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_FLUSH_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | \
209 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
210 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
211 #define WRITE_POST_FLUSH_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | \
212 REQ_POST_FLUSH_BARRIER | REQ_BARRIER)
213 #define WRITE_ORDERED_FLUSH_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | \
214 REQ_FLUSH | REQ_POST_FLUSH_BARRIER | \
218 * Attribute flags. These should be or-ed together to figure out what
221 #define ATTR_MODE (1 << 0)
222 #define ATTR_UID (1 << 1)
223 #define ATTR_GID (1 << 2)
224 #define ATTR_SIZE (1 << 3)
225 #define ATTR_ATIME (1 << 4)
226 #define ATTR_MTIME (1 << 5)
227 #define ATTR_CTIME (1 << 6)
228 #define ATTR_ATIME_SET (1 << 7)
229 #define ATTR_MTIME_SET (1 << 8)
230 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
231 #define ATTR_ATTR_FLAG (1 << 10)
232 #define ATTR_KILL_SUID (1 << 11)
233 #define ATTR_KILL_SGID (1 << 12)
234 #define ATTR_FILE (1 << 13)
235 #define ATTR_KILL_PRIV (1 << 14)
236 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
237 #define ATTR_TIMES_SET (1 << 16)
238 #define ATTR_TOUCH (1 << 17)
241 * Whiteout is represented by a char device. The following constants define the
242 * mode and device number to use.
244 #define WHITEOUT_MODE 0
245 #define WHITEOUT_DEV 0
248 * This is the Inode Attributes structure, used for notify_change(). It
249 * uses the above definitions as flags, to know which values have changed.
250 * Also, in this manner, a Filesystem can look at only the values it cares
251 * about. Basically, these are the attributes that the VFS layer can
252 * request to change from the FS layer.
254 * Derek Atkins <warlord@MIT.EDU> 94-10-20
257 unsigned int ia_valid;
262 struct timespec ia_atime;
263 struct timespec ia_mtime;
264 struct timespec ia_ctime;
267 * Not an attribute, but an auxiliary info for filesystems wanting to
268 * implement an ftruncate() like method. NOTE: filesystem should
269 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
271 struct file *ia_file;
275 * Includes for diskquotas.
277 #include <linux/quota.h>
280 * Maximum number of layers of fs stack. Needs to be limited to
281 * prevent kernel stack overflow
283 #define FILESYSTEM_MAX_STACK_DEPTH 2
286 * enum positive_aop_returns - aop return codes with specific semantics
288 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
289 * completed, that the page is still locked, and
290 * should be considered active. The VM uses this hint
291 * to return the page to the active list -- it won't
292 * be a candidate for writeback again in the near
293 * future. Other callers must be careful to unlock
294 * the page if they get this return. Returned by
297 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
298 * unlocked it and the page might have been truncated.
299 * The caller should back up to acquiring a new page and
300 * trying again. The aop will be taking reasonable
301 * precautions not to livelock. If the caller held a page
302 * reference, it should drop it before retrying. Returned
305 * address_space_operation functions return these large constants to indicate
306 * special semantics to the caller. These are much larger than the bytes in a
307 * page to allow for functions that return the number of bytes operated on in a
311 enum positive_aop_returns {
312 AOP_WRITEPAGE_ACTIVATE = 0x80000,
313 AOP_TRUNCATED_PAGE = 0x80001,
316 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
317 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
318 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
319 * helper code (eg buffer layer)
320 * to clear GFP_FS from alloc */
323 * oh the beauties of C type declarations.
326 struct address_space;
327 struct writeback_control;
329 #define IOCB_EVENTFD (1 << 0)
330 #define IOCB_APPEND (1 << 1)
331 #define IOCB_DIRECT (1 << 2)
334 struct file *ki_filp;
336 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
341 static inline bool is_sync_kiocb(struct kiocb *kiocb)
343 return kiocb->ki_complete == NULL;
346 static inline int iocb_flags(struct file *file);
348 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
350 *kiocb = (struct kiocb) {
352 .ki_flags = iocb_flags(filp),
357 * "descriptor" for what we're up to with a read.
358 * This allows us to use the same read code yet
359 * have multiple different users of the data that
360 * we read from a file.
362 * The simplest case just copies the data to user
375 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
376 unsigned long, unsigned long);
378 struct address_space_operations {
379 int (*writepage)(struct page *page, struct writeback_control *wbc);
380 int (*readpage)(struct file *, struct page *);
382 /* Write back some dirty pages from this mapping. */
383 int (*writepages)(struct address_space *, struct writeback_control *);
385 /* Set a page dirty. Return true if this dirtied it */
386 int (*set_page_dirty)(struct page *page);
388 int (*readpages)(struct file *filp, struct address_space *mapping,
389 struct list_head *pages, unsigned nr_pages);
391 int (*write_begin)(struct file *, struct address_space *mapping,
392 loff_t pos, unsigned len, unsigned flags,
393 struct page **pagep, void **fsdata);
394 int (*write_end)(struct file *, struct address_space *mapping,
395 loff_t pos, unsigned len, unsigned copied,
396 struct page *page, void *fsdata);
398 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
399 sector_t (*bmap)(struct address_space *, sector_t);
400 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
401 int (*releasepage) (struct page *, gfp_t);
402 void (*freepage)(struct page *);
403 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
405 * migrate the contents of a page to the specified target. If
406 * migrate_mode is MIGRATE_ASYNC, it must not block.
408 int (*migratepage) (struct address_space *,
409 struct page *, struct page *, enum migrate_mode);
410 bool (*isolate_page)(struct page *, isolate_mode_t);
411 void (*putback_page)(struct page *);
412 int (*launder_page) (struct page *);
413 int (*is_partially_uptodate) (struct page *, unsigned long,
415 void (*is_dirty_writeback) (struct page *, bool *, bool *);
416 int (*error_remove_page)(struct address_space *, struct page *);
418 /* swapfile support */
419 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
421 void (*swap_deactivate)(struct file *file);
424 extern const struct address_space_operations empty_aops;
427 * pagecache_write_begin/pagecache_write_end must be used by general code
428 * to write into the pagecache.
430 int pagecache_write_begin(struct file *, struct address_space *mapping,
431 loff_t pos, unsigned len, unsigned flags,
432 struct page **pagep, void **fsdata);
434 int pagecache_write_end(struct file *, struct address_space *mapping,
435 loff_t pos, unsigned len, unsigned copied,
436 struct page *page, void *fsdata);
438 struct address_space {
439 struct inode *host; /* owner: inode, block_device */
440 struct radix_tree_root page_tree; /* radix tree of all pages */
441 spinlock_t tree_lock; /* and lock protecting it */
442 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
443 struct rb_root i_mmap; /* tree of private and shared mappings */
444 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
445 /* Protected by tree_lock together with the radix tree */
446 unsigned long nrpages; /* number of total pages */
447 unsigned long nrshadows; /* number of shadow entries */
448 pgoff_t writeback_index;/* writeback starts here */
449 const struct address_space_operations *a_ops; /* methods */
450 unsigned long flags; /* error bits/gfp mask */
451 spinlock_t private_lock; /* for use by the address_space */
452 struct list_head private_list; /* ditto */
453 void *private_data; /* ditto */
454 } __attribute__((aligned(sizeof(long))));
456 * On most architectures that alignment is already the case; but
457 * must be enforced here for CRIS, to let the least significant bit
458 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
460 struct request_queue;
462 struct block_device {
463 dev_t bd_dev; /* not a kdev_t - it's a search key */
465 struct inode * bd_inode; /* will die */
466 struct super_block * bd_super;
467 struct mutex bd_mutex; /* open/close mutex */
468 struct list_head bd_inodes;
472 bool bd_write_holder;
474 struct list_head bd_holder_disks;
476 struct block_device * bd_contains;
477 unsigned bd_block_size;
478 struct hd_struct * bd_part;
479 /* number of times partitions within this device have been opened. */
480 unsigned bd_part_count;
482 struct gendisk * bd_disk;
483 struct request_queue * bd_queue;
484 struct backing_dev_info *bd_bdi;
485 struct list_head bd_list;
487 * Private data. You must have bd_claim'ed the block_device
488 * to use this. NOTE: bd_claim allows an owner to claim
489 * the same device multiple times, the owner must take special
490 * care to not mess up bd_private for that case.
492 unsigned long bd_private;
494 /* The counter of freeze processes */
495 int bd_fsfreeze_count;
496 /* Mutex for freeze */
497 struct mutex bd_fsfreeze_mutex;
501 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
504 #define PAGECACHE_TAG_DIRTY 0
505 #define PAGECACHE_TAG_WRITEBACK 1
506 #define PAGECACHE_TAG_TOWRITE 2
508 int mapping_tagged(struct address_space *mapping, int tag);
510 static inline void i_mmap_lock_write(struct address_space *mapping)
512 down_write(&mapping->i_mmap_rwsem);
515 static inline void i_mmap_unlock_write(struct address_space *mapping)
517 up_write(&mapping->i_mmap_rwsem);
520 static inline void i_mmap_lock_read(struct address_space *mapping)
522 down_read(&mapping->i_mmap_rwsem);
525 static inline void i_mmap_unlock_read(struct address_space *mapping)
527 up_read(&mapping->i_mmap_rwsem);
531 * Might pages of this file be mapped into userspace?
533 static inline int mapping_mapped(struct address_space *mapping)
535 return !RB_EMPTY_ROOT(&mapping->i_mmap);
539 * Might pages of this file have been modified in userspace?
540 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
541 * marks vma as VM_SHARED if it is shared, and the file was opened for
542 * writing i.e. vma may be mprotected writable even if now readonly.
544 * If i_mmap_writable is negative, no new writable mappings are allowed. You
545 * can only deny writable mappings, if none exists right now.
547 static inline int mapping_writably_mapped(struct address_space *mapping)
549 return atomic_read(&mapping->i_mmap_writable) > 0;
552 static inline int mapping_map_writable(struct address_space *mapping)
554 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
558 static inline void mapping_unmap_writable(struct address_space *mapping)
560 atomic_dec(&mapping->i_mmap_writable);
563 static inline int mapping_deny_writable(struct address_space *mapping)
565 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
569 static inline void mapping_allow_writable(struct address_space *mapping)
571 atomic_inc(&mapping->i_mmap_writable);
575 * Use sequence counter to get consistent i_size on 32-bit processors.
577 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
578 #include <linux/seqlock.h>
579 #define __NEED_I_SIZE_ORDERED
580 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
582 #define i_size_ordered_init(inode) do { } while (0)
586 #define ACL_NOT_CACHED ((void *)(-1))
588 #define IOP_FASTPERM 0x0001
589 #define IOP_LOOKUP 0x0002
590 #define IOP_NOFOLLOW 0x0004
593 * Keep mostly read-only and often accessed (especially for
594 * the RCU path lookup and 'stat' data) fields at the beginning
595 * of the 'struct inode'
599 unsigned short i_opflags;
602 unsigned int i_flags;
604 #ifdef CONFIG_FS_POSIX_ACL
605 struct posix_acl *i_acl;
606 struct posix_acl *i_default_acl;
609 const struct inode_operations *i_op;
610 struct super_block *i_sb;
611 struct address_space *i_mapping;
613 #ifdef CONFIG_SECURITY
617 /* Stat data, not accessed from path walking */
620 * Filesystems may only read i_nlink directly. They shall use the
621 * following functions for modification:
623 * (set|clear|inc|drop)_nlink
624 * inode_(inc|dec)_link_count
627 const unsigned int i_nlink;
628 unsigned int __i_nlink;
632 struct timespec i_atime;
633 struct timespec i_mtime;
634 struct timespec i_ctime;
635 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
636 unsigned short i_bytes;
637 unsigned int i_blkbits;
640 #ifdef __NEED_I_SIZE_ORDERED
641 seqcount_t i_size_seqcount;
645 unsigned long i_state;
646 struct mutex i_mutex;
648 unsigned long dirtied_when; /* jiffies of first dirtying */
649 unsigned long dirtied_time_when;
651 struct hlist_node i_hash;
652 struct list_head i_io_list; /* backing dev IO list */
653 #ifdef CONFIG_CGROUP_WRITEBACK
654 struct bdi_writeback *i_wb; /* the associated cgroup wb */
656 /* foreign inode detection, see wbc_detach_inode() */
658 u16 i_wb_frn_avg_time;
659 u16 i_wb_frn_history;
661 struct list_head i_lru; /* inode LRU list */
662 struct list_head i_sb_list;
664 struct hlist_head i_dentry;
665 struct rcu_head i_rcu;
669 atomic_t i_dio_count;
670 atomic_t i_writecount;
672 atomic_t i_readcount; /* struct files open RO */
674 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
675 struct file_lock_context *i_flctx;
676 struct address_space i_data;
677 struct list_head i_devices;
679 struct pipe_inode_info *i_pipe;
680 struct block_device *i_bdev;
687 #ifdef CONFIG_FSNOTIFY
688 __u32 i_fsnotify_mask; /* all events this inode cares about */
689 struct hlist_head i_fsnotify_marks;
692 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
693 struct fscrypt_info *i_crypt_info;
695 void *i_private; /* fs or device private pointer */
698 static inline unsigned int i_blocksize(const struct inode *node)
700 return (1 << node->i_blkbits);
703 static inline int inode_unhashed(struct inode *inode)
705 return hlist_unhashed(&inode->i_hash);
709 * inode->i_mutex nesting subclasses for the lock validator:
711 * 0: the object of the current VFS operation
715 * 4: second non-directory
716 * 5: second parent (when locking independent directories in rename)
718 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
719 * non-directories at once.
721 * The locking order between these classes is
722 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
724 enum inode_i_mutex_lock_class
734 static inline void inode_lock(struct inode *inode)
736 mutex_lock(&inode->i_mutex);
739 static inline void inode_unlock(struct inode *inode)
741 mutex_unlock(&inode->i_mutex);
744 static inline int inode_trylock(struct inode *inode)
746 return mutex_trylock(&inode->i_mutex);
749 static inline int inode_is_locked(struct inode *inode)
751 return mutex_is_locked(&inode->i_mutex);
754 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
756 mutex_lock_nested(&inode->i_mutex, subclass);
759 void lock_two_nondirectories(struct inode *, struct inode*);
760 void unlock_two_nondirectories(struct inode *, struct inode*);
763 * NOTE: in a 32bit arch with a preemptable kernel and
764 * an UP compile the i_size_read/write must be atomic
765 * with respect to the local cpu (unlike with preempt disabled),
766 * but they don't need to be atomic with respect to other cpus like in
767 * true SMP (so they need either to either locally disable irq around
768 * the read or for example on x86 they can be still implemented as a
769 * cmpxchg8b without the need of the lock prefix). For SMP compiles
770 * and 64bit archs it makes no difference if preempt is enabled or not.
772 static inline loff_t i_size_read(const struct inode *inode)
774 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
779 seq = read_seqcount_begin(&inode->i_size_seqcount);
780 i_size = inode->i_size;
781 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
783 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
787 i_size = inode->i_size;
791 return inode->i_size;
796 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
797 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
798 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
800 static inline void i_size_write(struct inode *inode, loff_t i_size)
802 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
804 write_seqcount_begin(&inode->i_size_seqcount);
805 inode->i_size = i_size;
806 write_seqcount_end(&inode->i_size_seqcount);
808 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
810 inode->i_size = i_size;
813 inode->i_size = i_size;
817 /* Helper functions so that in most cases filesystems will
818 * not need to deal directly with kuid_t and kgid_t and can
819 * instead deal with the raw numeric values that are stored
822 static inline uid_t i_uid_read(const struct inode *inode)
824 return from_kuid(&init_user_ns, inode->i_uid);
827 static inline gid_t i_gid_read(const struct inode *inode)
829 return from_kgid(&init_user_ns, inode->i_gid);
832 static inline void i_uid_write(struct inode *inode, uid_t uid)
834 inode->i_uid = make_kuid(&init_user_ns, uid);
837 static inline void i_gid_write(struct inode *inode, gid_t gid)
839 inode->i_gid = make_kgid(&init_user_ns, gid);
842 static inline unsigned iminor(const struct inode *inode)
844 return MINOR(inode->i_rdev);
847 static inline unsigned imajor(const struct inode *inode)
849 return MAJOR(inode->i_rdev);
852 extern struct block_device *I_BDEV(struct inode *inode);
855 rwlock_t lock; /* protects pid, uid, euid fields */
856 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
857 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
858 kuid_t uid, euid; /* uid/euid of process setting the owner */
859 int signum; /* posix.1b rt signal to be delivered on IO */
863 * Track a single file's readahead state
865 struct file_ra_state {
866 pgoff_t start; /* where readahead started */
867 unsigned int size; /* # of readahead pages */
868 unsigned int async_size; /* do asynchronous readahead when
869 there are only # of pages ahead */
871 unsigned int ra_pages; /* Maximum readahead window */
872 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
873 loff_t prev_pos; /* Cache last read() position */
877 * Check if @index falls in the readahead windows.
879 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
881 return (index >= ra->start &&
882 index < ra->start + ra->size);
887 struct llist_node fu_llist;
888 struct rcu_head fu_rcuhead;
891 struct inode *f_inode; /* cached value */
892 const struct file_operations *f_op;
895 * Protects f_ep_links, f_flags.
896 * Must not be taken from IRQ context.
899 atomic_long_t f_count;
900 unsigned int f_flags;
902 struct mutex f_pos_lock;
904 struct fown_struct f_owner;
905 const struct cred *f_cred;
906 struct file_ra_state f_ra;
909 #ifdef CONFIG_SECURITY
912 /* needed for tty driver, and maybe others */
916 /* Used by fs/eventpoll.c to link all the hooks to this file */
917 struct list_head f_ep_links;
918 struct list_head f_tfile_llink;
919 #endif /* #ifdef CONFIG_EPOLL */
920 struct address_space *f_mapping;
922 #ifdef CONFIG_FILE_TABLE_DEBUG
923 struct hlist_node f_hash;
924 #endif /* #ifdef CONFIG_FILE_TABLE_DEBUG */
925 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
930 /* file identifier */
931 unsigned char f_handle[0];
934 static inline struct file *get_file(struct file *f)
936 atomic_long_inc(&f->f_count);
939 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
940 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
941 #define file_count(x) atomic_long_read(&(x)->f_count)
943 #define MAX_NON_LFS ((1UL<<31) - 1)
945 /* Page cache limit. The filesystems should put that into their s_maxbytes
946 limits, otherwise bad things can happen in VM. */
947 #if BITS_PER_LONG==32
948 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
949 #elif BITS_PER_LONG==64
950 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
955 #define FL_DELEG 4 /* NFSv4 delegation */
956 #define FL_ACCESS 8 /* not trying to lock, just looking */
957 #define FL_EXISTS 16 /* when unlocking, test for existence */
958 #define FL_LEASE 32 /* lease held on this file */
959 #define FL_CLOSE 64 /* unlock on close */
960 #define FL_SLEEP 128 /* A blocking lock */
961 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
962 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
963 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
964 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
967 * Special return value from posix_lock_file() and vfs_lock_file() for
968 * asynchronous locking.
970 #define FILE_LOCK_DEFERRED 1
972 /* legacy typedef, should eventually be removed */
973 typedef void *fl_owner_t;
977 struct file_lock_operations {
978 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
979 void (*fl_release_private)(struct file_lock *);
982 struct lock_manager_operations {
983 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
984 unsigned long (*lm_owner_key)(struct file_lock *);
985 fl_owner_t (*lm_get_owner)(fl_owner_t);
986 void (*lm_put_owner)(fl_owner_t);
987 void (*lm_notify)(struct file_lock *); /* unblock callback */
988 int (*lm_grant)(struct file_lock *, int);
989 bool (*lm_break)(struct file_lock *);
990 int (*lm_change)(struct file_lock *, int, struct list_head *);
991 void (*lm_setup)(struct file_lock *, void **);
994 struct lock_manager {
995 struct list_head list;
997 * NFSv4 and up also want opens blocked during the grace period;
1004 void locks_start_grace(struct net *, struct lock_manager *);
1005 void locks_end_grace(struct lock_manager *);
1006 int locks_in_grace(struct net *);
1007 int opens_in_grace(struct net *);
1009 /* that will die - we need it for nfs_lock_info */
1010 #include <linux/nfs_fs_i.h>
1013 * struct file_lock represents a generic "file lock". It's used to represent
1014 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1015 * note that the same struct is used to represent both a request for a lock and
1016 * the lock itself, but the same object is never used for both.
1018 * FIXME: should we create a separate "struct lock_request" to help distinguish
1021 * The varous i_flctx lists are ordered by:
1024 * 2) lock range start
1027 * Obviously, the last two criteria only matter for POSIX locks.
1030 struct file_lock *fl_next; /* singly linked list for this inode */
1031 struct list_head fl_list; /* link into file_lock_context */
1032 struct hlist_node fl_link; /* node in global lists */
1033 struct list_head fl_block; /* circular list of blocked processes */
1034 fl_owner_t fl_owner;
1035 unsigned int fl_flags;
1036 unsigned char fl_type;
1037 unsigned int fl_pid;
1038 int fl_link_cpu; /* what cpu's list is this on? */
1039 struct pid *fl_nspid;
1040 wait_queue_head_t fl_wait;
1041 struct file *fl_file;
1045 struct fasync_struct * fl_fasync; /* for lease break notifications */
1046 /* for lease breaks: */
1047 unsigned long fl_break_time;
1048 unsigned long fl_downgrade_time;
1050 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1051 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1053 struct nfs_lock_info nfs_fl;
1054 struct nfs4_lock_info nfs4_fl;
1056 struct list_head link; /* link in AFS vnode's pending_locks list */
1057 int state; /* state of grant or error if -ve */
1062 struct file_lock_context {
1063 spinlock_t flc_lock;
1064 struct list_head flc_flock;
1065 struct list_head flc_posix;
1066 struct list_head flc_lease;
1069 /* The following constant reflects the upper bound of the file/locking space */
1071 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1072 #define OFFSET_MAX INT_LIMIT(loff_t)
1073 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1076 #include <linux/fcntl.h>
1078 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1080 #ifdef CONFIG_FILE_LOCKING
1081 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1082 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1083 struct flock __user *);
1085 #if BITS_PER_LONG == 32
1086 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1087 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1088 struct flock64 __user *);
1091 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1092 extern int fcntl_getlease(struct file *filp);
1095 void locks_free_lock_context(struct file_lock_context *ctx);
1096 void locks_free_lock(struct file_lock *fl);
1097 extern void locks_init_lock(struct file_lock *);
1098 extern struct file_lock * locks_alloc_lock(void);
1099 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1100 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1101 extern void locks_remove_posix(struct file *, fl_owner_t);
1102 extern void locks_remove_file(struct file *);
1103 extern void locks_release_private(struct file_lock *);
1104 extern void posix_test_lock(struct file *, struct file_lock *);
1105 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1106 extern int posix_unblock_lock(struct file_lock *);
1107 extern int vfs_test_lock(struct file *, struct file_lock *);
1108 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1109 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1110 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1111 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1112 extern void lease_get_mtime(struct inode *, struct timespec *time);
1113 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1114 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1115 extern int lease_modify(struct file_lock *, int, struct list_head *);
1116 struct files_struct;
1117 extern void show_fd_locks(struct seq_file *f,
1118 struct file *filp, struct files_struct *files);
1119 #else /* !CONFIG_FILE_LOCKING */
1120 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1121 struct flock __user *user)
1126 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1127 unsigned int cmd, struct flock __user *user)
1132 #if BITS_PER_LONG == 32
1133 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1134 struct flock64 __user *user)
1139 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1140 unsigned int cmd, struct flock64 __user *user)
1145 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1150 static inline int fcntl_getlease(struct file *filp)
1156 locks_free_lock_context(struct file_lock_context *ctx)
1160 static inline void locks_init_lock(struct file_lock *fl)
1165 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1170 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1175 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1180 static inline void locks_remove_file(struct file *filp)
1185 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1190 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1191 struct file_lock *conflock)
1196 static inline int posix_unblock_lock(struct file_lock *waiter)
1201 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1206 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1207 struct file_lock *fl, struct file_lock *conf)
1212 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1217 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1222 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1227 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1232 static inline int generic_setlease(struct file *filp, long arg,
1233 struct file_lock **flp, void **priv)
1238 static inline int vfs_setlease(struct file *filp, long arg,
1239 struct file_lock **lease, void **priv)
1244 static inline int lease_modify(struct file_lock *fl, int arg,
1245 struct list_head *dispose)
1250 struct files_struct;
1251 static inline void show_fd_locks(struct seq_file *f,
1252 struct file *filp, struct files_struct *files) {}
1253 #endif /* !CONFIG_FILE_LOCKING */
1255 static inline struct inode *file_inode(const struct file *f)
1260 static inline struct dentry *file_dentry(const struct file *file)
1262 struct dentry *dentry = file->f_path.dentry;
1264 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1265 return dentry->d_op->d_real(dentry, file_inode(file));
1270 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1272 return locks_lock_inode_wait(file_inode(filp), fl);
1275 struct fasync_struct {
1279 struct fasync_struct *fa_next; /* singly linked list */
1280 struct file *fa_file;
1281 struct rcu_head fa_rcu;
1284 #define FASYNC_MAGIC 0x4601
1286 /* SMP safe fasync helpers: */
1287 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1288 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1289 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1290 extern struct fasync_struct *fasync_alloc(void);
1291 extern void fasync_free(struct fasync_struct *);
1293 /* can be called from interrupts */
1294 extern void kill_fasync(struct fasync_struct **, int, int);
1296 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1297 extern void f_setown(struct file *filp, unsigned long arg, int force);
1298 extern void f_delown(struct file *filp);
1299 extern pid_t f_getown(struct file *filp);
1300 extern int send_sigurg(struct fown_struct *fown);
1308 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1309 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1310 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1311 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1312 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1315 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1316 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1318 /* Possible states of 'frozen' field */
1320 SB_UNFROZEN = 0, /* FS is unfrozen */
1321 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1322 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1323 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1324 * internal threads if needed) */
1325 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1328 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1331 int frozen; /* Is sb frozen? */
1332 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1333 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1336 struct super_block {
1337 struct list_head s_list; /* Keep this first */
1338 dev_t s_dev; /* search index; _not_ kdev_t */
1339 unsigned char s_blocksize_bits;
1340 unsigned long s_blocksize;
1341 loff_t s_maxbytes; /* Max file size */
1342 struct file_system_type *s_type;
1343 const struct super_operations *s_op;
1344 const struct dquot_operations *dq_op;
1345 const struct quotactl_ops *s_qcop;
1346 const struct export_operations *s_export_op;
1347 unsigned long s_flags;
1348 unsigned long s_iflags; /* internal SB_I_* flags */
1349 unsigned long s_magic;
1350 struct dentry *s_root;
1351 struct rw_semaphore s_umount;
1354 #ifdef CONFIG_SECURITY
1357 const struct xattr_handler **s_xattr;
1359 const struct fscrypt_operations *s_cop;
1361 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1362 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1363 struct block_device *s_bdev;
1364 struct backing_dev_info *s_bdi;
1365 struct mtd_info *s_mtd;
1366 struct hlist_node s_instances;
1367 unsigned int s_quota_types; /* Bitmask of supported quota types */
1368 struct quota_info s_dquot; /* Diskquota specific options */
1370 struct sb_writers s_writers;
1372 char s_id[32]; /* Informational name */
1373 u8 s_uuid[16]; /* UUID */
1375 void *s_fs_info; /* Filesystem private info */
1376 unsigned int s_max_links;
1379 /* Granularity of c/m/atime in ns.
1380 Cannot be worse than a second */
1384 * The next field is for VFS *only*. No filesystems have any business
1385 * even looking at it. You had been warned.
1387 struct mutex s_vfs_rename_mutex; /* Kludge */
1390 * Filesystem subtype. If non-empty the filesystem type field
1391 * in /proc/mounts will be "type.subtype"
1396 * Saved mount options for lazy filesystems using
1397 * generic_show_options()
1399 char __rcu *s_options;
1400 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1403 * Saved pool identifier for cleancache (-1 means none)
1405 int cleancache_poolid;
1407 struct shrinker s_shrink; /* per-sb shrinker handle */
1409 /* Number of inodes with nlink == 0 but still referenced */
1410 atomic_long_t s_remove_count;
1412 /* Being remounted read-only */
1413 int s_readonly_remount;
1415 /* AIO completions deferred from interrupt context */
1416 struct workqueue_struct *s_dio_done_wq;
1417 struct hlist_head s_pins;
1420 * Keep the lru lists last in the structure so they always sit on their
1421 * own individual cachelines.
1423 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1424 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1425 struct rcu_head rcu;
1426 struct work_struct destroy_work;
1428 struct mutex s_sync_lock; /* sync serialisation lock */
1431 * Indicates how deep in a filesystem stack this SB is
1435 /* s_inode_list_lock protects s_inodes */
1436 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1437 struct list_head s_inodes; /* all inodes */
1440 extern struct timespec current_fs_time(struct super_block *sb);
1443 * Snapshotting support.
1446 void __sb_end_write(struct super_block *sb, int level);
1447 int __sb_start_write(struct super_block *sb, int level, bool wait);
1449 #define __sb_writers_acquired(sb, lev) \
1450 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1451 #define __sb_writers_release(sb, lev) \
1452 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1455 * sb_end_write - drop write access to a superblock
1456 * @sb: the super we wrote to
1458 * Decrement number of writers to the filesystem. Wake up possible waiters
1459 * wanting to freeze the filesystem.
1461 static inline void sb_end_write(struct super_block *sb)
1463 __sb_end_write(sb, SB_FREEZE_WRITE);
1467 * sb_end_pagefault - drop write access to a superblock from a page fault
1468 * @sb: the super we wrote to
1470 * Decrement number of processes handling write page fault to the filesystem.
1471 * Wake up possible waiters wanting to freeze the filesystem.
1473 static inline void sb_end_pagefault(struct super_block *sb)
1475 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1479 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1480 * @sb: the super we wrote to
1482 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1483 * waiters wanting to freeze the filesystem.
1485 static inline void sb_end_intwrite(struct super_block *sb)
1487 __sb_end_write(sb, SB_FREEZE_FS);
1491 * sb_start_write - get write access to a superblock
1492 * @sb: the super we write to
1494 * When a process wants to write data or metadata to a file system (i.e. dirty
1495 * a page or an inode), it should embed the operation in a sb_start_write() -
1496 * sb_end_write() pair to get exclusion against file system freezing. This
1497 * function increments number of writers preventing freezing. If the file
1498 * system is already frozen, the function waits until the file system is
1501 * Since freeze protection behaves as a lock, users have to preserve
1502 * ordering of freeze protection and other filesystem locks. Generally,
1503 * freeze protection should be the outermost lock. In particular, we have:
1506 * -> i_mutex (write path, truncate, directory ops, ...)
1507 * -> s_umount (freeze_super, thaw_super)
1509 static inline void sb_start_write(struct super_block *sb)
1511 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1514 static inline int sb_start_write_trylock(struct super_block *sb)
1516 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1520 * sb_start_pagefault - get write access to a superblock from a page fault
1521 * @sb: the super we write to
1523 * When a process starts handling write page fault, it should embed the
1524 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1525 * exclusion against file system freezing. This is needed since the page fault
1526 * is going to dirty a page. This function increments number of running page
1527 * faults preventing freezing. If the file system is already frozen, the
1528 * function waits until the file system is thawed.
1530 * Since page fault freeze protection behaves as a lock, users have to preserve
1531 * ordering of freeze protection and other filesystem locks. It is advised to
1532 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1533 * handling code implies lock dependency:
1536 * -> sb_start_pagefault
1538 static inline void sb_start_pagefault(struct super_block *sb)
1540 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1544 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1545 * @sb: the super we write to
1547 * This is the third level of protection against filesystem freezing. It is
1548 * free for use by a filesystem. The only requirement is that it must rank
1549 * below sb_start_pagefault.
1551 * For example filesystem can call sb_start_intwrite() when starting a
1552 * transaction which somewhat eases handling of freezing for internal sources
1553 * of filesystem changes (internal fs threads, discarding preallocation on file
1556 static inline void sb_start_intwrite(struct super_block *sb)
1558 __sb_start_write(sb, SB_FREEZE_FS, true);
1562 extern bool inode_owner_or_capable(const struct inode *inode);
1565 * VFS helper functions..
1567 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1568 extern int vfs_create2(struct vfsmount *, struct inode *, struct dentry *, umode_t, bool);
1569 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1570 extern int vfs_mkdir2(struct vfsmount *, struct inode *, struct dentry *, umode_t);
1571 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1572 extern int vfs_mknod2(struct vfsmount *, struct inode *, struct dentry *, umode_t, dev_t);
1573 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1574 extern int vfs_symlink2(struct vfsmount *, struct inode *, struct dentry *, const char *);
1575 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1576 extern int vfs_link2(struct vfsmount *, struct dentry *, struct inode *, struct dentry *, struct inode **);
1577 extern int vfs_rmdir(struct inode *, struct dentry *);
1578 extern int vfs_rmdir2(struct vfsmount *, struct inode *, struct dentry *);
1579 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1580 extern int vfs_unlink2(struct vfsmount *, struct inode *, struct dentry *, struct inode **);
1581 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1582 extern int vfs_rename2(struct vfsmount *, struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1583 extern int vfs_whiteout(struct inode *, struct dentry *);
1586 * VFS dentry helper functions.
1588 extern void dentry_unhash(struct dentry *dentry);
1591 * VFS file helper functions.
1593 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1596 * VFS FS_IOC_FIEMAP helper definitions.
1598 struct fiemap_extent_info {
1599 unsigned int fi_flags; /* Flags as passed from user */
1600 unsigned int fi_extents_mapped; /* Number of mapped extents */
1601 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1602 struct fiemap_extent __user *fi_extents_start; /* Start of
1603 fiemap_extent array */
1605 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1606 u64 phys, u64 len, u32 flags);
1607 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1612 * NOTE! These match bits 12..15 of stat.st_mode
1613 * (ie "(i_mode >> 12) & 15").
1615 #define DT_UNKNOWN 0
1626 * This is the "filldir" function type, used by readdir() to let
1627 * the kernel specify what kind of dirent layout it wants to have.
1628 * This allows the kernel to read directories into kernel space or
1629 * to have different dirent layouts depending on the binary type.
1632 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1635 struct dir_context {
1636 const filldir_t actor;
1640 struct block_device_operations;
1642 /* These macros are for out of kernel modules to test that
1643 * the kernel supports the unlocked_ioctl and compat_ioctl
1644 * fields in struct file_operations. */
1645 #define HAVE_COMPAT_IOCTL 1
1646 #define HAVE_UNLOCKED_IOCTL 1
1649 * These flags let !MMU mmap() govern direct device mapping vs immediate
1650 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1652 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1653 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1654 * NOMMU_MAP_READ: Can be mapped for reading
1655 * NOMMU_MAP_WRITE: Can be mapped for writing
1656 * NOMMU_MAP_EXEC: Can be mapped for execution
1658 #define NOMMU_MAP_COPY 0x00000001
1659 #define NOMMU_MAP_DIRECT 0x00000008
1660 #define NOMMU_MAP_READ VM_MAYREAD
1661 #define NOMMU_MAP_WRITE VM_MAYWRITE
1662 #define NOMMU_MAP_EXEC VM_MAYEXEC
1664 #define NOMMU_VMFLAGS \
1665 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1670 struct file_operations {
1671 struct module *owner;
1672 loff_t (*llseek) (struct file *, loff_t, int);
1673 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1674 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1675 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1676 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1677 int (*iterate) (struct file *, struct dir_context *);
1678 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1679 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1680 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1681 int (*mmap) (struct file *, struct vm_area_struct *);
1682 int (*open) (struct inode *, struct file *);
1683 int (*flush) (struct file *, fl_owner_t id);
1684 int (*release) (struct inode *, struct file *);
1685 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1686 int (*aio_fsync) (struct kiocb *, int datasync);
1687 int (*fasync) (int, struct file *, int);
1688 int (*lock) (struct file *, int, struct file_lock *);
1689 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1690 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1691 int (*check_flags)(int);
1692 int (*flock) (struct file *, int, struct file_lock *);
1693 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1694 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1695 int (*setlease)(struct file *, long, struct file_lock **, void **);
1696 long (*fallocate)(struct file *file, int mode, loff_t offset,
1698 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1700 unsigned (*mmap_capabilities)(struct file *);
1704 struct inode_operations {
1705 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1706 const char * (*follow_link) (struct dentry *, void **);
1707 int (*permission) (struct inode *, int);
1708 int (*permission2) (struct vfsmount *, struct inode *, int);
1709 struct posix_acl * (*get_acl)(struct inode *, int);
1711 int (*readlink) (struct dentry *, char __user *,int);
1712 void (*put_link) (struct inode *, void *);
1714 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1715 int (*link) (struct dentry *,struct inode *,struct dentry *);
1716 int (*unlink) (struct inode *,struct dentry *);
1717 int (*symlink) (struct inode *,struct dentry *,const char *);
1718 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1719 int (*rmdir) (struct inode *,struct dentry *);
1720 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1721 int (*rename) (struct inode *, struct dentry *,
1722 struct inode *, struct dentry *);
1723 int (*rename2) (struct inode *, struct dentry *,
1724 struct inode *, struct dentry *, unsigned int);
1725 int (*setattr) (struct dentry *, struct iattr *);
1726 int (*setattr2) (struct vfsmount *, struct dentry *, struct iattr *);
1727 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1728 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1729 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1730 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1731 int (*removexattr) (struct dentry *, const char *);
1732 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1734 int (*update_time)(struct inode *, struct timespec *, int);
1735 int (*atomic_open)(struct inode *, struct dentry *,
1736 struct file *, unsigned open_flag,
1737 umode_t create_mode, int *opened);
1738 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1739 int (*set_acl)(struct inode *, struct posix_acl *, int);
1740 } ____cacheline_aligned;
1742 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1743 unsigned long nr_segs, unsigned long fast_segs,
1744 struct iovec *fast_pointer,
1745 struct iovec **ret_pointer);
1747 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1748 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1749 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1750 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1751 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1752 unsigned long, loff_t *);
1753 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1754 unsigned long, loff_t *);
1756 struct super_operations {
1757 struct inode *(*alloc_inode)(struct super_block *sb);
1758 void (*destroy_inode)(struct inode *);
1760 void (*dirty_inode) (struct inode *, int flags);
1761 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1762 int (*drop_inode) (struct inode *);
1763 void (*evict_inode) (struct inode *);
1764 void (*put_super) (struct super_block *);
1765 int (*sync_fs)(struct super_block *sb, int wait);
1766 int (*freeze_super) (struct super_block *);
1767 int (*freeze_fs) (struct super_block *);
1768 int (*thaw_super) (struct super_block *);
1769 int (*unfreeze_fs) (struct super_block *);
1770 int (*statfs) (struct dentry *, struct kstatfs *);
1771 int (*remount_fs) (struct super_block *, int *, char *);
1772 int (*remount_fs2) (struct vfsmount *, struct super_block *, int *, char *);
1773 void *(*clone_mnt_data) (void *);
1774 void (*copy_mnt_data) (void *, void *);
1775 void (*umount_begin) (struct super_block *);
1777 int (*show_options)(struct seq_file *, struct dentry *);
1778 int (*show_options2)(struct vfsmount *,struct seq_file *, struct dentry *);
1779 int (*show_devname)(struct seq_file *, struct dentry *);
1780 int (*show_path)(struct seq_file *, struct dentry *);
1781 int (*show_stats)(struct seq_file *, struct dentry *);
1783 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1784 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1785 struct dquot **(*get_dquots)(struct inode *);
1787 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1788 long (*nr_cached_objects)(struct super_block *,
1789 struct shrink_control *);
1790 long (*free_cached_objects)(struct super_block *,
1791 struct shrink_control *);
1795 * Inode flags - they have no relation to superblock flags now
1797 #define S_SYNC 1 /* Writes are synced at once */
1798 #define S_NOATIME 2 /* Do not update access times */
1799 #define S_APPEND 4 /* Append-only file */
1800 #define S_IMMUTABLE 8 /* Immutable file */
1801 #define S_DEAD 16 /* removed, but still open directory */
1802 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1803 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1804 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1805 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1806 #define S_PRIVATE 512 /* Inode is fs-internal */
1807 #define S_IMA 1024 /* Inode has an associated IMA struct */
1808 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1809 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1810 #ifdef CONFIG_FS_DAX
1811 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1813 #define S_DAX 0 /* Make all the DAX code disappear */
1817 * Note that nosuid etc flags are inode-specific: setting some file-system
1818 * flags just means all the inodes inherit those flags by default. It might be
1819 * possible to override it selectively if you really wanted to with some
1820 * ioctl() that is not currently implemented.
1822 * Exception: MS_RDONLY is always applied to the entire file system.
1824 * Unfortunately, it is possible to change a filesystems flags with it mounted
1825 * with files in use. This means that all of the inodes will not have their
1826 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1827 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1829 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1831 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1832 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1833 ((inode)->i_flags & S_SYNC))
1834 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1835 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1836 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1837 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1838 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1840 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1841 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1842 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1843 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1845 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1846 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1847 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1848 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1849 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1850 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1851 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1852 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1854 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1855 (inode)->i_rdev == WHITEOUT_DEV)
1858 * Inode state bits. Protected by inode->i_lock
1860 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1861 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1863 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1864 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1865 * various stages of removing an inode.
1867 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1869 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1870 * fdatasync(). i_atime is the usual cause.
1871 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1872 * these changes separately from I_DIRTY_SYNC so that we
1873 * don't have to write inode on fdatasync() when only
1874 * mtime has changed in it.
1875 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1876 * I_NEW Serves as both a mutex and completion notification.
1877 * New inodes set I_NEW. If two processes both create
1878 * the same inode, one of them will release its inode and
1879 * wait for I_NEW to be released before returning.
1880 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1881 * also cause waiting on I_NEW, without I_NEW actually
1882 * being set. find_inode() uses this to prevent returning
1883 * nearly-dead inodes.
1884 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1885 * is zero. I_FREEING must be set when I_WILL_FREE is
1887 * I_FREEING Set when inode is about to be freed but still has dirty
1888 * pages or buffers attached or the inode itself is still
1890 * I_CLEAR Added by clear_inode(). In this state the inode is
1891 * clean and can be destroyed. Inode keeps I_FREEING.
1893 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1894 * prohibited for many purposes. iget() must wait for
1895 * the inode to be completely released, then create it
1896 * anew. Other functions will just ignore such inodes,
1897 * if appropriate. I_NEW is used for waiting.
1899 * I_SYNC Writeback of inode is running. The bit is set during
1900 * data writeback, and cleared with a wakeup on the bit
1901 * address once it is done. The bit is also used to pin
1902 * the inode in memory for flusher thread.
1904 * I_REFERENCED Marks the inode as recently references on the LRU list.
1906 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1908 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1909 * synchronize competing switching instances and to tell
1910 * wb stat updates to grab mapping->tree_lock. See
1911 * inode_switch_wb_work_fn() for details.
1913 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1915 #define I_DIRTY_SYNC (1 << 0)
1916 #define I_DIRTY_DATASYNC (1 << 1)
1917 #define I_DIRTY_PAGES (1 << 2)
1919 #define I_NEW (1 << __I_NEW)
1920 #define I_WILL_FREE (1 << 4)
1921 #define I_FREEING (1 << 5)
1922 #define I_CLEAR (1 << 6)
1924 #define I_SYNC (1 << __I_SYNC)
1925 #define I_REFERENCED (1 << 8)
1926 #define __I_DIO_WAKEUP 9
1927 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1928 #define I_LINKABLE (1 << 10)
1929 #define I_DIRTY_TIME (1 << 11)
1930 #define __I_DIRTY_TIME_EXPIRED 12
1931 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1932 #define I_WB_SWITCH (1 << 13)
1934 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1935 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1937 extern void __mark_inode_dirty(struct inode *, int);
1938 static inline void mark_inode_dirty(struct inode *inode)
1940 __mark_inode_dirty(inode, I_DIRTY);
1943 static inline void mark_inode_dirty_sync(struct inode *inode)
1945 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1948 extern void inc_nlink(struct inode *inode);
1949 extern void drop_nlink(struct inode *inode);
1950 extern void clear_nlink(struct inode *inode);
1951 extern void set_nlink(struct inode *inode, unsigned int nlink);
1953 static inline void inode_inc_link_count(struct inode *inode)
1956 mark_inode_dirty(inode);
1959 static inline void inode_dec_link_count(struct inode *inode)
1962 mark_inode_dirty(inode);
1966 * inode_inc_iversion - increments i_version
1967 * @inode: inode that need to be updated
1969 * Every time the inode is modified, the i_version field will be incremented.
1970 * The filesystem has to be mounted with i_version flag
1973 static inline void inode_inc_iversion(struct inode *inode)
1975 spin_lock(&inode->i_lock);
1977 spin_unlock(&inode->i_lock);
1980 enum file_time_flags {
1987 extern bool atime_needs_update(const struct path *, struct inode *);
1988 extern void touch_atime(const struct path *);
1989 static inline void file_accessed(struct file *file)
1991 if (!(file->f_flags & O_NOATIME))
1992 touch_atime(&file->f_path);
1995 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1996 int sync_inode_metadata(struct inode *inode, int wait);
1998 struct file_system_type {
2001 #define FS_REQUIRES_DEV 1
2002 #define FS_BINARY_MOUNTDATA 2
2003 #define FS_HAS_SUBTYPE 4
2004 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2005 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
2006 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
2007 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2008 struct dentry *(*mount) (struct file_system_type *, int,
2009 const char *, void *);
2010 struct dentry *(*mount2) (struct vfsmount *, struct file_system_type *, int,
2011 const char *, void *);
2012 void *(*alloc_mnt_data) (void);
2013 void (*kill_sb) (struct super_block *);
2014 struct module *owner;
2015 struct file_system_type * next;
2016 struct hlist_head fs_supers;
2018 struct lock_class_key s_lock_key;
2019 struct lock_class_key s_umount_key;
2020 struct lock_class_key s_vfs_rename_key;
2021 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2023 struct lock_class_key i_lock_key;
2024 struct lock_class_key i_mutex_key;
2025 struct lock_class_key i_mutex_dir_key;
2028 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2030 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
2031 void *data, int (*fill_super)(struct super_block *, void *, int));
2032 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2033 int flags, const char *dev_name, void *data,
2034 int (*fill_super)(struct super_block *, void *, int));
2035 extern struct dentry *mount_single(struct file_system_type *fs_type,
2036 int flags, void *data,
2037 int (*fill_super)(struct super_block *, void *, int));
2038 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2039 int flags, void *data,
2040 int (*fill_super)(struct super_block *, void *, int));
2041 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2042 void generic_shutdown_super(struct super_block *sb);
2043 void kill_block_super(struct super_block *sb);
2044 void kill_anon_super(struct super_block *sb);
2045 void kill_litter_super(struct super_block *sb);
2046 void deactivate_super(struct super_block *sb);
2047 void deactivate_locked_super(struct super_block *sb);
2048 int set_anon_super(struct super_block *s, void *data);
2049 int get_anon_bdev(dev_t *);
2050 void free_anon_bdev(dev_t);
2051 struct super_block *sget(struct file_system_type *type,
2052 int (*test)(struct super_block *,void *),
2053 int (*set)(struct super_block *,void *),
2054 int flags, void *data);
2055 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2056 const struct super_operations *ops,
2057 const struct dentry_operations *dops,
2060 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2061 #define fops_get(fops) \
2062 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2063 #define fops_put(fops) \
2064 do { if (fops) module_put((fops)->owner); } while(0)
2066 * This one is to be used *ONLY* from ->open() instances.
2067 * fops must be non-NULL, pinned down *and* module dependencies
2068 * should be sufficient to pin the caller down as well.
2070 #define replace_fops(f, fops) \
2072 struct file *__file = (f); \
2073 fops_put(__file->f_op); \
2074 BUG_ON(!(__file->f_op = (fops))); \
2077 extern int register_filesystem(struct file_system_type *);
2078 extern int unregister_filesystem(struct file_system_type *);
2079 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2080 #define kern_mount(type) kern_mount_data(type, NULL)
2081 extern void kern_unmount(struct vfsmount *mnt);
2082 extern int may_umount_tree(struct vfsmount *);
2083 extern int may_umount(struct vfsmount *);
2084 extern long do_mount(const char *, const char __user *,
2085 const char *, unsigned long, void *);
2086 extern struct vfsmount *collect_mounts(struct path *);
2087 extern void drop_collected_mounts(struct vfsmount *);
2088 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2090 extern int vfs_statfs(struct path *, struct kstatfs *);
2091 extern int user_statfs(const char __user *, struct kstatfs *);
2092 extern int fd_statfs(int, struct kstatfs *);
2093 extern int vfs_ustat(dev_t, struct kstatfs *);
2094 extern int freeze_super(struct super_block *super);
2095 extern int thaw_super(struct super_block *super);
2096 extern bool our_mnt(struct vfsmount *mnt);
2098 extern int current_umask(void);
2100 extern void ihold(struct inode * inode);
2101 extern void iput(struct inode *);
2102 extern int generic_update_time(struct inode *, struct timespec *, int);
2105 extern struct kobject *fs_kobj;
2107 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2109 #define FLOCK_VERIFY_READ 1
2110 #define FLOCK_VERIFY_WRITE 2
2112 #ifdef CONFIG_FILE_LOCKING
2113 extern int locks_mandatory_locked(struct file *);
2114 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2117 * Candidates for mandatory locking have the setgid bit set
2118 * but no group execute bit - an otherwise meaningless combination.
2121 static inline int __mandatory_lock(struct inode *ino)
2123 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2127 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2128 * otherwise these will be advisory locks
2131 static inline int mandatory_lock(struct inode *ino)
2133 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2136 static inline int locks_verify_locked(struct file *file)
2138 if (mandatory_lock(file_inode(file)))
2139 return locks_mandatory_locked(file);
2143 static inline int locks_verify_truncate(struct inode *inode,
2147 if (inode->i_flctx && mandatory_lock(inode))
2148 return locks_mandatory_area(
2149 FLOCK_VERIFY_WRITE, inode, filp,
2150 size < inode->i_size ? size : inode->i_size,
2151 (size < inode->i_size ? inode->i_size - size
2152 : size - inode->i_size)
2157 static inline int break_lease(struct inode *inode, unsigned int mode)
2160 * Since this check is lockless, we must ensure that any refcounts
2161 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2162 * could end up racing with tasks trying to set a new lease on this
2166 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2167 return __break_lease(inode, mode, FL_LEASE);
2171 static inline int break_deleg(struct inode *inode, unsigned int mode)
2174 * Since this check is lockless, we must ensure that any refcounts
2175 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2176 * could end up racing with tasks trying to set a new lease on this
2180 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2181 return __break_lease(inode, mode, FL_DELEG);
2185 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2189 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2190 if (ret == -EWOULDBLOCK && delegated_inode) {
2191 *delegated_inode = inode;
2197 static inline int break_deleg_wait(struct inode **delegated_inode)
2201 ret = break_deleg(*delegated_inode, O_WRONLY);
2202 iput(*delegated_inode);
2203 *delegated_inode = NULL;
2207 static inline int break_layout(struct inode *inode, bool wait)
2210 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2211 return __break_lease(inode,
2212 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2217 #else /* !CONFIG_FILE_LOCKING */
2218 static inline int locks_mandatory_locked(struct file *file)
2223 static inline int locks_mandatory_area(int rw, struct inode *inode,
2224 struct file *filp, loff_t offset,
2230 static inline int __mandatory_lock(struct inode *inode)
2235 static inline int mandatory_lock(struct inode *inode)
2240 static inline int locks_verify_locked(struct file *file)
2245 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2251 static inline int break_lease(struct inode *inode, unsigned int mode)
2256 static inline int break_deleg(struct inode *inode, unsigned int mode)
2261 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2266 static inline int break_deleg_wait(struct inode **delegated_inode)
2272 static inline int break_layout(struct inode *inode, bool wait)
2277 #endif /* CONFIG_FILE_LOCKING */
2282 const char *name; /* pointer to actual string */
2283 const __user char *uptr; /* original userland pointer */
2284 struct audit_names *aname;
2289 extern long vfs_truncate(struct path *, loff_t);
2290 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2292 extern int do_truncate2(struct vfsmount *, struct dentry *, loff_t start,
2293 unsigned int time_attrs, struct file *filp);
2294 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2296 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2298 extern struct file *file_open_name(struct filename *, int, umode_t);
2299 extern struct file *filp_open(const char *, int, umode_t);
2300 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2301 const char *, int, umode_t);
2302 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2303 extern int filp_close(struct file *, fl_owner_t id);
2305 extern struct filename *getname_flags(const char __user *, int, int *);
2306 extern struct filename *getname(const char __user *);
2307 extern struct filename *getname_kernel(const char *);
2308 extern void putname(struct filename *name);
2314 extern int finish_open(struct file *file, struct dentry *dentry,
2315 int (*open)(struct inode *, struct file *),
2317 extern int finish_no_open(struct file *file, struct dentry *dentry);
2321 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2324 extern void __init vfs_caches_init_early(void);
2325 extern void __init vfs_caches_init(void);
2327 extern struct kmem_cache *names_cachep;
2329 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2330 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2333 extern int register_blkdev(unsigned int, const char *);
2334 extern void unregister_blkdev(unsigned int, const char *);
2335 extern void bdev_unhash_inode(dev_t dev);
2336 extern struct block_device *bdget(dev_t);
2337 extern struct block_device *bdgrab(struct block_device *bdev);
2338 extern void bd_set_size(struct block_device *, loff_t size);
2339 extern void bd_forget(struct inode *inode);
2340 extern void bdput(struct block_device *);
2341 extern void invalidate_bdev(struct block_device *);
2342 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2343 extern int sync_blockdev(struct block_device *bdev);
2344 extern void kill_bdev(struct block_device *);
2345 extern struct super_block *freeze_bdev(struct block_device *);
2346 extern void emergency_thaw_all(void);
2347 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2348 extern int fsync_bdev(struct block_device *);
2350 extern struct super_block *blockdev_superblock;
2352 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2354 return sb == blockdev_superblock;
2357 static inline void bd_forget(struct inode *inode) {}
2358 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2359 static inline void kill_bdev(struct block_device *bdev) {}
2360 static inline void invalidate_bdev(struct block_device *bdev) {}
2362 static inline struct super_block *freeze_bdev(struct block_device *sb)
2367 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2372 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2376 static inline int sb_is_blkdev_sb(struct super_block *sb)
2381 extern int sync_filesystem(struct super_block *);
2382 extern const struct file_operations def_blk_fops;
2383 extern const struct file_operations def_chr_fops;
2385 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2386 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2387 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2388 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2389 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2391 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2393 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2394 extern int __blkdev_reread_part(struct block_device *bdev);
2395 extern int blkdev_reread_part(struct block_device *bdev);
2398 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2399 extern void bd_unlink_disk_holder(struct block_device *bdev,
2400 struct gendisk *disk);
2402 static inline int bd_link_disk_holder(struct block_device *bdev,
2403 struct gendisk *disk)
2407 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2408 struct gendisk *disk)
2415 #define CHRDEV_MAJOR_HASH_SIZE 255
2416 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2417 extern int register_chrdev_region(dev_t, unsigned, const char *);
2418 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2419 unsigned int count, const char *name,
2420 const struct file_operations *fops);
2421 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2422 unsigned int count, const char *name);
2423 extern void unregister_chrdev_region(dev_t, unsigned);
2424 extern void chrdev_show(struct seq_file *,off_t);
2426 static inline int register_chrdev(unsigned int major, const char *name,
2427 const struct file_operations *fops)
2429 return __register_chrdev(major, 0, 256, name, fops);
2432 static inline void unregister_chrdev(unsigned int major, const char *name)
2434 __unregister_chrdev(major, 0, 256, name);
2437 /* fs/block_dev.c */
2438 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2439 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2442 #define BLKDEV_MAJOR_HASH_SIZE 255
2443 extern const char *__bdevname(dev_t, char *buffer);
2444 extern const char *bdevname(struct block_device *bdev, char *buffer);
2445 extern struct block_device *lookup_bdev(const char *);
2446 extern void blkdev_show(struct seq_file *,off_t);
2449 #define BLKDEV_MAJOR_HASH_SIZE 0
2452 extern void init_special_inode(struct inode *, umode_t, dev_t);
2454 /* Invalid inode operations -- fs/bad_inode.c */
2455 extern void make_bad_inode(struct inode *);
2456 extern int is_bad_inode(struct inode *);
2460 * return READ, READA, or WRITE
2462 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2465 * return data direction, READ or WRITE
2467 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2469 extern void check_disk_size_change(struct gendisk *disk,
2470 struct block_device *bdev);
2471 extern int revalidate_disk(struct gendisk *);
2472 extern int check_disk_change(struct block_device *);
2473 extern int __invalidate_device(struct block_device *, bool);
2474 extern int invalidate_partition(struct gendisk *, int);
2476 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2477 pgoff_t start, pgoff_t end);
2479 static inline void invalidate_remote_inode(struct inode *inode)
2481 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2482 S_ISLNK(inode->i_mode))
2483 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2485 extern int invalidate_inode_pages2(struct address_space *mapping);
2486 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2487 pgoff_t start, pgoff_t end);
2488 extern int write_inode_now(struct inode *, int);
2489 extern int filemap_fdatawrite(struct address_space *);
2490 extern int filemap_flush(struct address_space *);
2491 extern int filemap_fdatawait(struct address_space *);
2492 extern void filemap_fdatawait_keep_errors(struct address_space *);
2493 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2495 extern int filemap_write_and_wait(struct address_space *mapping);
2496 extern int filemap_write_and_wait_range(struct address_space *mapping,
2497 loff_t lstart, loff_t lend);
2498 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2499 loff_t start, loff_t end, int sync_mode);
2500 extern int filemap_fdatawrite_range(struct address_space *mapping,
2501 loff_t start, loff_t end);
2503 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2505 extern int vfs_fsync(struct file *file, int datasync);
2506 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2508 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2510 return vfs_fsync_range(file, pos, pos + count - 1,
2511 (file->f_flags & __O_SYNC) ? 0 : 1);
2513 extern void emergency_sync(void);
2514 extern void emergency_remount(void);
2516 extern sector_t bmap(struct inode *, sector_t);
2518 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2519 extern int notify_change2(struct vfsmount *, struct dentry *, struct iattr *, struct inode **);
2520 extern int inode_permission(struct inode *, int);
2521 extern int inode_permission2(struct vfsmount *, struct inode *, int);
2522 extern int __inode_permission(struct inode *, int);
2523 extern int __inode_permission2(struct vfsmount *, struct inode *, int);
2524 extern int generic_permission(struct inode *, int);
2525 extern int __check_sticky(struct inode *dir, struct inode *inode);
2527 static inline bool execute_ok(struct inode *inode)
2529 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2532 static inline void file_start_write(struct file *file)
2534 if (!S_ISREG(file_inode(file)->i_mode))
2536 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2539 static inline bool file_start_write_trylock(struct file *file)
2541 if (!S_ISREG(file_inode(file)->i_mode))
2543 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2546 static inline void file_end_write(struct file *file)
2548 if (!S_ISREG(file_inode(file)->i_mode))
2550 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2554 * get_write_access() gets write permission for a file.
2555 * put_write_access() releases this write permission.
2556 * This is used for regular files.
2557 * We cannot support write (and maybe mmap read-write shared) accesses and
2558 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2559 * can have the following values:
2560 * 0: no writers, no VM_DENYWRITE mappings
2561 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2562 * > 0: (i_writecount) users are writing to the file.
2564 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2565 * except for the cases where we don't hold i_writecount yet. Then we need to
2566 * use {get,deny}_write_access() - these functions check the sign and refuse
2567 * to do the change if sign is wrong.
2569 static inline int get_write_access(struct inode *inode)
2571 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2573 static inline int deny_write_access(struct file *file)
2575 struct inode *inode = file_inode(file);
2576 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2578 static inline void put_write_access(struct inode * inode)
2580 atomic_dec(&inode->i_writecount);
2582 static inline void allow_write_access(struct file *file)
2585 atomic_inc(&file_inode(file)->i_writecount);
2587 static inline bool inode_is_open_for_write(const struct inode *inode)
2589 return atomic_read(&inode->i_writecount) > 0;
2593 static inline void i_readcount_dec(struct inode *inode)
2595 BUG_ON(!atomic_read(&inode->i_readcount));
2596 atomic_dec(&inode->i_readcount);
2598 static inline void i_readcount_inc(struct inode *inode)
2600 atomic_inc(&inode->i_readcount);
2603 static inline void i_readcount_dec(struct inode *inode)
2607 static inline void i_readcount_inc(struct inode *inode)
2612 extern int do_pipe_flags(int *, int);
2614 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2615 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2616 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2617 extern struct file * open_exec(const char *);
2619 /* fs/dcache.c -- generic fs support functions */
2620 extern int is_subdir(struct dentry *, struct dentry *);
2621 extern int path_is_under(struct path *, struct path *);
2623 extern char *file_path(struct file *, char *, int);
2625 #include <linux/err.h>
2627 /* needed for stackable file system support */
2628 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2630 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2632 extern int inode_init_always(struct super_block *, struct inode *);
2633 extern void inode_init_once(struct inode *);
2634 extern void address_space_init_once(struct address_space *mapping);
2635 extern struct inode * igrab(struct inode *);
2636 extern ino_t iunique(struct super_block *, ino_t);
2637 extern int inode_needs_sync(struct inode *inode);
2638 extern int generic_delete_inode(struct inode *inode);
2639 static inline int generic_drop_inode(struct inode *inode)
2641 return !inode->i_nlink || inode_unhashed(inode);
2644 extern struct inode *ilookup5_nowait(struct super_block *sb,
2645 unsigned long hashval, int (*test)(struct inode *, void *),
2647 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2648 int (*test)(struct inode *, void *), void *data);
2649 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2651 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2652 extern struct inode * iget_locked(struct super_block *, unsigned long);
2653 extern struct inode *find_inode_nowait(struct super_block *,
2655 int (*match)(struct inode *,
2656 unsigned long, void *),
2658 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2659 extern int insert_inode_locked(struct inode *);
2660 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2661 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2663 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2665 extern void unlock_new_inode(struct inode *);
2666 extern unsigned int get_next_ino(void);
2668 extern void __iget(struct inode * inode);
2669 extern void iget_failed(struct inode *);
2670 extern void clear_inode(struct inode *);
2671 extern void __destroy_inode(struct inode *);
2672 extern struct inode *new_inode_pseudo(struct super_block *sb);
2673 extern struct inode *new_inode(struct super_block *sb);
2674 extern void free_inode_nonrcu(struct inode *inode);
2675 extern int should_remove_suid(struct dentry *);
2676 extern int file_remove_privs(struct file *);
2677 extern int dentry_needs_remove_privs(struct dentry *dentry);
2678 static inline int file_needs_remove_privs(struct file *file)
2680 return dentry_needs_remove_privs(file->f_path.dentry);
2683 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2684 static inline void insert_inode_hash(struct inode *inode)
2686 __insert_inode_hash(inode, inode->i_ino);
2689 extern void __remove_inode_hash(struct inode *);
2690 static inline void remove_inode_hash(struct inode *inode)
2692 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2693 __remove_inode_hash(inode);
2696 extern void inode_sb_list_add(struct inode *inode);
2699 extern blk_qc_t submit_bio(int, struct bio *);
2700 extern int bdev_read_only(struct block_device *);
2702 extern int set_blocksize(struct block_device *, int);
2703 extern int sb_set_blocksize(struct super_block *, int);
2704 extern int sb_min_blocksize(struct super_block *, int);
2706 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2707 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2708 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2709 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2710 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2711 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2712 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2713 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2715 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2716 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2718 /* fs/block_dev.c */
2719 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2720 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2721 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2723 extern void block_sync_page(struct page *page);
2726 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2727 struct pipe_inode_info *, size_t, unsigned int);
2728 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2729 struct pipe_inode_info *, size_t, unsigned int);
2730 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2731 struct file *, loff_t *, size_t, unsigned int);
2732 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2733 struct file *out, loff_t *, size_t len, unsigned int flags);
2734 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2735 loff_t *opos, size_t len, unsigned int flags);
2739 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2740 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2741 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2742 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2743 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2744 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2745 int whence, loff_t maxsize, loff_t eof);
2746 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2747 int whence, loff_t size);
2748 extern int generic_file_open(struct inode * inode, struct file * filp);
2749 extern int nonseekable_open(struct inode * inode, struct file * filp);
2752 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2753 loff_t file_offset);
2756 /* need locking between buffered and direct access */
2759 /* filesystem does not support filling holes */
2760 DIO_SKIP_HOLES = 0x02,
2762 /* filesystem can handle aio writes beyond i_size */
2763 DIO_ASYNC_EXTEND = 0x04,
2765 /* inode/fs/bdev does not need truncate protection */
2766 DIO_SKIP_DIO_COUNT = 0x08,
2769 void dio_end_io(struct bio *bio, int error);
2771 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2772 struct block_device *bdev, struct iov_iter *iter,
2773 loff_t offset, get_block_t get_block,
2774 dio_iodone_t end_io, dio_submit_t submit_io,
2777 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2778 struct inode *inode,
2779 struct iov_iter *iter, loff_t offset,
2780 get_block_t get_block)
2782 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2783 offset, get_block, NULL, NULL,
2784 DIO_LOCKING | DIO_SKIP_HOLES);
2788 void inode_dio_wait(struct inode *inode);
2791 * inode_dio_begin - signal start 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_begin(struct inode *inode)
2799 atomic_inc(&inode->i_dio_count);
2803 * inode_dio_end - signal finish of a direct I/O requests
2804 * @inode: inode the direct I/O happens on
2806 * This is called once we've finished processing a direct I/O request,
2807 * and is used to wake up callers waiting for direct I/O to be quiesced.
2809 static inline void inode_dio_end(struct inode *inode)
2811 if (atomic_dec_and_test(&inode->i_dio_count))
2812 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2815 struct inode *dio_bio_get_inode(struct bio *bio);
2817 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2820 extern const struct file_operations generic_ro_fops;
2822 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2824 extern int readlink_copy(char __user *, int, const char *);
2825 extern int page_readlink(struct dentry *, char __user *, int);
2826 extern const char *page_follow_link_light(struct dentry *, void **);
2827 extern void page_put_link(struct inode *, void *);
2828 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2830 extern int page_symlink(struct inode *inode, const char *symname, int len);
2831 extern const struct inode_operations page_symlink_inode_operations;
2832 extern void kfree_put_link(struct inode *, void *);
2833 extern void free_page_put_link(struct inode *, void *);
2834 extern int generic_readlink(struct dentry *, char __user *, int);
2835 extern void generic_fillattr(struct inode *, struct kstat *);
2836 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2837 extern int vfs_getattr(struct path *, struct kstat *);
2838 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2839 void inode_add_bytes(struct inode *inode, loff_t bytes);
2840 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2841 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2842 loff_t inode_get_bytes(struct inode *inode);
2843 void inode_set_bytes(struct inode *inode, loff_t bytes);
2844 const char *simple_follow_link(struct dentry *, void **);
2845 extern const struct inode_operations simple_symlink_inode_operations;
2847 extern int iterate_dir(struct file *, struct dir_context *);
2849 extern int vfs_stat(const char __user *, struct kstat *);
2850 extern int vfs_lstat(const char __user *, struct kstat *);
2851 extern int vfs_fstat(unsigned int, struct kstat *);
2852 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2854 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2856 extern int __generic_block_fiemap(struct inode *inode,
2857 struct fiemap_extent_info *fieinfo,
2858 loff_t start, loff_t len,
2859 get_block_t *get_block);
2860 extern int generic_block_fiemap(struct inode *inode,
2861 struct fiemap_extent_info *fieinfo, u64 start,
2862 u64 len, get_block_t *get_block);
2864 extern void get_filesystem(struct file_system_type *fs);
2865 extern void put_filesystem(struct file_system_type *fs);
2866 extern struct file_system_type *get_fs_type(const char *name);
2867 extern struct super_block *get_super(struct block_device *);
2868 extern struct super_block *get_super_thawed(struct block_device *);
2869 extern struct super_block *get_active_super(struct block_device *bdev);
2870 extern void drop_super(struct super_block *sb);
2871 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2872 extern void iterate_supers_type(struct file_system_type *,
2873 void (*)(struct super_block *, void *), void *);
2875 extern int dcache_dir_open(struct inode *, struct file *);
2876 extern int dcache_dir_close(struct inode *, struct file *);
2877 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2878 extern int dcache_readdir(struct file *, struct dir_context *);
2879 extern int simple_setattr(struct dentry *, struct iattr *);
2880 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2881 extern int simple_statfs(struct dentry *, struct kstatfs *);
2882 extern int simple_open(struct inode *inode, struct file *file);
2883 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2884 extern int simple_unlink(struct inode *, struct dentry *);
2885 extern int simple_rmdir(struct inode *, struct dentry *);
2886 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2887 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2888 extern int simple_empty(struct dentry *);
2889 extern int simple_readpage(struct file *file, struct page *page);
2890 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2891 loff_t pos, unsigned len, unsigned flags,
2892 struct page **pagep, void **fsdata);
2893 extern int simple_write_end(struct file *file, struct address_space *mapping,
2894 loff_t pos, unsigned len, unsigned copied,
2895 struct page *page, void *fsdata);
2896 extern int always_delete_dentry(const struct dentry *);
2897 extern struct inode *alloc_anon_inode(struct super_block *);
2898 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2899 extern const struct dentry_operations simple_dentry_operations;
2901 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2902 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2903 extern const struct file_operations simple_dir_operations;
2904 extern const struct inode_operations simple_dir_inode_operations;
2905 extern void make_empty_dir_inode(struct inode *inode);
2906 extern bool is_empty_dir_inode(struct inode *inode);
2907 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2908 struct dentry *d_alloc_name(struct dentry *, const char *);
2909 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2910 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2911 extern void simple_release_fs(struct vfsmount **mount, int *count);
2913 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2914 loff_t *ppos, const void *from, size_t available);
2915 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2916 const void __user *from, size_t count);
2918 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2919 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2921 extern int generic_check_addressable(unsigned, u64);
2923 #ifdef CONFIG_MIGRATION
2924 extern int buffer_migrate_page(struct address_space *,
2925 struct page *, struct page *,
2928 #define buffer_migrate_page NULL
2931 extern int inode_change_ok(const struct inode *, struct iattr *);
2932 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2933 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2935 extern int file_update_time(struct file *file);
2937 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2938 extern void save_mount_options(struct super_block *sb, char *options);
2939 extern void replace_mount_options(struct super_block *sb, char *options);
2941 static inline bool io_is_direct(struct file *filp)
2943 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2946 static inline int iocb_flags(struct file *file)
2949 if (file->f_flags & O_APPEND)
2951 if (io_is_direct(file))
2956 static inline ino_t parent_ino(struct dentry *dentry)
2961 * Don't strictly need d_lock here? If the parent ino could change
2962 * then surely we'd have a deeper race in the caller?
2964 spin_lock(&dentry->d_lock);
2965 res = dentry->d_parent->d_inode->i_ino;
2966 spin_unlock(&dentry->d_lock);
2970 /* Transaction based IO helpers */
2973 * An argresp is stored in an allocated page and holds the
2974 * size of the argument or response, along with its content
2976 struct simple_transaction_argresp {
2981 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2983 char *simple_transaction_get(struct file *file, const char __user *buf,
2985 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2986 size_t size, loff_t *pos);
2987 int simple_transaction_release(struct inode *inode, struct file *file);
2989 void simple_transaction_set(struct file *file, size_t n);
2992 * simple attribute files
2994 * These attributes behave similar to those in sysfs:
2996 * Writing to an attribute immediately sets a value, an open file can be
2997 * written to multiple times.
2999 * Reading from an attribute creates a buffer from the value that might get
3000 * read with multiple read calls. When the attribute has been read
3001 * completely, no further read calls are possible until the file is opened
3004 * All attributes contain a text representation of a numeric value
3005 * that are accessed with the get() and set() functions.
3007 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3008 static int __fops ## _open(struct inode *inode, struct file *file) \
3010 __simple_attr_check_format(__fmt, 0ull); \
3011 return simple_attr_open(inode, file, __get, __set, __fmt); \
3013 static const struct file_operations __fops = { \
3014 .owner = THIS_MODULE, \
3015 .open = __fops ## _open, \
3016 .release = simple_attr_release, \
3017 .read = simple_attr_read, \
3018 .write = simple_attr_write, \
3019 .llseek = generic_file_llseek, \
3022 static inline __printf(1, 2)
3023 void __simple_attr_check_format(const char *fmt, ...)
3025 /* don't do anything, just let the compiler check the arguments; */
3028 int simple_attr_open(struct inode *inode, struct file *file,
3029 int (*get)(void *, u64 *), int (*set)(void *, u64),
3031 int simple_attr_release(struct inode *inode, struct file *file);
3032 ssize_t simple_attr_read(struct file *file, char __user *buf,
3033 size_t len, loff_t *ppos);
3034 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3035 size_t len, loff_t *ppos);
3038 int proc_nr_files(struct ctl_table *table, int write,
3039 void __user *buffer, size_t *lenp, loff_t *ppos);
3040 int proc_nr_dentry(struct ctl_table *table, int write,
3041 void __user *buffer, size_t *lenp, loff_t *ppos);
3042 int proc_nr_inodes(struct ctl_table *table, int write,
3043 void __user *buffer, size_t *lenp, loff_t *ppos);
3044 int __init get_filesystem_list(char *buf);
3046 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3047 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3049 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3050 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3051 (flag & __FMODE_NONOTIFY)))
3053 static inline int is_sxid(umode_t mode)
3055 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3058 static inline int check_sticky(struct inode *dir, struct inode *inode)
3060 if (!(dir->i_mode & S_ISVTX))
3063 return __check_sticky(dir, inode);
3066 static inline void inode_has_no_xattr(struct inode *inode)
3068 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3069 inode->i_flags |= S_NOSEC;
3072 static inline bool is_root_inode(struct inode *inode)
3074 return inode == inode->i_sb->s_root->d_inode;
3077 static inline bool dir_emit(struct dir_context *ctx,
3078 const char *name, int namelen,
3079 u64 ino, unsigned type)
3081 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3083 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3085 return ctx->actor(ctx, ".", 1, ctx->pos,
3086 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3088 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3090 return ctx->actor(ctx, "..", 2, ctx->pos,
3091 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3093 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3095 if (ctx->pos == 0) {
3096 if (!dir_emit_dot(file, ctx))
3100 if (ctx->pos == 1) {
3101 if (!dir_emit_dotdot(file, ctx))
3107 static inline bool dir_relax(struct inode *inode)
3109 inode_unlock(inode);
3111 return !IS_DEADDIR(inode);
3114 extern bool path_noexec(const struct path *path);
3116 #endif /* _LINUX_FS_H */