Version:  2.0.40 2.2.26 2.4.37 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 4.0 4.1 4.2 4.3 4.4

Linux/include/linux/fs.h

  1 #ifndef _LINUX_FS_H
  2 #define _LINUX_FS_H
  3 
  4 #include <linux/linkage.h>
  5 #include <linux/wait.h>
  6 #include <linux/kdev_t.h>
  7 #include <linux/dcache.h>
  8 #include <linux/path.h>
  9 #include <linux/stat.h>
 10 #include <linux/cache.h>
 11 #include <linux/list.h>
 12 #include <linux/list_lru.h>
 13 #include <linux/llist.h>
 14 #include <linux/radix-tree.h>
 15 #include <linux/rbtree.h>
 16 #include <linux/init.h>
 17 #include <linux/pid.h>
 18 #include <linux/bug.h>
 19 #include <linux/mutex.h>
 20 #include <linux/rwsem.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 #include <linux/workqueue.h>
 33 #include <linux/percpu-rwsem.h>
 34 
 35 #include <asm/byteorder.h>
 36 #include <uapi/linux/fs.h>
 37 
 38 struct backing_dev_info;
 39 struct bdi_writeback;
 40 struct export_operations;
 41 struct hd_geometry;
 42 struct iovec;
 43 struct kiocb;
 44 struct kobject;
 45 struct pipe_inode_info;
 46 struct poll_table_struct;
 47 struct kstatfs;
 48 struct vm_area_struct;
 49 struct vfsmount;
 50 struct cred;
 51 struct swap_info_struct;
 52 struct seq_file;
 53 struct workqueue_struct;
 54 struct iov_iter;
 55 
 56 extern void __init inode_init(void);
 57 extern void __init inode_init_early(void);
 58 extern void __init files_init(void);
 59 extern void __init files_maxfiles_init(void);
 60 
 61 extern struct files_stat_struct files_stat;
 62 extern unsigned long get_max_files(void);
 63 extern int sysctl_nr_open;
 64 extern struct inodes_stat_t inodes_stat;
 65 extern int leases_enable, lease_break_time;
 66 extern int sysctl_protected_symlinks;
 67 extern int sysctl_protected_hardlinks;
 68 
 69 struct buffer_head;
 70 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 71                         struct buffer_head *bh_result, int create);
 72 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 73                         ssize_t bytes, void *private);
 74 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
 75 
 76 #define MAY_EXEC                0x00000001
 77 #define MAY_WRITE               0x00000002
 78 #define MAY_READ                0x00000004
 79 #define MAY_APPEND              0x00000008
 80 #define MAY_ACCESS              0x00000010
 81 #define MAY_OPEN                0x00000020
 82 #define MAY_CHDIR               0x00000040
 83 /* called from RCU mode, don't block */
 84 #define MAY_NOT_BLOCK           0x00000080
 85 
 86 /*
 87  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
 88  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
 89  */
 90 
 91 /* file is open for reading */
 92 #define FMODE_READ              ((__force fmode_t)0x1)
 93 /* file is open for writing */
 94 #define FMODE_WRITE             ((__force fmode_t)0x2)
 95 /* file is seekable */
 96 #define FMODE_LSEEK             ((__force fmode_t)0x4)
 97 /* file can be accessed using pread */
 98 #define FMODE_PREAD             ((__force fmode_t)0x8)
 99 /* file can be accessed using pwrite */
100 #define FMODE_PWRITE            ((__force fmode_t)0x10)
101 /* File is opened for execution with sys_execve / sys_uselib */
102 #define FMODE_EXEC              ((__force fmode_t)0x20)
103 /* File is opened with O_NDELAY (only set for block devices) */
104 #define FMODE_NDELAY            ((__force fmode_t)0x40)
105 /* File is opened with O_EXCL (only set for block devices) */
106 #define FMODE_EXCL              ((__force fmode_t)0x80)
107 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
108    (specialy hack for floppy.c) */
109 #define FMODE_WRITE_IOCTL       ((__force fmode_t)0x100)
110 /* 32bit hashes as llseek() offset (for directories) */
111 #define FMODE_32BITHASH         ((__force fmode_t)0x200)
112 /* 64bit hashes as llseek() offset (for directories) */
113 #define FMODE_64BITHASH         ((__force fmode_t)0x400)
114 
115 /*
116  * Don't update ctime and mtime.
117  *
118  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
119  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
120  */
121 #define FMODE_NOCMTIME          ((__force fmode_t)0x800)
122 
123 /* Expect random access pattern */
124 #define FMODE_RANDOM            ((__force fmode_t)0x1000)
125 
126 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
127 #define FMODE_UNSIGNED_OFFSET   ((__force fmode_t)0x2000)
128 
129 /* File is opened with O_PATH; almost nothing can be done with it */
130 #define FMODE_PATH              ((__force fmode_t)0x4000)
131 
132 /* File needs atomic accesses to f_pos */
133 #define FMODE_ATOMIC_POS        ((__force fmode_t)0x8000)
134 /* Write access to underlying fs */
135 #define FMODE_WRITER            ((__force fmode_t)0x10000)
136 /* Has read method(s) */
137 #define FMODE_CAN_READ          ((__force fmode_t)0x20000)
138 /* Has write method(s) */
139 #define FMODE_CAN_WRITE         ((__force fmode_t)0x40000)
140 
141 /* File was opened by fanotify and shouldn't generate fanotify events */
142 #define FMODE_NONOTIFY          ((__force fmode_t)0x4000000)
143 
144 /*
145  * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
146  * that indicates that they should check the contents of the iovec are
147  * valid, but not check the memory that the iovec elements
148  * points too.
149  */
150 #define CHECK_IOVEC_ONLY -1
151 
152 /*
153  * The below are the various read and write types that we support. Some of
154  * them include behavioral modifiers that send information down to the
155  * block layer and IO scheduler. Terminology:
156  *
157  *      The block layer uses device plugging to defer IO a little bit, in
158  *      the hope that we will see more IO very shortly. This increases
159  *      coalescing of adjacent IO and thus reduces the number of IOs we
160  *      have to send to the device. It also allows for better queuing,
161  *      if the IO isn't mergeable. If the caller is going to be waiting
162  *      for the IO, then he must ensure that the device is unplugged so
163  *      that the IO is dispatched to the driver.
164  *
165  *      All IO is handled async in Linux. This is fine for background
166  *      writes, but for reads or writes that someone waits for completion
167  *      on, we want to notify the block layer and IO scheduler so that they
168  *      know about it. That allows them to make better scheduling
169  *      decisions. So when the below references 'sync' and 'async', it
170  *      is referencing this priority hint.
171  *
172  * With that in mind, the available types are:
173  *
174  * READ                 A normal read operation. Device will be plugged.
175  * READ_SYNC            A synchronous read. Device is not plugged, caller can
176  *                      immediately wait on this read without caring about
177  *                      unplugging.
178  * READA                Used for read-ahead operations. Lower priority, and the
179  *                      block layer could (in theory) choose to ignore this
180  *                      request if it runs into resource problems.
181  * WRITE                A normal async write. Device will be plugged.
182  * WRITE_SYNC           Synchronous write. Identical to WRITE, but passes down
183  *                      the hint that someone will be waiting on this IO
184  *                      shortly. The write equivalent of READ_SYNC.
185  * WRITE_ODIRECT        Special case write for O_DIRECT only.
186  * WRITE_FLUSH          Like WRITE_SYNC but with preceding cache flush.
187  * WRITE_FUA            Like WRITE_SYNC but data is guaranteed to be on
188  *                      non-volatile media on completion.
189  * WRITE_FLUSH_FUA      Combination of WRITE_FLUSH and FUA. The IO is preceded
190  *                      by a cache flush and data is guaranteed to be on
191  *                      non-volatile media on completion.
192  *
193  */
194 #define RW_MASK                 REQ_WRITE
195 #define RWA_MASK                REQ_RAHEAD
196 
197 #define READ                    0
198 #define WRITE                   RW_MASK
199 #define READA                   RWA_MASK
200 
201 #define READ_SYNC               (READ | REQ_SYNC)
202 #define WRITE_SYNC              (WRITE | REQ_SYNC | REQ_NOIDLE)
203 #define WRITE_ODIRECT           (WRITE | REQ_SYNC)
204 #define WRITE_FLUSH             (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
205 #define WRITE_FUA               (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
206 #define WRITE_FLUSH_FUA         (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
207 
208 /*
209  * Attribute flags.  These should be or-ed together to figure out what
210  * has been changed!
211  */
212 #define ATTR_MODE       (1 << 0)
213 #define ATTR_UID        (1 << 1)
214 #define ATTR_GID        (1 << 2)
215 #define ATTR_SIZE       (1 << 3)
216 #define ATTR_ATIME      (1 << 4)
217 #define ATTR_MTIME      (1 << 5)
218 #define ATTR_CTIME      (1 << 6)
219 #define ATTR_ATIME_SET  (1 << 7)
220 #define ATTR_MTIME_SET  (1 << 8)
221 #define ATTR_FORCE      (1 << 9) /* Not a change, but a change it */
222 #define ATTR_ATTR_FLAG  (1 << 10)
223 #define ATTR_KILL_SUID  (1 << 11)
224 #define ATTR_KILL_SGID  (1 << 12)
225 #define ATTR_FILE       (1 << 13)
226 #define ATTR_KILL_PRIV  (1 << 14)
227 #define ATTR_OPEN       (1 << 15) /* Truncating from open(O_TRUNC) */
228 #define ATTR_TIMES_SET  (1 << 16)
229 
230 /*
231  * Whiteout is represented by a char device.  The following constants define the
232  * mode and device number to use.
233  */
234 #define WHITEOUT_MODE 0
235 #define WHITEOUT_DEV 0
236 
237 /*
238  * This is the Inode Attributes structure, used for notify_change().  It
239  * uses the above definitions as flags, to know which values have changed.
240  * Also, in this manner, a Filesystem can look at only the values it cares
241  * about.  Basically, these are the attributes that the VFS layer can
242  * request to change from the FS layer.
243  *
244  * Derek Atkins <warlord@MIT.EDU> 94-10-20
245  */
246 struct iattr {
247         unsigned int    ia_valid;
248         umode_t         ia_mode;
249         kuid_t          ia_uid;
250         kgid_t          ia_gid;
251         loff_t          ia_size;
252         struct timespec ia_atime;
253         struct timespec ia_mtime;
254         struct timespec ia_ctime;
255 
256         /*
257          * Not an attribute, but an auxiliary info for filesystems wanting to
258          * implement an ftruncate() like method.  NOTE: filesystem should
259          * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
260          */
261         struct file     *ia_file;
262 };
263 
264 /*
265  * Includes for diskquotas.
266  */
267 #include <linux/quota.h>
268 
269 /*
270  * Maximum number of layers of fs stack.  Needs to be limited to
271  * prevent kernel stack overflow
272  */
273 #define FILESYSTEM_MAX_STACK_DEPTH 2
274 
275 /** 
276  * enum positive_aop_returns - aop return codes with specific semantics
277  *
278  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
279  *                          completed, that the page is still locked, and
280  *                          should be considered active.  The VM uses this hint
281  *                          to return the page to the active list -- it won't
282  *                          be a candidate for writeback again in the near
283  *                          future.  Other callers must be careful to unlock
284  *                          the page if they get this return.  Returned by
285  *                          writepage(); 
286  *
287  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
288  *                      unlocked it and the page might have been truncated.
289  *                      The caller should back up to acquiring a new page and
290  *                      trying again.  The aop will be taking reasonable
291  *                      precautions not to livelock.  If the caller held a page
292  *                      reference, it should drop it before retrying.  Returned
293  *                      by readpage().
294  *
295  * address_space_operation functions return these large constants to indicate
296  * special semantics to the caller.  These are much larger than the bytes in a
297  * page to allow for functions that return the number of bytes operated on in a
298  * given page.
299  */
300 
301 enum positive_aop_returns {
302         AOP_WRITEPAGE_ACTIVATE  = 0x80000,
303         AOP_TRUNCATED_PAGE      = 0x80001,
304 };
305 
306 #define AOP_FLAG_UNINTERRUPTIBLE        0x0001 /* will not do a short write */
307 #define AOP_FLAG_CONT_EXPAND            0x0002 /* called from cont_expand */
308 #define AOP_FLAG_NOFS                   0x0004 /* used by filesystem to direct
309                                                 * helper code (eg buffer layer)
310                                                 * to clear GFP_FS from alloc */
311 
312 /*
313  * oh the beauties of C type declarations.
314  */
315 struct page;
316 struct address_space;
317 struct writeback_control;
318 
319 #define IOCB_EVENTFD            (1 << 0)
320 #define IOCB_APPEND             (1 << 1)
321 #define IOCB_DIRECT             (1 << 2)
322 
323 struct kiocb {
324         struct file             *ki_filp;
325         loff_t                  ki_pos;
326         void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
327         void                    *private;
328         int                     ki_flags;
329 };
330 
331 static inline bool is_sync_kiocb(struct kiocb *kiocb)
332 {
333         return kiocb->ki_complete == NULL;
334 }
335 
336 static inline int iocb_flags(struct file *file);
337 
338 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
339 {
340         *kiocb = (struct kiocb) {
341                 .ki_filp = filp,
342                 .ki_flags = iocb_flags(filp),
343         };
344 }
345 
346 /*
347  * "descriptor" for what we're up to with a read.
348  * This allows us to use the same read code yet
349  * have multiple different users of the data that
350  * we read from a file.
351  *
352  * The simplest case just copies the data to user
353  * mode.
354  */
355 typedef struct {
356         size_t written;
357         size_t count;
358         union {
359                 char __user *buf;
360                 void *data;
361         } arg;
362         int error;
363 } read_descriptor_t;
364 
365 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
366                 unsigned long, unsigned long);
367 
368 struct address_space_operations {
369         int (*writepage)(struct page *page, struct writeback_control *wbc);
370         int (*readpage)(struct file *, struct page *);
371 
372         /* Write back some dirty pages from this mapping. */
373         int (*writepages)(struct address_space *, struct writeback_control *);
374 
375         /* Set a page dirty.  Return true if this dirtied it */
376         int (*set_page_dirty)(struct page *page);
377 
378         int (*readpages)(struct file *filp, struct address_space *mapping,
379                         struct list_head *pages, unsigned nr_pages);
380 
381         int (*write_begin)(struct file *, struct address_space *mapping,
382                                 loff_t pos, unsigned len, unsigned flags,
383                                 struct page **pagep, void **fsdata);
384         int (*write_end)(struct file *, struct address_space *mapping,
385                                 loff_t pos, unsigned len, unsigned copied,
386                                 struct page *page, void *fsdata);
387 
388         /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
389         sector_t (*bmap)(struct address_space *, sector_t);
390         void (*invalidatepage) (struct page *, unsigned int, unsigned int);
391         int (*releasepage) (struct page *, gfp_t);
392         void (*freepage)(struct page *);
393         ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
394         /*
395          * migrate the contents of a page to the specified target. If
396          * migrate_mode is MIGRATE_ASYNC, it must not block.
397          */
398         int (*migratepage) (struct address_space *,
399                         struct page *, struct page *, enum migrate_mode);
400         int (*launder_page) (struct page *);
401         int (*is_partially_uptodate) (struct page *, unsigned long,
402                                         unsigned long);
403         void (*is_dirty_writeback) (struct page *, bool *, bool *);
404         int (*error_remove_page)(struct address_space *, struct page *);
405 
406         /* swapfile support */
407         int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
408                                 sector_t *span);
409         void (*swap_deactivate)(struct file *file);
410 };
411 
412 extern const struct address_space_operations empty_aops;
413 
414 /*
415  * pagecache_write_begin/pagecache_write_end must be used by general code
416  * to write into the pagecache.
417  */
418 int pagecache_write_begin(struct file *, struct address_space *mapping,
419                                 loff_t pos, unsigned len, unsigned flags,
420                                 struct page **pagep, void **fsdata);
421 
422 int pagecache_write_end(struct file *, struct address_space *mapping,
423                                 loff_t pos, unsigned len, unsigned copied,
424                                 struct page *page, void *fsdata);
425 
426 struct address_space {
427         struct inode            *host;          /* owner: inode, block_device */
428         struct radix_tree_root  page_tree;      /* radix tree of all pages */
429         spinlock_t              tree_lock;      /* and lock protecting it */
430         atomic_t                i_mmap_writable;/* count VM_SHARED mappings */
431         struct rb_root          i_mmap;         /* tree of private and shared mappings */
432         struct rw_semaphore     i_mmap_rwsem;   /* protect tree, count, list */
433         /* Protected by tree_lock together with the radix tree */
434         unsigned long           nrpages;        /* number of total pages */
435         unsigned long           nrshadows;      /* number of shadow entries */
436         pgoff_t                 writeback_index;/* writeback starts here */
437         const struct address_space_operations *a_ops;   /* methods */
438         unsigned long           flags;          /* error bits/gfp mask */
439         spinlock_t              private_lock;   /* for use by the address_space */
440         struct list_head        private_list;   /* ditto */
441         void                    *private_data;  /* ditto */
442 } __attribute__((aligned(sizeof(long))));
443         /*
444          * On most architectures that alignment is already the case; but
445          * must be enforced here for CRIS, to let the least significant bit
446          * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
447          */
448 struct request_queue;
449 
450 struct block_device {
451         dev_t                   bd_dev;  /* not a kdev_t - it's a search key */
452         int                     bd_openers;
453         struct inode *          bd_inode;       /* will die */
454         struct super_block *    bd_super;
455         struct mutex            bd_mutex;       /* open/close mutex */
456         struct list_head        bd_inodes;
457         void *                  bd_claiming;
458         void *                  bd_holder;
459         int                     bd_holders;
460         bool                    bd_write_holder;
461 #ifdef CONFIG_SYSFS
462         struct list_head        bd_holder_disks;
463 #endif
464         struct block_device *   bd_contains;
465         unsigned                bd_block_size;
466         struct hd_struct *      bd_part;
467         /* number of times partitions within this device have been opened. */
468         unsigned                bd_part_count;
469         int                     bd_invalidated;
470         struct gendisk *        bd_disk;
471         struct request_queue *  bd_queue;
472         struct list_head        bd_list;
473         /*
474          * Private data.  You must have bd_claim'ed the block_device
475          * to use this.  NOTE:  bd_claim allows an owner to claim
476          * the same device multiple times, the owner must take special
477          * care to not mess up bd_private for that case.
478          */
479         unsigned long           bd_private;
480 
481         /* The counter of freeze processes */
482         int                     bd_fsfreeze_count;
483         /* Mutex for freeze */
484         struct mutex            bd_fsfreeze_mutex;
485 };
486 
487 /*
488  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
489  * radix trees
490  */
491 #define PAGECACHE_TAG_DIRTY     0
492 #define PAGECACHE_TAG_WRITEBACK 1
493 #define PAGECACHE_TAG_TOWRITE   2
494 
495 int mapping_tagged(struct address_space *mapping, int tag);
496 
497 static inline void i_mmap_lock_write(struct address_space *mapping)
498 {
499         down_write(&mapping->i_mmap_rwsem);
500 }
501 
502 static inline void i_mmap_unlock_write(struct address_space *mapping)
503 {
504         up_write(&mapping->i_mmap_rwsem);
505 }
506 
507 static inline void i_mmap_lock_read(struct address_space *mapping)
508 {
509         down_read(&mapping->i_mmap_rwsem);
510 }
511 
512 static inline void i_mmap_unlock_read(struct address_space *mapping)
513 {
514         up_read(&mapping->i_mmap_rwsem);
515 }
516 
517 /*
518  * Might pages of this file be mapped into userspace?
519  */
520 static inline int mapping_mapped(struct address_space *mapping)
521 {
522         return  !RB_EMPTY_ROOT(&mapping->i_mmap);
523 }
524 
525 /*
526  * Might pages of this file have been modified in userspace?
527  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
528  * marks vma as VM_SHARED if it is shared, and the file was opened for
529  * writing i.e. vma may be mprotected writable even if now readonly.
530  *
531  * If i_mmap_writable is negative, no new writable mappings are allowed. You
532  * can only deny writable mappings, if none exists right now.
533  */
534 static inline int mapping_writably_mapped(struct address_space *mapping)
535 {
536         return atomic_read(&mapping->i_mmap_writable) > 0;
537 }
538 
539 static inline int mapping_map_writable(struct address_space *mapping)
540 {
541         return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
542                 0 : -EPERM;
543 }
544 
545 static inline void mapping_unmap_writable(struct address_space *mapping)
546 {
547         atomic_dec(&mapping->i_mmap_writable);
548 }
549 
550 static inline int mapping_deny_writable(struct address_space *mapping)
551 {
552         return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
553                 0 : -EBUSY;
554 }
555 
556 static inline void mapping_allow_writable(struct address_space *mapping)
557 {
558         atomic_inc(&mapping->i_mmap_writable);
559 }
560 
561 /*
562  * Use sequence counter to get consistent i_size on 32-bit processors.
563  */
564 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
565 #include <linux/seqlock.h>
566 #define __NEED_I_SIZE_ORDERED
567 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
568 #else
569 #define i_size_ordered_init(inode) do { } while (0)
570 #endif
571 
572 struct posix_acl;
573 #define ACL_NOT_CACHED ((void *)(-1))
574 
575 #define IOP_FASTPERM    0x0001
576 #define IOP_LOOKUP      0x0002
577 #define IOP_NOFOLLOW    0x0004
578 
579 /*
580  * Keep mostly read-only and often accessed (especially for
581  * the RCU path lookup and 'stat' data) fields at the beginning
582  * of the 'struct inode'
583  */
584 struct inode {
585         umode_t                 i_mode;
586         unsigned short          i_opflags;
587         kuid_t                  i_uid;
588         kgid_t                  i_gid;
589         unsigned int            i_flags;
590 
591 #ifdef CONFIG_FS_POSIX_ACL
592         struct posix_acl        *i_acl;
593         struct posix_acl        *i_default_acl;
594 #endif
595 
596         const struct inode_operations   *i_op;
597         struct super_block      *i_sb;
598         struct address_space    *i_mapping;
599 
600 #ifdef CONFIG_SECURITY
601         void                    *i_security;
602 #endif
603 
604         /* Stat data, not accessed from path walking */
605         unsigned long           i_ino;
606         /*
607          * Filesystems may only read i_nlink directly.  They shall use the
608          * following functions for modification:
609          *
610          *    (set|clear|inc|drop)_nlink
611          *    inode_(inc|dec)_link_count
612          */
613         union {
614                 const unsigned int i_nlink;
615                 unsigned int __i_nlink;
616         };
617         dev_t                   i_rdev;
618         loff_t                  i_size;
619         struct timespec         i_atime;
620         struct timespec         i_mtime;
621         struct timespec         i_ctime;
622         spinlock_t              i_lock; /* i_blocks, i_bytes, maybe i_size */
623         unsigned short          i_bytes;
624         unsigned int            i_blkbits;
625         blkcnt_t                i_blocks;
626 
627 #ifdef __NEED_I_SIZE_ORDERED
628         seqcount_t              i_size_seqcount;
629 #endif
630 
631         /* Misc */
632         unsigned long           i_state;
633         struct mutex            i_mutex;
634 
635         unsigned long           dirtied_when;   /* jiffies of first dirtying */
636         unsigned long           dirtied_time_when;
637 
638         struct hlist_node       i_hash;
639         struct list_head        i_io_list;      /* backing dev IO list */
640 #ifdef CONFIG_CGROUP_WRITEBACK
641         struct bdi_writeback    *i_wb;          /* the associated cgroup wb */
642 
643         /* foreign inode detection, see wbc_detach_inode() */
644         int                     i_wb_frn_winner;
645         u16                     i_wb_frn_avg_time;
646         u16                     i_wb_frn_history;
647 #endif
648         struct list_head        i_lru;          /* inode LRU list */
649         struct list_head        i_sb_list;
650         union {
651                 struct hlist_head       i_dentry;
652                 struct rcu_head         i_rcu;
653         };
654         u64                     i_version;
655         atomic_t                i_count;
656         atomic_t                i_dio_count;
657         atomic_t                i_writecount;
658 #ifdef CONFIG_IMA
659         atomic_t                i_readcount; /* struct files open RO */
660 #endif
661         const struct file_operations    *i_fop; /* former ->i_op->default_file_ops */
662         struct file_lock_context        *i_flctx;
663         struct address_space    i_data;
664         struct list_head        i_devices;
665         union {
666                 struct pipe_inode_info  *i_pipe;
667                 struct block_device     *i_bdev;
668                 struct cdev             *i_cdev;
669                 char                    *i_link;
670         };
671 
672         __u32                   i_generation;
673 
674 #ifdef CONFIG_FSNOTIFY
675         __u32                   i_fsnotify_mask; /* all events this inode cares about */
676         struct hlist_head       i_fsnotify_marks;
677 #endif
678 
679         void                    *i_private; /* fs or device private pointer */
680 };
681 
682 static inline int inode_unhashed(struct inode *inode)
683 {
684         return hlist_unhashed(&inode->i_hash);
685 }
686 
687 /*
688  * inode->i_mutex nesting subclasses for the lock validator:
689  *
690  * 0: the object of the current VFS operation
691  * 1: parent
692  * 2: child/target
693  * 3: xattr
694  * 4: second non-directory
695  * 5: second parent (when locking independent directories in rename)
696  *
697  * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
698  * non-directories at once.
699  *
700  * The locking order between these classes is
701  * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
702  */
703 enum inode_i_mutex_lock_class
704 {
705         I_MUTEX_NORMAL,
706         I_MUTEX_PARENT,
707         I_MUTEX_CHILD,
708         I_MUTEX_XATTR,
709         I_MUTEX_NONDIR2,
710         I_MUTEX_PARENT2,
711 };
712 
713 void lock_two_nondirectories(struct inode *, struct inode*);
714 void unlock_two_nondirectories(struct inode *, struct inode*);
715 
716 /*
717  * NOTE: in a 32bit arch with a preemptable kernel and
718  * an UP compile the i_size_read/write must be atomic
719  * with respect to the local cpu (unlike with preempt disabled),
720  * but they don't need to be atomic with respect to other cpus like in
721  * true SMP (so they need either to either locally disable irq around
722  * the read or for example on x86 they can be still implemented as a
723  * cmpxchg8b without the need of the lock prefix). For SMP compiles
724  * and 64bit archs it makes no difference if preempt is enabled or not.
725  */
726 static inline loff_t i_size_read(const struct inode *inode)
727 {
728 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
729         loff_t i_size;
730         unsigned int seq;
731 
732         do {
733                 seq = read_seqcount_begin(&inode->i_size_seqcount);
734                 i_size = inode->i_size;
735         } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
736         return i_size;
737 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
738         loff_t i_size;
739 
740         preempt_disable();
741         i_size = inode->i_size;
742         preempt_enable();
743         return i_size;
744 #else
745         return inode->i_size;
746 #endif
747 }
748 
749 /*
750  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
751  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
752  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
753  */
754 static inline void i_size_write(struct inode *inode, loff_t i_size)
755 {
756 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
757         preempt_disable();
758         write_seqcount_begin(&inode->i_size_seqcount);
759         inode->i_size = i_size;
760         write_seqcount_end(&inode->i_size_seqcount);
761         preempt_enable();
762 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
763         preempt_disable();
764         inode->i_size = i_size;
765         preempt_enable();
766 #else
767         inode->i_size = i_size;
768 #endif
769 }
770 
771 /* Helper functions so that in most cases filesystems will
772  * not need to deal directly with kuid_t and kgid_t and can
773  * instead deal with the raw numeric values that are stored
774  * in the filesystem.
775  */
776 static inline uid_t i_uid_read(const struct inode *inode)
777 {
778         return from_kuid(&init_user_ns, inode->i_uid);
779 }
780 
781 static inline gid_t i_gid_read(const struct inode *inode)
782 {
783         return from_kgid(&init_user_ns, inode->i_gid);
784 }
785 
786 static inline void i_uid_write(struct inode *inode, uid_t uid)
787 {
788         inode->i_uid = make_kuid(&init_user_ns, uid);
789 }
790 
791 static inline void i_gid_write(struct inode *inode, gid_t gid)
792 {
793         inode->i_gid = make_kgid(&init_user_ns, gid);
794 }
795 
796 static inline unsigned iminor(const struct inode *inode)
797 {
798         return MINOR(inode->i_rdev);
799 }
800 
801 static inline unsigned imajor(const struct inode *inode)
802 {
803         return MAJOR(inode->i_rdev);
804 }
805 
806 extern struct block_device *I_BDEV(struct inode *inode);
807 
808 struct fown_struct {
809         rwlock_t lock;          /* protects pid, uid, euid fields */
810         struct pid *pid;        /* pid or -pgrp where SIGIO should be sent */
811         enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
812         kuid_t uid, euid;       /* uid/euid of process setting the owner */
813         int signum;             /* posix.1b rt signal to be delivered on IO */
814 };
815 
816 /*
817  * Track a single file's readahead state
818  */
819 struct file_ra_state {
820         pgoff_t start;                  /* where readahead started */
821         unsigned int size;              /* # of readahead pages */
822         unsigned int async_size;        /* do asynchronous readahead when
823                                            there are only # of pages ahead */
824 
825         unsigned int ra_pages;          /* Maximum readahead window */
826         unsigned int mmap_miss;         /* Cache miss stat for mmap accesses */
827         loff_t prev_pos;                /* Cache last read() position */
828 };
829 
830 /*
831  * Check if @index falls in the readahead windows.
832  */
833 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
834 {
835         return (index >= ra->start &&
836                 index <  ra->start + ra->size);
837 }
838 
839 struct file {
840         union {
841                 struct llist_node       fu_llist;
842                 struct rcu_head         fu_rcuhead;
843         } f_u;
844         struct path             f_path;
845         struct inode            *f_inode;       /* cached value */
846         const struct file_operations    *f_op;
847 
848         /*
849          * Protects f_ep_links, f_flags.
850          * Must not be taken from IRQ context.
851          */
852         spinlock_t              f_lock;
853         atomic_long_t           f_count;
854         unsigned int            f_flags;
855         fmode_t                 f_mode;
856         struct mutex            f_pos_lock;
857         loff_t                  f_pos;
858         struct fown_struct      f_owner;
859         const struct cred       *f_cred;
860         struct file_ra_state    f_ra;
861 
862         u64                     f_version;
863 #ifdef CONFIG_SECURITY
864         void                    *f_security;
865 #endif
866         /* needed for tty driver, and maybe others */
867         void                    *private_data;
868 
869 #ifdef CONFIG_EPOLL
870         /* Used by fs/eventpoll.c to link all the hooks to this file */
871         struct list_head        f_ep_links;
872         struct list_head        f_tfile_llink;
873 #endif /* #ifdef CONFIG_EPOLL */
874         struct address_space    *f_mapping;
875 } __attribute__((aligned(4)));  /* lest something weird decides that 2 is OK */
876 
877 struct file_handle {
878         __u32 handle_bytes;
879         int handle_type;
880         /* file identifier */
881         unsigned char f_handle[0];
882 };
883 
884 static inline struct file *get_file(struct file *f)
885 {
886         atomic_long_inc(&f->f_count);
887         return f;
888 }
889 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
890 #define fput_atomic(x)  atomic_long_add_unless(&(x)->f_count, -1, 1)
891 #define file_count(x)   atomic_long_read(&(x)->f_count)
892 
893 #define MAX_NON_LFS     ((1UL<<31) - 1)
894 
895 /* Page cache limit. The filesystems should put that into their s_maxbytes 
896    limits, otherwise bad things can happen in VM. */ 
897 #if BITS_PER_LONG==32
898 #define MAX_LFS_FILESIZE        (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 
899 #elif BITS_PER_LONG==64
900 #define MAX_LFS_FILESIZE        ((loff_t)0x7fffffffffffffffLL)
901 #endif
902 
903 #define FL_POSIX        1
904 #define FL_FLOCK        2
905 #define FL_DELEG        4       /* NFSv4 delegation */
906 #define FL_ACCESS       8       /* not trying to lock, just looking */
907 #define FL_EXISTS       16      /* when unlocking, test for existence */
908 #define FL_LEASE        32      /* lease held on this file */
909 #define FL_CLOSE        64      /* unlock on close */
910 #define FL_SLEEP        128     /* A blocking lock */
911 #define FL_DOWNGRADE_PENDING    256 /* Lease is being downgraded */
912 #define FL_UNLOCK_PENDING       512 /* Lease is being broken */
913 #define FL_OFDLCK       1024    /* lock is "owned" by struct file */
914 #define FL_LAYOUT       2048    /* outstanding pNFS layout */
915 
916 /*
917  * Special return value from posix_lock_file() and vfs_lock_file() for
918  * asynchronous locking.
919  */
920 #define FILE_LOCK_DEFERRED 1
921 
922 /* legacy typedef, should eventually be removed */
923 typedef void *fl_owner_t;
924 
925 struct file_lock;
926 
927 struct file_lock_operations {
928         void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
929         void (*fl_release_private)(struct file_lock *);
930 };
931 
932 struct lock_manager_operations {
933         int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
934         unsigned long (*lm_owner_key)(struct file_lock *);
935         fl_owner_t (*lm_get_owner)(fl_owner_t);
936         void (*lm_put_owner)(fl_owner_t);
937         void (*lm_notify)(struct file_lock *);  /* unblock callback */
938         int (*lm_grant)(struct file_lock *, int);
939         bool (*lm_break)(struct file_lock *);
940         int (*lm_change)(struct file_lock *, int, struct list_head *);
941         void (*lm_setup)(struct file_lock *, void **);
942 };
943 
944 struct lock_manager {
945         struct list_head list;
946         /*
947          * NFSv4 and up also want opens blocked during the grace period;
948          * NLM doesn't care:
949          */
950         bool block_opens;
951 };
952 
953 struct net;
954 void locks_start_grace(struct net *, struct lock_manager *);
955 void locks_end_grace(struct lock_manager *);
956 int locks_in_grace(struct net *);
957 int opens_in_grace(struct net *);
958 
959 /* that will die - we need it for nfs_lock_info */
960 #include <linux/nfs_fs_i.h>
961 
962 /*
963  * struct file_lock represents a generic "file lock". It's used to represent
964  * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
965  * note that the same struct is used to represent both a request for a lock and
966  * the lock itself, but the same object is never used for both.
967  *
968  * FIXME: should we create a separate "struct lock_request" to help distinguish
969  * these two uses?
970  *
971  * The varous i_flctx lists are ordered by:
972  *
973  * 1) lock owner
974  * 2) lock range start
975  * 3) lock range end
976  *
977  * Obviously, the last two criteria only matter for POSIX locks.
978  */
979 struct file_lock {
980         struct file_lock *fl_next;      /* singly linked list for this inode  */
981         struct list_head fl_list;       /* link into file_lock_context */
982         struct hlist_node fl_link;      /* node in global lists */
983         struct list_head fl_block;      /* circular list of blocked processes */
984         fl_owner_t fl_owner;
985         unsigned int fl_flags;
986         unsigned char fl_type;
987         unsigned int fl_pid;
988         int fl_link_cpu;                /* what cpu's list is this on? */
989         struct pid *fl_nspid;
990         wait_queue_head_t fl_wait;
991         struct file *fl_file;
992         loff_t fl_start;
993         loff_t fl_end;
994 
995         struct fasync_struct *  fl_fasync; /* for lease break notifications */
996         /* for lease breaks: */
997         unsigned long fl_break_time;
998         unsigned long fl_downgrade_time;
999 
1000         const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
1001         const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1002         union {
1003                 struct nfs_lock_info    nfs_fl;
1004                 struct nfs4_lock_info   nfs4_fl;
1005                 struct {
1006                         struct list_head link;  /* link in AFS vnode's pending_locks list */
1007                         int state;              /* state of grant or error if -ve */
1008                 } afs;
1009         } fl_u;
1010 };
1011 
1012 struct file_lock_context {
1013         spinlock_t              flc_lock;
1014         struct list_head        flc_flock;
1015         struct list_head        flc_posix;
1016         struct list_head        flc_lease;
1017 };
1018 
1019 /* The following constant reflects the upper bound of the file/locking space */
1020 #ifndef OFFSET_MAX
1021 #define INT_LIMIT(x)    (~((x)1 << (sizeof(x)*8 - 1)))
1022 #define OFFSET_MAX      INT_LIMIT(loff_t)
1023 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1024 #endif
1025 
1026 #include <linux/fcntl.h>
1027 
1028 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1029 
1030 #ifdef CONFIG_FILE_LOCKING
1031 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1032 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1033                         struct flock __user *);
1034 
1035 #if BITS_PER_LONG == 32
1036 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1037 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1038                         struct flock64 __user *);
1039 #endif
1040 
1041 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1042 extern int fcntl_getlease(struct file *filp);
1043 
1044 /* fs/locks.c */
1045 void locks_free_lock_context(struct file_lock_context *ctx);
1046 void locks_free_lock(struct file_lock *fl);
1047 extern void locks_init_lock(struct file_lock *);
1048 extern struct file_lock * locks_alloc_lock(void);
1049 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1050 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1051 extern void locks_remove_posix(struct file *, fl_owner_t);
1052 extern void locks_remove_file(struct file *);
1053 extern void locks_release_private(struct file_lock *);
1054 extern void posix_test_lock(struct file *, struct file_lock *);
1055 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1056 extern int posix_unblock_lock(struct file_lock *);
1057 extern int vfs_test_lock(struct file *, struct file_lock *);
1058 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1059 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1060 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1061 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1062 extern void lease_get_mtime(struct inode *, struct timespec *time);
1063 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1064 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1065 extern int lease_modify(struct file_lock *, int, struct list_head *);
1066 struct files_struct;
1067 extern void show_fd_locks(struct seq_file *f,
1068                          struct file *filp, struct files_struct *files);
1069 #else /* !CONFIG_FILE_LOCKING */
1070 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1071                               struct flock __user *user)
1072 {
1073         return -EINVAL;
1074 }
1075 
1076 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1077                               unsigned int cmd, struct flock __user *user)
1078 {
1079         return -EACCES;
1080 }
1081 
1082 #if BITS_PER_LONG == 32
1083 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1084                                 struct flock64 __user *user)
1085 {
1086         return -EINVAL;
1087 }
1088 
1089 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1090                                 unsigned int cmd, struct flock64 __user *user)
1091 {
1092         return -EACCES;
1093 }
1094 #endif
1095 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1096 {
1097         return -EINVAL;
1098 }
1099 
1100 static inline int fcntl_getlease(struct file *filp)
1101 {
1102         return F_UNLCK;
1103 }
1104 
1105 static inline void
1106 locks_free_lock_context(struct file_lock_context *ctx)
1107 {
1108 }
1109 
1110 static inline void locks_init_lock(struct file_lock *fl)
1111 {
1112         return;
1113 }
1114 
1115 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1116 {
1117         return;
1118 }
1119 
1120 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1121 {
1122         return;
1123 }
1124 
1125 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1126 {
1127         return;
1128 }
1129 
1130 static inline void locks_remove_file(struct file *filp)
1131 {
1132         return;
1133 }
1134 
1135 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1136 {
1137         return;
1138 }
1139 
1140 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1141                                   struct file_lock *conflock)
1142 {
1143         return -ENOLCK;
1144 }
1145 
1146 static inline int posix_unblock_lock(struct file_lock *waiter)
1147 {
1148         return -ENOENT;
1149 }
1150 
1151 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1152 {
1153         return 0;
1154 }
1155 
1156 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1157                                 struct file_lock *fl, struct file_lock *conf)
1158 {
1159         return -ENOLCK;
1160 }
1161 
1162 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1163 {
1164         return 0;
1165 }
1166 
1167 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1168 {
1169         return -ENOLCK;
1170 }
1171 
1172 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1173 {
1174         return 0;
1175 }
1176 
1177 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1178 {
1179         return;
1180 }
1181 
1182 static inline int generic_setlease(struct file *filp, long arg,
1183                                     struct file_lock **flp, void **priv)
1184 {
1185         return -EINVAL;
1186 }
1187 
1188 static inline int vfs_setlease(struct file *filp, long arg,
1189                                struct file_lock **lease, void **priv)
1190 {
1191         return -EINVAL;
1192 }
1193 
1194 static inline int lease_modify(struct file_lock *fl, int arg,
1195                                struct list_head *dispose)
1196 {
1197         return -EINVAL;
1198 }
1199 
1200 struct files_struct;
1201 static inline void show_fd_locks(struct seq_file *f,
1202                         struct file *filp, struct files_struct *files) {}
1203 #endif /* !CONFIG_FILE_LOCKING */
1204 
1205 static inline struct inode *file_inode(const struct file *f)
1206 {
1207         return f->f_inode;
1208 }
1209 
1210 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1211 {
1212         return locks_lock_inode_wait(file_inode(filp), fl);
1213 }
1214 
1215 struct fasync_struct {
1216         spinlock_t              fa_lock;
1217         int                     magic;
1218         int                     fa_fd;
1219         struct fasync_struct    *fa_next; /* singly linked list */
1220         struct file             *fa_file;
1221         struct rcu_head         fa_rcu;
1222 };
1223 
1224 #define FASYNC_MAGIC 0x4601
1225 
1226 /* SMP safe fasync helpers: */
1227 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1228 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1229 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1230 extern struct fasync_struct *fasync_alloc(void);
1231 extern void fasync_free(struct fasync_struct *);
1232 
1233 /* can be called from interrupts */
1234 extern void kill_fasync(struct fasync_struct **, int, int);
1235 
1236 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1237 extern void f_setown(struct file *filp, unsigned long arg, int force);
1238 extern void f_delown(struct file *filp);
1239 extern pid_t f_getown(struct file *filp);
1240 extern int send_sigurg(struct fown_struct *fown);
1241 
1242 struct mm_struct;
1243 
1244 /*
1245  *      Umount options
1246  */
1247 
1248 #define MNT_FORCE       0x00000001      /* Attempt to forcibily umount */
1249 #define MNT_DETACH      0x00000002      /* Just detach from the tree */
1250 #define MNT_EXPIRE      0x00000004      /* Mark for expiry */
1251 #define UMOUNT_NOFOLLOW 0x00000008      /* Don't follow symlink on umount */
1252 #define UMOUNT_UNUSED   0x80000000      /* Flag guaranteed to be unused */
1253 
1254 /* sb->s_iflags */
1255 #define SB_I_CGROUPWB   0x00000001      /* cgroup-aware writeback enabled */
1256 #define SB_I_NOEXEC     0x00000002      /* Ignore executables on this fs */
1257 
1258 /* Possible states of 'frozen' field */
1259 enum {
1260         SB_UNFROZEN = 0,                /* FS is unfrozen */
1261         SB_FREEZE_WRITE = 1,            /* Writes, dir ops, ioctls frozen */
1262         SB_FREEZE_PAGEFAULT = 2,        /* Page faults stopped as well */
1263         SB_FREEZE_FS = 3,               /* For internal FS use (e.g. to stop
1264                                          * internal threads if needed) */
1265         SB_FREEZE_COMPLETE = 4,         /* ->freeze_fs finished successfully */
1266 };
1267 
1268 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1269 
1270 struct sb_writers {
1271         int                             frozen;         /* Is sb frozen? */
1272         wait_queue_head_t               wait_unfrozen;  /* for get_super_thawed() */
1273         struct percpu_rw_semaphore      rw_sem[SB_FREEZE_LEVELS];
1274 };
1275 
1276 struct super_block {
1277         struct list_head        s_list;         /* Keep this first */
1278         dev_t                   s_dev;          /* search index; _not_ kdev_t */
1279         unsigned char           s_blocksize_bits;
1280         unsigned long           s_blocksize;
1281         loff_t                  s_maxbytes;     /* Max file size */
1282         struct file_system_type *s_type;
1283         const struct super_operations   *s_op;
1284         const struct dquot_operations   *dq_op;
1285         const struct quotactl_ops       *s_qcop;
1286         const struct export_operations *s_export_op;
1287         unsigned long           s_flags;
1288         unsigned long           s_iflags;       /* internal SB_I_* flags */
1289         unsigned long           s_magic;
1290         struct dentry           *s_root;
1291         struct rw_semaphore     s_umount;
1292         int                     s_count;
1293         atomic_t                s_active;
1294 #ifdef CONFIG_SECURITY
1295         void                    *s_security;
1296 #endif
1297         const struct xattr_handler **s_xattr;
1298 
1299         struct hlist_bl_head    s_anon;         /* anonymous dentries for (nfs) exporting */
1300         struct list_head        s_mounts;       /* list of mounts; _not_ for fs use */
1301         struct block_device     *s_bdev;
1302         struct backing_dev_info *s_bdi;
1303         struct mtd_info         *s_mtd;
1304         struct hlist_node       s_instances;
1305         unsigned int            s_quota_types;  /* Bitmask of supported quota types */
1306         struct quota_info       s_dquot;        /* Diskquota specific options */
1307 
1308         struct sb_writers       s_writers;
1309 
1310         char s_id[32];                          /* Informational name */
1311         u8 s_uuid[16];                          /* UUID */
1312 
1313         void                    *s_fs_info;     /* Filesystem private info */
1314         unsigned int            s_max_links;
1315         fmode_t                 s_mode;
1316 
1317         /* Granularity of c/m/atime in ns.
1318            Cannot be worse than a second */
1319         u32                s_time_gran;
1320 
1321         /*
1322          * The next field is for VFS *only*. No filesystems have any business
1323          * even looking at it. You had been warned.
1324          */
1325         struct mutex s_vfs_rename_mutex;        /* Kludge */
1326 
1327         /*
1328          * Filesystem subtype.  If non-empty the filesystem type field
1329          * in /proc/mounts will be "type.subtype"
1330          */
1331         char *s_subtype;
1332 
1333         /*
1334          * Saved mount options for lazy filesystems using
1335          * generic_show_options()
1336          */
1337         char __rcu *s_options;
1338         const struct dentry_operations *s_d_op; /* default d_op for dentries */
1339 
1340         /*
1341          * Saved pool identifier for cleancache (-1 means none)
1342          */
1343         int cleancache_poolid;
1344 
1345         struct shrinker s_shrink;       /* per-sb shrinker handle */
1346 
1347         /* Number of inodes with nlink == 0 but still referenced */
1348         atomic_long_t s_remove_count;
1349 
1350         /* Being remounted read-only */
1351         int s_readonly_remount;
1352 
1353         /* AIO completions deferred from interrupt context */
1354         struct workqueue_struct *s_dio_done_wq;
1355         struct hlist_head s_pins;
1356 
1357         /*
1358          * Keep the lru lists last in the structure so they always sit on their
1359          * own individual cachelines.
1360          */
1361         struct list_lru         s_dentry_lru ____cacheline_aligned_in_smp;
1362         struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
1363         struct rcu_head         rcu;
1364         struct work_struct      destroy_work;
1365 
1366         struct mutex            s_sync_lock;    /* sync serialisation lock */
1367 
1368         /*
1369          * Indicates how deep in a filesystem stack this SB is
1370          */
1371         int s_stack_depth;
1372 
1373         /* s_inode_list_lock protects s_inodes */
1374         spinlock_t              s_inode_list_lock ____cacheline_aligned_in_smp;
1375         struct list_head        s_inodes;       /* all inodes */
1376 };
1377 
1378 extern struct timespec current_fs_time(struct super_block *sb);
1379 
1380 /*
1381  * Snapshotting support.
1382  */
1383 
1384 void __sb_end_write(struct super_block *sb, int level);
1385 int __sb_start_write(struct super_block *sb, int level, bool wait);
1386 
1387 #define __sb_writers_acquired(sb, lev)  \
1388         percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1389 #define __sb_writers_release(sb, lev)   \
1390         percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1391 
1392 /**
1393  * sb_end_write - drop write access to a superblock
1394  * @sb: the super we wrote to
1395  *
1396  * Decrement number of writers to the filesystem. Wake up possible waiters
1397  * wanting to freeze the filesystem.
1398  */
1399 static inline void sb_end_write(struct super_block *sb)
1400 {
1401         __sb_end_write(sb, SB_FREEZE_WRITE);
1402 }
1403 
1404 /**
1405  * sb_end_pagefault - drop write access to a superblock from a page fault
1406  * @sb: the super we wrote to
1407  *
1408  * Decrement number of processes handling write page fault to the filesystem.
1409  * Wake up possible waiters wanting to freeze the filesystem.
1410  */
1411 static inline void sb_end_pagefault(struct super_block *sb)
1412 {
1413         __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1414 }
1415 
1416 /**
1417  * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1418  * @sb: the super we wrote to
1419  *
1420  * Decrement fs-internal number of writers to the filesystem.  Wake up possible
1421  * waiters wanting to freeze the filesystem.
1422  */
1423 static inline void sb_end_intwrite(struct super_block *sb)
1424 {
1425         __sb_end_write(sb, SB_FREEZE_FS);
1426 }
1427 
1428 /**
1429  * sb_start_write - get write access to a superblock
1430  * @sb: the super we write to
1431  *
1432  * When a process wants to write data or metadata to a file system (i.e. dirty
1433  * a page or an inode), it should embed the operation in a sb_start_write() -
1434  * sb_end_write() pair to get exclusion against file system freezing. This
1435  * function increments number of writers preventing freezing. If the file
1436  * system is already frozen, the function waits until the file system is
1437  * thawed.
1438  *
1439  * Since freeze protection behaves as a lock, users have to preserve
1440  * ordering of freeze protection and other filesystem locks. Generally,
1441  * freeze protection should be the outermost lock. In particular, we have:
1442  *
1443  * sb_start_write
1444  *   -> i_mutex                 (write path, truncate, directory ops, ...)
1445  *   -> s_umount                (freeze_super, thaw_super)
1446  */
1447 static inline void sb_start_write(struct super_block *sb)
1448 {
1449         __sb_start_write(sb, SB_FREEZE_WRITE, true);
1450 }
1451 
1452 static inline int sb_start_write_trylock(struct super_block *sb)
1453 {
1454         return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1455 }
1456 
1457 /**
1458  * sb_start_pagefault - get write access to a superblock from a page fault
1459  * @sb: the super we write to
1460  *
1461  * When a process starts handling write page fault, it should embed the
1462  * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1463  * exclusion against file system freezing. This is needed since the page fault
1464  * is going to dirty a page. This function increments number of running page
1465  * faults preventing freezing. If the file system is already frozen, the
1466  * function waits until the file system is thawed.
1467  *
1468  * Since page fault freeze protection behaves as a lock, users have to preserve
1469  * ordering of freeze protection and other filesystem locks. It is advised to
1470  * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1471  * handling code implies lock dependency:
1472  *
1473  * mmap_sem
1474  *   -> sb_start_pagefault
1475  */
1476 static inline void sb_start_pagefault(struct super_block *sb)
1477 {
1478         __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1479 }
1480 
1481 /*
1482  * sb_start_intwrite - get write access to a superblock for internal fs purposes
1483  * @sb: the super we write to
1484  *
1485  * This is the third level of protection against filesystem freezing. It is
1486  * free for use by a filesystem. The only requirement is that it must rank
1487  * below sb_start_pagefault.
1488  *
1489  * For example filesystem can call sb_start_intwrite() when starting a
1490  * transaction which somewhat eases handling of freezing for internal sources
1491  * of filesystem changes (internal fs threads, discarding preallocation on file
1492  * close, etc.).
1493  */
1494 static inline void sb_start_intwrite(struct super_block *sb)
1495 {
1496         __sb_start_write(sb, SB_FREEZE_FS, true);
1497 }
1498 
1499 
1500 extern bool inode_owner_or_capable(const struct inode *inode);
1501 
1502 /*
1503  * VFS helper functions..
1504  */
1505 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1506 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1507 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1508 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1509 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1510 extern int vfs_rmdir(struct inode *, struct dentry *);
1511 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1512 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1513 extern int vfs_whiteout(struct inode *, struct dentry *);
1514 
1515 /*
1516  * VFS dentry helper functions.
1517  */
1518 extern void dentry_unhash(struct dentry *dentry);
1519 
1520 /*
1521  * VFS file helper functions.
1522  */
1523 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1524                         umode_t mode);
1525 /*
1526  * VFS FS_IOC_FIEMAP helper definitions.
1527  */
1528 struct fiemap_extent_info {
1529         unsigned int fi_flags;          /* Flags as passed from user */
1530         unsigned int fi_extents_mapped; /* Number of mapped extents */
1531         unsigned int fi_extents_max;    /* Size of fiemap_extent array */
1532         struct fiemap_extent __user *fi_extents_start; /* Start of
1533                                                         fiemap_extent array */
1534 };
1535 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1536                             u64 phys, u64 len, u32 flags);
1537 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1538 
1539 /*
1540  * File types
1541  *
1542  * NOTE! These match bits 12..15 of stat.st_mode
1543  * (ie "(i_mode >> 12) & 15").
1544  */
1545 #define DT_UNKNOWN      0
1546 #define DT_FIFO         1
1547 #define DT_CHR          2
1548 #define DT_DIR          4
1549 #define DT_BLK          6
1550 #define DT_REG          8
1551 #define DT_LNK          10
1552 #define DT_SOCK         12
1553 #define DT_WHT          14
1554 
1555 /*
1556  * This is the "filldir" function type, used by readdir() to let
1557  * the kernel specify what kind of dirent layout it wants to have.
1558  * This allows the kernel to read directories into kernel space or
1559  * to have different dirent layouts depending on the binary type.
1560  */
1561 struct dir_context;
1562 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1563                          unsigned);
1564 
1565 struct dir_context {
1566         const filldir_t actor;
1567         loff_t pos;
1568 };
1569 
1570 struct block_device_operations;
1571 
1572 /* These macros are for out of kernel modules to test that
1573  * the kernel supports the unlocked_ioctl and compat_ioctl
1574  * fields in struct file_operations. */
1575 #define HAVE_COMPAT_IOCTL 1
1576 #define HAVE_UNLOCKED_IOCTL 1
1577 
1578 /*
1579  * These flags let !MMU mmap() govern direct device mapping vs immediate
1580  * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1581  *
1582  * NOMMU_MAP_COPY:      Copy can be mapped (MAP_PRIVATE)
1583  * NOMMU_MAP_DIRECT:    Can be mapped directly (MAP_SHARED)
1584  * NOMMU_MAP_READ:      Can be mapped for reading
1585  * NOMMU_MAP_WRITE:     Can be mapped for writing
1586  * NOMMU_MAP_EXEC:      Can be mapped for execution
1587  */
1588 #define NOMMU_MAP_COPY          0x00000001
1589 #define NOMMU_MAP_DIRECT        0x00000008
1590 #define NOMMU_MAP_READ          VM_MAYREAD
1591 #define NOMMU_MAP_WRITE         VM_MAYWRITE
1592 #define NOMMU_MAP_EXEC          VM_MAYEXEC
1593 
1594 #define NOMMU_VMFLAGS \
1595         (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1596 
1597 
1598 struct iov_iter;
1599 
1600 struct file_operations {
1601         struct module *owner;
1602         loff_t (*llseek) (struct file *, loff_t, int);
1603         ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1604         ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1605         ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1606         ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1607         int (*iterate) (struct file *, struct dir_context *);
1608         unsigned int (*poll) (struct file *, struct poll_table_struct *);
1609         long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1610         long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1611         int (*mmap) (struct file *, struct vm_area_struct *);
1612         int (*open) (struct inode *, struct file *);
1613         int (*flush) (struct file *, fl_owner_t id);
1614         int (*release) (struct inode *, struct file *);
1615         int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1616         int (*aio_fsync) (struct kiocb *, int datasync);
1617         int (*fasync) (int, struct file *, int);
1618         int (*lock) (struct file *, int, struct file_lock *);
1619         ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1620         unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1621         int (*check_flags)(int);
1622         int (*flock) (struct file *, int, struct file_lock *);
1623         ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1624         ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1625         int (*setlease)(struct file *, long, struct file_lock **, void **);
1626         long (*fallocate)(struct file *file, int mode, loff_t offset,
1627                           loff_t len);
1628         void (*show_fdinfo)(struct seq_file *m, struct file *f);
1629 #ifndef CONFIG_MMU
1630         unsigned (*mmap_capabilities)(struct file *);
1631 #endif
1632 };
1633 
1634 struct inode_operations {
1635         struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1636         const char * (*follow_link) (struct dentry *, void **);
1637         int (*permission) (struct inode *, int);
1638         struct posix_acl * (*get_acl)(struct inode *, int);
1639 
1640         int (*readlink) (struct dentry *, char __user *,int);
1641         void (*put_link) (struct inode *, void *);
1642 
1643         int (*create) (struct inode *,struct dentry *, umode_t, bool);
1644         int (*link) (struct dentry *,struct inode *,struct dentry *);
1645         int (*unlink) (struct inode *,struct dentry *);
1646         int (*symlink) (struct inode *,struct dentry *,const char *);
1647         int (*mkdir) (struct inode *,struct dentry *,umode_t);
1648         int (*rmdir) (struct inode *,struct dentry *);
1649         int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1650         int (*rename) (struct inode *, struct dentry *,
1651                         struct inode *, struct dentry *);
1652         int (*rename2) (struct inode *, struct dentry *,
1653                         struct inode *, struct dentry *, unsigned int);
1654         int (*setattr) (struct dentry *, struct iattr *);
1655         int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1656         int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1657         ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1658         ssize_t (*listxattr) (struct dentry *, char *, size_t);
1659         int (*removexattr) (struct dentry *, const char *);
1660         int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1661                       u64 len);
1662         int (*update_time)(struct inode *, struct timespec *, int);
1663         int (*atomic_open)(struct inode *, struct dentry *,
1664                            struct file *, unsigned open_flag,
1665                            umode_t create_mode, int *opened);
1666         int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1667         int (*set_acl)(struct inode *, struct posix_acl *, int);
1668 } ____cacheline_aligned;
1669 
1670 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1671                               unsigned long nr_segs, unsigned long fast_segs,
1672                               struct iovec *fast_pointer,
1673                               struct iovec **ret_pointer);
1674 
1675 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1676 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1677 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1678 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1679 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1680                 unsigned long, loff_t *);
1681 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1682                 unsigned long, loff_t *);
1683 
1684 struct super_operations {
1685         struct inode *(*alloc_inode)(struct super_block *sb);
1686         void (*destroy_inode)(struct inode *);
1687 
1688         void (*dirty_inode) (struct inode *, int flags);
1689         int (*write_inode) (struct inode *, struct writeback_control *wbc);
1690         int (*drop_inode) (struct inode *);
1691         void (*evict_inode) (struct inode *);
1692         void (*put_super) (struct super_block *);
1693         int (*sync_fs)(struct super_block *sb, int wait);
1694         int (*freeze_super) (struct super_block *);
1695         int (*freeze_fs) (struct super_block *);
1696         int (*thaw_super) (struct super_block *);
1697         int (*unfreeze_fs) (struct super_block *);
1698         int (*statfs) (struct dentry *, struct kstatfs *);
1699         int (*remount_fs) (struct super_block *, int *, char *);
1700         void (*umount_begin) (struct super_block *);
1701 
1702         int (*show_options)(struct seq_file *, struct dentry *);
1703         int (*show_devname)(struct seq_file *, struct dentry *);
1704         int (*show_path)(struct seq_file *, struct dentry *);
1705         int (*show_stats)(struct seq_file *, struct dentry *);
1706 #ifdef CONFIG_QUOTA
1707         ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1708         ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1709         struct dquot **(*get_dquots)(struct inode *);
1710 #endif
1711         int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1712         long (*nr_cached_objects)(struct super_block *,
1713                                   struct shrink_control *);
1714         long (*free_cached_objects)(struct super_block *,
1715                                     struct shrink_control *);
1716 };
1717 
1718 /*
1719  * Inode flags - they have no relation to superblock flags now
1720  */
1721 #define S_SYNC          1       /* Writes are synced at once */
1722 #define S_NOATIME       2       /* Do not update access times */
1723 #define S_APPEND        4       /* Append-only file */
1724 #define S_IMMUTABLE     8       /* Immutable file */
1725 #define S_DEAD          16      /* removed, but still open directory */
1726 #define S_NOQUOTA       32      /* Inode is not counted to quota */
1727 #define S_DIRSYNC       64      /* Directory modifications are synchronous */
1728 #define S_NOCMTIME      128     /* Do not update file c/mtime */
1729 #define S_SWAPFILE      256     /* Do not truncate: swapon got its bmaps */
1730 #define S_PRIVATE       512     /* Inode is fs-internal */
1731 #define S_IMA           1024    /* Inode has an associated IMA struct */
1732 #define S_AUTOMOUNT     2048    /* Automount/referral quasi-directory */
1733 #define S_NOSEC         4096    /* no suid or xattr security attributes */
1734 #ifdef CONFIG_FS_DAX
1735 #define S_DAX           8192    /* Direct Access, avoiding the page cache */
1736 #else
1737 #define S_DAX           0       /* Make all the DAX code disappear */
1738 #endif
1739 
1740 /*
1741  * Note that nosuid etc flags are inode-specific: setting some file-system
1742  * flags just means all the inodes inherit those flags by default. It might be
1743  * possible to override it selectively if you really wanted to with some
1744  * ioctl() that is not currently implemented.
1745  *
1746  * Exception: MS_RDONLY is always applied to the entire file system.
1747  *
1748  * Unfortunately, it is possible to change a filesystems flags with it mounted
1749  * with files in use.  This means that all of the inodes will not have their
1750  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
1751  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1752  */
1753 #define __IS_FLG(inode, flg)    ((inode)->i_sb->s_flags & (flg))
1754 
1755 #define IS_RDONLY(inode)        ((inode)->i_sb->s_flags & MS_RDONLY)
1756 #define IS_SYNC(inode)          (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1757                                         ((inode)->i_flags & S_SYNC))
1758 #define IS_DIRSYNC(inode)       (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1759                                         ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1760 #define IS_MANDLOCK(inode)      __IS_FLG(inode, MS_MANDLOCK)
1761 #define IS_NOATIME(inode)       __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1762 #define IS_I_VERSION(inode)     __IS_FLG(inode, MS_I_VERSION)
1763 
1764 #define IS_NOQUOTA(inode)       ((inode)->i_flags & S_NOQUOTA)
1765 #define IS_APPEND(inode)        ((inode)->i_flags & S_APPEND)
1766 #define IS_IMMUTABLE(inode)     ((inode)->i_flags & S_IMMUTABLE)
1767 #define IS_POSIXACL(inode)      __IS_FLG(inode, MS_POSIXACL)
1768 
1769 #define IS_DEADDIR(inode)       ((inode)->i_flags & S_DEAD)
1770 #define IS_NOCMTIME(inode)      ((inode)->i_flags & S_NOCMTIME)
1771 #define IS_SWAPFILE(inode)      ((inode)->i_flags & S_SWAPFILE)
1772 #define IS_PRIVATE(inode)       ((inode)->i_flags & S_PRIVATE)
1773 #define IS_IMA(inode)           ((inode)->i_flags & S_IMA)
1774 #define IS_AUTOMOUNT(inode)     ((inode)->i_flags & S_AUTOMOUNT)
1775 #define IS_NOSEC(inode)         ((inode)->i_flags & S_NOSEC)
1776 #define IS_DAX(inode)           ((inode)->i_flags & S_DAX)
1777 
1778 #define IS_WHITEOUT(inode)      (S_ISCHR(inode->i_mode) && \
1779                                  (inode)->i_rdev == WHITEOUT_DEV)
1780 
1781 /*
1782  * Inode state bits.  Protected by inode->i_lock
1783  *
1784  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1785  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1786  *
1787  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1788  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1789  * various stages of removing an inode.
1790  *
1791  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1792  *
1793  * I_DIRTY_SYNC         Inode is dirty, but doesn't have to be written on
1794  *                      fdatasync().  i_atime is the usual cause.
1795  * I_DIRTY_DATASYNC     Data-related inode changes pending. We keep track of
1796  *                      these changes separately from I_DIRTY_SYNC so that we
1797  *                      don't have to write inode on fdatasync() when only
1798  *                      mtime has changed in it.
1799  * I_DIRTY_PAGES        Inode has dirty pages.  Inode itself may be clean.
1800  * I_NEW                Serves as both a mutex and completion notification.
1801  *                      New inodes set I_NEW.  If two processes both create
1802  *                      the same inode, one of them will release its inode and
1803  *                      wait for I_NEW to be released before returning.
1804  *                      Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1805  *                      also cause waiting on I_NEW, without I_NEW actually
1806  *                      being set.  find_inode() uses this to prevent returning
1807  *                      nearly-dead inodes.
1808  * I_WILL_FREE          Must be set when calling write_inode_now() if i_count
1809  *                      is zero.  I_FREEING must be set when I_WILL_FREE is
1810  *                      cleared.
1811  * I_FREEING            Set when inode is about to be freed but still has dirty
1812  *                      pages or buffers attached or the inode itself is still
1813  *                      dirty.
1814  * I_CLEAR              Added by clear_inode().  In this state the inode is
1815  *                      clean and can be destroyed.  Inode keeps I_FREEING.
1816  *
1817  *                      Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1818  *                      prohibited for many purposes.  iget() must wait for
1819  *                      the inode to be completely released, then create it
1820  *                      anew.  Other functions will just ignore such inodes,
1821  *                      if appropriate.  I_NEW is used for waiting.
1822  *
1823  * I_SYNC               Writeback of inode is running. The bit is set during
1824  *                      data writeback, and cleared with a wakeup on the bit
1825  *                      address once it is done. The bit is also used to pin
1826  *                      the inode in memory for flusher thread.
1827  *
1828  * I_REFERENCED         Marks the inode as recently references on the LRU list.
1829  *
1830  * I_DIO_WAKEUP         Never set.  Only used as a key for wait_on_bit().
1831  *
1832  * I_WB_SWITCH          Cgroup bdi_writeback switching in progress.  Used to
1833  *                      synchronize competing switching instances and to tell
1834  *                      wb stat updates to grab mapping->tree_lock.  See
1835  *                      inode_switch_wb_work_fn() for details.
1836  *
1837  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1838  */
1839 #define I_DIRTY_SYNC            (1 << 0)
1840 #define I_DIRTY_DATASYNC        (1 << 1)
1841 #define I_DIRTY_PAGES           (1 << 2)
1842 #define __I_NEW                 3
1843 #define I_NEW                   (1 << __I_NEW)
1844 #define I_WILL_FREE             (1 << 4)
1845 #define I_FREEING               (1 << 5)
1846 #define I_CLEAR                 (1 << 6)
1847 #define __I_SYNC                7
1848 #define I_SYNC                  (1 << __I_SYNC)
1849 #define I_REFERENCED            (1 << 8)
1850 #define __I_DIO_WAKEUP          9
1851 #define I_DIO_WAKEUP            (1 << __I_DIO_WAKEUP)
1852 #define I_LINKABLE              (1 << 10)
1853 #define I_DIRTY_TIME            (1 << 11)
1854 #define __I_DIRTY_TIME_EXPIRED  12
1855 #define I_DIRTY_TIME_EXPIRED    (1 << __I_DIRTY_TIME_EXPIRED)
1856 #define I_WB_SWITCH             (1 << 13)
1857 
1858 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1859 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1860 
1861 extern void __mark_inode_dirty(struct inode *, int);
1862 static inline void mark_inode_dirty(struct inode *inode)
1863 {
1864         __mark_inode_dirty(inode, I_DIRTY);
1865 }
1866 
1867 static inline void mark_inode_dirty_sync(struct inode *inode)
1868 {
1869         __mark_inode_dirty(inode, I_DIRTY_SYNC);
1870 }
1871 
1872 extern void inc_nlink(struct inode *inode);
1873 extern void drop_nlink(struct inode *inode);
1874 extern void clear_nlink(struct inode *inode);
1875 extern void set_nlink(struct inode *inode, unsigned int nlink);
1876 
1877 static inline void inode_inc_link_count(struct inode *inode)
1878 {
1879         inc_nlink(inode);
1880         mark_inode_dirty(inode);
1881 }
1882 
1883 static inline void inode_dec_link_count(struct inode *inode)
1884 {
1885         drop_nlink(inode);
1886         mark_inode_dirty(inode);
1887 }
1888 
1889 /**
1890  * inode_inc_iversion - increments i_version
1891  * @inode: inode that need to be updated
1892  *
1893  * Every time the inode is modified, the i_version field will be incremented.
1894  * The filesystem has to be mounted with i_version flag
1895  */
1896 
1897 static inline void inode_inc_iversion(struct inode *inode)
1898 {
1899        spin_lock(&inode->i_lock);
1900        inode->i_version++;
1901        spin_unlock(&inode->i_lock);
1902 }
1903 
1904 enum file_time_flags {
1905         S_ATIME = 1,
1906         S_MTIME = 2,
1907         S_CTIME = 4,
1908         S_VERSION = 8,
1909 };
1910 
1911 extern bool atime_needs_update(const struct path *, struct inode *);
1912 extern void touch_atime(const struct path *);
1913 static inline void file_accessed(struct file *file)
1914 {
1915         if (!(file->f_flags & O_NOATIME))
1916                 touch_atime(&file->f_path);
1917 }
1918 
1919 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1920 int sync_inode_metadata(struct inode *inode, int wait);
1921 
1922 struct file_system_type {
1923         const char *name;
1924         int fs_flags;
1925 #define FS_REQUIRES_DEV         1 
1926 #define FS_BINARY_MOUNTDATA     2
1927 #define FS_HAS_SUBTYPE          4
1928 #define FS_USERNS_MOUNT         8       /* Can be mounted by userns root */
1929 #define FS_USERNS_DEV_MOUNT     16 /* A userns mount does not imply MNT_NODEV */
1930 #define FS_USERNS_VISIBLE       32      /* FS must already be visible */
1931 #define FS_RENAME_DOES_D_MOVE   32768   /* FS will handle d_move() during rename() internally. */
1932         struct dentry *(*mount) (struct file_system_type *, int,
1933                        const char *, void *);
1934         void (*kill_sb) (struct super_block *);
1935         struct module *owner;
1936         struct file_system_type * next;
1937         struct hlist_head fs_supers;
1938 
1939         struct lock_class_key s_lock_key;
1940         struct lock_class_key s_umount_key;
1941         struct lock_class_key s_vfs_rename_key;
1942         struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1943 
1944         struct lock_class_key i_lock_key;
1945         struct lock_class_key i_mutex_key;
1946         struct lock_class_key i_mutex_dir_key;
1947 };
1948 
1949 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1950 
1951 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1952         void *data, int (*fill_super)(struct super_block *, void *, int));
1953 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1954         int flags, const char *dev_name, void *data,
1955         int (*fill_super)(struct super_block *, void *, int));
1956 extern struct dentry *mount_single(struct file_system_type *fs_type,
1957         int flags, void *data,
1958         int (*fill_super)(struct super_block *, void *, int));
1959 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1960         int flags, void *data,
1961         int (*fill_super)(struct super_block *, void *, int));
1962 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1963 void generic_shutdown_super(struct super_block *sb);
1964 void kill_block_super(struct super_block *sb);
1965 void kill_anon_super(struct super_block *sb);
1966 void kill_litter_super(struct super_block *sb);
1967 void deactivate_super(struct super_block *sb);
1968 void deactivate_locked_super(struct super_block *sb);
1969 int set_anon_super(struct super_block *s, void *data);
1970 int get_anon_bdev(dev_t *);
1971 void free_anon_bdev(dev_t);
1972 struct super_block *sget(struct file_system_type *type,
1973                         int (*test)(struct super_block *,void *),
1974                         int (*set)(struct super_block *,void *),
1975                         int flags, void *data);
1976 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1977         const struct super_operations *ops,
1978         const struct dentry_operations *dops,
1979         unsigned long);
1980 
1981 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1982 #define fops_get(fops) \
1983         (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1984 #define fops_put(fops) \
1985         do { if (fops) module_put((fops)->owner); } while(0)
1986 /*
1987  * This one is to be used *ONLY* from ->open() instances.
1988  * fops must be non-NULL, pinned down *and* module dependencies
1989  * should be sufficient to pin the caller down as well.
1990  */
1991 #define replace_fops(f, fops) \
1992         do {    \
1993                 struct file *__file = (f); \
1994                 fops_put(__file->f_op); \
1995                 BUG_ON(!(__file->f_op = (fops))); \
1996         } while(0)
1997 
1998 extern int register_filesystem(struct file_system_type *);
1999 extern int unregister_filesystem(struct file_system_type *);
2000 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2001 #define kern_mount(type) kern_mount_data(type, NULL)
2002 extern void kern_unmount(struct vfsmount *mnt);
2003 extern int may_umount_tree(struct vfsmount *);
2004 extern int may_umount(struct vfsmount *);
2005 extern long do_mount(const char *, const char __user *,
2006                      const char *, unsigned long, void *);
2007 extern struct vfsmount *collect_mounts(struct path *);
2008 extern void drop_collected_mounts(struct vfsmount *);
2009 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2010                           struct vfsmount *);
2011 extern int vfs_statfs(struct path *, struct kstatfs *);
2012 extern int user_statfs(const char __user *, struct kstatfs *);
2013 extern int fd_statfs(int, struct kstatfs *);
2014 extern int vfs_ustat(dev_t, struct kstatfs *);
2015 extern int freeze_super(struct super_block *super);
2016 extern int thaw_super(struct super_block *super);
2017 extern bool our_mnt(struct vfsmount *mnt);
2018 
2019 extern int current_umask(void);
2020 
2021 extern void ihold(struct inode * inode);
2022 extern void iput(struct inode *);
2023 extern int generic_update_time(struct inode *, struct timespec *, int);
2024 
2025 /* /sys/fs */
2026 extern struct kobject *fs_kobj;
2027 
2028 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2029 
2030 #define FLOCK_VERIFY_READ  1
2031 #define FLOCK_VERIFY_WRITE 2
2032 
2033 #ifdef CONFIG_FILE_LOCKING
2034 extern int locks_mandatory_locked(struct file *);
2035 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2036 
2037 /*
2038  * Candidates for mandatory locking have the setgid bit set
2039  * but no group execute bit -  an otherwise meaningless combination.
2040  */
2041 
2042 static inline int __mandatory_lock(struct inode *ino)
2043 {
2044         return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2045 }
2046 
2047 /*
2048  * ... and these candidates should be on MS_MANDLOCK mounted fs,
2049  * otherwise these will be advisory locks
2050  */
2051 
2052 static inline int mandatory_lock(struct inode *ino)
2053 {
2054         return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2055 }
2056 
2057 static inline int locks_verify_locked(struct file *file)
2058 {
2059         if (mandatory_lock(file_inode(file)))
2060                 return locks_mandatory_locked(file);
2061         return 0;
2062 }
2063 
2064 static inline int locks_verify_truncate(struct inode *inode,
2065                                     struct file *filp,
2066                                     loff_t size)
2067 {
2068         if (inode->i_flctx && mandatory_lock(inode))
2069                 return locks_mandatory_area(
2070                         FLOCK_VERIFY_WRITE, inode, filp,
2071                         size < inode->i_size ? size : inode->i_size,
2072                         (size < inode->i_size ? inode->i_size - size
2073                          : size - inode->i_size)
2074                 );
2075         return 0;
2076 }
2077 
2078 static inline int break_lease(struct inode *inode, unsigned int mode)
2079 {
2080         /*
2081          * Since this check is lockless, we must ensure that any refcounts
2082          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2083          * could end up racing with tasks trying to set a new lease on this
2084          * file.
2085          */
2086         smp_mb();
2087         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2088                 return __break_lease(inode, mode, FL_LEASE);
2089         return 0;
2090 }
2091 
2092 static inline int break_deleg(struct inode *inode, unsigned int mode)
2093 {
2094         /*
2095          * Since this check is lockless, we must ensure that any refcounts
2096          * taken are done before checking i_flctx->flc_lease. Otherwise, we
2097          * could end up racing with tasks trying to set a new lease on this
2098          * file.
2099          */
2100         smp_mb();
2101         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2102                 return __break_lease(inode, mode, FL_DELEG);
2103         return 0;
2104 }
2105 
2106 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2107 {
2108         int ret;
2109 
2110         ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2111         if (ret == -EWOULDBLOCK && delegated_inode) {
2112                 *delegated_inode = inode;
2113                 ihold(inode);
2114         }
2115         return ret;
2116 }
2117 
2118 static inline int break_deleg_wait(struct inode **delegated_inode)
2119 {
2120         int ret;
2121 
2122         ret = break_deleg(*delegated_inode, O_WRONLY);
2123         iput(*delegated_inode);
2124         *delegated_inode = NULL;
2125         return ret;
2126 }
2127 
2128 static inline int break_layout(struct inode *inode, bool wait)
2129 {
2130         smp_mb();
2131         if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2132                 return __break_lease(inode,
2133                                 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2134                                 FL_LAYOUT);
2135         return 0;
2136 }
2137 
2138 #else /* !CONFIG_FILE_LOCKING */
2139 static inline int locks_mandatory_locked(struct file *file)
2140 {
2141         return 0;
2142 }
2143 
2144 static inline int locks_mandatory_area(int rw, struct inode *inode,
2145                                        struct file *filp, loff_t offset,
2146                                        size_t count)
2147 {
2148         return 0;
2149 }
2150 
2151 static inline int __mandatory_lock(struct inode *inode)
2152 {
2153         return 0;
2154 }
2155 
2156 static inline int mandatory_lock(struct inode *inode)
2157 {
2158         return 0;
2159 }
2160 
2161 static inline int locks_verify_locked(struct file *file)
2162 {
2163         return 0;
2164 }
2165 
2166 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2167                                         size_t size)
2168 {
2169         return 0;
2170 }
2171 
2172 static inline int break_lease(struct inode *inode, unsigned int mode)
2173 {
2174         return 0;
2175 }
2176 
2177 static inline int break_deleg(struct inode *inode, unsigned int mode)
2178 {
2179         return 0;
2180 }
2181 
2182 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2183 {
2184         return 0;
2185 }
2186 
2187 static inline int break_deleg_wait(struct inode **delegated_inode)
2188 {
2189         BUG();
2190         return 0;
2191 }
2192 
2193 static inline int break_layout(struct inode *inode, bool wait)
2194 {
2195         return 0;
2196 }
2197 
2198 #endif /* CONFIG_FILE_LOCKING */
2199 
2200 /* fs/open.c */
2201 struct audit_names;
2202 struct filename {
2203         const char              *name;  /* pointer to actual string */
2204         const __user char       *uptr;  /* original userland pointer */
2205         struct audit_names      *aname;
2206         int                     refcnt;
2207         const char              iname[];
2208 };
2209 
2210 extern long vfs_truncate(struct path *, loff_t);
2211 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2212                        struct file *filp);
2213 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2214                         loff_t len);
2215 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2216                         umode_t mode);
2217 extern struct file *file_open_name(struct filename *, int, umode_t);
2218 extern struct file *filp_open(const char *, int, umode_t);
2219 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2220                                    const char *, int);
2221 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2222 extern int filp_close(struct file *, fl_owner_t id);
2223 
2224 extern struct filename *getname_flags(const char __user *, int, int *);
2225 extern struct filename *getname(const char __user *);
2226 extern struct filename *getname_kernel(const char *);
2227 extern void putname(struct filename *name);
2228 
2229 enum {
2230         FILE_CREATED = 1,
2231         FILE_OPENED = 2
2232 };
2233 extern int finish_open(struct file *file, struct dentry *dentry,
2234                         int (*open)(struct inode *, struct file *),
2235                         int *opened);
2236 extern int finish_no_open(struct file *file, struct dentry *dentry);
2237 
2238 /* fs/ioctl.c */
2239 
2240 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2241 
2242 /* fs/dcache.c */
2243 extern void __init vfs_caches_init_early(void);
2244 extern void __init vfs_caches_init(void);
2245 
2246 extern struct kmem_cache *names_cachep;
2247 
2248 #define __getname()             kmem_cache_alloc(names_cachep, GFP_KERNEL)
2249 #define __putname(name)         kmem_cache_free(names_cachep, (void *)(name))
2250 
2251 #ifdef CONFIG_BLOCK
2252 extern int register_blkdev(unsigned int, const char *);
2253 extern void unregister_blkdev(unsigned int, const char *);
2254 extern struct block_device *bdget(dev_t);
2255 extern struct block_device *bdgrab(struct block_device *bdev);
2256 extern void bd_set_size(struct block_device *, loff_t size);
2257 extern void bd_forget(struct inode *inode);
2258 extern void bdput(struct block_device *);
2259 extern void invalidate_bdev(struct block_device *);
2260 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2261 extern int sync_blockdev(struct block_device *bdev);
2262 extern void kill_bdev(struct block_device *);
2263 extern struct super_block *freeze_bdev(struct block_device *);
2264 extern void emergency_thaw_all(void);
2265 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2266 extern int fsync_bdev(struct block_device *);
2267 
2268 extern struct super_block *blockdev_superblock;
2269 
2270 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2271 {
2272         return sb == blockdev_superblock;
2273 }
2274 #else
2275 static inline void bd_forget(struct inode *inode) {}
2276 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2277 static inline void kill_bdev(struct block_device *bdev) {}
2278 static inline void invalidate_bdev(struct block_device *bdev) {}
2279 
2280 static inline struct super_block *freeze_bdev(struct block_device *sb)
2281 {
2282         return NULL;
2283 }
2284 
2285 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2286 {
2287         return 0;
2288 }
2289 
2290 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2291 {
2292 }
2293 
2294 static inline int sb_is_blkdev_sb(struct super_block *sb)
2295 {
2296         return 0;
2297 }
2298 #endif
2299 extern int sync_filesystem(struct super_block *);
2300 extern const struct file_operations def_blk_fops;
2301 extern const struct file_operations def_chr_fops;
2302 #ifdef CONFIG_BLOCK
2303 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2304 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2305 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2306 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2307 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2308                                                void *holder);
2309 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2310                                               void *holder);
2311 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2312 extern int __blkdev_reread_part(struct block_device *bdev);
2313 extern int blkdev_reread_part(struct block_device *bdev);
2314 
2315 #ifdef CONFIG_SYSFS
2316 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2317 extern void bd_unlink_disk_holder(struct block_device *bdev,
2318                                   struct gendisk *disk);
2319 #else
2320 static inline int bd_link_disk_holder(struct block_device *bdev,
2321                                       struct gendisk *disk)
2322 {
2323         return 0;
2324 }
2325 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2326                                          struct gendisk *disk)
2327 {
2328 }
2329 #endif
2330 #endif
2331 
2332 /* fs/char_dev.c */
2333 #define CHRDEV_MAJOR_HASH_SIZE  255
2334 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2335 extern int register_chrdev_region(dev_t, unsigned, const char *);
2336 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2337                              unsigned int count, const char *name,
2338                              const struct file_operations *fops);
2339 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2340                                 unsigned int count, const char *name);
2341 extern void unregister_chrdev_region(dev_t, unsigned);
2342 extern void chrdev_show(struct seq_file *,off_t);
2343 
2344 static inline int register_chrdev(unsigned int major, const char *name,
2345                                   const struct file_operations *fops)
2346 {
2347         return __register_chrdev(major, 0, 256, name, fops);
2348 }
2349 
2350 static inline void unregister_chrdev(unsigned int major, const char *name)
2351 {
2352         __unregister_chrdev(major, 0, 256, name);
2353 }
2354 
2355 /* fs/block_dev.c */
2356 #define BDEVNAME_SIZE   32      /* Largest string for a blockdev identifier */
2357 #define BDEVT_SIZE      10      /* Largest string for MAJ:MIN for blkdev */
2358 
2359 #ifdef CONFIG_BLOCK
2360 #define BLKDEV_MAJOR_HASH_SIZE  255
2361 extern const char *__bdevname(dev_t, char *buffer);
2362 extern const char *bdevname(struct block_device *bdev, char *buffer);
2363 extern struct block_device *lookup_bdev(const char *);
2364 extern void blkdev_show(struct seq_file *,off_t);
2365 
2366 #else
2367 #define BLKDEV_MAJOR_HASH_SIZE  0
2368 #endif
2369 
2370 extern void init_special_inode(struct inode *, umode_t, dev_t);
2371 
2372 /* Invalid inode operations -- fs/bad_inode.c */
2373 extern void make_bad_inode(struct inode *);
2374 extern int is_bad_inode(struct inode *);
2375 
2376 #ifdef CONFIG_BLOCK
2377 /*
2378  * return READ, READA, or WRITE
2379  */
2380 #define bio_rw(bio)             ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2381 
2382 /*
2383  * return data direction, READ or WRITE
2384  */
2385 #define bio_data_dir(bio)       ((bio)->bi_rw & 1)
2386 
2387 extern void check_disk_size_change(struct gendisk *disk,
2388                                    struct block_device *bdev);
2389 extern int revalidate_disk(struct gendisk *);
2390 extern int check_disk_change(struct block_device *);
2391 extern int __invalidate_device(struct block_device *, bool);
2392 extern int invalidate_partition(struct gendisk *, int);
2393 #endif
2394 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2395                                         pgoff_t start, pgoff_t end);
2396 
2397 static inline void invalidate_remote_inode(struct inode *inode)
2398 {
2399         if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2400             S_ISLNK(inode->i_mode))
2401                 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2402 }
2403 extern int invalidate_inode_pages2(struct address_space *mapping);
2404 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2405                                          pgoff_t start, pgoff_t end);
2406 extern int write_inode_now(struct inode *, int);
2407 extern int filemap_fdatawrite(struct address_space *);
2408 extern int filemap_flush(struct address_space *);
2409 extern int filemap_fdatawait(struct address_space *);
2410 extern void filemap_fdatawait_keep_errors(struct address_space *);
2411 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2412                                    loff_t lend);
2413 extern int filemap_write_and_wait(struct address_space *mapping);
2414 extern int filemap_write_and_wait_range(struct address_space *mapping,
2415                                         loff_t lstart, loff_t lend);
2416 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2417                                 loff_t start, loff_t end, int sync_mode);
2418 extern int filemap_fdatawrite_range(struct address_space *mapping,
2419                                 loff_t start, loff_t end);
2420 
2421 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2422                            int datasync);
2423 extern int vfs_fsync(struct file *file, int datasync);
2424 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2425 {
2426         if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2427                 return 0;
2428         return vfs_fsync_range(file, pos, pos + count - 1,
2429                                (file->f_flags & __O_SYNC) ? 0 : 1);
2430 }
2431 extern void emergency_sync(void);
2432 extern void emergency_remount(void);
2433 #ifdef CONFIG_BLOCK
2434 extern sector_t bmap(struct inode *, sector_t);
2435 #endif
2436 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2437 extern int inode_permission(struct inode *, int);
2438 extern int __inode_permission(struct inode *, int);
2439 extern int generic_permission(struct inode *, int);
2440 extern int __check_sticky(struct inode *dir, struct inode *inode);
2441 
2442 static inline bool execute_ok(struct inode *inode)
2443 {
2444         return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2445 }
2446 
2447 static inline void file_start_write(struct file *file)
2448 {
2449         if (!S_ISREG(file_inode(file)->i_mode))
2450                 return;
2451         __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2452 }
2453 
2454 static inline bool file_start_write_trylock(struct file *file)
2455 {
2456         if (!S_ISREG(file_inode(file)->i_mode))
2457                 return true;
2458         return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2459 }
2460 
2461 static inline void file_end_write(struct file *file)
2462 {
2463         if (!S_ISREG(file_inode(file)->i_mode))
2464                 return;
2465         __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2466 }
2467 
2468 /*
2469  * get_write_access() gets write permission for a file.
2470  * put_write_access() releases this write permission.
2471  * This is used for regular files.
2472  * We cannot support write (and maybe mmap read-write shared) accesses and
2473  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2474  * can have the following values:
2475  * 0: no writers, no VM_DENYWRITE mappings
2476  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2477  * > 0: (i_writecount) users are writing to the file.
2478  *
2479  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2480  * except for the cases where we don't hold i_writecount yet. Then we need to
2481  * use {get,deny}_write_access() - these functions check the sign and refuse
2482  * to do the change if sign is wrong.
2483  */
2484 static inline int get_write_access(struct inode *inode)
2485 {
2486         return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2487 }
2488 static inline int deny_write_access(struct file *file)
2489 {
2490         struct inode *inode = file_inode(file);
2491         return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2492 }
2493 static inline void put_write_access(struct inode * inode)
2494 {
2495         atomic_dec(&inode->i_writecount);
2496 }
2497 static inline void allow_write_access(struct file *file)
2498 {
2499         if (file)
2500                 atomic_inc(&file_inode(file)->i_writecount);
2501 }
2502 static inline bool inode_is_open_for_write(const struct inode *inode)
2503 {
2504         return atomic_read(&inode->i_writecount) > 0;
2505 }
2506 
2507 #ifdef CONFIG_IMA
2508 static inline void i_readcount_dec(struct inode *inode)
2509 {
2510         BUG_ON(!atomic_read(&inode->i_readcount));
2511         atomic_dec(&inode->i_readcount);
2512 }
2513 static inline void i_readcount_inc(struct inode *inode)
2514 {
2515         atomic_inc(&inode->i_readcount);
2516 }
2517 #else
2518 static inline void i_readcount_dec(struct inode *inode)
2519 {
2520         return;
2521 }
2522 static inline void i_readcount_inc(struct inode *inode)
2523 {
2524         return;
2525 }
2526 #endif
2527 extern int do_pipe_flags(int *, int);
2528 
2529 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2530 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2531 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2532 extern struct file * open_exec(const char *);
2533  
2534 /* fs/dcache.c -- generic fs support functions */
2535 extern int is_subdir(struct dentry *, struct dentry *);
2536 extern int path_is_under(struct path *, struct path *);
2537 
2538 extern char *file_path(struct file *, char *, int);
2539 
2540 #include <linux/err.h>
2541 
2542 /* needed for stackable file system support */
2543 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2544 
2545 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2546 
2547 extern int inode_init_always(struct super_block *, struct inode *);
2548 extern void inode_init_once(struct inode *);
2549 extern void address_space_init_once(struct address_space *mapping);
2550 extern struct inode * igrab(struct inode *);
2551 extern ino_t iunique(struct super_block *, ino_t);
2552 extern int inode_needs_sync(struct inode *inode);
2553 extern int generic_delete_inode(struct inode *inode);
2554 static inline int generic_drop_inode(struct inode *inode)
2555 {
2556         return !inode->i_nlink || inode_unhashed(inode);
2557 }
2558 
2559 extern struct inode *ilookup5_nowait(struct super_block *sb,
2560                 unsigned long hashval, int (*test)(struct inode *, void *),
2561                 void *data);
2562 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2563                 int (*test)(struct inode *, void *), void *data);
2564 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2565 
2566 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2567 extern struct inode * iget_locked(struct super_block *, unsigned long);
2568 extern struct inode *find_inode_nowait(struct super_block *,
2569                                        unsigned long,
2570                                        int (*match)(struct inode *,
2571                                                     unsigned long, void *),
2572                                        void *data);
2573 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2574 extern int insert_inode_locked(struct inode *);
2575 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2576 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2577 #else
2578 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2579 #endif
2580 extern void unlock_new_inode(struct inode *);
2581 extern unsigned int get_next_ino(void);
2582 
2583 extern void __iget(struct inode * inode);
2584 extern void iget_failed(struct inode *);
2585 extern void clear_inode(struct inode *);
2586 extern void __destroy_inode(struct inode *);
2587 extern struct inode *new_inode_pseudo(struct super_block *sb);
2588 extern struct inode *new_inode(struct super_block *sb);
2589 extern void free_inode_nonrcu(struct inode *inode);
2590 extern int should_remove_suid(struct dentry *);
2591 extern int file_remove_privs(struct file *);
2592 extern int dentry_needs_remove_privs(struct dentry *dentry);
2593 static inline int file_needs_remove_privs(struct file *file)
2594 {
2595         return dentry_needs_remove_privs(file->f_path.dentry);
2596 }
2597 
2598 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2599 static inline void insert_inode_hash(struct inode *inode)
2600 {
2601         __insert_inode_hash(inode, inode->i_ino);
2602 }
2603 
2604 extern void __remove_inode_hash(struct inode *);
2605 static inline void remove_inode_hash(struct inode *inode)
2606 {
2607         if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2608                 __remove_inode_hash(inode);
2609 }
2610 
2611 extern void inode_sb_list_add(struct inode *inode);
2612 
2613 #ifdef CONFIG_BLOCK
2614 extern blk_qc_t submit_bio(int, struct bio *);
2615 extern int bdev_read_only(struct block_device *);
2616 #endif
2617 extern int set_blocksize(struct block_device *, int);
2618 extern int sb_set_blocksize(struct super_block *, int);
2619 extern int sb_min_blocksize(struct super_block *, int);
2620 
2621 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2622 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2623 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2624 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2625 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2626 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2627 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2628 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2629 
2630 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2631 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2632 
2633 /* fs/block_dev.c */
2634 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2635 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2636 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2637                         int datasync);
2638 extern void block_sync_page(struct page *page);
2639 
2640 /* fs/splice.c */
2641 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2642                 struct pipe_inode_info *, size_t, unsigned int);
2643 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2644                 struct pipe_inode_info *, size_t, unsigned int);
2645 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2646                 struct file *, loff_t *, size_t, unsigned int);
2647 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2648                 struct file *out, loff_t *, size_t len, unsigned int flags);
2649 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2650                 loff_t *opos, size_t len, unsigned int flags);
2651 
2652 
2653 extern void
2654 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2655 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2656 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2657 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2658 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2659 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2660                 int whence, loff_t maxsize, loff_t eof);
2661 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2662                 int whence, loff_t size);
2663 extern int generic_file_open(struct inode * inode, struct file * filp);
2664 extern int nonseekable_open(struct inode * inode, struct file * filp);
2665 
2666 #ifdef CONFIG_BLOCK
2667 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2668                             loff_t file_offset);
2669 
2670 enum {
2671         /* need locking between buffered and direct access */
2672         DIO_LOCKING     = 0x01,
2673 
2674         /* filesystem does not support filling holes */
2675         DIO_SKIP_HOLES  = 0x02,
2676 
2677         /* filesystem can handle aio writes beyond i_size */
2678         DIO_ASYNC_EXTEND = 0x04,
2679 
2680         /* inode/fs/bdev does not need truncate protection */
2681         DIO_SKIP_DIO_COUNT = 0x08,
2682 };
2683 
2684 void dio_end_io(struct bio *bio, int error);
2685 
2686 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2687                              struct block_device *bdev, struct iov_iter *iter,
2688                              loff_t offset, get_block_t get_block,
2689                              dio_iodone_t end_io, dio_submit_t submit_io,
2690                              int flags);
2691 
2692 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2693                                          struct inode *inode,
2694                                          struct iov_iter *iter, loff_t offset,
2695                                          get_block_t get_block)
2696 {
2697         return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2698                                     offset, get_block, NULL, NULL,
2699                                     DIO_LOCKING | DIO_SKIP_HOLES);
2700 }
2701 #endif
2702 
2703 void inode_dio_wait(struct inode *inode);
2704 
2705 /*
2706  * inode_dio_begin - signal start of a direct I/O requests
2707  * @inode: inode the direct I/O happens on
2708  *
2709  * This is called once we've finished processing a direct I/O request,
2710  * and is used to wake up callers waiting for direct I/O to be quiesced.
2711  */
2712 static inline void inode_dio_begin(struct inode *inode)
2713 {
2714         atomic_inc(&inode->i_dio_count);
2715 }
2716 
2717 /*
2718  * inode_dio_end - signal finish of a direct I/O requests
2719  * @inode: inode the direct I/O happens on
2720  *
2721  * This is called once we've finished processing a direct I/O request,
2722  * and is used to wake up callers waiting for direct I/O to be quiesced.
2723  */
2724 static inline void inode_dio_end(struct inode *inode)
2725 {
2726         if (atomic_dec_and_test(&inode->i_dio_count))
2727                 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2728 }
2729 
2730 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2731                             unsigned int mask);
2732 
2733 extern const struct file_operations generic_ro_fops;
2734 
2735 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2736 
2737 extern int readlink_copy(char __user *, int, const char *);
2738 extern int page_readlink(struct dentry *, char __user *, int);
2739 extern const char *page_follow_link_light(struct dentry *, void **);
2740 extern void page_put_link(struct inode *, void *);
2741 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2742                 int nofs);
2743 extern int page_symlink(struct inode *inode, const char *symname, int len);
2744 extern const struct inode_operations page_symlink_inode_operations;
2745 extern void kfree_put_link(struct inode *, void *);
2746 extern void free_page_put_link(struct inode *, void *);
2747 extern int generic_readlink(struct dentry *, char __user *, int);
2748 extern void generic_fillattr(struct inode *, struct kstat *);
2749 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2750 extern int vfs_getattr(struct path *, struct kstat *);
2751 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2752 void inode_add_bytes(struct inode *inode, loff_t bytes);
2753 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2754 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2755 loff_t inode_get_bytes(struct inode *inode);
2756 void inode_set_bytes(struct inode *inode, loff_t bytes);
2757 const char *simple_follow_link(struct dentry *, void **);
2758 extern const struct inode_operations simple_symlink_inode_operations;
2759 
2760 extern int iterate_dir(struct file *, struct dir_context *);
2761 
2762 extern int vfs_stat(const char __user *, struct kstat *);
2763 extern int vfs_lstat(const char __user *, struct kstat *);
2764 extern int vfs_fstat(unsigned int, struct kstat *);
2765 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2766 
2767 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2768                     unsigned long arg);
2769 extern int __generic_block_fiemap(struct inode *inode,
2770                                   struct fiemap_extent_info *fieinfo,
2771                                   loff_t start, loff_t len,
2772                                   get_block_t *get_block);
2773 extern int generic_block_fiemap(struct inode *inode,
2774                                 struct fiemap_extent_info *fieinfo, u64 start,
2775                                 u64 len, get_block_t *get_block);
2776 
2777 extern void get_filesystem(struct file_system_type *fs);
2778 extern void put_filesystem(struct file_system_type *fs);
2779 extern struct file_system_type *get_fs_type(const char *name);
2780 extern struct super_block *get_super(struct block_device *);
2781 extern struct super_block *get_super_thawed(struct block_device *);
2782 extern struct super_block *get_active_super(struct block_device *bdev);
2783 extern void drop_super(struct super_block *sb);
2784 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2785 extern void iterate_supers_type(struct file_system_type *,
2786                                 void (*)(struct super_block *, void *), void *);
2787 
2788 extern int dcache_dir_open(struct inode *, struct file *);
2789 extern int dcache_dir_close(struct inode *, struct file *);
2790 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2791 extern int dcache_readdir(struct file *, struct dir_context *);
2792 extern int simple_setattr(struct dentry *, struct iattr *);
2793 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2794 extern int simple_statfs(struct dentry *, struct kstatfs *);
2795 extern int simple_open(struct inode *inode, struct file *file);
2796 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2797 extern int simple_unlink(struct inode *, struct dentry *);
2798 extern int simple_rmdir(struct inode *, struct dentry *);
2799 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2800 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2801 extern int simple_empty(struct dentry *);
2802 extern int simple_readpage(struct file *file, struct page *page);
2803 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2804                         loff_t pos, unsigned len, unsigned flags,
2805                         struct page **pagep, void **fsdata);
2806 extern int simple_write_end(struct file *file, struct address_space *mapping,
2807                         loff_t pos, unsigned len, unsigned copied,
2808                         struct page *page, void *fsdata);
2809 extern int always_delete_dentry(const struct dentry *);
2810 extern struct inode *alloc_anon_inode(struct super_block *);
2811 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2812 extern const struct dentry_operations simple_dentry_operations;
2813 
2814 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2815 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2816 extern const struct file_operations simple_dir_operations;
2817 extern const struct inode_operations simple_dir_inode_operations;
2818 extern void make_empty_dir_inode(struct inode *inode);
2819 extern bool is_empty_dir_inode(struct inode *inode);
2820 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2821 struct dentry *d_alloc_name(struct dentry *, const char *);
2822 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2823 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2824 extern void simple_release_fs(struct vfsmount **mount, int *count);
2825 
2826 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2827                         loff_t *ppos, const void *from, size_t available);
2828 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2829                 const void __user *from, size_t count);
2830 
2831 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2832 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2833 
2834 extern int generic_check_addressable(unsigned, u64);
2835 
2836 #ifdef CONFIG_MIGRATION
2837 extern int buffer_migrate_page(struct address_space *,
2838                                 struct page *, struct page *,
2839                                 enum migrate_mode);
2840 #else
2841 #define buffer_migrate_page NULL
2842 #endif
2843 
2844 extern int inode_change_ok(const struct inode *, struct iattr *);
2845 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2846 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2847 
2848 extern int file_update_time(struct file *file);
2849 
2850 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2851 extern void save_mount_options(struct super_block *sb, char *options);
2852 extern void replace_mount_options(struct super_block *sb, char *options);
2853 
2854 static inline bool io_is_direct(struct file *filp)
2855 {
2856         return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2857 }
2858 
2859 static inline int iocb_flags(struct file *file)
2860 {
2861         int res = 0;
2862         if (file->f_flags & O_APPEND)
2863                 res |= IOCB_APPEND;
2864         if (io_is_direct(file))
2865                 res |= IOCB_DIRECT;
2866         return res;
2867 }
2868 
2869 static inline ino_t parent_ino(struct dentry *dentry)
2870 {
2871         ino_t res;
2872 
2873         /*
2874          * Don't strictly need d_lock here? If the parent ino could change
2875          * then surely we'd have a deeper race in the caller?
2876          */
2877         spin_lock(&dentry->d_lock);
2878         res = dentry->d_parent->d_inode->i_ino;
2879         spin_unlock(&dentry->d_lock);
2880         return res;
2881 }
2882 
2883 /* Transaction based IO helpers */
2884 
2885 /*
2886  * An argresp is stored in an allocated page and holds the
2887  * size of the argument or response, along with its content
2888  */
2889 struct simple_transaction_argresp {
2890         ssize_t size;
2891         char data[0];
2892 };
2893 
2894 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2895 
2896 char *simple_transaction_get(struct file *file, const char __user *buf,
2897                                 size_t size);
2898 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2899                                 size_t size, loff_t *pos);
2900 int simple_transaction_release(struct inode *inode, struct file *file);
2901 
2902 void simple_transaction_set(struct file *file, size_t n);
2903 
2904 /*
2905  * simple attribute files
2906  *
2907  * These attributes behave similar to those in sysfs:
2908  *
2909  * Writing to an attribute immediately sets a value, an open file can be
2910  * written to multiple times.
2911  *
2912  * Reading from an attribute creates a buffer from the value that might get
2913  * read with multiple read calls. When the attribute has been read
2914  * completely, no further read calls are possible until the file is opened
2915  * again.
2916  *
2917  * All attributes contain a text representation of a numeric value
2918  * that are accessed with the get() and set() functions.
2919  */
2920 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)            \
2921 static int __fops ## _open(struct inode *inode, struct file *file)      \
2922 {                                                                       \
2923         __simple_attr_check_format(__fmt, 0ull);                        \
2924         return simple_attr_open(inode, file, __get, __set, __fmt);      \
2925 }                                                                       \
2926 static const struct file_operations __fops = {                          \
2927         .owner   = THIS_MODULE,                                         \
2928         .open    = __fops ## _open,                                     \
2929         .release = simple_attr_release,                                 \
2930         .read    = simple_attr_read,                                    \
2931         .write   = simple_attr_write,                                   \
2932         .llseek  = generic_file_llseek,                                 \
2933 }
2934 
2935 static inline __printf(1, 2)
2936 void __simple_attr_check_format(const char *fmt, ...)
2937 {
2938         /* don't do anything, just let the compiler check the arguments; */
2939 }
2940 
2941 int simple_attr_open(struct inode *inode, struct file *file,
2942                      int (*get)(void *, u64 *), int (*set)(void *, u64),
2943                      const char *fmt);
2944 int simple_attr_release(struct inode *inode, struct file *file);
2945 ssize_t simple_attr_read(struct file *file, char __user *buf,
2946                          size_t len, loff_t *ppos);
2947 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2948                           size_t len, loff_t *ppos);
2949 
2950 struct ctl_table;
2951 int proc_nr_files(struct ctl_table *table, int write,
2952                   void __user *buffer, size_t *lenp, loff_t *ppos);
2953 int proc_nr_dentry(struct ctl_table *table, int write,
2954                   void __user *buffer, size_t *lenp, loff_t *ppos);
2955 int proc_nr_inodes(struct ctl_table *table, int write,
2956                    void __user *buffer, size_t *lenp, loff_t *ppos);
2957 int __init get_filesystem_list(char *buf);
2958 
2959 #define __FMODE_EXEC            ((__force int) FMODE_EXEC)
2960 #define __FMODE_NONOTIFY        ((__force int) FMODE_NONOTIFY)
2961 
2962 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2963 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2964                                             (flag & __FMODE_NONOTIFY)))
2965 
2966 static inline int is_sxid(umode_t mode)
2967 {
2968         return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2969 }
2970 
2971 static inline int check_sticky(struct inode *dir, struct inode *inode)
2972 {
2973         if (!(dir->i_mode & S_ISVTX))
2974                 return 0;
2975 
2976         return __check_sticky(dir, inode);
2977 }
2978 
2979 static inline void inode_has_no_xattr(struct inode *inode)
2980 {
2981         if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2982                 inode->i_flags |= S_NOSEC;
2983 }
2984 
2985 static inline bool is_root_inode(struct inode *inode)
2986 {
2987         return inode == inode->i_sb->s_root->d_inode;
2988 }
2989 
2990 static inline bool dir_emit(struct dir_context *ctx,
2991                             const char *name, int namelen,
2992                             u64 ino, unsigned type)
2993 {
2994         return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2995 }
2996 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2997 {
2998         return ctx->actor(ctx, ".", 1, ctx->pos,
2999                           file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3000 }
3001 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3002 {
3003         return ctx->actor(ctx, "..", 2, ctx->pos,
3004                           parent_ino(file->f_path.dentry), DT_DIR) == 0;
3005 }
3006 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3007 {
3008         if (ctx->pos == 0) {
3009                 if (!dir_emit_dot(file, ctx))
3010                         return false;
3011                 ctx->pos = 1;
3012         }
3013         if (ctx->pos == 1) {
3014                 if (!dir_emit_dotdot(file, ctx))
3015                         return false;
3016                 ctx->pos = 2;
3017         }
3018         return true;
3019 }
3020 static inline bool dir_relax(struct inode *inode)
3021 {
3022         mutex_unlock(&inode->i_mutex);
3023         mutex_lock(&inode->i_mutex);
3024         return !IS_DEADDIR(inode);
3025 }
3026 
3027 extern bool path_noexec(const struct path *path);
3028 
3029 #endif /* _LINUX_FS_H */
3030 

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