2642 #endif
2643 
2644 extern int inode_change_ok(const struct inode *, struct iattr *);
2645 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2646 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2647 
2648 extern int file_update_time(struct file *file);
2649 
2650 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2651 extern void save_mount_options(struct super_block *sb, char *options);
2652 extern void replace_mount_options(struct super_block *sb, char *options);
2653 
2654 static inline ino_t parent_ino(struct dentry *dentry)
2655 {
2656         ino_t res;
2657 
2658         /*
2659          * Don't strictly need d_lock here? If the parent ino could change
2660          * then surely we'd have a deeper race in the caller?
2661          */
2662         spin_lock(&dentry->d_lock);
2663         res = dentry->d_parent->d_inode->i_ino;
2664         spin_unlock(&dentry->d_lock);
2665         return res;
2666 }
2667 
2668 /* Transaction based IO helpers */
2669 
2670 /*
2671  * An argresp is stored in an allocated page and holds the
2672  * size of the argument or response, along with its content
2673  */
2674 struct simple_transaction_argresp {
2675         ssize_t size;
2676         char data[0];
2677 };
2678 
2679 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2680 
2681 char *simple_transaction_get(struct file *file, const char __user *buf,
2682                                 size_t size);
2683 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2684                                 size_t size, loff_t *pos);
2685 int simple_transaction_release(struct inode *inode, struct file *file);
2686 
2687 void simple_transaction_set(struct file *file, size_t n);
2688 
2689 /*
2690  * simple attribute files
2691  *
2692  * These attributes behave similar to those in sysfs:
2693  *
2694  * Writing to an attribute immediately sets a value, an open file can be
2695  * written to multiple times.
2696  *
2697  * Reading from an attribute creates a buffer from the value that might get
2698  * read with multiple read calls. When the attribute has been read
2699  * completely, no further read calls are possible until the file is opened
2700  * again.
2701  *
2702  * All attributes contain a text representation of a numeric value
2703  * that are accessed with the get() and set() functions.
2704  */
2705 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2706 static int __fops ## _open(struct inode *inode, struct file *file)      \
2707 {                                                                       \
2708         __simple_attr_check_format(__fmt, 0ull);                        \
2709         return simple_attr_open(inode, file, __get, __set, __fmt);      \
2710 }                                                                       \
2711 static const struct file_operations __fops = {                          \
2712         .owner   = THIS_MODULE,                                         \
2713         .open    = __fops ## _open,                                     \
2714         .release = simple_attr_release,                                 \
2715         .read    = simple_attr_read,                                    \
2716         .write   = simple_attr_write,                                   \
2717         .llseek  = generic_file_llseek,                                 \
2718 }
2719 
2720 static inline __printf(1, 2)
2721 void __simple_attr_check_format(const char *fmt, ...)
2722 {
2723         /* don't do anything, just let the compiler check the arguments; */
2724 }
2725 
2726 int simple_attr_open(struct inode *inode, struct file *file,
2727                      int (*get)(void *, u64 *), int (*set)(void *, u64),
2728                      const char *fmt);
2729 int simple_attr_release(struct inode *inode, struct file *file);
2730 ssize_t simple_attr_read(struct file *file, char __user *buf,
2731                          size_t len, loff_t *ppos);
2732 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2733                           size_t len, loff_t *ppos);
2734 
2735 struct ctl_table;
2736 int proc_nr_files(struct ctl_table *table, int write,
2737                   void __user *buffer, size_t *lenp, loff_t *ppos);
2738 int proc_nr_dentry(struct ctl_table *table, int write,
2739                   void __user *buffer, size_t *lenp, loff_t *ppos);
2740 int proc_nr_inodes(struct ctl_table *table, int write,
2741                    void __user *buffer, size_t *lenp, loff_t *ppos);
2742 int __init get_filesystem_list(char *buf);
2743 
2744 #define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2745 #define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2746 
2747 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2748 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2749                                             (flag & __FMODE_NONOTIFY)))
2750 
2751 static inline int is_sxid(umode_t mode)
2752 {
2753         return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2754 }
2755 
2756 static inline void inode_has_no_xattr(struct inode *inode)
2757 {
2758         if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2759                 inode->i_flags |= S_NOSEC;
2760 }
2761 
2762 static inline bool dir_emit(struct dir_context *ctx,
2763                             const char *name, int namelen,
2764                             u64 ino, unsigned type)
2765 {
2766         return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2767 }
2768 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2769 {
2770         return ctx->actor(ctx, ".", 1, ctx->pos,
2771                           file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2772 }
2773 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2774 {
2775         return ctx->actor(ctx, "..", 2, ctx->pos,
2776                           parent_ino(file->f_path.dentry), DT_DIR) == 0;
2777 }
2778 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2779 {
2780         if (ctx->pos == 0) {
2781                 if (!dir_emit_dot(file, ctx))
2782                         return false;
2783                 ctx->pos = 1;
2784         }
2785         if (ctx->pos == 1) {
2786                 if (!dir_emit_dotdot(file, ctx))
2787                         return false;
2788                 ctx->pos = 2;
2789         }
2790         return true;
2791 }
2792 static inline bool dir_relax(struct inode *inode)
2793 {
2794         mutex_unlock(&inode->i_mutex);
2795         mutex_lock(&inode->i_mutex);
2796         return !IS_DEADDIR(inode);
2797 }
2798 
2799 #endif /* _LINUX_FS_H */
2800 

This page was automatically generated by LXR 0.3.1 (source).  •  Linux is a registered trademark of Linus Torvalds  •  Contact us