5 #include <linux/linkage.h>
6 #include <linux/wait.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/rwsem.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
34 #include <asm/byteorder.h>
35 #include <uapi/linux/fs.h>
37 struct backing_dev_info;
39 struct export_operations;
45 struct pipe_inode_info;
46 struct poll_table_struct;
48 struct vm_area_struct;
51 struct swap_info_struct;
53 struct workqueue_struct;
57 extern void __init inode_init(void);
58 extern void __init inode_init_early(void);
59 extern void __init files_init(unsigned long);
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;
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);
75 #define MAY_EXEC 0x00000001
76 #define MAY_WRITE 0x00000002
77 #define MAY_READ 0x00000004
78 #define MAY_APPEND 0x00000008
79 #define MAY_ACCESS 0x00000010
80 #define MAY_OPEN 0x00000020
81 #define MAY_CHDIR 0x00000040
82 /* called from RCU mode, don't block */
83 #define MAY_NOT_BLOCK 0x00000080
86 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
87 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
90 /* file is open for reading */
91 #define FMODE_READ ((__force fmode_t)0x1)
92 /* file is open for writing */
93 #define FMODE_WRITE ((__force fmode_t)0x2)
94 /* file is seekable */
95 #define FMODE_LSEEK ((__force fmode_t)0x4)
96 /* file can be accessed using pread */
97 #define FMODE_PREAD ((__force fmode_t)0x8)
98 /* file can be accessed using pwrite */
99 #define FMODE_PWRITE ((__force fmode_t)0x10)
100 /* File is opened for execution with sys_execve / sys_uselib */
101 #define FMODE_EXEC ((__force fmode_t)0x20)
102 /* File is opened with O_NDELAY (only set for block devices) */
103 #define FMODE_NDELAY ((__force fmode_t)0x40)
104 /* File is opened with O_EXCL (only set for block devices) */
105 #define FMODE_EXCL ((__force fmode_t)0x80)
106 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
107 (specialy hack for floppy.c) */
108 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
109 /* 32bit hashes as llseek() offset (for directories) */
110 #define FMODE_32BITHASH ((__force fmode_t)0x200)
111 /* 64bit hashes as llseek() offset (for directories) */
112 #define FMODE_64BITHASH ((__force fmode_t)0x400)
115 * Don't update ctime and mtime.
117 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
118 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
120 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
122 /* Expect random access pattern */
123 #define FMODE_RANDOM ((__force fmode_t)0x1000)
125 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
126 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
128 /* File is opened with O_PATH; almost nothing can be done with it */
129 #define FMODE_PATH ((__force fmode_t)0x4000)
131 /* File needs atomic accesses to f_pos */
132 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
133 /* Write access to underlying fs */
134 #define FMODE_WRITER ((__force fmode_t)0x10000)
135 /* Has read method(s) */
136 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
137 /* Has write method(s) */
138 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
140 /* File was opened by fanotify and shouldn't generate fanotify events */
141 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
144 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
145 * that indicates that they should check the contents of the iovec are
146 * valid, but not check the memory that the iovec elements
149 #define CHECK_IOVEC_ONLY -1
152 * The below are the various read and write types that we support. Some of
153 * them include behavioral modifiers that send information down to the
154 * block layer and IO scheduler. Terminology:
156 * The block layer uses device plugging to defer IO a little bit, in
157 * the hope that we will see more IO very shortly. This increases
158 * coalescing of adjacent IO and thus reduces the number of IOs we
159 * have to send to the device. It also allows for better queuing,
160 * if the IO isn't mergeable. If the caller is going to be waiting
161 * for the IO, then he must ensure that the device is unplugged so
162 * that the IO is dispatched to the driver.
164 * All IO is handled async in Linux. This is fine for background
165 * writes, but for reads or writes that someone waits for completion
166 * on, we want to notify the block layer and IO scheduler so that they
167 * know about it. That allows them to make better scheduling
168 * decisions. So when the below references 'sync' and 'async', it
169 * is referencing this priority hint.
171 * With that in mind, the available types are:
173 * READ A normal read operation. Device will be plugged.
174 * READ_SYNC A synchronous read. Device is not plugged, caller can
175 * immediately wait on this read without caring about
177 * READA Used for read-ahead operations. Lower priority, and the
178 * block layer could (in theory) choose to ignore this
179 * request if it runs into resource problems.
180 * WRITE A normal async write. Device will be plugged.
181 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
182 * the hint that someone will be waiting on this IO
183 * shortly. The write equivalent of READ_SYNC.
184 * WRITE_ODIRECT Special case write for O_DIRECT only.
185 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
186 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
187 * non-volatile media on completion.
188 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
189 * by a cache flush and data is guaranteed to be on
190 * non-volatile media on completion.
193 #define RW_MASK REQ_WRITE
194 #define RWA_MASK REQ_RAHEAD
197 #define WRITE RW_MASK
198 #define READA RWA_MASK
200 #define READ_SYNC (READ | REQ_SYNC)
201 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
202 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
203 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
204 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
205 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
208 * Attribute flags. These should be or-ed together to figure out what
211 #define ATTR_MODE (1 << 0)
212 #define ATTR_UID (1 << 1)
213 #define ATTR_GID (1 << 2)
214 #define ATTR_SIZE (1 << 3)
215 #define ATTR_ATIME (1 << 4)
216 #define ATTR_MTIME (1 << 5)
217 #define ATTR_CTIME (1 << 6)
218 #define ATTR_ATIME_SET (1 << 7)
219 #define ATTR_MTIME_SET (1 << 8)
220 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
221 #define ATTR_ATTR_FLAG (1 << 10)
222 #define ATTR_KILL_SUID (1 << 11)
223 #define ATTR_KILL_SGID (1 << 12)
224 #define ATTR_FILE (1 << 13)
225 #define ATTR_KILL_PRIV (1 << 14)
226 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
227 #define ATTR_TIMES_SET (1 << 16)
230 * Whiteout is represented by a char device. The following constants define the
231 * mode and device number to use.
233 #define WHITEOUT_MODE 0
234 #define WHITEOUT_DEV 0
237 * This is the Inode Attributes structure, used for notify_change(). It
238 * uses the above definitions as flags, to know which values have changed.
239 * Also, in this manner, a Filesystem can look at only the values it cares
240 * about. Basically, these are the attributes that the VFS layer can
241 * request to change from the FS layer.
243 * Derek Atkins <warlord@MIT.EDU> 94-10-20
246 unsigned int ia_valid;
251 struct timespec ia_atime;
252 struct timespec ia_mtime;
253 struct timespec ia_ctime;
256 * Not an attribute, but an auxiliary info for filesystems wanting to
257 * implement an ftruncate() like method. NOTE: filesystem should
258 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
260 struct file *ia_file;
264 * Includes for diskquotas.
266 #include <linux/quota.h>
269 * Maximum number of layers of fs stack. Needs to be limited to
270 * prevent kernel stack overflow
272 #define FILESYSTEM_MAX_STACK_DEPTH 2
275 * enum positive_aop_returns - aop return codes with specific semantics
277 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
278 * completed, that the page is still locked, and
279 * should be considered active. The VM uses this hint
280 * to return the page to the active list -- it won't
281 * be a candidate for writeback again in the near
282 * future. Other callers must be careful to unlock
283 * the page if they get this return. Returned by
286 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
287 * unlocked it and the page might have been truncated.
288 * The caller should back up to acquiring a new page and
289 * trying again. The aop will be taking reasonable
290 * precautions not to livelock. If the caller held a page
291 * reference, it should drop it before retrying. Returned
294 * address_space_operation functions return these large constants to indicate
295 * special semantics to the caller. These are much larger than the bytes in a
296 * page to allow for functions that return the number of bytes operated on in a
300 enum positive_aop_returns {
301 AOP_WRITEPAGE_ACTIVATE = 0x80000,
302 AOP_TRUNCATED_PAGE = 0x80001,
305 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
306 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
307 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
308 * helper code (eg buffer layer)
309 * to clear GFP_FS from alloc */
312 * oh the beauties of C type declarations.
315 struct address_space;
316 struct writeback_control;
318 #define IOCB_EVENTFD (1 << 0)
319 #define IOCB_APPEND (1 << 1)
320 #define IOCB_DIRECT (1 << 2)
323 struct file *ki_filp;
325 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
330 static inline bool is_sync_kiocb(struct kiocb *kiocb)
332 return kiocb->ki_complete == NULL;
335 static inline int iocb_flags(struct file *file);
337 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
339 *kiocb = (struct kiocb) {
341 .ki_flags = iocb_flags(filp),
346 * "descriptor" for what we're up to with a read.
347 * This allows us to use the same read code yet
348 * have multiple different users of the data that
349 * we read from a file.
351 * The simplest case just copies the data to user
364 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
365 unsigned long, unsigned long);
367 struct address_space_operations {
368 int (*writepage)(struct page *page, struct writeback_control *wbc);
369 int (*readpage)(struct file *, struct page *);
371 /* Write back some dirty pages from this mapping. */
372 int (*writepages)(struct address_space *, struct writeback_control *);
374 /* Set a page dirty. Return true if this dirtied it */
375 int (*set_page_dirty)(struct page *page);
377 int (*readpages)(struct file *filp, struct address_space *mapping,
378 struct list_head *pages, unsigned nr_pages);
380 int (*write_begin)(struct file *, struct address_space *mapping,
381 loff_t pos, unsigned len, unsigned flags,
382 struct page **pagep, void **fsdata);
383 int (*write_end)(struct file *, struct address_space *mapping,
384 loff_t pos, unsigned len, unsigned copied,
385 struct page *page, void *fsdata);
387 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
388 sector_t (*bmap)(struct address_space *, sector_t);
389 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
390 int (*releasepage) (struct page *, gfp_t);
391 void (*freepage)(struct page *);
392 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
394 * migrate the contents of a page to the specified target. If
395 * migrate_mode is MIGRATE_ASYNC, it must not block.
397 int (*migratepage) (struct address_space *,
398 struct page *, struct page *, enum migrate_mode);
399 int (*launder_page) (struct page *);
400 int (*is_partially_uptodate) (struct page *, unsigned long,
402 void (*is_dirty_writeback) (struct page *, bool *, bool *);
403 int (*error_remove_page)(struct address_space *, struct page *);
405 /* swapfile support */
406 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
408 void (*swap_deactivate)(struct file *file);
411 extern const struct address_space_operations empty_aops;
414 * pagecache_write_begin/pagecache_write_end must be used by general code
415 * to write into the pagecache.
417 int pagecache_write_begin(struct file *, struct address_space *mapping,
418 loff_t pos, unsigned len, unsigned flags,
419 struct page **pagep, void **fsdata);
421 int pagecache_write_end(struct file *, struct address_space *mapping,
422 loff_t pos, unsigned len, unsigned copied,
423 struct page *page, void *fsdata);
425 struct address_space {
426 struct inode *host; /* owner: inode, block_device */
427 struct radix_tree_root page_tree; /* radix tree of all pages */
428 spinlock_t tree_lock; /* and lock protecting it */
429 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
430 struct rb_root i_mmap; /* tree of private and shared mappings */
431 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
432 /* Protected by tree_lock together with the radix tree */
433 unsigned long nrpages; /* number of total pages */
434 unsigned long nrshadows; /* number of shadow entries */
435 pgoff_t writeback_index;/* writeback starts here */
436 const struct address_space_operations *a_ops; /* methods */
437 unsigned long flags; /* error bits/gfp mask */
438 spinlock_t private_lock; /* for use by the address_space */
439 struct list_head private_list; /* ditto */
440 void *private_data; /* ditto */
441 } __attribute__((aligned(sizeof(long))));
443 * On most architectures that alignment is already the case; but
444 * must be enforced here for CRIS, to let the least significant bit
445 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
447 struct request_queue;
449 struct block_device {
450 dev_t bd_dev; /* not a kdev_t - it's a search key */
452 struct inode * bd_inode; /* will die */
453 struct super_block * bd_super;
454 struct mutex bd_mutex; /* open/close mutex */
455 struct list_head bd_inodes;
459 bool bd_write_holder;
461 struct list_head bd_holder_disks;
463 struct block_device * bd_contains;
464 unsigned bd_block_size;
465 struct hd_struct * bd_part;
466 /* number of times partitions within this device have been opened. */
467 unsigned bd_part_count;
469 struct gendisk * bd_disk;
470 struct request_queue * bd_queue;
471 struct list_head bd_list;
473 * Private data. You must have bd_claim'ed the block_device
474 * to use this. NOTE: bd_claim allows an owner to claim
475 * the same device multiple times, the owner must take special
476 * care to not mess up bd_private for that case.
478 unsigned long bd_private;
480 /* The counter of freeze processes */
481 int bd_fsfreeze_count;
482 /* Mutex for freeze */
483 struct mutex bd_fsfreeze_mutex;
487 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
490 #define PAGECACHE_TAG_DIRTY 0
491 #define PAGECACHE_TAG_WRITEBACK 1
492 #define PAGECACHE_TAG_TOWRITE 2
494 int mapping_tagged(struct address_space *mapping, int tag);
496 static inline void i_mmap_lock_write(struct address_space *mapping)
498 down_write(&mapping->i_mmap_rwsem);
501 static inline void i_mmap_unlock_write(struct address_space *mapping)
503 up_write(&mapping->i_mmap_rwsem);
506 static inline void i_mmap_lock_read(struct address_space *mapping)
508 down_read(&mapping->i_mmap_rwsem);
511 static inline void i_mmap_unlock_read(struct address_space *mapping)
513 up_read(&mapping->i_mmap_rwsem);
517 * Might pages of this file be mapped into userspace?
519 static inline int mapping_mapped(struct address_space *mapping)
521 return !RB_EMPTY_ROOT(&mapping->i_mmap);
525 * Might pages of this file have been modified in userspace?
526 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
527 * marks vma as VM_SHARED if it is shared, and the file was opened for
528 * writing i.e. vma may be mprotected writable even if now readonly.
530 * If i_mmap_writable is negative, no new writable mappings are allowed. You
531 * can only deny writable mappings, if none exists right now.
533 static inline int mapping_writably_mapped(struct address_space *mapping)
535 return atomic_read(&mapping->i_mmap_writable) > 0;
538 static inline int mapping_map_writable(struct address_space *mapping)
540 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
544 static inline void mapping_unmap_writable(struct address_space *mapping)
546 atomic_dec(&mapping->i_mmap_writable);
549 static inline int mapping_deny_writable(struct address_space *mapping)
551 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
555 static inline void mapping_allow_writable(struct address_space *mapping)
557 atomic_inc(&mapping->i_mmap_writable);
561 * Use sequence counter to get consistent i_size on 32-bit processors.
563 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
564 #include <linux/seqlock.h>
565 #define __NEED_I_SIZE_ORDERED
566 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
568 #define i_size_ordered_init(inode) do { } while (0)
572 #define ACL_NOT_CACHED ((void *)(-1))
574 #define IOP_FASTPERM 0x0001
575 #define IOP_LOOKUP 0x0002
576 #define IOP_NOFOLLOW 0x0004
579 * Keep mostly read-only and often accessed (especially for
580 * the RCU path lookup and 'stat' data) fields at the beginning
581 * of the 'struct inode'
585 unsigned short i_opflags;
588 unsigned int i_flags;
590 #ifdef CONFIG_FS_POSIX_ACL
591 struct posix_acl *i_acl;
592 struct posix_acl *i_default_acl;
595 const struct inode_operations *i_op;
596 struct super_block *i_sb;
597 struct address_space *i_mapping;
599 #ifdef CONFIG_SECURITY
603 /* Stat data, not accessed from path walking */
606 * Filesystems may only read i_nlink directly. They shall use the
607 * following functions for modification:
609 * (set|clear|inc|drop)_nlink
610 * inode_(inc|dec)_link_count
613 const unsigned int i_nlink;
614 unsigned int __i_nlink;
618 struct timespec i_atime;
619 struct timespec i_mtime;
620 struct timespec i_ctime;
621 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
622 unsigned short i_bytes;
623 unsigned int i_blkbits;
626 #ifdef __NEED_I_SIZE_ORDERED
627 seqcount_t i_size_seqcount;
631 unsigned long i_state;
632 struct mutex i_mutex;
634 unsigned long dirtied_when; /* jiffies of first dirtying */
635 unsigned long dirtied_time_when;
637 struct hlist_node i_hash;
638 struct list_head i_wb_list; /* backing dev IO list */
639 #ifdef CONFIG_CGROUP_WRITEBACK
640 struct bdi_writeback *i_wb; /* the associated cgroup wb */
642 struct list_head i_lru; /* inode LRU list */
643 struct list_head i_sb_list;
645 struct hlist_head i_dentry;
646 struct rcu_head i_rcu;
650 atomic_t i_dio_count;
651 atomic_t i_writecount;
653 atomic_t i_readcount; /* struct files open RO */
655 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
656 struct file_lock_context *i_flctx;
657 struct address_space i_data;
658 struct list_head i_devices;
660 struct pipe_inode_info *i_pipe;
661 struct block_device *i_bdev;
667 #ifdef CONFIG_FSNOTIFY
668 __u32 i_fsnotify_mask; /* all events this inode cares about */
669 struct hlist_head i_fsnotify_marks;
672 void *i_private; /* fs or device private pointer */
675 static inline int inode_unhashed(struct inode *inode)
677 return hlist_unhashed(&inode->i_hash);
681 * inode->i_mutex nesting subclasses for the lock validator:
683 * 0: the object of the current VFS operation
687 * 4: second non-directory
688 * 5: second parent (when locking independent directories in rename)
690 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
691 * non-directories at once.
693 * The locking order between these classes is
694 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
696 enum inode_i_mutex_lock_class
706 void lock_two_nondirectories(struct inode *, struct inode*);
707 void unlock_two_nondirectories(struct inode *, struct inode*);
710 * NOTE: in a 32bit arch with a preemptable kernel and
711 * an UP compile the i_size_read/write must be atomic
712 * with respect to the local cpu (unlike with preempt disabled),
713 * but they don't need to be atomic with respect to other cpus like in
714 * true SMP (so they need either to either locally disable irq around
715 * the read or for example on x86 they can be still implemented as a
716 * cmpxchg8b without the need of the lock prefix). For SMP compiles
717 * and 64bit archs it makes no difference if preempt is enabled or not.
719 static inline loff_t i_size_read(const struct inode *inode)
721 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
726 seq = read_seqcount_begin(&inode->i_size_seqcount);
727 i_size = inode->i_size;
728 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
730 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
734 i_size = inode->i_size;
738 return inode->i_size;
743 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
744 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
745 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
747 static inline void i_size_write(struct inode *inode, loff_t i_size)
749 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
751 write_seqcount_begin(&inode->i_size_seqcount);
752 inode->i_size = i_size;
753 write_seqcount_end(&inode->i_size_seqcount);
755 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
757 inode->i_size = i_size;
760 inode->i_size = i_size;
764 /* Helper functions so that in most cases filesystems will
765 * not need to deal directly with kuid_t and kgid_t and can
766 * instead deal with the raw numeric values that are stored
769 static inline uid_t i_uid_read(const struct inode *inode)
771 return from_kuid(&init_user_ns, inode->i_uid);
774 static inline gid_t i_gid_read(const struct inode *inode)
776 return from_kgid(&init_user_ns, inode->i_gid);
779 static inline void i_uid_write(struct inode *inode, uid_t uid)
781 inode->i_uid = make_kuid(&init_user_ns, uid);
784 static inline void i_gid_write(struct inode *inode, gid_t gid)
786 inode->i_gid = make_kgid(&init_user_ns, gid);
789 static inline unsigned iminor(const struct inode *inode)
791 return MINOR(inode->i_rdev);
794 static inline unsigned imajor(const struct inode *inode)
796 return MAJOR(inode->i_rdev);
799 extern struct block_device *I_BDEV(struct inode *inode);
802 rwlock_t lock; /* protects pid, uid, euid fields */
803 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
804 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
805 kuid_t uid, euid; /* uid/euid of process setting the owner */
806 int signum; /* posix.1b rt signal to be delivered on IO */
810 * Track a single file's readahead state
812 struct file_ra_state {
813 pgoff_t start; /* where readahead started */
814 unsigned int size; /* # of readahead pages */
815 unsigned int async_size; /* do asynchronous readahead when
816 there are only # of pages ahead */
818 unsigned int ra_pages; /* Maximum readahead window */
819 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
820 loff_t prev_pos; /* Cache last read() position */
824 * Check if @index falls in the readahead windows.
826 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
828 return (index >= ra->start &&
829 index < ra->start + ra->size);
834 struct llist_node fu_llist;
835 struct rcu_head fu_rcuhead;
838 struct inode *f_inode; /* cached value */
839 const struct file_operations *f_op;
842 * Protects f_ep_links, f_flags.
843 * Must not be taken from IRQ context.
846 atomic_long_t f_count;
847 unsigned int f_flags;
849 struct mutex f_pos_lock;
851 struct fown_struct f_owner;
852 const struct cred *f_cred;
853 struct file_ra_state f_ra;
856 #ifdef CONFIG_SECURITY
859 /* needed for tty driver, and maybe others */
863 /* Used by fs/eventpoll.c to link all the hooks to this file */
864 struct list_head f_ep_links;
865 struct list_head f_tfile_llink;
866 #endif /* #ifdef CONFIG_EPOLL */
867 struct address_space *f_mapping;
868 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
873 /* file identifier */
874 unsigned char f_handle[0];
877 static inline struct file *get_file(struct file *f)
879 atomic_long_inc(&f->f_count);
882 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
883 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
884 #define file_count(x) atomic_long_read(&(x)->f_count)
886 #define MAX_NON_LFS ((1UL<<31) - 1)
888 /* Page cache limit. The filesystems should put that into their s_maxbytes
889 limits, otherwise bad things can happen in VM. */
890 #if BITS_PER_LONG==32
891 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
892 #elif BITS_PER_LONG==64
893 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
898 #define FL_DELEG 4 /* NFSv4 delegation */
899 #define FL_ACCESS 8 /* not trying to lock, just looking */
900 #define FL_EXISTS 16 /* when unlocking, test for existence */
901 #define FL_LEASE 32 /* lease held on this file */
902 #define FL_CLOSE 64 /* unlock on close */
903 #define FL_SLEEP 128 /* A blocking lock */
904 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
905 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
906 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
907 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
910 * Special return value from posix_lock_file() and vfs_lock_file() for
911 * asynchronous locking.
913 #define FILE_LOCK_DEFERRED 1
915 /* legacy typedef, should eventually be removed */
916 typedef void *fl_owner_t;
920 struct file_lock_operations {
921 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
922 void (*fl_release_private)(struct file_lock *);
925 struct lock_manager_operations {
926 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
927 unsigned long (*lm_owner_key)(struct file_lock *);
928 fl_owner_t (*lm_get_owner)(fl_owner_t);
929 void (*lm_put_owner)(fl_owner_t);
930 void (*lm_notify)(struct file_lock *); /* unblock callback */
931 int (*lm_grant)(struct file_lock *, int);
932 bool (*lm_break)(struct file_lock *);
933 int (*lm_change)(struct file_lock *, int, struct list_head *);
934 void (*lm_setup)(struct file_lock *, void **);
937 struct lock_manager {
938 struct list_head list;
942 void locks_start_grace(struct net *, struct lock_manager *);
943 void locks_end_grace(struct lock_manager *);
944 int locks_in_grace(struct net *);
946 /* that will die - we need it for nfs_lock_info */
947 #include <linux/nfs_fs_i.h>
950 * struct file_lock represents a generic "file lock". It's used to represent
951 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
952 * note that the same struct is used to represent both a request for a lock and
953 * the lock itself, but the same object is never used for both.
955 * FIXME: should we create a separate "struct lock_request" to help distinguish
958 * The varous i_flctx lists are ordered by:
961 * 2) lock range start
964 * Obviously, the last two criteria only matter for POSIX locks.
967 struct file_lock *fl_next; /* singly linked list for this inode */
968 struct list_head fl_list; /* link into file_lock_context */
969 struct hlist_node fl_link; /* node in global lists */
970 struct list_head fl_block; /* circular list of blocked processes */
972 unsigned int fl_flags;
973 unsigned char fl_type;
975 int fl_link_cpu; /* what cpu's list is this on? */
976 struct pid *fl_nspid;
977 wait_queue_head_t fl_wait;
978 struct file *fl_file;
982 struct fasync_struct * fl_fasync; /* for lease break notifications */
983 /* for lease breaks: */
984 unsigned long fl_break_time;
985 unsigned long fl_downgrade_time;
987 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
988 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
990 struct nfs_lock_info nfs_fl;
991 struct nfs4_lock_info nfs4_fl;
993 struct list_head link; /* link in AFS vnode's pending_locks list */
994 int state; /* state of grant or error if -ve */
999 struct file_lock_context {
1000 spinlock_t flc_lock;
1001 struct list_head flc_flock;
1002 struct list_head flc_posix;
1003 struct list_head flc_lease;
1006 /* The following constant reflects the upper bound of the file/locking space */
1008 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1009 #define OFFSET_MAX INT_LIMIT(loff_t)
1010 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1013 #include <linux/fcntl.h>
1015 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1017 #ifdef CONFIG_FILE_LOCKING
1018 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1019 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1020 struct flock __user *);
1022 #if BITS_PER_LONG == 32
1023 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1024 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1025 struct flock64 __user *);
1028 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1029 extern int fcntl_getlease(struct file *filp);
1032 void locks_free_lock_context(struct file_lock_context *ctx);
1033 void locks_free_lock(struct file_lock *fl);
1034 extern void locks_init_lock(struct file_lock *);
1035 extern struct file_lock * locks_alloc_lock(void);
1036 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1037 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1038 extern void locks_remove_posix(struct file *, fl_owner_t);
1039 extern void locks_remove_file(struct file *);
1040 extern void locks_release_private(struct file_lock *);
1041 extern void posix_test_lock(struct file *, struct file_lock *);
1042 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1043 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1044 extern int posix_unblock_lock(struct file_lock *);
1045 extern int vfs_test_lock(struct file *, struct file_lock *);
1046 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1047 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1048 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1049 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1050 extern void lease_get_mtime(struct inode *, struct timespec *time);
1051 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1052 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1053 extern int lease_modify(struct file_lock *, int, struct list_head *);
1054 struct files_struct;
1055 extern void show_fd_locks(struct seq_file *f,
1056 struct file *filp, struct files_struct *files);
1057 #else /* !CONFIG_FILE_LOCKING */
1058 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1059 struct flock __user *user)
1064 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1065 unsigned int cmd, struct flock __user *user)
1070 #if BITS_PER_LONG == 32
1071 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1072 struct flock64 __user *user)
1077 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1078 unsigned int cmd, struct flock64 __user *user)
1083 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1088 static inline int fcntl_getlease(struct file *filp)
1094 locks_free_lock_context(struct file_lock_context *ctx)
1098 static inline void locks_init_lock(struct file_lock *fl)
1103 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1108 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1113 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1118 static inline void locks_remove_file(struct file *filp)
1123 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1128 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1129 struct file_lock *conflock)
1134 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1139 static inline int posix_unblock_lock(struct file_lock *waiter)
1144 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1149 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1150 struct file_lock *fl, struct file_lock *conf)
1155 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1160 static inline int flock_lock_file_wait(struct file *filp,
1161 struct file_lock *request)
1166 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1171 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1176 static inline int generic_setlease(struct file *filp, long arg,
1177 struct file_lock **flp, void **priv)
1182 static inline int vfs_setlease(struct file *filp, long arg,
1183 struct file_lock **lease, void **priv)
1188 static inline int lease_modify(struct file_lock *fl, int arg,
1189 struct list_head *dispose)
1194 struct files_struct;
1195 static inline void show_fd_locks(struct seq_file *f,
1196 struct file *filp, struct files_struct *files) {}
1197 #endif /* !CONFIG_FILE_LOCKING */
1200 struct fasync_struct {
1204 struct fasync_struct *fa_next; /* singly linked list */
1205 struct file *fa_file;
1206 struct rcu_head fa_rcu;
1209 #define FASYNC_MAGIC 0x4601
1211 /* SMP safe fasync helpers: */
1212 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1213 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1214 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1215 extern struct fasync_struct *fasync_alloc(void);
1216 extern void fasync_free(struct fasync_struct *);
1218 /* can be called from interrupts */
1219 extern void kill_fasync(struct fasync_struct **, int, int);
1221 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1222 extern void f_setown(struct file *filp, unsigned long arg, int force);
1223 extern void f_delown(struct file *filp);
1224 extern pid_t f_getown(struct file *filp);
1225 extern int send_sigurg(struct fown_struct *fown);
1233 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1234 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1235 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1236 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1237 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1240 /* Possible states of 'frozen' field */
1242 SB_UNFROZEN = 0, /* FS is unfrozen */
1243 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1244 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1245 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1246 * internal threads if needed) */
1247 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1250 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1253 /* Counters for counting writers at each level */
1254 struct percpu_counter counter[SB_FREEZE_LEVELS];
1255 wait_queue_head_t wait; /* queue for waiting for
1256 writers / faults to finish */
1257 int frozen; /* Is sb frozen? */
1258 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1260 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1261 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1265 struct super_block {
1266 struct list_head s_list; /* Keep this first */
1267 dev_t s_dev; /* search index; _not_ kdev_t */
1268 unsigned char s_blocksize_bits;
1269 unsigned long s_blocksize;
1270 loff_t s_maxbytes; /* Max file size */
1271 struct file_system_type *s_type;
1272 const struct super_operations *s_op;
1273 const struct dquot_operations *dq_op;
1274 const struct quotactl_ops *s_qcop;
1275 const struct export_operations *s_export_op;
1276 unsigned long s_flags;
1277 unsigned long s_magic;
1278 struct dentry *s_root;
1279 struct rw_semaphore s_umount;
1282 #ifdef CONFIG_SECURITY
1285 const struct xattr_handler **s_xattr;
1287 struct list_head s_inodes; /* all inodes */
1288 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1289 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1290 struct block_device *s_bdev;
1291 struct backing_dev_info *s_bdi;
1292 struct mtd_info *s_mtd;
1293 struct hlist_node s_instances;
1294 unsigned int s_quota_types; /* Bitmask of supported quota types */
1295 struct quota_info s_dquot; /* Diskquota specific options */
1297 struct sb_writers s_writers;
1299 char s_id[32]; /* Informational name */
1300 u8 s_uuid[16]; /* UUID */
1302 void *s_fs_info; /* Filesystem private info */
1303 unsigned int s_max_links;
1306 /* Granularity of c/m/atime in ns.
1307 Cannot be worse than a second */
1311 * The next field is for VFS *only*. No filesystems have any business
1312 * even looking at it. You had been warned.
1314 struct mutex s_vfs_rename_mutex; /* Kludge */
1317 * Filesystem subtype. If non-empty the filesystem type field
1318 * in /proc/mounts will be "type.subtype"
1323 * Saved mount options for lazy filesystems using
1324 * generic_show_options()
1326 char __rcu *s_options;
1327 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1330 * Saved pool identifier for cleancache (-1 means none)
1332 int cleancache_poolid;
1334 struct shrinker s_shrink; /* per-sb shrinker handle */
1336 /* Number of inodes with nlink == 0 but still referenced */
1337 atomic_long_t s_remove_count;
1339 /* Being remounted read-only */
1340 int s_readonly_remount;
1342 /* AIO completions deferred from interrupt context */
1343 struct workqueue_struct *s_dio_done_wq;
1344 struct hlist_head s_pins;
1347 * Keep the lru lists last in the structure so they always sit on their
1348 * own individual cachelines.
1350 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1351 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1352 struct rcu_head rcu;
1355 * Indicates how deep in a filesystem stack this SB is
1360 extern struct timespec current_fs_time(struct super_block *sb);
1363 * Snapshotting support.
1366 void __sb_end_write(struct super_block *sb, int level);
1367 int __sb_start_write(struct super_block *sb, int level, bool wait);
1370 * sb_end_write - drop write access to a superblock
1371 * @sb: the super we wrote to
1373 * Decrement number of writers to the filesystem. Wake up possible waiters
1374 * wanting to freeze the filesystem.
1376 static inline void sb_end_write(struct super_block *sb)
1378 __sb_end_write(sb, SB_FREEZE_WRITE);
1382 * sb_end_pagefault - drop write access to a superblock from a page fault
1383 * @sb: the super we wrote to
1385 * Decrement number of processes handling write page fault to the filesystem.
1386 * Wake up possible waiters wanting to freeze the filesystem.
1388 static inline void sb_end_pagefault(struct super_block *sb)
1390 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1394 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1395 * @sb: the super we wrote to
1397 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1398 * waiters wanting to freeze the filesystem.
1400 static inline void sb_end_intwrite(struct super_block *sb)
1402 __sb_end_write(sb, SB_FREEZE_FS);
1406 * sb_start_write - get write access to a superblock
1407 * @sb: the super we write to
1409 * When a process wants to write data or metadata to a file system (i.e. dirty
1410 * a page or an inode), it should embed the operation in a sb_start_write() -
1411 * sb_end_write() pair to get exclusion against file system freezing. This
1412 * function increments number of writers preventing freezing. If the file
1413 * system is already frozen, the function waits until the file system is
1416 * Since freeze protection behaves as a lock, users have to preserve
1417 * ordering of freeze protection and other filesystem locks. Generally,
1418 * freeze protection should be the outermost lock. In particular, we have:
1421 * -> i_mutex (write path, truncate, directory ops, ...)
1422 * -> s_umount (freeze_super, thaw_super)
1424 static inline void sb_start_write(struct super_block *sb)
1426 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1429 static inline int sb_start_write_trylock(struct super_block *sb)
1431 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1435 * sb_start_pagefault - get write access to a superblock from a page fault
1436 * @sb: the super we write to
1438 * When a process starts handling write page fault, it should embed the
1439 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1440 * exclusion against file system freezing. This is needed since the page fault
1441 * is going to dirty a page. This function increments number of running page
1442 * faults preventing freezing. If the file system is already frozen, the
1443 * function waits until the file system is thawed.
1445 * Since page fault freeze protection behaves as a lock, users have to preserve
1446 * ordering of freeze protection and other filesystem locks. It is advised to
1447 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1448 * handling code implies lock dependency:
1451 * -> sb_start_pagefault
1453 static inline void sb_start_pagefault(struct super_block *sb)
1455 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1459 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1460 * @sb: the super we write to
1462 * This is the third level of protection against filesystem freezing. It is
1463 * free for use by a filesystem. The only requirement is that it must rank
1464 * below sb_start_pagefault.
1466 * For example filesystem can call sb_start_intwrite() when starting a
1467 * transaction which somewhat eases handling of freezing for internal sources
1468 * of filesystem changes (internal fs threads, discarding preallocation on file
1471 static inline void sb_start_intwrite(struct super_block *sb)
1473 __sb_start_write(sb, SB_FREEZE_FS, true);
1477 extern bool inode_owner_or_capable(const struct inode *inode);
1480 * VFS helper functions..
1482 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1483 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1484 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1485 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1486 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1487 extern int vfs_rmdir(struct inode *, struct dentry *);
1488 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1489 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1490 extern int vfs_whiteout(struct inode *, struct dentry *);
1493 * VFS dentry helper functions.
1495 extern void dentry_unhash(struct dentry *dentry);
1498 * VFS file helper functions.
1500 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1503 * VFS FS_IOC_FIEMAP helper definitions.
1505 struct fiemap_extent_info {
1506 unsigned int fi_flags; /* Flags as passed from user */
1507 unsigned int fi_extents_mapped; /* Number of mapped extents */
1508 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1509 struct fiemap_extent __user *fi_extents_start; /* Start of
1510 fiemap_extent array */
1512 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1513 u64 phys, u64 len, u32 flags);
1514 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1519 * NOTE! These match bits 12..15 of stat.st_mode
1520 * (ie "(i_mode >> 12) & 15").
1522 #define DT_UNKNOWN 0
1533 * This is the "filldir" function type, used by readdir() to let
1534 * the kernel specify what kind of dirent layout it wants to have.
1535 * This allows the kernel to read directories into kernel space or
1536 * to have different dirent layouts depending on the binary type.
1539 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1542 struct dir_context {
1543 const filldir_t actor;
1547 struct block_device_operations;
1549 /* These macros are for out of kernel modules to test that
1550 * the kernel supports the unlocked_ioctl and compat_ioctl
1551 * fields in struct file_operations. */
1552 #define HAVE_COMPAT_IOCTL 1
1553 #define HAVE_UNLOCKED_IOCTL 1
1556 * These flags let !MMU mmap() govern direct device mapping vs immediate
1557 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1559 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1560 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1561 * NOMMU_MAP_READ: Can be mapped for reading
1562 * NOMMU_MAP_WRITE: Can be mapped for writing
1563 * NOMMU_MAP_EXEC: Can be mapped for execution
1565 #define NOMMU_MAP_COPY 0x00000001
1566 #define NOMMU_MAP_DIRECT 0x00000008
1567 #define NOMMU_MAP_READ VM_MAYREAD
1568 #define NOMMU_MAP_WRITE VM_MAYWRITE
1569 #define NOMMU_MAP_EXEC VM_MAYEXEC
1571 #define NOMMU_VMFLAGS \
1572 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1577 struct file_operations {
1578 struct module *owner;
1579 loff_t (*llseek) (struct file *, loff_t, int);
1580 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1581 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1582 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1583 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1584 int (*iterate) (struct file *, struct dir_context *);
1585 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1586 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1587 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1588 int (*mmap) (struct file *, struct vm_area_struct *);
1589 int (*mremap)(struct file *, struct vm_area_struct *);
1590 int (*open) (struct inode *, struct file *);
1591 int (*flush) (struct file *, fl_owner_t id);
1592 int (*release) (struct inode *, struct file *);
1593 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1594 int (*aio_fsync) (struct kiocb *, int datasync);
1595 int (*fasync) (int, struct file *, int);
1596 int (*lock) (struct file *, int, struct file_lock *);
1597 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1598 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1599 int (*check_flags)(int);
1600 int (*flock) (struct file *, int, struct file_lock *);
1601 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1602 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1603 int (*setlease)(struct file *, long, struct file_lock **, void **);
1604 long (*fallocate)(struct file *file, int mode, loff_t offset,
1606 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1608 unsigned (*mmap_capabilities)(struct file *);
1612 struct inode_operations {
1613 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1614 void * (*follow_link) (struct dentry *, struct nameidata *);
1615 int (*permission) (struct inode *, int);
1616 struct posix_acl * (*get_acl)(struct inode *, int);
1618 int (*readlink) (struct dentry *, char __user *,int);
1619 void (*put_link) (struct dentry *, struct nameidata *, void *);
1621 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1622 int (*link) (struct dentry *,struct inode *,struct dentry *);
1623 int (*unlink) (struct inode *,struct dentry *);
1624 int (*symlink) (struct inode *,struct dentry *,const char *);
1625 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1626 int (*rmdir) (struct inode *,struct dentry *);
1627 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1628 int (*rename) (struct inode *, struct dentry *,
1629 struct inode *, struct dentry *);
1630 int (*rename2) (struct inode *, struct dentry *,
1631 struct inode *, struct dentry *, unsigned int);
1632 int (*setattr) (struct dentry *, struct iattr *);
1633 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1634 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1635 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1636 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1637 int (*removexattr) (struct dentry *, const char *);
1638 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1640 int (*update_time)(struct inode *, struct timespec *, int);
1641 int (*atomic_open)(struct inode *, struct dentry *,
1642 struct file *, unsigned open_flag,
1643 umode_t create_mode, int *opened);
1644 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1645 int (*set_acl)(struct inode *, struct posix_acl *, int);
1647 /* WARNING: probably going away soon, do not use! */
1648 int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
1649 } ____cacheline_aligned;
1651 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1652 unsigned long nr_segs, unsigned long fast_segs,
1653 struct iovec *fast_pointer,
1654 struct iovec **ret_pointer);
1656 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1657 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1658 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1659 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1660 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1661 unsigned long, loff_t *);
1662 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1663 unsigned long, loff_t *);
1665 struct super_operations {
1666 struct inode *(*alloc_inode)(struct super_block *sb);
1667 void (*destroy_inode)(struct inode *);
1669 void (*dirty_inode) (struct inode *, int flags);
1670 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1671 int (*drop_inode) (struct inode *);
1672 void (*evict_inode) (struct inode *);
1673 void (*put_super) (struct super_block *);
1674 int (*sync_fs)(struct super_block *sb, int wait);
1675 int (*freeze_super) (struct super_block *);
1676 int (*freeze_fs) (struct super_block *);
1677 int (*thaw_super) (struct super_block *);
1678 int (*unfreeze_fs) (struct super_block *);
1679 int (*statfs) (struct dentry *, struct kstatfs *);
1680 int (*remount_fs) (struct super_block *, int *, char *);
1681 void (*umount_begin) (struct super_block *);
1683 int (*show_options)(struct seq_file *, struct dentry *);
1684 int (*show_devname)(struct seq_file *, struct dentry *);
1685 int (*show_path)(struct seq_file *, struct dentry *);
1686 int (*show_stats)(struct seq_file *, struct dentry *);
1688 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1689 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1690 struct dquot **(*get_dquots)(struct inode *);
1692 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1693 long (*nr_cached_objects)(struct super_block *,
1694 struct shrink_control *);
1695 long (*free_cached_objects)(struct super_block *,
1696 struct shrink_control *);
1700 * Inode flags - they have no relation to superblock flags now
1702 #define S_SYNC 1 /* Writes are synced at once */
1703 #define S_NOATIME 2 /* Do not update access times */
1704 #define S_APPEND 4 /* Append-only file */
1705 #define S_IMMUTABLE 8 /* Immutable file */
1706 #define S_DEAD 16 /* removed, but still open directory */
1707 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1708 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1709 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1710 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1711 #define S_PRIVATE 512 /* Inode is fs-internal */
1712 #define S_IMA 1024 /* Inode has an associated IMA struct */
1713 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1714 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1715 #ifdef CONFIG_FS_DAX
1716 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1718 #define S_DAX 0 /* Make all the DAX code disappear */
1722 * Note that nosuid etc flags are inode-specific: setting some file-system
1723 * flags just means all the inodes inherit those flags by default. It might be
1724 * possible to override it selectively if you really wanted to with some
1725 * ioctl() that is not currently implemented.
1727 * Exception: MS_RDONLY is always applied to the entire file system.
1729 * Unfortunately, it is possible to change a filesystems flags with it mounted
1730 * with files in use. This means that all of the inodes will not have their
1731 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1732 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1734 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1736 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1737 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1738 ((inode)->i_flags & S_SYNC))
1739 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1740 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1741 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1742 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1743 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1745 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1746 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1747 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1748 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1750 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1751 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1752 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1753 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1754 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1755 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1756 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1757 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1759 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1760 (inode)->i_rdev == WHITEOUT_DEV)
1763 * Inode state bits. Protected by inode->i_lock
1765 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1766 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1768 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1769 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1770 * various stages of removing an inode.
1772 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1774 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1775 * fdatasync(). i_atime is the usual cause.
1776 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1777 * these changes separately from I_DIRTY_SYNC so that we
1778 * don't have to write inode on fdatasync() when only
1779 * mtime has changed in it.
1780 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1781 * I_NEW Serves as both a mutex and completion notification.
1782 * New inodes set I_NEW. If two processes both create
1783 * the same inode, one of them will release its inode and
1784 * wait for I_NEW to be released before returning.
1785 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1786 * also cause waiting on I_NEW, without I_NEW actually
1787 * being set. find_inode() uses this to prevent returning
1788 * nearly-dead inodes.
1789 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1790 * is zero. I_FREEING must be set when I_WILL_FREE is
1792 * I_FREEING Set when inode is about to be freed but still has dirty
1793 * pages or buffers attached or the inode itself is still
1795 * I_CLEAR Added by clear_inode(). In this state the inode is
1796 * clean and can be destroyed. Inode keeps I_FREEING.
1798 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1799 * prohibited for many purposes. iget() must wait for
1800 * the inode to be completely released, then create it
1801 * anew. Other functions will just ignore such inodes,
1802 * if appropriate. I_NEW is used for waiting.
1804 * I_SYNC Writeback of inode is running. The bit is set during
1805 * data writeback, and cleared with a wakeup on the bit
1806 * address once it is done. The bit is also used to pin
1807 * the inode in memory for flusher thread.
1809 * I_REFERENCED Marks the inode as recently references on the LRU list.
1811 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1813 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1815 #define I_DIRTY_SYNC (1 << 0)
1816 #define I_DIRTY_DATASYNC (1 << 1)
1817 #define I_DIRTY_PAGES (1 << 2)
1819 #define I_NEW (1 << __I_NEW)
1820 #define I_WILL_FREE (1 << 4)
1821 #define I_FREEING (1 << 5)
1822 #define I_CLEAR (1 << 6)
1824 #define I_SYNC (1 << __I_SYNC)
1825 #define I_REFERENCED (1 << 8)
1826 #define __I_DIO_WAKEUP 9
1827 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1828 #define I_LINKABLE (1 << 10)
1829 #define I_DIRTY_TIME (1 << 11)
1830 #define __I_DIRTY_TIME_EXPIRED 12
1831 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1833 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1834 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1836 extern void __mark_inode_dirty(struct inode *, int);
1837 static inline void mark_inode_dirty(struct inode *inode)
1839 __mark_inode_dirty(inode, I_DIRTY);
1842 static inline void mark_inode_dirty_sync(struct inode *inode)
1844 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1847 extern void inc_nlink(struct inode *inode);
1848 extern void drop_nlink(struct inode *inode);
1849 extern void clear_nlink(struct inode *inode);
1850 extern void set_nlink(struct inode *inode, unsigned int nlink);
1852 static inline void inode_inc_link_count(struct inode *inode)
1855 mark_inode_dirty(inode);
1858 static inline void inode_dec_link_count(struct inode *inode)
1861 mark_inode_dirty(inode);
1865 * inode_inc_iversion - increments i_version
1866 * @inode: inode that need to be updated
1868 * Every time the inode is modified, the i_version field will be incremented.
1869 * The filesystem has to be mounted with i_version flag
1872 static inline void inode_inc_iversion(struct inode *inode)
1874 spin_lock(&inode->i_lock);
1876 spin_unlock(&inode->i_lock);
1879 enum file_time_flags {
1886 extern void touch_atime(const struct path *);
1887 static inline void file_accessed(struct file *file)
1889 if (!(file->f_flags & O_NOATIME))
1890 touch_atime(&file->f_path);
1893 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1894 int sync_inode_metadata(struct inode *inode, int wait);
1896 struct file_system_type {
1899 #define FS_REQUIRES_DEV 1
1900 #define FS_BINARY_MOUNTDATA 2
1901 #define FS_HAS_SUBTYPE 4
1902 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1903 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1904 #define FS_CGROUP_WRITEBACK 32 /* Supports cgroup-aware writeback */
1905 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1906 struct dentry *(*mount) (struct file_system_type *, int,
1907 const char *, void *);
1908 void (*kill_sb) (struct super_block *);
1909 struct module *owner;
1910 struct file_system_type * next;
1911 struct hlist_head fs_supers;
1913 struct lock_class_key s_lock_key;
1914 struct lock_class_key s_umount_key;
1915 struct lock_class_key s_vfs_rename_key;
1916 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1918 struct lock_class_key i_lock_key;
1919 struct lock_class_key i_mutex_key;
1920 struct lock_class_key i_mutex_dir_key;
1923 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1925 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1926 void *data, int (*fill_super)(struct super_block *, void *, int));
1927 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1928 int flags, const char *dev_name, void *data,
1929 int (*fill_super)(struct super_block *, void *, int));
1930 extern struct dentry *mount_single(struct file_system_type *fs_type,
1931 int flags, void *data,
1932 int (*fill_super)(struct super_block *, void *, int));
1933 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1934 int flags, void *data,
1935 int (*fill_super)(struct super_block *, void *, int));
1936 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1937 void generic_shutdown_super(struct super_block *sb);
1938 void kill_block_super(struct super_block *sb);
1939 void kill_anon_super(struct super_block *sb);
1940 void kill_litter_super(struct super_block *sb);
1941 void deactivate_super(struct super_block *sb);
1942 void deactivate_locked_super(struct super_block *sb);
1943 int set_anon_super(struct super_block *s, void *data);
1944 int get_anon_bdev(dev_t *);
1945 void free_anon_bdev(dev_t);
1946 struct super_block *sget(struct file_system_type *type,
1947 int (*test)(struct super_block *,void *),
1948 int (*set)(struct super_block *,void *),
1949 int flags, void *data);
1950 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1951 const struct super_operations *ops,
1952 const struct dentry_operations *dops,
1955 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1956 #define fops_get(fops) \
1957 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1958 #define fops_put(fops) \
1959 do { if (fops) module_put((fops)->owner); } while(0)
1961 * This one is to be used *ONLY* from ->open() instances.
1962 * fops must be non-NULL, pinned down *and* module dependencies
1963 * should be sufficient to pin the caller down as well.
1965 #define replace_fops(f, fops) \
1967 struct file *__file = (f); \
1968 fops_put(__file->f_op); \
1969 BUG_ON(!(__file->f_op = (fops))); \
1972 extern int register_filesystem(struct file_system_type *);
1973 extern int unregister_filesystem(struct file_system_type *);
1974 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1975 #define kern_mount(type) kern_mount_data(type, NULL)
1976 extern void kern_unmount(struct vfsmount *mnt);
1977 extern int may_umount_tree(struct vfsmount *);
1978 extern int may_umount(struct vfsmount *);
1979 extern long do_mount(const char *, const char __user *,
1980 const char *, unsigned long, void *);
1981 extern struct vfsmount *collect_mounts(struct path *);
1982 extern void drop_collected_mounts(struct vfsmount *);
1983 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1985 extern int vfs_statfs(struct path *, struct kstatfs *);
1986 extern int user_statfs(const char __user *, struct kstatfs *);
1987 extern int fd_statfs(int, struct kstatfs *);
1988 extern int vfs_ustat(dev_t, struct kstatfs *);
1989 extern int freeze_super(struct super_block *super);
1990 extern int thaw_super(struct super_block *super);
1991 extern bool our_mnt(struct vfsmount *mnt);
1992 extern bool fs_fully_visible(struct file_system_type *);
1994 extern int current_umask(void);
1996 extern void ihold(struct inode * inode);
1997 extern void iput(struct inode *);
1998 extern int generic_update_time(struct inode *, struct timespec *, int);
2000 static inline struct inode *file_inode(const struct file *f)
2006 extern struct kobject *fs_kobj;
2008 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2010 #define FLOCK_VERIFY_READ 1
2011 #define FLOCK_VERIFY_WRITE 2
2013 #ifdef CONFIG_FILE_LOCKING
2014 extern int locks_mandatory_locked(struct file *);
2015 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2018 * Candidates for mandatory locking have the setgid bit set
2019 * but no group execute bit - an otherwise meaningless combination.
2022 static inline int __mandatory_lock(struct inode *ino)
2024 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2028 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2029 * otherwise these will be advisory locks
2032 static inline int mandatory_lock(struct inode *ino)
2034 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2037 static inline int locks_verify_locked(struct file *file)
2039 if (mandatory_lock(file_inode(file)))
2040 return locks_mandatory_locked(file);
2044 static inline int locks_verify_truncate(struct inode *inode,
2048 if (inode->i_flctx && mandatory_lock(inode))
2049 return locks_mandatory_area(
2050 FLOCK_VERIFY_WRITE, inode, filp,
2051 size < inode->i_size ? size : inode->i_size,
2052 (size < inode->i_size ? inode->i_size - size
2053 : size - inode->i_size)
2058 static inline int break_lease(struct inode *inode, unsigned int mode)
2061 * Since this check is lockless, we must ensure that any refcounts
2062 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2063 * could end up racing with tasks trying to set a new lease on this
2067 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2068 return __break_lease(inode, mode, FL_LEASE);
2072 static inline int break_deleg(struct inode *inode, unsigned int mode)
2075 * Since this check is lockless, we must ensure that any refcounts
2076 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2077 * could end up racing with tasks trying to set a new lease on this
2081 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2082 return __break_lease(inode, mode, FL_DELEG);
2086 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2090 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2091 if (ret == -EWOULDBLOCK && delegated_inode) {
2092 *delegated_inode = inode;
2098 static inline int break_deleg_wait(struct inode **delegated_inode)
2102 ret = break_deleg(*delegated_inode, O_WRONLY);
2103 iput(*delegated_inode);
2104 *delegated_inode = NULL;
2108 static inline int break_layout(struct inode *inode, bool wait)
2111 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2112 return __break_lease(inode,
2113 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2118 #else /* !CONFIG_FILE_LOCKING */
2119 static inline int locks_mandatory_locked(struct file *file)
2124 static inline int locks_mandatory_area(int rw, struct inode *inode,
2125 struct file *filp, loff_t offset,
2131 static inline int __mandatory_lock(struct inode *inode)
2136 static inline int mandatory_lock(struct inode *inode)
2141 static inline int locks_verify_locked(struct file *file)
2146 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2152 static inline int break_lease(struct inode *inode, unsigned int mode)
2157 static inline int break_deleg(struct inode *inode, unsigned int mode)
2162 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2167 static inline int break_deleg_wait(struct inode **delegated_inode)
2173 static inline int break_layout(struct inode *inode, bool wait)
2178 #endif /* CONFIG_FILE_LOCKING */
2183 const char *name; /* pointer to actual string */
2184 const __user char *uptr; /* original userland pointer */
2185 struct audit_names *aname;
2190 extern long vfs_truncate(struct path *, loff_t);
2191 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2193 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2195 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2197 extern struct file *file_open_name(struct filename *, int, umode_t);
2198 extern struct file *filp_open(const char *, int, umode_t);
2199 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2201 extern int vfs_open(const struct path *, struct file *, const struct cred *);
2202 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2203 extern int filp_close(struct file *, fl_owner_t id);
2205 extern struct filename *getname_flags(const char __user *, int, int *);
2206 extern struct filename *getname(const char __user *);
2207 extern struct filename *getname_kernel(const char *);
2208 extern void putname(struct filename *name);
2214 extern int finish_open(struct file *file, struct dentry *dentry,
2215 int (*open)(struct inode *, struct file *),
2217 extern int finish_no_open(struct file *file, struct dentry *dentry);
2221 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2224 extern void __init vfs_caches_init_early(void);
2225 extern void __init vfs_caches_init(unsigned long);
2227 extern struct kmem_cache *names_cachep;
2229 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2230 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2233 extern int register_blkdev(unsigned int, const char *);
2234 extern void unregister_blkdev(unsigned int, const char *);
2235 extern struct block_device *bdget(dev_t);
2236 extern struct block_device *bdgrab(struct block_device *bdev);
2237 extern void bd_set_size(struct block_device *, loff_t size);
2238 extern void bd_forget(struct inode *inode);
2239 extern void bdput(struct block_device *);
2240 extern void invalidate_bdev(struct block_device *);
2241 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2242 extern int sync_blockdev(struct block_device *bdev);
2243 extern void kill_bdev(struct block_device *);
2244 extern struct super_block *freeze_bdev(struct block_device *);
2245 extern void emergency_thaw_all(void);
2246 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2247 extern int fsync_bdev(struct block_device *);
2249 extern struct super_block *blockdev_superblock;
2251 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2253 return sb == blockdev_superblock;
2256 static inline void bd_forget(struct inode *inode) {}
2257 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2258 static inline void kill_bdev(struct block_device *bdev) {}
2259 static inline void invalidate_bdev(struct block_device *bdev) {}
2261 static inline struct super_block *freeze_bdev(struct block_device *sb)
2266 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2271 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2275 static inline int sb_is_blkdev_sb(struct super_block *sb)
2280 extern int sync_filesystem(struct super_block *);
2281 extern const struct file_operations def_blk_fops;
2282 extern const struct file_operations def_chr_fops;
2284 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2285 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2286 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2287 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2288 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2290 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2292 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2293 extern int __blkdev_reread_part(struct block_device *bdev);
2294 extern int blkdev_reread_part(struct block_device *bdev);
2297 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2298 extern void bd_unlink_disk_holder(struct block_device *bdev,
2299 struct gendisk *disk);
2301 static inline int bd_link_disk_holder(struct block_device *bdev,
2302 struct gendisk *disk)
2306 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2307 struct gendisk *disk)
2314 #define CHRDEV_MAJOR_HASH_SIZE 255
2315 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2316 extern int register_chrdev_region(dev_t, unsigned, const char *);
2317 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2318 unsigned int count, const char *name,
2319 const struct file_operations *fops);
2320 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2321 unsigned int count, const char *name);
2322 extern void unregister_chrdev_region(dev_t, unsigned);
2323 extern void chrdev_show(struct seq_file *,off_t);
2325 static inline int register_chrdev(unsigned int major, const char *name,
2326 const struct file_operations *fops)
2328 return __register_chrdev(major, 0, 256, name, fops);
2331 static inline void unregister_chrdev(unsigned int major, const char *name)
2333 __unregister_chrdev(major, 0, 256, name);
2336 /* fs/block_dev.c */
2337 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2338 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2341 #define BLKDEV_MAJOR_HASH_SIZE 255
2342 extern const char *__bdevname(dev_t, char *buffer);
2343 extern const char *bdevname(struct block_device *bdev, char *buffer);
2344 extern struct block_device *lookup_bdev(const char *);
2345 extern void blkdev_show(struct seq_file *,off_t);
2348 #define BLKDEV_MAJOR_HASH_SIZE 0
2351 extern void init_special_inode(struct inode *, umode_t, dev_t);
2353 /* Invalid inode operations -- fs/bad_inode.c */
2354 extern void make_bad_inode(struct inode *);
2355 extern int is_bad_inode(struct inode *);
2359 * return READ, READA, or WRITE
2361 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2364 * return data direction, READ or WRITE
2366 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2368 extern void check_disk_size_change(struct gendisk *disk,
2369 struct block_device *bdev);
2370 extern int revalidate_disk(struct gendisk *);
2371 extern int check_disk_change(struct block_device *);
2372 extern int __invalidate_device(struct block_device *, bool);
2373 extern int invalidate_partition(struct gendisk *, int);
2375 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2376 pgoff_t start, pgoff_t end);
2378 static inline void invalidate_remote_inode(struct inode *inode)
2380 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2381 S_ISLNK(inode->i_mode))
2382 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2384 extern int invalidate_inode_pages2(struct address_space *mapping);
2385 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2386 pgoff_t start, pgoff_t end);
2387 extern int write_inode_now(struct inode *, int);
2388 extern int filemap_fdatawrite(struct address_space *);
2389 extern int filemap_flush(struct address_space *);
2390 extern int filemap_fdatawait(struct address_space *);
2391 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2393 extern int filemap_write_and_wait(struct address_space *mapping);
2394 extern int filemap_write_and_wait_range(struct address_space *mapping,
2395 loff_t lstart, loff_t lend);
2396 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2397 loff_t start, loff_t end, int sync_mode);
2398 extern int filemap_fdatawrite_range(struct address_space *mapping,
2399 loff_t start, loff_t end);
2401 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2403 extern int vfs_fsync(struct file *file, int datasync);
2404 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2406 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2408 return vfs_fsync_range(file, pos, pos + count - 1,
2409 (file->f_flags & __O_SYNC) ? 0 : 1);
2411 extern void emergency_sync(void);
2412 extern void emergency_remount(void);
2414 extern sector_t bmap(struct inode *, sector_t);
2416 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2417 extern int inode_permission(struct inode *, int);
2418 extern int __inode_permission(struct inode *, int);
2419 extern int generic_permission(struct inode *, int);
2420 extern int __check_sticky(struct inode *dir, struct inode *inode);
2422 static inline bool execute_ok(struct inode *inode)
2424 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2427 static inline void file_start_write(struct file *file)
2429 if (!S_ISREG(file_inode(file)->i_mode))
2431 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2434 static inline bool file_start_write_trylock(struct file *file)
2436 if (!S_ISREG(file_inode(file)->i_mode))
2438 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2441 static inline void file_end_write(struct file *file)
2443 if (!S_ISREG(file_inode(file)->i_mode))
2445 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2449 * get_write_access() gets write permission for a file.
2450 * put_write_access() releases this write permission.
2451 * This is used for regular files.
2452 * We cannot support write (and maybe mmap read-write shared) accesses and
2453 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2454 * can have the following values:
2455 * 0: no writers, no VM_DENYWRITE mappings
2456 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2457 * > 0: (i_writecount) users are writing to the file.
2459 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2460 * except for the cases where we don't hold i_writecount yet. Then we need to
2461 * use {get,deny}_write_access() - these functions check the sign and refuse
2462 * to do the change if sign is wrong.
2464 static inline int get_write_access(struct inode *inode)
2466 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2468 static inline int deny_write_access(struct file *file)
2470 struct inode *inode = file_inode(file);
2471 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2473 static inline void put_write_access(struct inode * inode)
2475 atomic_dec(&inode->i_writecount);
2477 static inline void allow_write_access(struct file *file)
2480 atomic_inc(&file_inode(file)->i_writecount);
2482 static inline bool inode_is_open_for_write(const struct inode *inode)
2484 return atomic_read(&inode->i_writecount) > 0;
2488 static inline void i_readcount_dec(struct inode *inode)
2490 BUG_ON(!atomic_read(&inode->i_readcount));
2491 atomic_dec(&inode->i_readcount);
2493 static inline void i_readcount_inc(struct inode *inode)
2495 atomic_inc(&inode->i_readcount);
2498 static inline void i_readcount_dec(struct inode *inode)
2502 static inline void i_readcount_inc(struct inode *inode)
2507 extern int do_pipe_flags(int *, int);
2509 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2510 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2511 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2512 extern struct file * open_exec(const char *);
2514 /* fs/dcache.c -- generic fs support functions */
2515 extern int is_subdir(struct dentry *, struct dentry *);
2516 extern int path_is_under(struct path *, struct path *);
2518 #include <linux/err.h>
2520 /* needed for stackable file system support */
2521 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2523 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2525 extern int inode_init_always(struct super_block *, struct inode *);
2526 extern void inode_init_once(struct inode *);
2527 extern void address_space_init_once(struct address_space *mapping);
2528 extern struct inode * igrab(struct inode *);
2529 extern ino_t iunique(struct super_block *, ino_t);
2530 extern int inode_needs_sync(struct inode *inode);
2531 extern int generic_delete_inode(struct inode *inode);
2532 static inline int generic_drop_inode(struct inode *inode)
2534 return !inode->i_nlink || inode_unhashed(inode);
2537 extern struct inode *ilookup5_nowait(struct super_block *sb,
2538 unsigned long hashval, int (*test)(struct inode *, void *),
2540 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2541 int (*test)(struct inode *, void *), void *data);
2542 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2544 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2545 extern struct inode * iget_locked(struct super_block *, unsigned long);
2546 extern struct inode *find_inode_nowait(struct super_block *,
2548 int (*match)(struct inode *,
2549 unsigned long, void *),
2551 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2552 extern int insert_inode_locked(struct inode *);
2553 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2554 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2556 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2558 extern void unlock_new_inode(struct inode *);
2559 extern unsigned int get_next_ino(void);
2561 extern void __iget(struct inode * inode);
2562 extern void iget_failed(struct inode *);
2563 extern void clear_inode(struct inode *);
2564 extern void __destroy_inode(struct inode *);
2565 extern struct inode *new_inode_pseudo(struct super_block *sb);
2566 extern struct inode *new_inode(struct super_block *sb);
2567 extern void free_inode_nonrcu(struct inode *inode);
2568 extern int should_remove_suid(struct dentry *);
2569 extern int file_remove_suid(struct file *);
2571 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2572 static inline void insert_inode_hash(struct inode *inode)
2574 __insert_inode_hash(inode, inode->i_ino);
2577 extern void __remove_inode_hash(struct inode *);
2578 static inline void remove_inode_hash(struct inode *inode)
2580 if (!inode_unhashed(inode))
2581 __remove_inode_hash(inode);
2584 extern void inode_sb_list_add(struct inode *inode);
2587 extern void submit_bio(int, struct bio *);
2588 extern int bdev_read_only(struct block_device *);
2590 extern int set_blocksize(struct block_device *, int);
2591 extern int sb_set_blocksize(struct super_block *, int);
2592 extern int sb_min_blocksize(struct super_block *, int);
2594 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2595 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2596 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2597 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2598 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2599 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2600 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2601 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2603 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2604 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2606 /* fs/block_dev.c */
2607 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2608 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2609 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2611 extern void block_sync_page(struct page *page);
2614 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2615 struct pipe_inode_info *, size_t, unsigned int);
2616 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2617 struct pipe_inode_info *, size_t, unsigned int);
2618 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2619 struct file *, loff_t *, size_t, unsigned int);
2620 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2621 struct file *out, loff_t *, size_t len, unsigned int flags);
2622 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2623 loff_t *opos, size_t len, unsigned int flags);
2627 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2628 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2629 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2630 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2631 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2632 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2633 int whence, loff_t maxsize, loff_t eof);
2634 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2635 int whence, loff_t size);
2636 extern int generic_file_open(struct inode * inode, struct file * filp);
2637 extern int nonseekable_open(struct inode * inode, struct file * filp);
2639 ssize_t dax_do_io(struct kiocb *, struct inode *, struct iov_iter *, loff_t,
2640 get_block_t, dio_iodone_t, int flags);
2641 int dax_clear_blocks(struct inode *, sector_t block, long size);
2642 int dax_zero_page_range(struct inode *, loff_t from, unsigned len, get_block_t);
2643 int dax_truncate_page(struct inode *, loff_t from, get_block_t);
2644 int dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t);
2645 int dax_pfn_mkwrite(struct vm_area_struct *, struct vm_fault *);
2646 #define dax_mkwrite(vma, vmf, gb) dax_fault(vma, vmf, gb)
2649 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2650 loff_t file_offset);
2653 /* need locking between buffered and direct access */
2656 /* filesystem does not support filling holes */
2657 DIO_SKIP_HOLES = 0x02,
2659 /* filesystem can handle aio writes beyond i_size */
2660 DIO_ASYNC_EXTEND = 0x04,
2662 /* inode/fs/bdev does not need truncate protection */
2663 DIO_SKIP_DIO_COUNT = 0x08,
2666 void dio_end_io(struct bio *bio, int error);
2668 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2669 struct block_device *bdev, struct iov_iter *iter,
2670 loff_t offset, get_block_t get_block,
2671 dio_iodone_t end_io, dio_submit_t submit_io,
2674 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2675 struct inode *inode,
2676 struct iov_iter *iter, loff_t offset,
2677 get_block_t get_block)
2679 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2680 offset, get_block, NULL, NULL,
2681 DIO_LOCKING | DIO_SKIP_HOLES);
2685 void inode_dio_wait(struct inode *inode);
2688 * inode_dio_begin - signal start of a direct I/O requests
2689 * @inode: inode the direct I/O happens on
2691 * This is called once we've finished processing a direct I/O request,
2692 * and is used to wake up callers waiting for direct I/O to be quiesced.
2694 static inline void inode_dio_begin(struct inode *inode)
2696 atomic_inc(&inode->i_dio_count);
2700 * inode_dio_end - signal finish of a direct I/O requests
2701 * @inode: inode the direct I/O happens on
2703 * This is called once we've finished processing a direct I/O request,
2704 * and is used to wake up callers waiting for direct I/O to be quiesced.
2706 static inline void inode_dio_end(struct inode *inode)
2708 if (atomic_dec_and_test(&inode->i_dio_count))
2709 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2712 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2715 extern const struct file_operations generic_ro_fops;
2717 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2719 extern int readlink_copy(char __user *, int, const char *);
2720 extern int page_readlink(struct dentry *, char __user *, int);
2721 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2722 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2723 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2725 extern int page_symlink(struct inode *inode, const char *symname, int len);
2726 extern const struct inode_operations page_symlink_inode_operations;
2727 extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
2728 extern int generic_readlink(struct dentry *, char __user *, int);
2729 extern void generic_fillattr(struct inode *, struct kstat *);
2730 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2731 extern int vfs_getattr(struct path *, struct kstat *);
2732 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2733 void inode_add_bytes(struct inode *inode, loff_t bytes);
2734 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2735 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2736 loff_t inode_get_bytes(struct inode *inode);
2737 void inode_set_bytes(struct inode *inode, loff_t bytes);
2739 extern int iterate_dir(struct file *, struct dir_context *);
2741 extern int vfs_stat(const char __user *, struct kstat *);
2742 extern int vfs_lstat(const char __user *, struct kstat *);
2743 extern int vfs_fstat(unsigned int, struct kstat *);
2744 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2746 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2748 extern int __generic_block_fiemap(struct inode *inode,
2749 struct fiemap_extent_info *fieinfo,
2750 loff_t start, loff_t len,
2751 get_block_t *get_block);
2752 extern int generic_block_fiemap(struct inode *inode,
2753 struct fiemap_extent_info *fieinfo, u64 start,
2754 u64 len, get_block_t *get_block);
2756 extern void get_filesystem(struct file_system_type *fs);
2757 extern void put_filesystem(struct file_system_type *fs);
2758 extern struct file_system_type *get_fs_type(const char *name);
2759 extern struct super_block *get_super(struct block_device *);
2760 extern struct super_block *get_super_thawed(struct block_device *);
2761 extern struct super_block *get_active_super(struct block_device *bdev);
2762 extern void drop_super(struct super_block *sb);
2763 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2764 extern void iterate_supers_type(struct file_system_type *,
2765 void (*)(struct super_block *, void *), void *);
2767 extern int dcache_dir_open(struct inode *, struct file *);
2768 extern int dcache_dir_close(struct inode *, struct file *);
2769 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2770 extern int dcache_readdir(struct file *, struct dir_context *);
2771 extern int simple_setattr(struct dentry *, struct iattr *);
2772 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2773 extern int simple_statfs(struct dentry *, struct kstatfs *);
2774 extern int simple_open(struct inode *inode, struct file *file);
2775 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2776 extern int simple_unlink(struct inode *, struct dentry *);
2777 extern int simple_rmdir(struct inode *, struct dentry *);
2778 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2779 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2780 extern int simple_empty(struct dentry *);
2781 extern int simple_readpage(struct file *file, struct page *page);
2782 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2783 loff_t pos, unsigned len, unsigned flags,
2784 struct page **pagep, void **fsdata);
2785 extern int simple_write_end(struct file *file, struct address_space *mapping,
2786 loff_t pos, unsigned len, unsigned copied,
2787 struct page *page, void *fsdata);
2788 extern int always_delete_dentry(const struct dentry *);
2789 extern struct inode *alloc_anon_inode(struct super_block *);
2790 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2791 extern const struct dentry_operations simple_dentry_operations;
2793 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2794 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2795 extern const struct file_operations simple_dir_operations;
2796 extern const struct inode_operations simple_dir_inode_operations;
2797 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2798 struct dentry *d_alloc_name(struct dentry *, const char *);
2799 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2800 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2801 extern void simple_release_fs(struct vfsmount **mount, int *count);
2803 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2804 loff_t *ppos, const void *from, size_t available);
2805 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2806 const void __user *from, size_t count);
2808 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2809 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2811 extern int generic_check_addressable(unsigned, u64);
2813 #ifdef CONFIG_MIGRATION
2814 extern int buffer_migrate_page(struct address_space *,
2815 struct page *, struct page *,
2818 #define buffer_migrate_page NULL
2821 extern int inode_change_ok(const struct inode *, struct iattr *);
2822 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2823 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2825 extern int file_update_time(struct file *file);
2827 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2828 extern void save_mount_options(struct super_block *sb, char *options);
2829 extern void replace_mount_options(struct super_block *sb, char *options);
2831 static inline bool io_is_direct(struct file *filp)
2833 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2836 static inline int iocb_flags(struct file *file)
2839 if (file->f_flags & O_APPEND)
2841 if (io_is_direct(file))
2846 static inline ino_t parent_ino(struct dentry *dentry)
2851 * Don't strictly need d_lock here? If the parent ino could change
2852 * then surely we'd have a deeper race in the caller?
2854 spin_lock(&dentry->d_lock);
2855 res = dentry->d_parent->d_inode->i_ino;
2856 spin_unlock(&dentry->d_lock);
2860 /* Transaction based IO helpers */
2863 * An argresp is stored in an allocated page and holds the
2864 * size of the argument or response, along with its content
2866 struct simple_transaction_argresp {
2871 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2873 char *simple_transaction_get(struct file *file, const char __user *buf,
2875 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2876 size_t size, loff_t *pos);
2877 int simple_transaction_release(struct inode *inode, struct file *file);
2879 void simple_transaction_set(struct file *file, size_t n);
2882 * simple attribute files
2884 * These attributes behave similar to those in sysfs:
2886 * Writing to an attribute immediately sets a value, an open file can be
2887 * written to multiple times.
2889 * Reading from an attribute creates a buffer from the value that might get
2890 * read with multiple read calls. When the attribute has been read
2891 * completely, no further read calls are possible until the file is opened
2894 * All attributes contain a text representation of a numeric value
2895 * that are accessed with the get() and set() functions.
2897 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2898 static int __fops ## _open(struct inode *inode, struct file *file) \
2900 __simple_attr_check_format(__fmt, 0ull); \
2901 return simple_attr_open(inode, file, __get, __set, __fmt); \
2903 static const struct file_operations __fops = { \
2904 .owner = THIS_MODULE, \
2905 .open = __fops ## _open, \
2906 .release = simple_attr_release, \
2907 .read = simple_attr_read, \
2908 .write = simple_attr_write, \
2909 .llseek = generic_file_llseek, \
2912 static inline __printf(1, 2)
2913 void __simple_attr_check_format(const char *fmt, ...)
2915 /* don't do anything, just let the compiler check the arguments; */
2918 int simple_attr_open(struct inode *inode, struct file *file,
2919 int (*get)(void *, u64 *), int (*set)(void *, u64),
2921 int simple_attr_release(struct inode *inode, struct file *file);
2922 ssize_t simple_attr_read(struct file *file, char __user *buf,
2923 size_t len, loff_t *ppos);
2924 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2925 size_t len, loff_t *ppos);
2928 int proc_nr_files(struct ctl_table *table, int write,
2929 void __user *buffer, size_t *lenp, loff_t *ppos);
2930 int proc_nr_dentry(struct ctl_table *table, int write,
2931 void __user *buffer, size_t *lenp, loff_t *ppos);
2932 int proc_nr_inodes(struct ctl_table *table, int write,
2933 void __user *buffer, size_t *lenp, loff_t *ppos);
2934 int __init get_filesystem_list(char *buf);
2936 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2937 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2939 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2940 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2941 (flag & __FMODE_NONOTIFY)))
2943 static inline int is_sxid(umode_t mode)
2945 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2948 static inline int check_sticky(struct inode *dir, struct inode *inode)
2950 if (!(dir->i_mode & S_ISVTX))
2953 return __check_sticky(dir, inode);
2956 static inline void inode_has_no_xattr(struct inode *inode)
2958 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2959 inode->i_flags |= S_NOSEC;
2962 static inline bool is_root_inode(struct inode *inode)
2964 return inode == inode->i_sb->s_root->d_inode;
2967 static inline bool dir_emit(struct dir_context *ctx,
2968 const char *name, int namelen,
2969 u64 ino, unsigned type)
2971 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2973 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2975 return ctx->actor(ctx, ".", 1, ctx->pos,
2976 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2978 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2980 return ctx->actor(ctx, "..", 2, ctx->pos,
2981 parent_ino(file->f_path.dentry), DT_DIR) == 0;
2983 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2985 if (ctx->pos == 0) {
2986 if (!dir_emit_dot(file, ctx))
2990 if (ctx->pos == 1) {
2991 if (!dir_emit_dotdot(file, ctx))
2997 static inline bool dir_relax(struct inode *inode)
2999 mutex_unlock(&inode->i_mutex);
3000 mutex_lock(&inode->i_mutex);
3001 return !IS_DEADDIR(inode);
3004 #endif /* _LINUX_FS_H */