4 #include <linux/linkage.h>
5 #include <linux/wait.h>
6 #include <linux/kdev_t.h>
7 #include <linux/dcache.h>
8 #include <linux/path.h>
9 #include <linux/stat.h>
10 #include <linux/cache.h>
11 #include <linux/list.h>
12 #include <linux/list_lru.h>
13 #include <linux/llist.h>
14 #include <linux/radix-tree.h>
15 #include <linux/rbtree.h>
16 #include <linux/init.h>
17 #include <linux/pid.h>
18 #include <linux/bug.h>
19 #include <linux/mutex.h>
20 #include <linux/rwsem.h>
21 #include <linux/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
32 #include <linux/workqueue.h>
33 #include <linux/percpu-rwsem.h>
35 #include <asm/byteorder.h>
36 #include <uapi/linux/fs.h>
38 struct backing_dev_info;
40 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;
56 struct fscrypt_operations;
58 extern void __init inode_init(void);
59 extern void __init inode_init_early(void);
60 extern void __init files_init(void);
61 extern void __init files_maxfiles_init(void);
63 extern struct files_stat_struct files_stat;
64 extern unsigned long get_max_files(void);
65 extern int sysctl_nr_open;
66 extern struct inodes_stat_t inodes_stat;
67 extern int leases_enable, lease_break_time;
68 extern int sysctl_protected_symlinks;
69 extern int sysctl_protected_hardlinks;
72 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
73 struct buffer_head *bh_result, int create);
74 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
75 ssize_t bytes, void *private);
76 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
78 #define MAY_EXEC 0x00000001
79 #define MAY_WRITE 0x00000002
80 #define MAY_READ 0x00000004
81 #define MAY_APPEND 0x00000008
82 #define MAY_ACCESS 0x00000010
83 #define MAY_OPEN 0x00000020
84 #define MAY_CHDIR 0x00000040
85 /* called from RCU mode, don't block */
86 #define MAY_NOT_BLOCK 0x00000080
89 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
90 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
93 /* file is open for reading */
94 #define FMODE_READ ((__force fmode_t)0x1)
95 /* file is open for writing */
96 #define FMODE_WRITE ((__force fmode_t)0x2)
97 /* file is seekable */
98 #define FMODE_LSEEK ((__force fmode_t)0x4)
99 /* file can be accessed using pread */
100 #define FMODE_PREAD ((__force fmode_t)0x8)
101 /* file can be accessed using pwrite */
102 #define FMODE_PWRITE ((__force fmode_t)0x10)
103 /* File is opened for execution with sys_execve / sys_uselib */
104 #define FMODE_EXEC ((__force fmode_t)0x20)
105 /* File is opened with O_NDELAY (only set for block devices) */
106 #define FMODE_NDELAY ((__force fmode_t)0x40)
107 /* File is opened with O_EXCL (only set for block devices) */
108 #define FMODE_EXCL ((__force fmode_t)0x80)
109 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
110 (specialy hack for floppy.c) */
111 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
112 /* 32bit hashes as llseek() offset (for directories) */
113 #define FMODE_32BITHASH ((__force fmode_t)0x200)
114 /* 64bit hashes as llseek() offset (for directories) */
115 #define FMODE_64BITHASH ((__force fmode_t)0x400)
118 * Don't update ctime and mtime.
120 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
121 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
123 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
125 /* Expect random access pattern */
126 #define FMODE_RANDOM ((__force fmode_t)0x1000)
128 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
129 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
131 /* File is opened with O_PATH; almost nothing can be done with it */
132 #define FMODE_PATH ((__force fmode_t)0x4000)
134 /* File needs atomic accesses to f_pos */
135 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
136 /* Write access to underlying fs */
137 #define FMODE_WRITER ((__force fmode_t)0x10000)
138 /* Has read method(s) */
139 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
140 /* Has write method(s) */
141 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
143 /* File was opened by fanotify and shouldn't generate fanotify events */
144 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
147 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
148 * that indicates that they should check the contents of the iovec are
149 * valid, but not check the memory that the iovec elements
152 #define CHECK_IOVEC_ONLY -1
155 * The below are the various read and write types that we support. Some of
156 * them include behavioral modifiers that send information down to the
157 * block layer and IO scheduler. Terminology:
159 * The block layer uses device plugging to defer IO a little bit, in
160 * the hope that we will see more IO very shortly. This increases
161 * coalescing of adjacent IO and thus reduces the number of IOs we
162 * have to send to the device. It also allows for better queuing,
163 * if the IO isn't mergeable. If the caller is going to be waiting
164 * for the IO, then he must ensure that the device is unplugged so
165 * that the IO is dispatched to the driver.
167 * All IO is handled async in Linux. This is fine for background
168 * writes, but for reads or writes that someone waits for completion
169 * on, we want to notify the block layer and IO scheduler so that they
170 * know about it. That allows them to make better scheduling
171 * decisions. So when the below references 'sync' and 'async', it
172 * is referencing this priority hint.
174 * With that in mind, the available types are:
176 * READ A normal read operation. Device will be plugged.
177 * READ_SYNC A synchronous read. Device is not plugged, caller can
178 * immediately wait on this read without caring about
180 * READA Used for read-ahead operations. Lower priority, and the
181 * block layer could (in theory) choose to ignore this
182 * request if it runs into resource problems.
183 * WRITE A normal async write. Device will be plugged.
184 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
185 * the hint that someone will be waiting on this IO
186 * shortly. The write equivalent of READ_SYNC.
187 * WRITE_ODIRECT Special case write for O_DIRECT only.
188 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
189 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
190 * non-volatile media on completion.
191 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
192 * by a cache flush and data is guaranteed to be on
193 * non-volatile media on completion.
196 #define RW_MASK REQ_WRITE
197 #define RWA_MASK REQ_RAHEAD
200 #define WRITE RW_MASK
201 #define READA RWA_MASK
203 #define READ_SYNC (READ | REQ_SYNC)
204 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
205 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
206 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
207 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
208 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
211 * Attribute flags. These should be or-ed together to figure out what
214 #define ATTR_MODE (1 << 0)
215 #define ATTR_UID (1 << 1)
216 #define ATTR_GID (1 << 2)
217 #define ATTR_SIZE (1 << 3)
218 #define ATTR_ATIME (1 << 4)
219 #define ATTR_MTIME (1 << 5)
220 #define ATTR_CTIME (1 << 6)
221 #define ATTR_ATIME_SET (1 << 7)
222 #define ATTR_MTIME_SET (1 << 8)
223 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
224 #define ATTR_ATTR_FLAG (1 << 10)
225 #define ATTR_KILL_SUID (1 << 11)
226 #define ATTR_KILL_SGID (1 << 12)
227 #define ATTR_FILE (1 << 13)
228 #define ATTR_KILL_PRIV (1 << 14)
229 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
230 #define ATTR_TIMES_SET (1 << 16)
231 #define ATTR_TOUCH (1 << 17)
234 * Whiteout is represented by a char device. The following constants define the
235 * mode and device number to use.
237 #define WHITEOUT_MODE 0
238 #define WHITEOUT_DEV 0
241 * This is the Inode Attributes structure, used for notify_change(). It
242 * uses the above definitions as flags, to know which values have changed.
243 * Also, in this manner, a Filesystem can look at only the values it cares
244 * about. Basically, these are the attributes that the VFS layer can
245 * request to change from the FS layer.
247 * Derek Atkins <warlord@MIT.EDU> 94-10-20
250 unsigned int ia_valid;
255 struct timespec ia_atime;
256 struct timespec ia_mtime;
257 struct timespec ia_ctime;
260 * Not an attribute, but an auxiliary info for filesystems wanting to
261 * implement an ftruncate() like method. NOTE: filesystem should
262 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
264 struct file *ia_file;
268 * Includes for diskquotas.
270 #include <linux/quota.h>
273 * Maximum number of layers of fs stack. Needs to be limited to
274 * prevent kernel stack overflow
276 #define FILESYSTEM_MAX_STACK_DEPTH 2
279 * enum positive_aop_returns - aop return codes with specific semantics
281 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
282 * completed, that the page is still locked, and
283 * should be considered active. The VM uses this hint
284 * to return the page to the active list -- it won't
285 * be a candidate for writeback again in the near
286 * future. Other callers must be careful to unlock
287 * the page if they get this return. Returned by
290 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
291 * unlocked it and the page might have been truncated.
292 * The caller should back up to acquiring a new page and
293 * trying again. The aop will be taking reasonable
294 * precautions not to livelock. If the caller held a page
295 * reference, it should drop it before retrying. Returned
298 * address_space_operation functions return these large constants to indicate
299 * special semantics to the caller. These are much larger than the bytes in a
300 * page to allow for functions that return the number of bytes operated on in a
304 enum positive_aop_returns {
305 AOP_WRITEPAGE_ACTIVATE = 0x80000,
306 AOP_TRUNCATED_PAGE = 0x80001,
309 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
310 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
311 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
312 * helper code (eg buffer layer)
313 * to clear GFP_FS from alloc */
316 * oh the beauties of C type declarations.
319 struct address_space;
320 struct writeback_control;
322 #define IOCB_EVENTFD (1 << 0)
323 #define IOCB_APPEND (1 << 1)
324 #define IOCB_DIRECT (1 << 2)
327 struct file *ki_filp;
329 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
334 static inline bool is_sync_kiocb(struct kiocb *kiocb)
336 return kiocb->ki_complete == NULL;
339 static inline int iocb_flags(struct file *file);
341 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
343 *kiocb = (struct kiocb) {
345 .ki_flags = iocb_flags(filp),
350 * "descriptor" for what we're up to with a read.
351 * This allows us to use the same read code yet
352 * have multiple different users of the data that
353 * we read from a file.
355 * The simplest case just copies the data to user
368 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
369 unsigned long, unsigned long);
371 struct address_space_operations {
372 int (*writepage)(struct page *page, struct writeback_control *wbc);
373 int (*readpage)(struct file *, struct page *);
375 /* Write back some dirty pages from this mapping. */
376 int (*writepages)(struct address_space *, struct writeback_control *);
378 /* Set a page dirty. Return true if this dirtied it */
379 int (*set_page_dirty)(struct page *page);
381 int (*readpages)(struct file *filp, struct address_space *mapping,
382 struct list_head *pages, unsigned nr_pages);
384 int (*write_begin)(struct file *, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned flags,
386 struct page **pagep, void **fsdata);
387 int (*write_end)(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned copied,
389 struct page *page, void *fsdata);
391 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
392 sector_t (*bmap)(struct address_space *, sector_t);
393 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
394 int (*releasepage) (struct page *, gfp_t);
395 void (*freepage)(struct page *);
396 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
398 * migrate the contents of a page to the specified target. If
399 * migrate_mode is MIGRATE_ASYNC, it must not block.
401 int (*migratepage) (struct address_space *,
402 struct page *, struct page *, enum migrate_mode);
403 int (*launder_page) (struct page *);
404 int (*is_partially_uptodate) (struct page *, unsigned long,
406 void (*is_dirty_writeback) (struct page *, bool *, bool *);
407 int (*error_remove_page)(struct address_space *, struct page *);
409 /* swapfile support */
410 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
412 void (*swap_deactivate)(struct file *file);
415 extern const struct address_space_operations empty_aops;
418 * pagecache_write_begin/pagecache_write_end must be used by general code
419 * to write into the pagecache.
421 int pagecache_write_begin(struct file *, struct address_space *mapping,
422 loff_t pos, unsigned len, unsigned flags,
423 struct page **pagep, void **fsdata);
425 int pagecache_write_end(struct file *, struct address_space *mapping,
426 loff_t pos, unsigned len, unsigned copied,
427 struct page *page, void *fsdata);
429 struct address_space {
430 struct inode *host; /* owner: inode, block_device */
431 struct radix_tree_root page_tree; /* radix tree of all pages */
432 spinlock_t tree_lock; /* and lock protecting it */
433 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
434 struct rb_root i_mmap; /* tree of private and shared mappings */
435 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
436 /* Protected by tree_lock together with the radix tree */
437 unsigned long nrpages; /* number of total pages */
438 unsigned long nrshadows; /* number of shadow entries */
439 pgoff_t writeback_index;/* writeback starts here */
440 const struct address_space_operations *a_ops; /* methods */
441 unsigned long flags; /* error bits/gfp mask */
442 spinlock_t private_lock; /* for use by the address_space */
443 struct list_head private_list; /* ditto */
444 void *private_data; /* ditto */
445 } __attribute__((aligned(sizeof(long))));
447 * On most architectures that alignment is already the case; but
448 * must be enforced here for CRIS, to let the least significant bit
449 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
451 struct request_queue;
453 struct block_device {
454 dev_t bd_dev; /* not a kdev_t - it's a search key */
456 struct inode * bd_inode; /* will die */
457 struct super_block * bd_super;
458 struct mutex bd_mutex; /* open/close mutex */
459 struct list_head bd_inodes;
463 bool bd_write_holder;
465 struct list_head bd_holder_disks;
467 struct block_device * bd_contains;
468 unsigned bd_block_size;
469 struct hd_struct * bd_part;
470 /* number of times partitions within this device have been opened. */
471 unsigned bd_part_count;
473 struct gendisk * bd_disk;
474 struct request_queue * bd_queue;
475 struct list_head bd_list;
477 * Private data. You must have bd_claim'ed the block_device
478 * to use this. NOTE: bd_claim allows an owner to claim
479 * the same device multiple times, the owner must take special
480 * care to not mess up bd_private for that case.
482 unsigned long bd_private;
484 /* The counter of freeze processes */
485 int bd_fsfreeze_count;
486 /* Mutex for freeze */
487 struct mutex bd_fsfreeze_mutex;
491 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
494 #define PAGECACHE_TAG_DIRTY 0
495 #define PAGECACHE_TAG_WRITEBACK 1
496 #define PAGECACHE_TAG_TOWRITE 2
498 int mapping_tagged(struct address_space *mapping, int tag);
500 static inline void i_mmap_lock_write(struct address_space *mapping)
502 down_write(&mapping->i_mmap_rwsem);
505 static inline void i_mmap_unlock_write(struct address_space *mapping)
507 up_write(&mapping->i_mmap_rwsem);
510 static inline void i_mmap_lock_read(struct address_space *mapping)
512 down_read(&mapping->i_mmap_rwsem);
515 static inline void i_mmap_unlock_read(struct address_space *mapping)
517 up_read(&mapping->i_mmap_rwsem);
521 * Might pages of this file be mapped into userspace?
523 static inline int mapping_mapped(struct address_space *mapping)
525 return !RB_EMPTY_ROOT(&mapping->i_mmap);
529 * Might pages of this file have been modified in userspace?
530 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
531 * marks vma as VM_SHARED if it is shared, and the file was opened for
532 * writing i.e. vma may be mprotected writable even if now readonly.
534 * If i_mmap_writable is negative, no new writable mappings are allowed. You
535 * can only deny writable mappings, if none exists right now.
537 static inline int mapping_writably_mapped(struct address_space *mapping)
539 return atomic_read(&mapping->i_mmap_writable) > 0;
542 static inline int mapping_map_writable(struct address_space *mapping)
544 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
548 static inline void mapping_unmap_writable(struct address_space *mapping)
550 atomic_dec(&mapping->i_mmap_writable);
553 static inline int mapping_deny_writable(struct address_space *mapping)
555 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
559 static inline void mapping_allow_writable(struct address_space *mapping)
561 atomic_inc(&mapping->i_mmap_writable);
565 * Use sequence counter to get consistent i_size on 32-bit processors.
567 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
568 #include <linux/seqlock.h>
569 #define __NEED_I_SIZE_ORDERED
570 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
572 #define i_size_ordered_init(inode) do { } while (0)
576 #define ACL_NOT_CACHED ((void *)(-1))
578 #define IOP_FASTPERM 0x0001
579 #define IOP_LOOKUP 0x0002
580 #define IOP_NOFOLLOW 0x0004
583 * Keep mostly read-only and often accessed (especially for
584 * the RCU path lookup and 'stat' data) fields at the beginning
585 * of the 'struct inode'
589 unsigned short i_opflags;
592 unsigned int i_flags;
594 #ifdef CONFIG_FS_POSIX_ACL
595 struct posix_acl *i_acl;
596 struct posix_acl *i_default_acl;
599 const struct inode_operations *i_op;
600 struct super_block *i_sb;
601 struct address_space *i_mapping;
603 #ifdef CONFIG_SECURITY
607 /* Stat data, not accessed from path walking */
610 * Filesystems may only read i_nlink directly. They shall use the
611 * following functions for modification:
613 * (set|clear|inc|drop)_nlink
614 * inode_(inc|dec)_link_count
617 const unsigned int i_nlink;
618 unsigned int __i_nlink;
622 struct timespec i_atime;
623 struct timespec i_mtime;
624 struct timespec i_ctime;
625 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
626 unsigned short i_bytes;
627 unsigned int i_blkbits;
630 #ifdef __NEED_I_SIZE_ORDERED
631 seqcount_t i_size_seqcount;
635 unsigned long i_state;
636 struct mutex i_mutex;
638 unsigned long dirtied_when; /* jiffies of first dirtying */
639 unsigned long dirtied_time_when;
641 struct hlist_node i_hash;
642 struct list_head i_io_list; /* backing dev IO list */
643 #ifdef CONFIG_CGROUP_WRITEBACK
644 struct bdi_writeback *i_wb; /* the associated cgroup wb */
646 /* foreign inode detection, see wbc_detach_inode() */
648 u16 i_wb_frn_avg_time;
649 u16 i_wb_frn_history;
651 struct list_head i_lru; /* inode LRU list */
652 struct list_head i_sb_list;
654 struct hlist_head i_dentry;
655 struct rcu_head i_rcu;
659 atomic_t i_dio_count;
660 atomic_t i_writecount;
662 atomic_t i_readcount; /* struct files open RO */
664 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
665 struct file_lock_context *i_flctx;
666 struct address_space i_data;
667 struct list_head i_devices;
669 struct pipe_inode_info *i_pipe;
670 struct block_device *i_bdev;
677 #ifdef CONFIG_FSNOTIFY
678 __u32 i_fsnotify_mask; /* all events this inode cares about */
679 struct hlist_head i_fsnotify_marks;
682 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
683 struct fscrypt_info *i_crypt_info;
685 void *i_private; /* fs or device private pointer */
688 static inline unsigned int i_blocksize(const struct inode *node)
690 return (1 << node->i_blkbits);
693 static inline int inode_unhashed(struct inode *inode)
695 return hlist_unhashed(&inode->i_hash);
699 * inode->i_mutex nesting subclasses for the lock validator:
701 * 0: the object of the current VFS operation
705 * 4: second non-directory
706 * 5: second parent (when locking independent directories in rename)
708 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
709 * non-directories at once.
711 * The locking order between these classes is
712 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
714 enum inode_i_mutex_lock_class
724 static inline void inode_lock(struct inode *inode)
726 mutex_lock(&inode->i_mutex);
729 static inline void inode_unlock(struct inode *inode)
731 mutex_unlock(&inode->i_mutex);
734 static inline int inode_trylock(struct inode *inode)
736 return mutex_trylock(&inode->i_mutex);
739 static inline int inode_is_locked(struct inode *inode)
741 return mutex_is_locked(&inode->i_mutex);
744 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
746 mutex_lock_nested(&inode->i_mutex, subclass);
749 void lock_two_nondirectories(struct inode *, struct inode*);
750 void unlock_two_nondirectories(struct inode *, struct inode*);
753 * NOTE: in a 32bit arch with a preemptable kernel and
754 * an UP compile the i_size_read/write must be atomic
755 * with respect to the local cpu (unlike with preempt disabled),
756 * but they don't need to be atomic with respect to other cpus like in
757 * true SMP (so they need either to either locally disable irq around
758 * the read or for example on x86 they can be still implemented as a
759 * cmpxchg8b without the need of the lock prefix). For SMP compiles
760 * and 64bit archs it makes no difference if preempt is enabled or not.
762 static inline loff_t i_size_read(const struct inode *inode)
764 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
769 seq = read_seqcount_begin(&inode->i_size_seqcount);
770 i_size = inode->i_size;
771 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
773 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
777 i_size = inode->i_size;
781 return inode->i_size;
786 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
787 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
788 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
790 static inline void i_size_write(struct inode *inode, loff_t i_size)
792 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
794 write_seqcount_begin(&inode->i_size_seqcount);
795 inode->i_size = i_size;
796 write_seqcount_end(&inode->i_size_seqcount);
798 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
800 inode->i_size = i_size;
803 inode->i_size = i_size;
807 /* Helper functions so that in most cases filesystems will
808 * not need to deal directly with kuid_t and kgid_t and can
809 * instead deal with the raw numeric values that are stored
812 static inline uid_t i_uid_read(const struct inode *inode)
814 return from_kuid(&init_user_ns, inode->i_uid);
817 static inline gid_t i_gid_read(const struct inode *inode)
819 return from_kgid(&init_user_ns, inode->i_gid);
822 static inline void i_uid_write(struct inode *inode, uid_t uid)
824 inode->i_uid = make_kuid(&init_user_ns, uid);
827 static inline void i_gid_write(struct inode *inode, gid_t gid)
829 inode->i_gid = make_kgid(&init_user_ns, gid);
832 static inline unsigned iminor(const struct inode *inode)
834 return MINOR(inode->i_rdev);
837 static inline unsigned imajor(const struct inode *inode)
839 return MAJOR(inode->i_rdev);
842 extern struct block_device *I_BDEV(struct inode *inode);
845 rwlock_t lock; /* protects pid, uid, euid fields */
846 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
847 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
848 kuid_t uid, euid; /* uid/euid of process setting the owner */
849 int signum; /* posix.1b rt signal to be delivered on IO */
853 * Track a single file's readahead state
855 struct file_ra_state {
856 pgoff_t start; /* where readahead started */
857 unsigned int size; /* # of readahead pages */
858 unsigned int async_size; /* do asynchronous readahead when
859 there are only # of pages ahead */
861 unsigned int ra_pages; /* Maximum readahead window */
862 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
863 loff_t prev_pos; /* Cache last read() position */
867 * Check if @index falls in the readahead windows.
869 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
871 return (index >= ra->start &&
872 index < ra->start + ra->size);
877 struct llist_node fu_llist;
878 struct rcu_head fu_rcuhead;
881 struct inode *f_inode; /* cached value */
882 const struct file_operations *f_op;
885 * Protects f_ep_links, f_flags.
886 * Must not be taken from IRQ context.
889 atomic_long_t f_count;
890 unsigned int f_flags;
892 struct mutex f_pos_lock;
894 struct fown_struct f_owner;
895 const struct cred *f_cred;
896 struct file_ra_state f_ra;
899 #ifdef CONFIG_SECURITY
902 /* needed for tty driver, and maybe others */
906 /* Used by fs/eventpoll.c to link all the hooks to this file */
907 struct list_head f_ep_links;
908 struct list_head f_tfile_llink;
909 #endif /* #ifdef CONFIG_EPOLL */
910 struct address_space *f_mapping;
911 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
916 /* file identifier */
917 unsigned char f_handle[0];
920 static inline struct file *get_file(struct file *f)
922 atomic_long_inc(&f->f_count);
925 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
926 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
927 #define file_count(x) atomic_long_read(&(x)->f_count)
929 #define MAX_NON_LFS ((1UL<<31) - 1)
931 /* Page cache limit. The filesystems should put that into their s_maxbytes
932 limits, otherwise bad things can happen in VM. */
933 #if BITS_PER_LONG==32
934 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
935 #elif BITS_PER_LONG==64
936 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
941 #define FL_DELEG 4 /* NFSv4 delegation */
942 #define FL_ACCESS 8 /* not trying to lock, just looking */
943 #define FL_EXISTS 16 /* when unlocking, test for existence */
944 #define FL_LEASE 32 /* lease held on this file */
945 #define FL_CLOSE 64 /* unlock on close */
946 #define FL_SLEEP 128 /* A blocking lock */
947 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
948 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
949 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
950 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
953 * Special return value from posix_lock_file() and vfs_lock_file() for
954 * asynchronous locking.
956 #define FILE_LOCK_DEFERRED 1
958 /* legacy typedef, should eventually be removed */
959 typedef void *fl_owner_t;
963 struct file_lock_operations {
964 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
965 void (*fl_release_private)(struct file_lock *);
968 struct lock_manager_operations {
969 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
970 unsigned long (*lm_owner_key)(struct file_lock *);
971 fl_owner_t (*lm_get_owner)(fl_owner_t);
972 void (*lm_put_owner)(fl_owner_t);
973 void (*lm_notify)(struct file_lock *); /* unblock callback */
974 int (*lm_grant)(struct file_lock *, int);
975 bool (*lm_break)(struct file_lock *);
976 int (*lm_change)(struct file_lock *, int, struct list_head *);
977 void (*lm_setup)(struct file_lock *, void **);
980 struct lock_manager {
981 struct list_head list;
983 * NFSv4 and up also want opens blocked during the grace period;
990 void locks_start_grace(struct net *, struct lock_manager *);
991 void locks_end_grace(struct lock_manager *);
992 int locks_in_grace(struct net *);
993 int opens_in_grace(struct net *);
995 /* that will die - we need it for nfs_lock_info */
996 #include <linux/nfs_fs_i.h>
999 * struct file_lock represents a generic "file lock". It's used to represent
1000 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1001 * note that the same struct is used to represent both a request for a lock and
1002 * the lock itself, but the same object is never used for both.
1004 * FIXME: should we create a separate "struct lock_request" to help distinguish
1007 * The varous i_flctx lists are ordered by:
1010 * 2) lock range start
1013 * Obviously, the last two criteria only matter for POSIX locks.
1016 struct file_lock *fl_next; /* singly linked list for this inode */
1017 struct list_head fl_list; /* link into file_lock_context */
1018 struct hlist_node fl_link; /* node in global lists */
1019 struct list_head fl_block; /* circular list of blocked processes */
1020 fl_owner_t fl_owner;
1021 unsigned int fl_flags;
1022 unsigned char fl_type;
1023 unsigned int fl_pid;
1024 int fl_link_cpu; /* what cpu's list is this on? */
1025 struct pid *fl_nspid;
1026 wait_queue_head_t fl_wait;
1027 struct file *fl_file;
1031 struct fasync_struct * fl_fasync; /* for lease break notifications */
1032 /* for lease breaks: */
1033 unsigned long fl_break_time;
1034 unsigned long fl_downgrade_time;
1036 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1037 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1039 struct nfs_lock_info nfs_fl;
1040 struct nfs4_lock_info nfs4_fl;
1042 struct list_head link; /* link in AFS vnode's pending_locks list */
1043 int state; /* state of grant or error if -ve */
1048 struct file_lock_context {
1049 spinlock_t flc_lock;
1050 struct list_head flc_flock;
1051 struct list_head flc_posix;
1052 struct list_head flc_lease;
1055 /* The following constant reflects the upper bound of the file/locking space */
1057 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1058 #define OFFSET_MAX INT_LIMIT(loff_t)
1059 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1062 #include <linux/fcntl.h>
1064 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1066 #ifdef CONFIG_FILE_LOCKING
1067 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1068 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1069 struct flock __user *);
1071 #if BITS_PER_LONG == 32
1072 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1073 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1074 struct flock64 __user *);
1077 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1078 extern int fcntl_getlease(struct file *filp);
1081 void locks_free_lock_context(struct file_lock_context *ctx);
1082 void locks_free_lock(struct file_lock *fl);
1083 extern void locks_init_lock(struct file_lock *);
1084 extern struct file_lock * locks_alloc_lock(void);
1085 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1086 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1087 extern void locks_remove_posix(struct file *, fl_owner_t);
1088 extern void locks_remove_file(struct file *);
1089 extern void locks_release_private(struct file_lock *);
1090 extern void posix_test_lock(struct file *, struct file_lock *);
1091 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1092 extern int posix_unblock_lock(struct file_lock *);
1093 extern int vfs_test_lock(struct file *, struct file_lock *);
1094 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1095 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1096 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1097 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1098 extern void lease_get_mtime(struct inode *, struct timespec *time);
1099 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1100 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1101 extern int lease_modify(struct file_lock *, int, struct list_head *);
1102 struct files_struct;
1103 extern void show_fd_locks(struct seq_file *f,
1104 struct file *filp, struct files_struct *files);
1105 #else /* !CONFIG_FILE_LOCKING */
1106 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1107 struct flock __user *user)
1112 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1113 unsigned int cmd, struct flock __user *user)
1118 #if BITS_PER_LONG == 32
1119 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1120 struct flock64 __user *user)
1125 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1126 unsigned int cmd, struct flock64 __user *user)
1131 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1136 static inline int fcntl_getlease(struct file *filp)
1142 locks_free_lock_context(struct file_lock_context *ctx)
1146 static inline void locks_init_lock(struct file_lock *fl)
1151 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1156 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1161 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1166 static inline void locks_remove_file(struct file *filp)
1171 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1176 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1177 struct file_lock *conflock)
1182 static inline int posix_unblock_lock(struct file_lock *waiter)
1187 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1192 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1193 struct file_lock *fl, struct file_lock *conf)
1198 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1203 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1208 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1213 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1218 static inline int generic_setlease(struct file *filp, long arg,
1219 struct file_lock **flp, void **priv)
1224 static inline int vfs_setlease(struct file *filp, long arg,
1225 struct file_lock **lease, void **priv)
1230 static inline int lease_modify(struct file_lock *fl, int arg,
1231 struct list_head *dispose)
1236 struct files_struct;
1237 static inline void show_fd_locks(struct seq_file *f,
1238 struct file *filp, struct files_struct *files) {}
1239 #endif /* !CONFIG_FILE_LOCKING */
1241 static inline struct inode *file_inode(const struct file *f)
1246 static inline struct dentry *file_dentry(const struct file *file)
1248 struct dentry *dentry = file->f_path.dentry;
1250 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1251 return dentry->d_op->d_real(dentry, file_inode(file));
1256 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1258 return locks_lock_inode_wait(file_inode(filp), fl);
1261 struct fasync_struct {
1265 struct fasync_struct *fa_next; /* singly linked list */
1266 struct file *fa_file;
1267 struct rcu_head fa_rcu;
1270 #define FASYNC_MAGIC 0x4601
1272 /* SMP safe fasync helpers: */
1273 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1274 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1275 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1276 extern struct fasync_struct *fasync_alloc(void);
1277 extern void fasync_free(struct fasync_struct *);
1279 /* can be called from interrupts */
1280 extern void kill_fasync(struct fasync_struct **, int, int);
1282 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1283 extern void f_setown(struct file *filp, unsigned long arg, int force);
1284 extern void f_delown(struct file *filp);
1285 extern pid_t f_getown(struct file *filp);
1286 extern int send_sigurg(struct fown_struct *fown);
1294 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1295 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1296 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1297 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1298 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1301 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1302 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1303 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1305 /* Possible states of 'frozen' field */
1307 SB_UNFROZEN = 0, /* FS is unfrozen */
1308 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1309 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1310 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1311 * internal threads if needed) */
1312 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1315 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1318 int frozen; /* Is sb frozen? */
1319 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1320 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1323 struct super_block {
1324 struct list_head s_list; /* Keep this first */
1325 dev_t s_dev; /* search index; _not_ kdev_t */
1326 unsigned char s_blocksize_bits;
1327 unsigned long s_blocksize;
1328 loff_t s_maxbytes; /* Max file size */
1329 struct file_system_type *s_type;
1330 const struct super_operations *s_op;
1331 const struct dquot_operations *dq_op;
1332 const struct quotactl_ops *s_qcop;
1333 const struct export_operations *s_export_op;
1334 unsigned long s_flags;
1335 unsigned long s_iflags; /* internal SB_I_* flags */
1336 unsigned long s_magic;
1337 struct dentry *s_root;
1338 struct rw_semaphore s_umount;
1341 #ifdef CONFIG_SECURITY
1344 const struct xattr_handler **s_xattr;
1346 const struct fscrypt_operations *s_cop;
1348 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1349 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1350 struct block_device *s_bdev;
1351 struct backing_dev_info *s_bdi;
1352 struct mtd_info *s_mtd;
1353 struct hlist_node s_instances;
1354 unsigned int s_quota_types; /* Bitmask of supported quota types */
1355 struct quota_info s_dquot; /* Diskquota specific options */
1357 struct sb_writers s_writers;
1359 char s_id[32]; /* Informational name */
1360 u8 s_uuid[16]; /* UUID */
1362 void *s_fs_info; /* Filesystem private info */
1363 unsigned int s_max_links;
1366 /* Granularity of c/m/atime in ns.
1367 Cannot be worse than a second */
1371 * The next field is for VFS *only*. No filesystems have any business
1372 * even looking at it. You had been warned.
1374 struct mutex s_vfs_rename_mutex; /* Kludge */
1377 * Filesystem subtype. If non-empty the filesystem type field
1378 * in /proc/mounts will be "type.subtype"
1383 * Saved mount options for lazy filesystems using
1384 * generic_show_options()
1386 char __rcu *s_options;
1387 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1390 * Saved pool identifier for cleancache (-1 means none)
1392 int cleancache_poolid;
1394 struct shrinker s_shrink; /* per-sb shrinker handle */
1396 /* Number of inodes with nlink == 0 but still referenced */
1397 atomic_long_t s_remove_count;
1399 /* Being remounted read-only */
1400 int s_readonly_remount;
1402 /* AIO completions deferred from interrupt context */
1403 struct workqueue_struct *s_dio_done_wq;
1404 struct hlist_head s_pins;
1407 * Keep the lru lists last in the structure so they always sit on their
1408 * own individual cachelines.
1410 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1411 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1412 struct rcu_head rcu;
1413 struct work_struct destroy_work;
1415 struct mutex s_sync_lock; /* sync serialisation lock */
1418 * Indicates how deep in a filesystem stack this SB is
1422 /* s_inode_list_lock protects s_inodes */
1423 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1424 struct list_head s_inodes; /* all inodes */
1427 extern struct timespec current_fs_time(struct super_block *sb);
1430 * Snapshotting support.
1433 void __sb_end_write(struct super_block *sb, int level);
1434 int __sb_start_write(struct super_block *sb, int level, bool wait);
1436 #define __sb_writers_acquired(sb, lev) \
1437 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1438 #define __sb_writers_release(sb, lev) \
1439 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1442 * sb_end_write - drop write access to a superblock
1443 * @sb: the super we wrote to
1445 * Decrement number of writers to the filesystem. Wake up possible waiters
1446 * wanting to freeze the filesystem.
1448 static inline void sb_end_write(struct super_block *sb)
1450 __sb_end_write(sb, SB_FREEZE_WRITE);
1454 * sb_end_pagefault - drop write access to a superblock from a page fault
1455 * @sb: the super we wrote to
1457 * Decrement number of processes handling write page fault to the filesystem.
1458 * Wake up possible waiters wanting to freeze the filesystem.
1460 static inline void sb_end_pagefault(struct super_block *sb)
1462 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1466 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1467 * @sb: the super we wrote to
1469 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1470 * waiters wanting to freeze the filesystem.
1472 static inline void sb_end_intwrite(struct super_block *sb)
1474 __sb_end_write(sb, SB_FREEZE_FS);
1478 * sb_start_write - get write access to a superblock
1479 * @sb: the super we write to
1481 * When a process wants to write data or metadata to a file system (i.e. dirty
1482 * a page or an inode), it should embed the operation in a sb_start_write() -
1483 * sb_end_write() pair to get exclusion against file system freezing. This
1484 * function increments number of writers preventing freezing. If the file
1485 * system is already frozen, the function waits until the file system is
1488 * Since freeze protection behaves as a lock, users have to preserve
1489 * ordering of freeze protection and other filesystem locks. Generally,
1490 * freeze protection should be the outermost lock. In particular, we have:
1493 * -> i_mutex (write path, truncate, directory ops, ...)
1494 * -> s_umount (freeze_super, thaw_super)
1496 static inline void sb_start_write(struct super_block *sb)
1498 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1501 static inline int sb_start_write_trylock(struct super_block *sb)
1503 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1507 * sb_start_pagefault - get write access to a superblock from a page fault
1508 * @sb: the super we write to
1510 * When a process starts handling write page fault, it should embed the
1511 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1512 * exclusion against file system freezing. This is needed since the page fault
1513 * is going to dirty a page. This function increments number of running page
1514 * faults preventing freezing. If the file system is already frozen, the
1515 * function waits until the file system is thawed.
1517 * Since page fault freeze protection behaves as a lock, users have to preserve
1518 * ordering of freeze protection and other filesystem locks. It is advised to
1519 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1520 * handling code implies lock dependency:
1523 * -> sb_start_pagefault
1525 static inline void sb_start_pagefault(struct super_block *sb)
1527 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1531 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1532 * @sb: the super we write to
1534 * This is the third level of protection against filesystem freezing. It is
1535 * free for use by a filesystem. The only requirement is that it must rank
1536 * below sb_start_pagefault.
1538 * For example filesystem can call sb_start_intwrite() when starting a
1539 * transaction which somewhat eases handling of freezing for internal sources
1540 * of filesystem changes (internal fs threads, discarding preallocation on file
1543 static inline void sb_start_intwrite(struct super_block *sb)
1545 __sb_start_write(sb, SB_FREEZE_FS, true);
1549 extern bool inode_owner_or_capable(const struct inode *inode);
1552 * VFS helper functions..
1554 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1555 extern int vfs_create2(struct vfsmount *, struct inode *, struct dentry *, umode_t, bool);
1556 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1557 extern int vfs_mkdir2(struct vfsmount *, struct inode *, struct dentry *, umode_t);
1558 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1559 extern int vfs_mknod2(struct vfsmount *, struct inode *, struct dentry *, umode_t, dev_t);
1560 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1561 extern int vfs_symlink2(struct vfsmount *, struct inode *, struct dentry *, const char *);
1562 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1563 extern int vfs_link2(struct vfsmount *, struct dentry *, struct inode *, struct dentry *, struct inode **);
1564 extern int vfs_rmdir(struct inode *, struct dentry *);
1565 extern int vfs_rmdir2(struct vfsmount *, struct inode *, struct dentry *);
1566 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1567 extern int vfs_unlink2(struct vfsmount *, struct inode *, struct dentry *, struct inode **);
1568 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1569 extern int vfs_rename2(struct vfsmount *, struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1570 extern int vfs_whiteout(struct inode *, struct dentry *);
1573 * VFS dentry helper functions.
1575 extern void dentry_unhash(struct dentry *dentry);
1578 * VFS file helper functions.
1580 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1583 * VFS FS_IOC_FIEMAP helper definitions.
1585 struct fiemap_extent_info {
1586 unsigned int fi_flags; /* Flags as passed from user */
1587 unsigned int fi_extents_mapped; /* Number of mapped extents */
1588 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1589 struct fiemap_extent __user *fi_extents_start; /* Start of
1590 fiemap_extent array */
1592 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1593 u64 phys, u64 len, u32 flags);
1594 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1599 * NOTE! These match bits 12..15 of stat.st_mode
1600 * (ie "(i_mode >> 12) & 15").
1602 #define DT_UNKNOWN 0
1613 * This is the "filldir" function type, used by readdir() to let
1614 * the kernel specify what kind of dirent layout it wants to have.
1615 * This allows the kernel to read directories into kernel space or
1616 * to have different dirent layouts depending on the binary type.
1619 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1622 struct dir_context {
1623 const filldir_t actor;
1627 struct block_device_operations;
1629 /* These macros are for out of kernel modules to test that
1630 * the kernel supports the unlocked_ioctl and compat_ioctl
1631 * fields in struct file_operations. */
1632 #define HAVE_COMPAT_IOCTL 1
1633 #define HAVE_UNLOCKED_IOCTL 1
1636 * These flags let !MMU mmap() govern direct device mapping vs immediate
1637 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1639 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1640 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1641 * NOMMU_MAP_READ: Can be mapped for reading
1642 * NOMMU_MAP_WRITE: Can be mapped for writing
1643 * NOMMU_MAP_EXEC: Can be mapped for execution
1645 #define NOMMU_MAP_COPY 0x00000001
1646 #define NOMMU_MAP_DIRECT 0x00000008
1647 #define NOMMU_MAP_READ VM_MAYREAD
1648 #define NOMMU_MAP_WRITE VM_MAYWRITE
1649 #define NOMMU_MAP_EXEC VM_MAYEXEC
1651 #define NOMMU_VMFLAGS \
1652 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1657 struct file_operations {
1658 struct module *owner;
1659 loff_t (*llseek) (struct file *, loff_t, int);
1660 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1661 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1662 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1663 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1664 int (*iterate) (struct file *, struct dir_context *);
1665 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1666 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1667 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1668 int (*mmap) (struct file *, struct vm_area_struct *);
1669 int (*open) (struct inode *, struct file *);
1670 int (*flush) (struct file *, fl_owner_t id);
1671 int (*release) (struct inode *, struct file *);
1672 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1673 int (*aio_fsync) (struct kiocb *, int datasync);
1674 int (*fasync) (int, struct file *, int);
1675 int (*lock) (struct file *, int, struct file_lock *);
1676 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1677 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1678 int (*check_flags)(int);
1679 int (*flock) (struct file *, int, struct file_lock *);
1680 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1681 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1682 int (*setlease)(struct file *, long, struct file_lock **, void **);
1683 long (*fallocate)(struct file *file, int mode, loff_t offset,
1685 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1687 unsigned (*mmap_capabilities)(struct file *);
1691 struct inode_operations {
1692 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1693 const char * (*follow_link) (struct dentry *, void **);
1694 int (*permission) (struct inode *, int);
1695 int (*permission2) (struct vfsmount *, struct inode *, int);
1696 struct posix_acl * (*get_acl)(struct inode *, int);
1698 int (*readlink) (struct dentry *, char __user *,int);
1699 void (*put_link) (struct inode *, void *);
1701 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1702 int (*link) (struct dentry *,struct inode *,struct dentry *);
1703 int (*unlink) (struct inode *,struct dentry *);
1704 int (*symlink) (struct inode *,struct dentry *,const char *);
1705 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1706 int (*rmdir) (struct inode *,struct dentry *);
1707 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1708 int (*rename) (struct inode *, struct dentry *,
1709 struct inode *, struct dentry *);
1710 int (*rename2) (struct inode *, struct dentry *,
1711 struct inode *, struct dentry *, unsigned int);
1712 int (*setattr) (struct dentry *, struct iattr *);
1713 int (*setattr2) (struct vfsmount *, struct dentry *, struct iattr *);
1714 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1715 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1716 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1717 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1718 int (*removexattr) (struct dentry *, const char *);
1719 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1721 int (*update_time)(struct inode *, struct timespec *, int);
1722 int (*atomic_open)(struct inode *, struct dentry *,
1723 struct file *, unsigned open_flag,
1724 umode_t create_mode, int *opened);
1725 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1726 int (*set_acl)(struct inode *, struct posix_acl *, int);
1727 } ____cacheline_aligned;
1729 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1730 unsigned long nr_segs, unsigned long fast_segs,
1731 struct iovec *fast_pointer,
1732 struct iovec **ret_pointer);
1734 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1735 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1736 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1737 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1738 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1739 unsigned long, loff_t *);
1740 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1741 unsigned long, loff_t *);
1743 struct super_operations {
1744 struct inode *(*alloc_inode)(struct super_block *sb);
1745 void (*destroy_inode)(struct inode *);
1747 void (*dirty_inode) (struct inode *, int flags);
1748 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1749 int (*drop_inode) (struct inode *);
1750 void (*evict_inode) (struct inode *);
1751 void (*put_super) (struct super_block *);
1752 int (*sync_fs)(struct super_block *sb, int wait);
1753 int (*freeze_super) (struct super_block *);
1754 int (*freeze_fs) (struct super_block *);
1755 int (*thaw_super) (struct super_block *);
1756 int (*unfreeze_fs) (struct super_block *);
1757 int (*statfs) (struct dentry *, struct kstatfs *);
1758 int (*remount_fs) (struct super_block *, int *, char *);
1759 int (*remount_fs2) (struct vfsmount *, struct super_block *, int *, char *);
1760 void *(*clone_mnt_data) (void *);
1761 void (*copy_mnt_data) (void *, void *);
1762 void (*umount_begin) (struct super_block *);
1764 int (*show_options)(struct seq_file *, struct dentry *);
1765 int (*show_options2)(struct vfsmount *,struct seq_file *, struct dentry *);
1766 int (*show_devname)(struct seq_file *, struct dentry *);
1767 int (*show_path)(struct seq_file *, struct dentry *);
1768 int (*show_stats)(struct seq_file *, struct dentry *);
1770 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1771 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1772 struct dquot **(*get_dquots)(struct inode *);
1774 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1775 long (*nr_cached_objects)(struct super_block *,
1776 struct shrink_control *);
1777 long (*free_cached_objects)(struct super_block *,
1778 struct shrink_control *);
1782 * Inode flags - they have no relation to superblock flags now
1784 #define S_SYNC 1 /* Writes are synced at once */
1785 #define S_NOATIME 2 /* Do not update access times */
1786 #define S_APPEND 4 /* Append-only file */
1787 #define S_IMMUTABLE 8 /* Immutable file */
1788 #define S_DEAD 16 /* removed, but still open directory */
1789 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1790 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1791 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1792 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1793 #define S_PRIVATE 512 /* Inode is fs-internal */
1794 #define S_IMA 1024 /* Inode has an associated IMA struct */
1795 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1796 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1797 #ifdef CONFIG_FS_DAX
1798 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1800 #define S_DAX 0 /* Make all the DAX code disappear */
1802 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1805 * Note that nosuid etc flags are inode-specific: setting some file-system
1806 * flags just means all the inodes inherit those flags by default. It might be
1807 * possible to override it selectively if you really wanted to with some
1808 * ioctl() that is not currently implemented.
1810 * Exception: MS_RDONLY is always applied to the entire file system.
1812 * Unfortunately, it is possible to change a filesystems flags with it mounted
1813 * with files in use. This means that all of the inodes will not have their
1814 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1815 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1817 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1819 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1820 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1821 ((inode)->i_flags & S_SYNC))
1822 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1823 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1824 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1825 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1826 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1828 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1829 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1830 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1831 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1833 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1834 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1835 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1836 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1837 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1838 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1839 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1840 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1841 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1843 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1844 (inode)->i_rdev == WHITEOUT_DEV)
1847 * Inode state bits. Protected by inode->i_lock
1849 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1850 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1852 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1853 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1854 * various stages of removing an inode.
1856 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1858 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1859 * fdatasync(). i_atime is the usual cause.
1860 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1861 * these changes separately from I_DIRTY_SYNC so that we
1862 * don't have to write inode on fdatasync() when only
1863 * mtime has changed in it.
1864 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1865 * I_NEW Serves as both a mutex and completion notification.
1866 * New inodes set I_NEW. If two processes both create
1867 * the same inode, one of them will release its inode and
1868 * wait for I_NEW to be released before returning.
1869 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1870 * also cause waiting on I_NEW, without I_NEW actually
1871 * being set. find_inode() uses this to prevent returning
1872 * nearly-dead inodes.
1873 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1874 * is zero. I_FREEING must be set when I_WILL_FREE is
1876 * I_FREEING Set when inode is about to be freed but still has dirty
1877 * pages or buffers attached or the inode itself is still
1879 * I_CLEAR Added by clear_inode(). In this state the inode is
1880 * clean and can be destroyed. Inode keeps I_FREEING.
1882 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1883 * prohibited for many purposes. iget() must wait for
1884 * the inode to be completely released, then create it
1885 * anew. Other functions will just ignore such inodes,
1886 * if appropriate. I_NEW is used for waiting.
1888 * I_SYNC Writeback of inode is running. The bit is set during
1889 * data writeback, and cleared with a wakeup on the bit
1890 * address once it is done. The bit is also used to pin
1891 * the inode in memory for flusher thread.
1893 * I_REFERENCED Marks the inode as recently references on the LRU list.
1895 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1897 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1898 * synchronize competing switching instances and to tell
1899 * wb stat updates to grab mapping->tree_lock. See
1900 * inode_switch_wb_work_fn() for details.
1902 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1904 #define I_DIRTY_SYNC (1 << 0)
1905 #define I_DIRTY_DATASYNC (1 << 1)
1906 #define I_DIRTY_PAGES (1 << 2)
1908 #define I_NEW (1 << __I_NEW)
1909 #define I_WILL_FREE (1 << 4)
1910 #define I_FREEING (1 << 5)
1911 #define I_CLEAR (1 << 6)
1913 #define I_SYNC (1 << __I_SYNC)
1914 #define I_REFERENCED (1 << 8)
1915 #define __I_DIO_WAKEUP 9
1916 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1917 #define I_LINKABLE (1 << 10)
1918 #define I_DIRTY_TIME (1 << 11)
1919 #define __I_DIRTY_TIME_EXPIRED 12
1920 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1921 #define I_WB_SWITCH (1 << 13)
1923 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1924 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1926 extern void __mark_inode_dirty(struct inode *, int);
1927 static inline void mark_inode_dirty(struct inode *inode)
1929 __mark_inode_dirty(inode, I_DIRTY);
1932 static inline void mark_inode_dirty_sync(struct inode *inode)
1934 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1937 extern void inc_nlink(struct inode *inode);
1938 extern void drop_nlink(struct inode *inode);
1939 extern void clear_nlink(struct inode *inode);
1940 extern void set_nlink(struct inode *inode, unsigned int nlink);
1942 static inline void inode_inc_link_count(struct inode *inode)
1945 mark_inode_dirty(inode);
1948 static inline void inode_dec_link_count(struct inode *inode)
1951 mark_inode_dirty(inode);
1955 * inode_inc_iversion - increments i_version
1956 * @inode: inode that need to be updated
1958 * Every time the inode is modified, the i_version field will be incremented.
1959 * The filesystem has to be mounted with i_version flag
1962 static inline void inode_inc_iversion(struct inode *inode)
1964 spin_lock(&inode->i_lock);
1966 spin_unlock(&inode->i_lock);
1969 enum file_time_flags {
1976 extern bool atime_needs_update(const struct path *, struct inode *);
1977 extern void touch_atime(const struct path *);
1978 static inline void file_accessed(struct file *file)
1980 if (!(file->f_flags & O_NOATIME))
1981 touch_atime(&file->f_path);
1984 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1985 int sync_inode_metadata(struct inode *inode, int wait);
1987 struct file_system_type {
1990 #define FS_REQUIRES_DEV 1
1991 #define FS_BINARY_MOUNTDATA 2
1992 #define FS_HAS_SUBTYPE 4
1993 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1994 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1995 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1996 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1997 struct dentry *(*mount) (struct file_system_type *, int,
1998 const char *, void *);
1999 struct dentry *(*mount2) (struct vfsmount *, struct file_system_type *, int,
2000 const char *, void *);
2001 void *(*alloc_mnt_data) (void);
2002 void (*kill_sb) (struct super_block *);
2003 struct module *owner;
2004 struct file_system_type * next;
2005 struct hlist_head fs_supers;
2007 struct lock_class_key s_lock_key;
2008 struct lock_class_key s_umount_key;
2009 struct lock_class_key s_vfs_rename_key;
2010 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2012 struct lock_class_key i_lock_key;
2013 struct lock_class_key i_mutex_key;
2014 struct lock_class_key i_mutex_dir_key;
2017 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2019 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
2020 void *data, int (*fill_super)(struct super_block *, void *, int));
2021 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2022 int flags, const char *dev_name, void *data,
2023 int (*fill_super)(struct super_block *, void *, int));
2024 extern struct dentry *mount_single(struct file_system_type *fs_type,
2025 int flags, void *data,
2026 int (*fill_super)(struct super_block *, void *, int));
2027 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2028 int flags, void *data,
2029 int (*fill_super)(struct super_block *, void *, int));
2030 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2031 void generic_shutdown_super(struct super_block *sb);
2032 void kill_block_super(struct super_block *sb);
2033 void kill_anon_super(struct super_block *sb);
2034 void kill_litter_super(struct super_block *sb);
2035 void deactivate_super(struct super_block *sb);
2036 void deactivate_locked_super(struct super_block *sb);
2037 int set_anon_super(struct super_block *s, void *data);
2038 int get_anon_bdev(dev_t *);
2039 void free_anon_bdev(dev_t);
2040 struct super_block *sget(struct file_system_type *type,
2041 int (*test)(struct super_block *,void *),
2042 int (*set)(struct super_block *,void *),
2043 int flags, void *data);
2044 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2045 const struct super_operations *ops,
2046 const struct dentry_operations *dops,
2049 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2050 #define fops_get(fops) \
2051 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2052 #define fops_put(fops) \
2053 do { if (fops) module_put((fops)->owner); } while(0)
2055 * This one is to be used *ONLY* from ->open() instances.
2056 * fops must be non-NULL, pinned down *and* module dependencies
2057 * should be sufficient to pin the caller down as well.
2059 #define replace_fops(f, fops) \
2061 struct file *__file = (f); \
2062 fops_put(__file->f_op); \
2063 BUG_ON(!(__file->f_op = (fops))); \
2066 extern int register_filesystem(struct file_system_type *);
2067 extern int unregister_filesystem(struct file_system_type *);
2068 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2069 #define kern_mount(type) kern_mount_data(type, NULL)
2070 extern void kern_unmount(struct vfsmount *mnt);
2071 extern int may_umount_tree(struct vfsmount *);
2072 extern int may_umount(struct vfsmount *);
2073 extern long do_mount(const char *, const char __user *,
2074 const char *, unsigned long, void *);
2075 extern struct vfsmount *collect_mounts(struct path *);
2076 extern void drop_collected_mounts(struct vfsmount *);
2077 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2079 extern int vfs_statfs(struct path *, struct kstatfs *);
2080 extern int user_statfs(const char __user *, struct kstatfs *);
2081 extern int fd_statfs(int, struct kstatfs *);
2082 extern int vfs_ustat(dev_t, struct kstatfs *);
2083 extern int freeze_super(struct super_block *super);
2084 extern int thaw_super(struct super_block *super);
2085 extern bool our_mnt(struct vfsmount *mnt);
2087 extern int current_umask(void);
2089 extern void ihold(struct inode * inode);
2090 extern void iput(struct inode *);
2091 extern int generic_update_time(struct inode *, struct timespec *, int);
2094 extern struct kobject *fs_kobj;
2096 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2098 #define FLOCK_VERIFY_READ 1
2099 #define FLOCK_VERIFY_WRITE 2
2101 #ifdef CONFIG_FILE_LOCKING
2102 extern int locks_mandatory_locked(struct file *);
2103 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2106 * Candidates for mandatory locking have the setgid bit set
2107 * but no group execute bit - an otherwise meaningless combination.
2110 static inline int __mandatory_lock(struct inode *ino)
2112 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2116 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2117 * otherwise these will be advisory locks
2120 static inline int mandatory_lock(struct inode *ino)
2122 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2125 static inline int locks_verify_locked(struct file *file)
2127 if (mandatory_lock(file_inode(file)))
2128 return locks_mandatory_locked(file);
2132 static inline int locks_verify_truncate(struct inode *inode,
2136 if (inode->i_flctx && mandatory_lock(inode))
2137 return locks_mandatory_area(
2138 FLOCK_VERIFY_WRITE, inode, filp,
2139 size < inode->i_size ? size : inode->i_size,
2140 (size < inode->i_size ? inode->i_size - size
2141 : size - inode->i_size)
2146 static inline int break_lease(struct inode *inode, unsigned int mode)
2149 * Since this check is lockless, we must ensure that any refcounts
2150 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2151 * could end up racing with tasks trying to set a new lease on this
2155 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2156 return __break_lease(inode, mode, FL_LEASE);
2160 static inline int break_deleg(struct inode *inode, unsigned int mode)
2163 * Since this check is lockless, we must ensure that any refcounts
2164 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2165 * could end up racing with tasks trying to set a new lease on this
2169 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2170 return __break_lease(inode, mode, FL_DELEG);
2174 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2178 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2179 if (ret == -EWOULDBLOCK && delegated_inode) {
2180 *delegated_inode = inode;
2186 static inline int break_deleg_wait(struct inode **delegated_inode)
2190 ret = break_deleg(*delegated_inode, O_WRONLY);
2191 iput(*delegated_inode);
2192 *delegated_inode = NULL;
2196 static inline int break_layout(struct inode *inode, bool wait)
2199 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2200 return __break_lease(inode,
2201 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2206 #else /* !CONFIG_FILE_LOCKING */
2207 static inline int locks_mandatory_locked(struct file *file)
2212 static inline int locks_mandatory_area(int rw, struct inode *inode,
2213 struct file *filp, loff_t offset,
2219 static inline int __mandatory_lock(struct inode *inode)
2224 static inline int mandatory_lock(struct inode *inode)
2229 static inline int locks_verify_locked(struct file *file)
2234 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2240 static inline int break_lease(struct inode *inode, unsigned int mode)
2245 static inline int break_deleg(struct inode *inode, unsigned int mode)
2250 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2255 static inline int break_deleg_wait(struct inode **delegated_inode)
2261 static inline int break_layout(struct inode *inode, bool wait)
2266 #endif /* CONFIG_FILE_LOCKING */
2271 const char *name; /* pointer to actual string */
2272 const __user char *uptr; /* original userland pointer */
2273 struct audit_names *aname;
2278 extern long vfs_truncate(struct path *, loff_t);
2279 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2281 extern int do_truncate2(struct vfsmount *, struct dentry *, loff_t start,
2282 unsigned int time_attrs, struct file *filp);
2283 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2285 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2287 extern struct file *file_open_name(struct filename *, int, umode_t);
2288 extern struct file *filp_open(const char *, int, umode_t);
2289 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2290 const char *, int, umode_t);
2291 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2292 extern int filp_close(struct file *, fl_owner_t id);
2294 extern struct filename *getname_flags(const char __user *, int, int *);
2295 extern struct filename *getname(const char __user *);
2296 extern struct filename *getname_kernel(const char *);
2297 extern void putname(struct filename *name);
2303 extern int finish_open(struct file *file, struct dentry *dentry,
2304 int (*open)(struct inode *, struct file *),
2306 extern int finish_no_open(struct file *file, struct dentry *dentry);
2310 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2313 extern void __init vfs_caches_init_early(void);
2314 extern void __init vfs_caches_init(void);
2316 extern struct kmem_cache *names_cachep;
2318 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2319 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2322 extern int register_blkdev(unsigned int, const char *);
2323 extern void unregister_blkdev(unsigned int, const char *);
2324 extern struct block_device *bdget(dev_t);
2325 extern struct block_device *bdgrab(struct block_device *bdev);
2326 extern void bd_set_size(struct block_device *, loff_t size);
2327 extern void bd_forget(struct inode *inode);
2328 extern void bdput(struct block_device *);
2329 extern void invalidate_bdev(struct block_device *);
2330 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2331 extern int sync_blockdev(struct block_device *bdev);
2332 extern void kill_bdev(struct block_device *);
2333 extern struct super_block *freeze_bdev(struct block_device *);
2334 extern void emergency_thaw_all(void);
2335 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2336 extern int fsync_bdev(struct block_device *);
2338 extern struct super_block *blockdev_superblock;
2340 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2342 return sb == blockdev_superblock;
2345 static inline void bd_forget(struct inode *inode) {}
2346 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2347 static inline void kill_bdev(struct block_device *bdev) {}
2348 static inline void invalidate_bdev(struct block_device *bdev) {}
2350 static inline struct super_block *freeze_bdev(struct block_device *sb)
2355 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2360 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2364 static inline int sb_is_blkdev_sb(struct super_block *sb)
2369 extern int sync_filesystem(struct super_block *);
2370 extern const struct file_operations def_blk_fops;
2371 extern const struct file_operations def_chr_fops;
2373 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2374 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2375 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2376 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2377 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2379 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2381 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2382 extern int __blkdev_reread_part(struct block_device *bdev);
2383 extern int blkdev_reread_part(struct block_device *bdev);
2386 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2387 extern void bd_unlink_disk_holder(struct block_device *bdev,
2388 struct gendisk *disk);
2390 static inline int bd_link_disk_holder(struct block_device *bdev,
2391 struct gendisk *disk)
2395 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2396 struct gendisk *disk)
2403 #define CHRDEV_MAJOR_HASH_SIZE 255
2404 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2405 extern int register_chrdev_region(dev_t, unsigned, const char *);
2406 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2407 unsigned int count, const char *name,
2408 const struct file_operations *fops);
2409 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2410 unsigned int count, const char *name);
2411 extern void unregister_chrdev_region(dev_t, unsigned);
2412 extern void chrdev_show(struct seq_file *,off_t);
2414 static inline int register_chrdev(unsigned int major, const char *name,
2415 const struct file_operations *fops)
2417 return __register_chrdev(major, 0, 256, name, fops);
2420 static inline void unregister_chrdev(unsigned int major, const char *name)
2422 __unregister_chrdev(major, 0, 256, name);
2425 /* fs/block_dev.c */
2426 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2427 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2430 #define BLKDEV_MAJOR_HASH_SIZE 255
2431 extern const char *__bdevname(dev_t, char *buffer);
2432 extern const char *bdevname(struct block_device *bdev, char *buffer);
2433 extern struct block_device *lookup_bdev(const char *);
2434 extern void blkdev_show(struct seq_file *,off_t);
2437 #define BLKDEV_MAJOR_HASH_SIZE 0
2440 extern void init_special_inode(struct inode *, umode_t, dev_t);
2442 /* Invalid inode operations -- fs/bad_inode.c */
2443 extern void make_bad_inode(struct inode *);
2444 extern int is_bad_inode(struct inode *);
2448 * return READ, READA, or WRITE
2450 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2453 * return data direction, READ or WRITE
2455 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2457 extern void check_disk_size_change(struct gendisk *disk,
2458 struct block_device *bdev);
2459 extern int revalidate_disk(struct gendisk *);
2460 extern int check_disk_change(struct block_device *);
2461 extern int __invalidate_device(struct block_device *, bool);
2462 extern int invalidate_partition(struct gendisk *, int);
2464 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2465 pgoff_t start, pgoff_t end);
2467 static inline void invalidate_remote_inode(struct inode *inode)
2469 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2470 S_ISLNK(inode->i_mode))
2471 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2473 extern int invalidate_inode_pages2(struct address_space *mapping);
2474 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2475 pgoff_t start, pgoff_t end);
2476 extern int write_inode_now(struct inode *, int);
2477 extern int filemap_fdatawrite(struct address_space *);
2478 extern int filemap_flush(struct address_space *);
2479 extern int filemap_fdatawait(struct address_space *);
2480 extern void filemap_fdatawait_keep_errors(struct address_space *);
2481 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2483 extern int filemap_write_and_wait(struct address_space *mapping);
2484 extern int filemap_write_and_wait_range(struct address_space *mapping,
2485 loff_t lstart, loff_t lend);
2486 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2487 loff_t start, loff_t end, int sync_mode);
2488 extern int filemap_fdatawrite_range(struct address_space *mapping,
2489 loff_t start, loff_t end);
2491 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2493 extern int vfs_fsync(struct file *file, int datasync);
2494 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2496 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2498 return vfs_fsync_range(file, pos, pos + count - 1,
2499 (file->f_flags & __O_SYNC) ? 0 : 1);
2501 extern void emergency_sync(void);
2502 extern void emergency_remount(void);
2504 extern sector_t bmap(struct inode *, sector_t);
2506 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2507 extern int notify_change2(struct vfsmount *, struct dentry *, struct iattr *, struct inode **);
2508 extern int inode_permission(struct inode *, int);
2509 extern int inode_permission2(struct vfsmount *, struct inode *, int);
2510 extern int __inode_permission(struct inode *, int);
2511 extern int __inode_permission2(struct vfsmount *, struct inode *, int);
2512 extern int generic_permission(struct inode *, int);
2513 extern int __check_sticky(struct inode *dir, struct inode *inode);
2515 static inline bool execute_ok(struct inode *inode)
2517 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2520 static inline void file_start_write(struct file *file)
2522 if (!S_ISREG(file_inode(file)->i_mode))
2524 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2527 static inline bool file_start_write_trylock(struct file *file)
2529 if (!S_ISREG(file_inode(file)->i_mode))
2531 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2534 static inline void file_end_write(struct file *file)
2536 if (!S_ISREG(file_inode(file)->i_mode))
2538 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2542 * get_write_access() gets write permission for a file.
2543 * put_write_access() releases this write permission.
2544 * This is used for regular files.
2545 * We cannot support write (and maybe mmap read-write shared) accesses and
2546 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2547 * can have the following values:
2548 * 0: no writers, no VM_DENYWRITE mappings
2549 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2550 * > 0: (i_writecount) users are writing to the file.
2552 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2553 * except for the cases where we don't hold i_writecount yet. Then we need to
2554 * use {get,deny}_write_access() - these functions check the sign and refuse
2555 * to do the change if sign is wrong.
2557 static inline int get_write_access(struct inode *inode)
2559 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2561 static inline int deny_write_access(struct file *file)
2563 struct inode *inode = file_inode(file);
2564 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2566 static inline void put_write_access(struct inode * inode)
2568 atomic_dec(&inode->i_writecount);
2570 static inline void allow_write_access(struct file *file)
2573 atomic_inc(&file_inode(file)->i_writecount);
2575 static inline bool inode_is_open_for_write(const struct inode *inode)
2577 return atomic_read(&inode->i_writecount) > 0;
2581 static inline void i_readcount_dec(struct inode *inode)
2583 BUG_ON(!atomic_read(&inode->i_readcount));
2584 atomic_dec(&inode->i_readcount);
2586 static inline void i_readcount_inc(struct inode *inode)
2588 atomic_inc(&inode->i_readcount);
2591 static inline void i_readcount_dec(struct inode *inode)
2595 static inline void i_readcount_inc(struct inode *inode)
2600 extern int do_pipe_flags(int *, int);
2602 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2603 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2604 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2605 extern struct file * open_exec(const char *);
2607 /* fs/dcache.c -- generic fs support functions */
2608 extern int is_subdir(struct dentry *, struct dentry *);
2609 extern int path_is_under(struct path *, struct path *);
2611 extern char *file_path(struct file *, char *, int);
2613 #include <linux/err.h>
2615 /* needed for stackable file system support */
2616 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2618 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2620 extern int inode_init_always(struct super_block *, struct inode *);
2621 extern void inode_init_once(struct inode *);
2622 extern void address_space_init_once(struct address_space *mapping);
2623 extern struct inode * igrab(struct inode *);
2624 extern ino_t iunique(struct super_block *, ino_t);
2625 extern int inode_needs_sync(struct inode *inode);
2626 extern int generic_delete_inode(struct inode *inode);
2627 static inline int generic_drop_inode(struct inode *inode)
2629 return !inode->i_nlink || inode_unhashed(inode);
2632 extern struct inode *ilookup5_nowait(struct super_block *sb,
2633 unsigned long hashval, int (*test)(struct inode *, void *),
2635 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2636 int (*test)(struct inode *, void *), void *data);
2637 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2639 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2640 extern struct inode * iget_locked(struct super_block *, unsigned long);
2641 extern struct inode *find_inode_nowait(struct super_block *,
2643 int (*match)(struct inode *,
2644 unsigned long, void *),
2646 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2647 extern int insert_inode_locked(struct inode *);
2648 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2649 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2651 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2653 extern void unlock_new_inode(struct inode *);
2654 extern unsigned int get_next_ino(void);
2656 extern void __iget(struct inode * inode);
2657 extern void iget_failed(struct inode *);
2658 extern void clear_inode(struct inode *);
2659 extern void __destroy_inode(struct inode *);
2660 extern struct inode *new_inode_pseudo(struct super_block *sb);
2661 extern struct inode *new_inode(struct super_block *sb);
2662 extern void free_inode_nonrcu(struct inode *inode);
2663 extern int should_remove_suid(struct dentry *);
2664 extern int file_remove_privs(struct file *);
2665 extern int dentry_needs_remove_privs(struct dentry *dentry);
2666 static inline int file_needs_remove_privs(struct file *file)
2668 return dentry_needs_remove_privs(file->f_path.dentry);
2671 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2672 static inline void insert_inode_hash(struct inode *inode)
2674 __insert_inode_hash(inode, inode->i_ino);
2677 extern void __remove_inode_hash(struct inode *);
2678 static inline void remove_inode_hash(struct inode *inode)
2680 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2681 __remove_inode_hash(inode);
2684 extern void inode_sb_list_add(struct inode *inode);
2687 extern blk_qc_t submit_bio(int, struct bio *);
2688 extern int bdev_read_only(struct block_device *);
2690 extern int set_blocksize(struct block_device *, int);
2691 extern int sb_set_blocksize(struct super_block *, int);
2692 extern int sb_min_blocksize(struct super_block *, int);
2694 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2695 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2696 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2697 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2698 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2699 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2700 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2701 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2703 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2704 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2706 /* fs/block_dev.c */
2707 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2708 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2709 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2711 extern void block_sync_page(struct page *page);
2714 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2715 struct pipe_inode_info *, size_t, unsigned int);
2716 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2717 struct pipe_inode_info *, size_t, unsigned int);
2718 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2719 struct file *, loff_t *, size_t, unsigned int);
2720 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2721 struct file *out, loff_t *, size_t len, unsigned int flags);
2722 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2723 loff_t *opos, size_t len, unsigned int flags);
2727 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2728 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2729 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2730 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2731 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2732 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2733 int whence, loff_t maxsize, loff_t eof);
2734 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2735 int whence, loff_t size);
2736 extern int generic_file_open(struct inode * inode, struct file * filp);
2737 extern int nonseekable_open(struct inode * inode, struct file * filp);
2740 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2741 loff_t file_offset);
2744 /* need locking between buffered and direct access */
2747 /* filesystem does not support filling holes */
2748 DIO_SKIP_HOLES = 0x02,
2750 /* filesystem can handle aio writes beyond i_size */
2751 DIO_ASYNC_EXTEND = 0x04,
2753 /* inode/fs/bdev does not need truncate protection */
2754 DIO_SKIP_DIO_COUNT = 0x08,
2757 void dio_end_io(struct bio *bio, int error);
2759 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2760 struct block_device *bdev, struct iov_iter *iter,
2761 loff_t offset, get_block_t get_block,
2762 dio_iodone_t end_io, dio_submit_t submit_io,
2765 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2766 struct inode *inode,
2767 struct iov_iter *iter, loff_t offset,
2768 get_block_t get_block)
2770 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2771 offset, get_block, NULL, NULL,
2772 DIO_LOCKING | DIO_SKIP_HOLES);
2776 void inode_dio_wait(struct inode *inode);
2779 * inode_dio_begin - signal start of a direct I/O requests
2780 * @inode: inode the direct I/O happens on
2782 * This is called once we've finished processing a direct I/O request,
2783 * and is used to wake up callers waiting for direct I/O to be quiesced.
2785 static inline void inode_dio_begin(struct inode *inode)
2787 atomic_inc(&inode->i_dio_count);
2791 * inode_dio_end - signal finish of a direct I/O requests
2792 * @inode: inode the direct I/O happens on
2794 * This is called once we've finished processing a direct I/O request,
2795 * and is used to wake up callers waiting for direct I/O to be quiesced.
2797 static inline void inode_dio_end(struct inode *inode)
2799 if (atomic_dec_and_test(&inode->i_dio_count))
2800 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2803 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2806 extern const struct file_operations generic_ro_fops;
2808 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2810 extern int readlink_copy(char __user *, int, const char *);
2811 extern int page_readlink(struct dentry *, char __user *, int);
2812 extern const char *page_follow_link_light(struct dentry *, void **);
2813 extern void page_put_link(struct inode *, void *);
2814 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2816 extern int page_symlink(struct inode *inode, const char *symname, int len);
2817 extern const struct inode_operations page_symlink_inode_operations;
2818 extern void kfree_put_link(struct inode *, void *);
2819 extern void free_page_put_link(struct inode *, void *);
2820 extern int generic_readlink(struct dentry *, char __user *, int);
2821 extern void generic_fillattr(struct inode *, struct kstat *);
2822 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2823 extern int vfs_getattr(struct path *, struct kstat *);
2824 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2825 void inode_add_bytes(struct inode *inode, loff_t bytes);
2826 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2827 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2828 loff_t inode_get_bytes(struct inode *inode);
2829 void inode_set_bytes(struct inode *inode, loff_t bytes);
2830 const char *simple_follow_link(struct dentry *, void **);
2831 extern const struct inode_operations simple_symlink_inode_operations;
2833 extern int iterate_dir(struct file *, struct dir_context *);
2835 extern int vfs_stat(const char __user *, struct kstat *);
2836 extern int vfs_lstat(const char __user *, struct kstat *);
2837 extern int vfs_fstat(unsigned int, struct kstat *);
2838 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2840 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2842 extern int __generic_block_fiemap(struct inode *inode,
2843 struct fiemap_extent_info *fieinfo,
2844 loff_t start, loff_t len,
2845 get_block_t *get_block);
2846 extern int generic_block_fiemap(struct inode *inode,
2847 struct fiemap_extent_info *fieinfo, u64 start,
2848 u64 len, get_block_t *get_block);
2850 extern void get_filesystem(struct file_system_type *fs);
2851 extern void put_filesystem(struct file_system_type *fs);
2852 extern struct file_system_type *get_fs_type(const char *name);
2853 extern struct super_block *get_super(struct block_device *);
2854 extern struct super_block *get_super_thawed(struct block_device *);
2855 extern struct super_block *get_active_super(struct block_device *bdev);
2856 extern void drop_super(struct super_block *sb);
2857 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2858 extern void iterate_supers_type(struct file_system_type *,
2859 void (*)(struct super_block *, void *), void *);
2861 extern int dcache_dir_open(struct inode *, struct file *);
2862 extern int dcache_dir_close(struct inode *, struct file *);
2863 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2864 extern int dcache_readdir(struct file *, struct dir_context *);
2865 extern int simple_setattr(struct dentry *, struct iattr *);
2866 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2867 extern int simple_statfs(struct dentry *, struct kstatfs *);
2868 extern int simple_open(struct inode *inode, struct file *file);
2869 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2870 extern int simple_unlink(struct inode *, struct dentry *);
2871 extern int simple_rmdir(struct inode *, struct dentry *);
2872 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2873 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2874 extern int simple_empty(struct dentry *);
2875 extern int simple_readpage(struct file *file, struct page *page);
2876 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2877 loff_t pos, unsigned len, unsigned flags,
2878 struct page **pagep, void **fsdata);
2879 extern int simple_write_end(struct file *file, struct address_space *mapping,
2880 loff_t pos, unsigned len, unsigned copied,
2881 struct page *page, void *fsdata);
2882 extern int always_delete_dentry(const struct dentry *);
2883 extern struct inode *alloc_anon_inode(struct super_block *);
2884 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2885 extern const struct dentry_operations simple_dentry_operations;
2887 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2888 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2889 extern const struct file_operations simple_dir_operations;
2890 extern const struct inode_operations simple_dir_inode_operations;
2891 extern void make_empty_dir_inode(struct inode *inode);
2892 extern bool is_empty_dir_inode(struct inode *inode);
2893 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2894 struct dentry *d_alloc_name(struct dentry *, const char *);
2895 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2896 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2897 extern void simple_release_fs(struct vfsmount **mount, int *count);
2899 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2900 loff_t *ppos, const void *from, size_t available);
2901 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2902 const void __user *from, size_t count);
2904 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2905 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2907 extern int generic_check_addressable(unsigned, u64);
2909 #ifdef CONFIG_MIGRATION
2910 extern int buffer_migrate_page(struct address_space *,
2911 struct page *, struct page *,
2914 #define buffer_migrate_page NULL
2917 extern int inode_change_ok(const struct inode *, struct iattr *);
2918 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2919 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2921 extern int file_update_time(struct file *file);
2923 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2924 extern void save_mount_options(struct super_block *sb, char *options);
2925 extern void replace_mount_options(struct super_block *sb, char *options);
2927 static inline bool io_is_direct(struct file *filp)
2929 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2932 static inline int iocb_flags(struct file *file)
2935 if (file->f_flags & O_APPEND)
2937 if (io_is_direct(file))
2942 static inline ino_t parent_ino(struct dentry *dentry)
2947 * Don't strictly need d_lock here? If the parent ino could change
2948 * then surely we'd have a deeper race in the caller?
2950 spin_lock(&dentry->d_lock);
2951 res = dentry->d_parent->d_inode->i_ino;
2952 spin_unlock(&dentry->d_lock);
2956 /* Transaction based IO helpers */
2959 * An argresp is stored in an allocated page and holds the
2960 * size of the argument or response, along with its content
2962 struct simple_transaction_argresp {
2967 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2969 char *simple_transaction_get(struct file *file, const char __user *buf,
2971 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2972 size_t size, loff_t *pos);
2973 int simple_transaction_release(struct inode *inode, struct file *file);
2975 void simple_transaction_set(struct file *file, size_t n);
2978 * simple attribute files
2980 * These attributes behave similar to those in sysfs:
2982 * Writing to an attribute immediately sets a value, an open file can be
2983 * written to multiple times.
2985 * Reading from an attribute creates a buffer from the value that might get
2986 * read with multiple read calls. When the attribute has been read
2987 * completely, no further read calls are possible until the file is opened
2990 * All attributes contain a text representation of a numeric value
2991 * that are accessed with the get() and set() functions.
2993 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2994 static int __fops ## _open(struct inode *inode, struct file *file) \
2996 __simple_attr_check_format(__fmt, 0ull); \
2997 return simple_attr_open(inode, file, __get, __set, __fmt); \
2999 static const struct file_operations __fops = { \
3000 .owner = THIS_MODULE, \
3001 .open = __fops ## _open, \
3002 .release = simple_attr_release, \
3003 .read = simple_attr_read, \
3004 .write = simple_attr_write, \
3005 .llseek = generic_file_llseek, \
3008 static inline __printf(1, 2)
3009 void __simple_attr_check_format(const char *fmt, ...)
3011 /* don't do anything, just let the compiler check the arguments; */
3014 int simple_attr_open(struct inode *inode, struct file *file,
3015 int (*get)(void *, u64 *), int (*set)(void *, u64),
3017 int simple_attr_release(struct inode *inode, struct file *file);
3018 ssize_t simple_attr_read(struct file *file, char __user *buf,
3019 size_t len, loff_t *ppos);
3020 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3021 size_t len, loff_t *ppos);
3024 int proc_nr_files(struct ctl_table *table, int write,
3025 void __user *buffer, size_t *lenp, loff_t *ppos);
3026 int proc_nr_dentry(struct ctl_table *table, int write,
3027 void __user *buffer, size_t *lenp, loff_t *ppos);
3028 int proc_nr_inodes(struct ctl_table *table, int write,
3029 void __user *buffer, size_t *lenp, loff_t *ppos);
3030 int __init get_filesystem_list(char *buf);
3032 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3033 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3035 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3036 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3037 (flag & __FMODE_NONOTIFY)))
3039 static inline int is_sxid(umode_t mode)
3041 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3044 static inline int check_sticky(struct inode *dir, struct inode *inode)
3046 if (!(dir->i_mode & S_ISVTX))
3049 return __check_sticky(dir, inode);
3052 static inline void inode_has_no_xattr(struct inode *inode)
3054 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3055 inode->i_flags |= S_NOSEC;
3058 static inline bool is_root_inode(struct inode *inode)
3060 return inode == inode->i_sb->s_root->d_inode;
3063 static inline bool dir_emit(struct dir_context *ctx,
3064 const char *name, int namelen,
3065 u64 ino, unsigned type)
3067 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3069 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3071 return ctx->actor(ctx, ".", 1, ctx->pos,
3072 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3074 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3076 return ctx->actor(ctx, "..", 2, ctx->pos,
3077 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3079 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3081 if (ctx->pos == 0) {
3082 if (!dir_emit_dot(file, ctx))
3086 if (ctx->pos == 1) {
3087 if (!dir_emit_dotdot(file, ctx))
3093 static inline bool dir_relax(struct inode *inode)
3095 inode_unlock(inode);
3097 return !IS_DEADDIR(inode);
3100 extern bool path_noexec(const struct path *path);
3101 extern void inode_nohighmem(struct inode *inode);
3103 #endif /* _LINUX_FS_H */