4 #include <linux/linkage.h>
5 #include <linux/wait.h>
6 #include <linux/kdev_t.h>
7 #include <linux/dcache.h>
8 #include <linux/path.h>
9 #include <linux/stat.h>
10 #include <linux/cache.h>
11 #include <linux/list.h>
12 #include <linux/list_lru.h>
13 #include <linux/llist.h>
14 #include <linux/radix-tree.h>
15 #include <linux/rbtree.h>
16 #include <linux/init.h>
17 #include <linux/pid.h>
18 #include <linux/bug.h>
19 #include <linux/mutex.h>
20 #include <linux/rwsem.h>
21 #include <linux/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
32 #include <linux/workqueue.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/delayed_call.h>
36 #include <asm/byteorder.h>
37 #include <uapi/linux/fs.h>
39 struct backing_dev_info;
41 struct export_operations;
46 struct pipe_inode_info;
47 struct poll_table_struct;
49 struct vm_area_struct;
52 struct swap_info_struct;
54 struct workqueue_struct;
57 struct fscrypt_operations;
59 extern void __init inode_init(void);
60 extern void __init inode_init_early(void);
61 extern void __init files_init(void);
62 extern void __init files_maxfiles_init(void);
64 extern struct files_stat_struct files_stat;
65 extern unsigned long get_max_files(void);
66 extern int sysctl_nr_open;
67 extern struct inodes_stat_t inodes_stat;
68 extern int leases_enable, lease_break_time;
69 extern int sysctl_protected_symlinks;
70 extern int sysctl_protected_hardlinks;
73 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
74 struct buffer_head *bh_result, int create);
75 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
76 ssize_t bytes, void *private);
77 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
79 #define MAY_EXEC 0x00000001
80 #define MAY_WRITE 0x00000002
81 #define MAY_READ 0x00000004
82 #define MAY_APPEND 0x00000008
83 #define MAY_ACCESS 0x00000010
84 #define MAY_OPEN 0x00000020
85 #define MAY_CHDIR 0x00000040
86 /* called from RCU mode, don't block */
87 #define MAY_NOT_BLOCK 0x00000080
90 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
91 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
94 /* file is open for reading */
95 #define FMODE_READ ((__force fmode_t)0x1)
96 /* file is open for writing */
97 #define FMODE_WRITE ((__force fmode_t)0x2)
98 /* file is seekable */
99 #define FMODE_LSEEK ((__force fmode_t)0x4)
100 /* file can be accessed using pread */
101 #define FMODE_PREAD ((__force fmode_t)0x8)
102 /* file can be accessed using pwrite */
103 #define FMODE_PWRITE ((__force fmode_t)0x10)
104 /* File is opened for execution with sys_execve / sys_uselib */
105 #define FMODE_EXEC ((__force fmode_t)0x20)
106 /* File is opened with O_NDELAY (only set for block devices) */
107 #define FMODE_NDELAY ((__force fmode_t)0x40)
108 /* File is opened with O_EXCL (only set for block devices) */
109 #define FMODE_EXCL ((__force fmode_t)0x80)
110 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
111 (specialy hack for floppy.c) */
112 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
113 /* 32bit hashes as llseek() offset (for directories) */
114 #define FMODE_32BITHASH ((__force fmode_t)0x200)
115 /* 64bit hashes as llseek() offset (for directories) */
116 #define FMODE_64BITHASH ((__force fmode_t)0x400)
119 * Don't update ctime and mtime.
121 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
122 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
124 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
126 /* Expect random access pattern */
127 #define FMODE_RANDOM ((__force fmode_t)0x1000)
129 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
130 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
132 /* File is opened with O_PATH; almost nothing can be done with it */
133 #define FMODE_PATH ((__force fmode_t)0x4000)
135 /* File needs atomic accesses to f_pos */
136 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
137 /* Write access to underlying fs */
138 #define FMODE_WRITER ((__force fmode_t)0x10000)
139 /* Has read method(s) */
140 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
141 /* Has write method(s) */
142 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
144 /* File was opened by fanotify and shouldn't generate fanotify events */
145 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
148 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
149 * that indicates that they should check the contents of the iovec are
150 * valid, but not check the memory that the iovec elements
153 #define CHECK_IOVEC_ONLY -1
156 * The below are the various read and write types that we support. Some of
157 * them include behavioral modifiers that send information down to the
158 * block layer and IO scheduler. Terminology:
160 * The block layer uses device plugging to defer IO a little bit, in
161 * the hope that we will see more IO very shortly. This increases
162 * coalescing of adjacent IO and thus reduces the number of IOs we
163 * have to send to the device. It also allows for better queuing,
164 * if the IO isn't mergeable. If the caller is going to be waiting
165 * for the IO, then he must ensure that the device is unplugged so
166 * that the IO is dispatched to the driver.
168 * All IO is handled async in Linux. This is fine for background
169 * writes, but for reads or writes that someone waits for completion
170 * on, we want to notify the block layer and IO scheduler so that they
171 * know about it. That allows them to make better scheduling
172 * decisions. So when the below references 'sync' and 'async', it
173 * is referencing this priority hint.
175 * With that in mind, the available types are:
177 * READ A normal read operation. Device will be plugged.
178 * READ_SYNC A synchronous read. Device is not plugged, caller can
179 * immediately wait on this read without caring about
181 * READA Used for read-ahead operations. Lower priority, and the
182 * block layer could (in theory) choose to ignore this
183 * request if it runs into resource problems.
184 * WRITE A normal async write. Device will be plugged.
185 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
186 * the hint that someone will be waiting on this IO
187 * shortly. The write equivalent of READ_SYNC.
188 * WRITE_ODIRECT Special case write for O_DIRECT only.
189 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
190 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
191 * non-volatile media on completion.
192 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
193 * by a cache flush and data is guaranteed to be on
194 * non-volatile media on completion.
197 #define RW_MASK REQ_WRITE
198 #define RWA_MASK REQ_RAHEAD
201 #define WRITE RW_MASK
202 #define READA RWA_MASK
204 #define READ_SYNC (READ | REQ_SYNC)
205 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
206 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
207 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
208 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
209 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
212 * Attribute flags. These should be or-ed together to figure out what
215 #define ATTR_MODE (1 << 0)
216 #define ATTR_UID (1 << 1)
217 #define ATTR_GID (1 << 2)
218 #define ATTR_SIZE (1 << 3)
219 #define ATTR_ATIME (1 << 4)
220 #define ATTR_MTIME (1 << 5)
221 #define ATTR_CTIME (1 << 6)
222 #define ATTR_ATIME_SET (1 << 7)
223 #define ATTR_MTIME_SET (1 << 8)
224 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
225 #define ATTR_ATTR_FLAG (1 << 10)
226 #define ATTR_KILL_SUID (1 << 11)
227 #define ATTR_KILL_SGID (1 << 12)
228 #define ATTR_FILE (1 << 13)
229 #define ATTR_KILL_PRIV (1 << 14)
230 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
231 #define ATTR_TIMES_SET (1 << 16)
234 * Whiteout is represented by a char device. The following constants define the
235 * mode and device number to use.
237 #define WHITEOUT_MODE 0
238 #define WHITEOUT_DEV 0
241 * This is the Inode Attributes structure, used for notify_change(). It
242 * uses the above definitions as flags, to know which values have changed.
243 * Also, in this manner, a Filesystem can look at only the values it cares
244 * about. Basically, these are the attributes that the VFS layer can
245 * request to change from the FS layer.
247 * Derek Atkins <warlord@MIT.EDU> 94-10-20
250 unsigned int ia_valid;
255 struct timespec ia_atime;
256 struct timespec ia_mtime;
257 struct timespec ia_ctime;
260 * Not an attribute, but an auxiliary info for filesystems wanting to
261 * implement an ftruncate() like method. NOTE: filesystem should
262 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
264 struct file *ia_file;
268 * Includes for diskquotas.
270 #include <linux/quota.h>
273 * Maximum number of layers of fs stack. Needs to be limited to
274 * prevent kernel stack overflow
276 #define FILESYSTEM_MAX_STACK_DEPTH 2
279 * enum positive_aop_returns - aop return codes with specific semantics
281 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
282 * completed, that the page is still locked, and
283 * should be considered active. The VM uses this hint
284 * to return the page to the active list -- it won't
285 * be a candidate for writeback again in the near
286 * future. Other callers must be careful to unlock
287 * the page if they get this return. Returned by
290 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
291 * unlocked it and the page might have been truncated.
292 * The caller should back up to acquiring a new page and
293 * trying again. The aop will be taking reasonable
294 * precautions not to livelock. If the caller held a page
295 * reference, it should drop it before retrying. Returned
298 * address_space_operation functions return these large constants to indicate
299 * special semantics to the caller. These are much larger than the bytes in a
300 * page to allow for functions that return the number of bytes operated on in a
304 enum positive_aop_returns {
305 AOP_WRITEPAGE_ACTIVATE = 0x80000,
306 AOP_TRUNCATED_PAGE = 0x80001,
309 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
310 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
311 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
312 * helper code (eg buffer layer)
313 * to clear GFP_FS from alloc */
316 * oh the beauties of C type declarations.
319 struct address_space;
320 struct writeback_control;
322 #define IOCB_EVENTFD (1 << 0)
323 #define IOCB_APPEND (1 << 1)
324 #define IOCB_DIRECT (1 << 2)
325 #define IOCB_HIPRI (1 << 3)
328 struct file *ki_filp;
330 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
335 static inline bool is_sync_kiocb(struct kiocb *kiocb)
337 return kiocb->ki_complete == NULL;
340 static inline int iocb_flags(struct file *file);
342 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
344 *kiocb = (struct kiocb) {
346 .ki_flags = iocb_flags(filp),
351 * "descriptor" for what we're up to with a read.
352 * This allows us to use the same read code yet
353 * have multiple different users of the data that
354 * we read from a file.
356 * The simplest case just copies the data to user
369 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
370 unsigned long, unsigned long);
372 struct address_space_operations {
373 int (*writepage)(struct page *page, struct writeback_control *wbc);
374 int (*readpage)(struct file *, struct page *);
376 /* Write back some dirty pages from this mapping. */
377 int (*writepages)(struct address_space *, struct writeback_control *);
379 /* Set a page dirty. Return true if this dirtied it */
380 int (*set_page_dirty)(struct page *page);
382 int (*readpages)(struct file *filp, struct address_space *mapping,
383 struct list_head *pages, unsigned nr_pages);
385 int (*write_begin)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned flags,
387 struct page **pagep, void **fsdata);
388 int (*write_end)(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned copied,
390 struct page *page, void *fsdata);
392 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
393 sector_t (*bmap)(struct address_space *, sector_t);
394 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
395 int (*releasepage) (struct page *, gfp_t);
396 void (*freepage)(struct page *);
397 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
399 * migrate the contents of a page to the specified target. If
400 * migrate_mode is MIGRATE_ASYNC, it must not block.
402 int (*migratepage) (struct address_space *,
403 struct page *, struct page *, enum migrate_mode);
404 int (*launder_page) (struct page *);
405 int (*is_partially_uptodate) (struct page *, unsigned long,
407 void (*is_dirty_writeback) (struct page *, bool *, bool *);
408 int (*error_remove_page)(struct address_space *, struct page *);
410 /* swapfile support */
411 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
413 void (*swap_deactivate)(struct file *file);
416 extern const struct address_space_operations empty_aops;
419 * pagecache_write_begin/pagecache_write_end must be used by general code
420 * to write into the pagecache.
422 int pagecache_write_begin(struct file *, struct address_space *mapping,
423 loff_t pos, unsigned len, unsigned flags,
424 struct page **pagep, void **fsdata);
426 int pagecache_write_end(struct file *, struct address_space *mapping,
427 loff_t pos, unsigned len, unsigned copied,
428 struct page *page, void *fsdata);
430 struct address_space {
431 struct inode *host; /* owner: inode, block_device */
432 struct radix_tree_root page_tree; /* radix tree of all pages */
433 spinlock_t tree_lock; /* and lock protecting it */
434 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
435 struct rb_root i_mmap; /* tree of private and shared mappings */
436 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
437 /* Protected by tree_lock together with the radix tree */
438 unsigned long nrpages; /* number of total pages */
439 /* number of shadow or DAX exceptional entries */
440 unsigned long nrexceptional;
441 pgoff_t writeback_index;/* writeback starts here */
442 const struct address_space_operations *a_ops; /* methods */
443 unsigned long flags; /* error bits/gfp mask */
444 spinlock_t private_lock; /* for use by the address_space */
445 struct list_head private_list; /* ditto */
446 void *private_data; /* ditto */
447 } __attribute__((aligned(sizeof(long))));
449 * On most architectures that alignment is already the case; but
450 * must be enforced here for CRIS, to let the least significant bit
451 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
453 struct request_queue;
455 struct block_device {
456 dev_t bd_dev; /* not a kdev_t - it's a search key */
458 struct inode * bd_inode; /* will die */
459 struct super_block * bd_super;
460 struct mutex bd_mutex; /* open/close mutex */
461 struct list_head bd_inodes;
465 bool bd_write_holder;
467 struct list_head bd_holder_disks;
469 struct block_device * bd_contains;
470 unsigned bd_block_size;
471 struct hd_struct * bd_part;
472 /* number of times partitions within this device have been opened. */
473 unsigned bd_part_count;
475 struct gendisk * bd_disk;
476 struct request_queue * bd_queue;
477 struct list_head bd_list;
479 * Private data. You must have bd_claim'ed the block_device
480 * to use this. NOTE: bd_claim allows an owner to claim
481 * the same device multiple times, the owner must take special
482 * care to not mess up bd_private for that case.
484 unsigned long bd_private;
486 /* The counter of freeze processes */
487 int bd_fsfreeze_count;
488 /* Mutex for freeze */
489 struct mutex bd_fsfreeze_mutex;
493 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
496 #define PAGECACHE_TAG_DIRTY 0
497 #define PAGECACHE_TAG_WRITEBACK 1
498 #define PAGECACHE_TAG_TOWRITE 2
500 int mapping_tagged(struct address_space *mapping, int tag);
502 static inline void i_mmap_lock_write(struct address_space *mapping)
504 down_write(&mapping->i_mmap_rwsem);
507 static inline void i_mmap_unlock_write(struct address_space *mapping)
509 up_write(&mapping->i_mmap_rwsem);
512 static inline void i_mmap_lock_read(struct address_space *mapping)
514 down_read(&mapping->i_mmap_rwsem);
517 static inline void i_mmap_unlock_read(struct address_space *mapping)
519 up_read(&mapping->i_mmap_rwsem);
523 * Might pages of this file be mapped into userspace?
525 static inline int mapping_mapped(struct address_space *mapping)
527 return !RB_EMPTY_ROOT(&mapping->i_mmap);
531 * Might pages of this file have been modified in userspace?
532 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
533 * marks vma as VM_SHARED if it is shared, and the file was opened for
534 * writing i.e. vma may be mprotected writable even if now readonly.
536 * If i_mmap_writable is negative, no new writable mappings are allowed. You
537 * can only deny writable mappings, if none exists right now.
539 static inline int mapping_writably_mapped(struct address_space *mapping)
541 return atomic_read(&mapping->i_mmap_writable) > 0;
544 static inline int mapping_map_writable(struct address_space *mapping)
546 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
550 static inline void mapping_unmap_writable(struct address_space *mapping)
552 atomic_dec(&mapping->i_mmap_writable);
555 static inline int mapping_deny_writable(struct address_space *mapping)
557 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
561 static inline void mapping_allow_writable(struct address_space *mapping)
563 atomic_inc(&mapping->i_mmap_writable);
567 * Use sequence counter to get consistent i_size on 32-bit processors.
569 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
570 #include <linux/seqlock.h>
571 #define __NEED_I_SIZE_ORDERED
572 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
574 #define i_size_ordered_init(inode) do { } while (0)
578 #define ACL_NOT_CACHED ((void *)(-1))
580 static inline struct posix_acl *
581 uncached_acl_sentinel(struct task_struct *task)
583 return (void *)task + 1;
587 is_uncached_acl(struct posix_acl *acl)
589 return (long)acl & 1;
592 #define IOP_FASTPERM 0x0001
593 #define IOP_LOOKUP 0x0002
594 #define IOP_NOFOLLOW 0x0004
597 * Keep mostly read-only and often accessed (especially for
598 * the RCU path lookup and 'stat' data) fields at the beginning
599 * of the 'struct inode'
603 unsigned short i_opflags;
606 unsigned int i_flags;
608 #ifdef CONFIG_FS_POSIX_ACL
609 struct posix_acl *i_acl;
610 struct posix_acl *i_default_acl;
613 const struct inode_operations *i_op;
614 struct super_block *i_sb;
615 struct address_space *i_mapping;
617 #ifdef CONFIG_SECURITY
621 /* Stat data, not accessed from path walking */
624 * Filesystems may only read i_nlink directly. They shall use the
625 * following functions for modification:
627 * (set|clear|inc|drop)_nlink
628 * inode_(inc|dec)_link_count
631 const unsigned int i_nlink;
632 unsigned int __i_nlink;
636 struct timespec i_atime;
637 struct timespec i_mtime;
638 struct timespec i_ctime;
639 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
640 unsigned short i_bytes;
641 unsigned int i_blkbits;
644 #ifdef __NEED_I_SIZE_ORDERED
645 seqcount_t i_size_seqcount;
649 unsigned long i_state;
650 struct mutex i_mutex;
652 unsigned long dirtied_when; /* jiffies of first dirtying */
653 unsigned long dirtied_time_when;
655 struct hlist_node i_hash;
656 struct list_head i_io_list; /* backing dev IO list */
657 #ifdef CONFIG_CGROUP_WRITEBACK
658 struct bdi_writeback *i_wb; /* the associated cgroup wb */
660 /* foreign inode detection, see wbc_detach_inode() */
662 u16 i_wb_frn_avg_time;
663 u16 i_wb_frn_history;
665 struct list_head i_lru; /* inode LRU list */
666 struct list_head i_sb_list;
668 struct hlist_head i_dentry;
669 struct rcu_head i_rcu;
673 atomic_t i_dio_count;
674 atomic_t i_writecount;
676 atomic_t i_readcount; /* struct files open RO */
678 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
679 struct file_lock_context *i_flctx;
680 struct address_space i_data;
681 struct list_head i_devices;
683 struct pipe_inode_info *i_pipe;
684 struct block_device *i_bdev;
691 #ifdef CONFIG_FSNOTIFY
692 __u32 i_fsnotify_mask; /* all events this inode cares about */
693 struct hlist_head i_fsnotify_marks;
696 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
697 struct fscrypt_info *i_crypt_info;
700 void *i_private; /* fs or device private pointer */
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;
921 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
926 /* file identifier */
927 unsigned char f_handle[0];
930 static inline struct file *get_file(struct file *f)
932 atomic_long_inc(&f->f_count);
935 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
936 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
937 #define file_count(x) atomic_long_read(&(x)->f_count)
939 #define MAX_NON_LFS ((1UL<<31) - 1)
941 /* Page cache limit. The filesystems should put that into their s_maxbytes
942 limits, otherwise bad things can happen in VM. */
943 #if BITS_PER_LONG==32
944 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
945 #elif BITS_PER_LONG==64
946 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
951 #define FL_DELEG 4 /* NFSv4 delegation */
952 #define FL_ACCESS 8 /* not trying to lock, just looking */
953 #define FL_EXISTS 16 /* when unlocking, test for existence */
954 #define FL_LEASE 32 /* lease held on this file */
955 #define FL_CLOSE 64 /* unlock on close */
956 #define FL_SLEEP 128 /* A blocking lock */
957 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
958 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
959 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
960 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
963 * Special return value from posix_lock_file() and vfs_lock_file() for
964 * asynchronous locking.
966 #define FILE_LOCK_DEFERRED 1
968 /* legacy typedef, should eventually be removed */
969 typedef void *fl_owner_t;
973 struct file_lock_operations {
974 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
975 void (*fl_release_private)(struct file_lock *);
978 struct lock_manager_operations {
979 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
980 unsigned long (*lm_owner_key)(struct file_lock *);
981 fl_owner_t (*lm_get_owner)(fl_owner_t);
982 void (*lm_put_owner)(fl_owner_t);
983 void (*lm_notify)(struct file_lock *); /* unblock callback */
984 int (*lm_grant)(struct file_lock *, int);
985 bool (*lm_break)(struct file_lock *);
986 int (*lm_change)(struct file_lock *, int, struct list_head *);
987 void (*lm_setup)(struct file_lock *, void **);
990 struct lock_manager {
991 struct list_head list;
993 * NFSv4 and up also want opens blocked during the grace period;
1000 void locks_start_grace(struct net *, struct lock_manager *);
1001 void locks_end_grace(struct lock_manager *);
1002 int locks_in_grace(struct net *);
1003 int opens_in_grace(struct net *);
1005 /* that will die - we need it for nfs_lock_info */
1006 #include <linux/nfs_fs_i.h>
1009 * struct file_lock represents a generic "file lock". It's used to represent
1010 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1011 * note that the same struct is used to represent both a request for a lock and
1012 * the lock itself, but the same object is never used for both.
1014 * FIXME: should we create a separate "struct lock_request" to help distinguish
1017 * The varous i_flctx lists are ordered by:
1020 * 2) lock range start
1023 * Obviously, the last two criteria only matter for POSIX locks.
1026 struct file_lock *fl_next; /* singly linked list for this inode */
1027 struct list_head fl_list; /* link into file_lock_context */
1028 struct hlist_node fl_link; /* node in global lists */
1029 struct list_head fl_block; /* circular list of blocked processes */
1030 fl_owner_t fl_owner;
1031 unsigned int fl_flags;
1032 unsigned char fl_type;
1033 unsigned int fl_pid;
1034 int fl_link_cpu; /* what cpu's list is this on? */
1035 struct pid *fl_nspid;
1036 wait_queue_head_t fl_wait;
1037 struct file *fl_file;
1041 struct fasync_struct * fl_fasync; /* for lease break notifications */
1042 /* for lease breaks: */
1043 unsigned long fl_break_time;
1044 unsigned long fl_downgrade_time;
1046 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1047 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1049 struct nfs_lock_info nfs_fl;
1050 struct nfs4_lock_info nfs4_fl;
1052 struct list_head link; /* link in AFS vnode's pending_locks list */
1053 int state; /* state of grant or error if -ve */
1058 struct file_lock_context {
1059 spinlock_t flc_lock;
1060 struct list_head flc_flock;
1061 struct list_head flc_posix;
1062 struct list_head flc_lease;
1065 /* The following constant reflects the upper bound of the file/locking space */
1067 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1068 #define OFFSET_MAX INT_LIMIT(loff_t)
1069 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1072 #include <linux/fcntl.h>
1074 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1076 #ifdef CONFIG_FILE_LOCKING
1077 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1078 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1079 struct flock __user *);
1081 #if BITS_PER_LONG == 32
1082 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1083 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1084 struct flock64 __user *);
1087 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1088 extern int fcntl_getlease(struct file *filp);
1091 void locks_free_lock_context(struct inode *inode);
1092 void locks_free_lock(struct file_lock *fl);
1093 extern void locks_init_lock(struct file_lock *);
1094 extern struct file_lock * locks_alloc_lock(void);
1095 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1096 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1097 extern void locks_remove_posix(struct file *, fl_owner_t);
1098 extern void locks_remove_file(struct file *);
1099 extern void locks_release_private(struct file_lock *);
1100 extern void posix_test_lock(struct file *, struct file_lock *);
1101 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1102 extern int posix_unblock_lock(struct file_lock *);
1103 extern int vfs_test_lock(struct file *, struct file_lock *);
1104 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1105 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1106 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1107 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1108 extern void lease_get_mtime(struct inode *, struct timespec *time);
1109 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1110 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1111 extern int lease_modify(struct file_lock *, int, struct list_head *);
1112 struct files_struct;
1113 extern void show_fd_locks(struct seq_file *f,
1114 struct file *filp, struct files_struct *files);
1115 #else /* !CONFIG_FILE_LOCKING */
1116 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1117 struct flock __user *user)
1122 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1123 unsigned int cmd, struct flock __user *user)
1128 #if BITS_PER_LONG == 32
1129 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1130 struct flock64 __user *user)
1135 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1136 unsigned int cmd, struct flock64 __user *user)
1141 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1146 static inline int fcntl_getlease(struct file *filp)
1152 locks_free_lock_context(struct inode *inode)
1156 static inline void locks_init_lock(struct file_lock *fl)
1161 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1166 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1171 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1176 static inline void locks_remove_file(struct file *filp)
1181 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1186 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1187 struct file_lock *conflock)
1192 static inline int posix_unblock_lock(struct file_lock *waiter)
1197 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1202 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1203 struct file_lock *fl, struct file_lock *conf)
1208 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1213 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1218 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1223 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1228 static inline int generic_setlease(struct file *filp, long arg,
1229 struct file_lock **flp, void **priv)
1234 static inline int vfs_setlease(struct file *filp, long arg,
1235 struct file_lock **lease, void **priv)
1240 static inline int lease_modify(struct file_lock *fl, int arg,
1241 struct list_head *dispose)
1246 struct files_struct;
1247 static inline void show_fd_locks(struct seq_file *f,
1248 struct file *filp, struct files_struct *files) {}
1249 #endif /* !CONFIG_FILE_LOCKING */
1251 static inline struct inode *file_inode(const struct file *f)
1256 static inline struct dentry *file_dentry(const struct file *file)
1258 struct dentry *dentry = file->f_path.dentry;
1260 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1261 return dentry->d_op->d_real(dentry, file_inode(file));
1266 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1268 return locks_lock_inode_wait(file_inode(filp), fl);
1271 struct fasync_struct {
1275 struct fasync_struct *fa_next; /* singly linked list */
1276 struct file *fa_file;
1277 struct rcu_head fa_rcu;
1280 #define FASYNC_MAGIC 0x4601
1282 /* SMP safe fasync helpers: */
1283 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1284 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1285 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1286 extern struct fasync_struct *fasync_alloc(void);
1287 extern void fasync_free(struct fasync_struct *);
1289 /* can be called from interrupts */
1290 extern void kill_fasync(struct fasync_struct **, int, int);
1292 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1293 extern void f_setown(struct file *filp, unsigned long arg, int force);
1294 extern void f_delown(struct file *filp);
1295 extern pid_t f_getown(struct file *filp);
1296 extern int send_sigurg(struct fown_struct *fown);
1304 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1305 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1306 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1307 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1308 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1311 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1312 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1314 /* Possible states of 'frozen' field */
1316 SB_UNFROZEN = 0, /* FS is unfrozen */
1317 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1318 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1319 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1320 * internal threads if needed) */
1321 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1324 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1327 int frozen; /* Is sb frozen? */
1328 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1329 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1332 struct super_block {
1333 struct list_head s_list; /* Keep this first */
1334 dev_t s_dev; /* search index; _not_ kdev_t */
1335 unsigned char s_blocksize_bits;
1336 unsigned long s_blocksize;
1337 loff_t s_maxbytes; /* Max file size */
1338 struct file_system_type *s_type;
1339 const struct super_operations *s_op;
1340 const struct dquot_operations *dq_op;
1341 const struct quotactl_ops *s_qcop;
1342 const struct export_operations *s_export_op;
1343 unsigned long s_flags;
1344 unsigned long s_iflags; /* internal SB_I_* flags */
1345 unsigned long s_magic;
1346 struct dentry *s_root;
1347 struct rw_semaphore s_umount;
1350 #ifdef CONFIG_SECURITY
1353 const struct xattr_handler **s_xattr;
1355 const struct fscrypt_operations *s_cop;
1357 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1358 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1359 struct block_device *s_bdev;
1360 struct backing_dev_info *s_bdi;
1361 struct mtd_info *s_mtd;
1362 struct hlist_node s_instances;
1363 unsigned int s_quota_types; /* Bitmask of supported quota types */
1364 struct quota_info s_dquot; /* Diskquota specific options */
1366 struct sb_writers s_writers;
1368 char s_id[32]; /* Informational name */
1369 u8 s_uuid[16]; /* UUID */
1371 void *s_fs_info; /* Filesystem private info */
1372 unsigned int s_max_links;
1375 /* Granularity of c/m/atime in ns.
1376 Cannot be worse than a second */
1380 * The next field is for VFS *only*. No filesystems have any business
1381 * even looking at it. You had been warned.
1383 struct mutex s_vfs_rename_mutex; /* Kludge */
1386 * Filesystem subtype. If non-empty the filesystem type field
1387 * in /proc/mounts will be "type.subtype"
1392 * Saved mount options for lazy filesystems using
1393 * generic_show_options()
1395 char __rcu *s_options;
1396 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1399 * Saved pool identifier for cleancache (-1 means none)
1401 int cleancache_poolid;
1403 struct shrinker s_shrink; /* per-sb shrinker handle */
1405 /* Number of inodes with nlink == 0 but still referenced */
1406 atomic_long_t s_remove_count;
1408 /* Being remounted read-only */
1409 int s_readonly_remount;
1411 /* AIO completions deferred from interrupt context */
1412 struct workqueue_struct *s_dio_done_wq;
1413 struct hlist_head s_pins;
1416 * Keep the lru lists last in the structure so they always sit on their
1417 * own individual cachelines.
1419 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1420 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1421 struct rcu_head rcu;
1422 struct work_struct destroy_work;
1424 struct mutex s_sync_lock; /* sync serialisation lock */
1427 * Indicates how deep in a filesystem stack this SB is
1431 /* s_inode_list_lock protects s_inodes */
1432 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1433 struct list_head s_inodes; /* all inodes */
1436 extern struct timespec current_fs_time(struct super_block *sb);
1439 * Snapshotting support.
1442 void __sb_end_write(struct super_block *sb, int level);
1443 int __sb_start_write(struct super_block *sb, int level, bool wait);
1445 #define __sb_writers_acquired(sb, lev) \
1446 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1447 #define __sb_writers_release(sb, lev) \
1448 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1451 * sb_end_write - drop write access to a superblock
1452 * @sb: the super we wrote to
1454 * Decrement number of writers to the filesystem. Wake up possible waiters
1455 * wanting to freeze the filesystem.
1457 static inline void sb_end_write(struct super_block *sb)
1459 __sb_end_write(sb, SB_FREEZE_WRITE);
1463 * sb_end_pagefault - drop write access to a superblock from a page fault
1464 * @sb: the super we wrote to
1466 * Decrement number of processes handling write page fault to the filesystem.
1467 * Wake up possible waiters wanting to freeze the filesystem.
1469 static inline void sb_end_pagefault(struct super_block *sb)
1471 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1475 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1476 * @sb: the super we wrote to
1478 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1479 * waiters wanting to freeze the filesystem.
1481 static inline void sb_end_intwrite(struct super_block *sb)
1483 __sb_end_write(sb, SB_FREEZE_FS);
1487 * sb_start_write - get write access to a superblock
1488 * @sb: the super we write to
1490 * When a process wants to write data or metadata to a file system (i.e. dirty
1491 * a page or an inode), it should embed the operation in a sb_start_write() -
1492 * sb_end_write() pair to get exclusion against file system freezing. This
1493 * function increments number of writers preventing freezing. If the file
1494 * system is already frozen, the function waits until the file system is
1497 * Since freeze protection behaves as a lock, users have to preserve
1498 * ordering of freeze protection and other filesystem locks. Generally,
1499 * freeze protection should be the outermost lock. In particular, we have:
1502 * -> i_mutex (write path, truncate, directory ops, ...)
1503 * -> s_umount (freeze_super, thaw_super)
1505 static inline void sb_start_write(struct super_block *sb)
1507 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1510 static inline int sb_start_write_trylock(struct super_block *sb)
1512 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1516 * sb_start_pagefault - get write access to a superblock from a page fault
1517 * @sb: the super we write to
1519 * When a process starts handling write page fault, it should embed the
1520 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1521 * exclusion against file system freezing. This is needed since the page fault
1522 * is going to dirty a page. This function increments number of running page
1523 * faults preventing freezing. If the file system is already frozen, the
1524 * function waits until the file system is thawed.
1526 * Since page fault freeze protection behaves as a lock, users have to preserve
1527 * ordering of freeze protection and other filesystem locks. It is advised to
1528 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1529 * handling code implies lock dependency:
1532 * -> sb_start_pagefault
1534 static inline void sb_start_pagefault(struct super_block *sb)
1536 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1540 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1541 * @sb: the super we write to
1543 * This is the third level of protection against filesystem freezing. It is
1544 * free for use by a filesystem. The only requirement is that it must rank
1545 * below sb_start_pagefault.
1547 * For example filesystem can call sb_start_intwrite() when starting a
1548 * transaction which somewhat eases handling of freezing for internal sources
1549 * of filesystem changes (internal fs threads, discarding preallocation on file
1552 static inline void sb_start_intwrite(struct super_block *sb)
1554 __sb_start_write(sb, SB_FREEZE_FS, true);
1558 extern bool inode_owner_or_capable(const struct inode *inode);
1561 * VFS helper functions..
1563 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1564 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1565 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1566 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1567 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1568 extern int vfs_rmdir(struct inode *, struct dentry *);
1569 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1570 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1571 extern int vfs_whiteout(struct inode *, struct dentry *);
1574 * VFS file helper functions.
1576 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1579 * VFS FS_IOC_FIEMAP helper definitions.
1581 struct fiemap_extent_info {
1582 unsigned int fi_flags; /* Flags as passed from user */
1583 unsigned int fi_extents_mapped; /* Number of mapped extents */
1584 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1585 struct fiemap_extent __user *fi_extents_start; /* Start of
1586 fiemap_extent array */
1588 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1589 u64 phys, u64 len, u32 flags);
1590 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1595 * NOTE! These match bits 12..15 of stat.st_mode
1596 * (ie "(i_mode >> 12) & 15").
1598 #define DT_UNKNOWN 0
1609 * This is the "filldir" function type, used by readdir() to let
1610 * the kernel specify what kind of dirent layout it wants to have.
1611 * This allows the kernel to read directories into kernel space or
1612 * to have different dirent layouts depending on the binary type.
1615 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1618 struct dir_context {
1619 const filldir_t actor;
1623 struct block_device_operations;
1625 /* These macros are for out of kernel modules to test that
1626 * the kernel supports the unlocked_ioctl and compat_ioctl
1627 * fields in struct file_operations. */
1628 #define HAVE_COMPAT_IOCTL 1
1629 #define HAVE_UNLOCKED_IOCTL 1
1632 * These flags let !MMU mmap() govern direct device mapping vs immediate
1633 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1635 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1636 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1637 * NOMMU_MAP_READ: Can be mapped for reading
1638 * NOMMU_MAP_WRITE: Can be mapped for writing
1639 * NOMMU_MAP_EXEC: Can be mapped for execution
1641 #define NOMMU_MAP_COPY 0x00000001
1642 #define NOMMU_MAP_DIRECT 0x00000008
1643 #define NOMMU_MAP_READ VM_MAYREAD
1644 #define NOMMU_MAP_WRITE VM_MAYWRITE
1645 #define NOMMU_MAP_EXEC VM_MAYEXEC
1647 #define NOMMU_VMFLAGS \
1648 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1653 struct file_operations {
1654 struct module *owner;
1655 loff_t (*llseek) (struct file *, loff_t, int);
1656 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1657 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1658 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1659 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1660 int (*iterate) (struct file *, struct dir_context *);
1661 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1662 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1663 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1664 int (*mmap) (struct file *, struct vm_area_struct *);
1665 int (*open) (struct inode *, struct file *);
1666 int (*flush) (struct file *, fl_owner_t id);
1667 int (*release) (struct inode *, struct file *);
1668 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1669 int (*aio_fsync) (struct kiocb *, int datasync);
1670 int (*fasync) (int, struct file *, int);
1671 int (*lock) (struct file *, int, struct file_lock *);
1672 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1673 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1674 int (*check_flags)(int);
1675 int (*flock) (struct file *, int, struct file_lock *);
1676 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1677 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1678 int (*setlease)(struct file *, long, struct file_lock **, void **);
1679 long (*fallocate)(struct file *file, int mode, loff_t offset,
1681 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1683 unsigned (*mmap_capabilities)(struct file *);
1685 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1686 loff_t, size_t, unsigned int);
1687 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1689 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1693 struct inode_operations {
1694 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1695 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1696 int (*permission) (struct inode *, int);
1697 struct posix_acl * (*get_acl)(struct inode *, int);
1699 int (*readlink) (struct dentry *, char __user *,int);
1701 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1702 int (*link) (struct dentry *,struct inode *,struct dentry *);
1703 int (*unlink) (struct inode *,struct dentry *);
1704 int (*symlink) (struct inode *,struct dentry *,const char *);
1705 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1706 int (*rmdir) (struct inode *,struct dentry *);
1707 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1708 int (*rename) (struct inode *, struct dentry *,
1709 struct inode *, struct dentry *);
1710 int (*rename2) (struct inode *, struct dentry *,
1711 struct inode *, struct dentry *, unsigned int);
1712 int (*setattr) (struct dentry *, struct iattr *);
1713 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1714 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1715 ssize_t (*getxattr) (struct dentry *, struct inode *,
1716 const char *, void *, size_t);
1717 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1718 int (*removexattr) (struct dentry *, const char *);
1719 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1721 int (*update_time)(struct inode *, struct timespec *, int);
1722 int (*atomic_open)(struct inode *, struct dentry *,
1723 struct file *, unsigned open_flag,
1724 umode_t create_mode, int *opened);
1725 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1726 int (*set_acl)(struct inode *, struct posix_acl *, int);
1727 } ____cacheline_aligned;
1729 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1730 unsigned long nr_segs, unsigned long fast_segs,
1731 struct iovec *fast_pointer,
1732 struct iovec **ret_pointer);
1734 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1735 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1736 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1737 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1738 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1739 unsigned long, loff_t *, int);
1740 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1741 unsigned long, loff_t *, int);
1742 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1743 loff_t, size_t, unsigned int);
1744 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1745 struct file *file_out, loff_t pos_out, u64 len);
1746 extern int vfs_dedupe_file_range(struct file *file,
1747 struct file_dedupe_range *same);
1749 struct super_operations {
1750 struct inode *(*alloc_inode)(struct super_block *sb);
1751 void (*destroy_inode)(struct inode *);
1753 void (*dirty_inode) (struct inode *, int flags);
1754 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1755 int (*drop_inode) (struct inode *);
1756 void (*evict_inode) (struct inode *);
1757 void (*put_super) (struct super_block *);
1758 int (*sync_fs)(struct super_block *sb, int wait);
1759 int (*freeze_super) (struct super_block *);
1760 int (*freeze_fs) (struct super_block *);
1761 int (*thaw_super) (struct super_block *);
1762 int (*unfreeze_fs) (struct super_block *);
1763 int (*statfs) (struct dentry *, struct kstatfs *);
1764 int (*remount_fs) (struct super_block *, int *, char *);
1765 void (*umount_begin) (struct super_block *);
1767 int (*show_options)(struct seq_file *, struct dentry *);
1768 int (*show_devname)(struct seq_file *, struct dentry *);
1769 int (*show_path)(struct seq_file *, struct dentry *);
1770 int (*show_stats)(struct seq_file *, struct dentry *);
1772 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1773 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1774 struct dquot **(*get_dquots)(struct inode *);
1776 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1777 long (*nr_cached_objects)(struct super_block *,
1778 struct shrink_control *);
1779 long (*free_cached_objects)(struct super_block *,
1780 struct shrink_control *);
1784 * Inode flags - they have no relation to superblock flags now
1786 #define S_SYNC 1 /* Writes are synced at once */
1787 #define S_NOATIME 2 /* Do not update access times */
1788 #define S_APPEND 4 /* Append-only file */
1789 #define S_IMMUTABLE 8 /* Immutable file */
1790 #define S_DEAD 16 /* removed, but still open directory */
1791 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1792 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1793 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1794 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1795 #define S_PRIVATE 512 /* Inode is fs-internal */
1796 #define S_IMA 1024 /* Inode has an associated IMA struct */
1797 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1798 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1799 #ifdef CONFIG_FS_DAX
1800 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1802 #define S_DAX 0 /* Make all the DAX code disappear */
1806 * Note that nosuid etc flags are inode-specific: setting some file-system
1807 * flags just means all the inodes inherit those flags by default. It might be
1808 * possible to override it selectively if you really wanted to with some
1809 * ioctl() that is not currently implemented.
1811 * Exception: MS_RDONLY is always applied to the entire file system.
1813 * Unfortunately, it is possible to change a filesystems flags with it mounted
1814 * with files in use. This means that all of the inodes will not have their
1815 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1816 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1818 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1820 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1821 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1822 ((inode)->i_flags & S_SYNC))
1823 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1824 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1825 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1826 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1827 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1829 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1830 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1831 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1832 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1834 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1835 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1836 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1837 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1838 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1839 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1840 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1841 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1843 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1844 (inode)->i_rdev == WHITEOUT_DEV)
1847 * Inode state bits. Protected by inode->i_lock
1849 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1850 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1852 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1853 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1854 * various stages of removing an inode.
1856 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1858 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1859 * fdatasync(). i_atime is the usual cause.
1860 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1861 * these changes separately from I_DIRTY_SYNC so that we
1862 * don't have to write inode on fdatasync() when only
1863 * mtime has changed in it.
1864 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1865 * I_NEW Serves as both a mutex and completion notification.
1866 * New inodes set I_NEW. If two processes both create
1867 * the same inode, one of them will release its inode and
1868 * wait for I_NEW to be released before returning.
1869 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1870 * also cause waiting on I_NEW, without I_NEW actually
1871 * being set. find_inode() uses this to prevent returning
1872 * nearly-dead inodes.
1873 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1874 * is zero. I_FREEING must be set when I_WILL_FREE is
1876 * I_FREEING Set when inode is about to be freed but still has dirty
1877 * pages or buffers attached or the inode itself is still
1879 * I_CLEAR Added by clear_inode(). In this state the inode is
1880 * clean and can be destroyed. Inode keeps I_FREEING.
1882 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1883 * prohibited for many purposes. iget() must wait for
1884 * the inode to be completely released, then create it
1885 * anew. Other functions will just ignore such inodes,
1886 * if appropriate. I_NEW is used for waiting.
1888 * I_SYNC Writeback of inode is running. The bit is set during
1889 * data writeback, and cleared with a wakeup on the bit
1890 * address once it is done. The bit is also used to pin
1891 * the inode in memory for flusher thread.
1893 * I_REFERENCED Marks the inode as recently references on the LRU list.
1895 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1897 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1898 * synchronize competing switching instances and to tell
1899 * wb stat updates to grab mapping->tree_lock. See
1900 * inode_switch_wb_work_fn() for details.
1902 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1904 #define I_DIRTY_SYNC (1 << 0)
1905 #define I_DIRTY_DATASYNC (1 << 1)
1906 #define I_DIRTY_PAGES (1 << 2)
1908 #define I_NEW (1 << __I_NEW)
1909 #define I_WILL_FREE (1 << 4)
1910 #define I_FREEING (1 << 5)
1911 #define I_CLEAR (1 << 6)
1913 #define I_SYNC (1 << __I_SYNC)
1914 #define I_REFERENCED (1 << 8)
1915 #define __I_DIO_WAKEUP 9
1916 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1917 #define I_LINKABLE (1 << 10)
1918 #define I_DIRTY_TIME (1 << 11)
1919 #define __I_DIRTY_TIME_EXPIRED 12
1920 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1921 #define I_WB_SWITCH (1 << 13)
1923 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1924 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1926 extern void __mark_inode_dirty(struct inode *, int);
1927 static inline void mark_inode_dirty(struct inode *inode)
1929 __mark_inode_dirty(inode, I_DIRTY);
1932 static inline void mark_inode_dirty_sync(struct inode *inode)
1934 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1937 extern void inc_nlink(struct inode *inode);
1938 extern void drop_nlink(struct inode *inode);
1939 extern void clear_nlink(struct inode *inode);
1940 extern void set_nlink(struct inode *inode, unsigned int nlink);
1942 static inline void inode_inc_link_count(struct inode *inode)
1945 mark_inode_dirty(inode);
1948 static inline void inode_dec_link_count(struct inode *inode)
1951 mark_inode_dirty(inode);
1955 * inode_inc_iversion - increments i_version
1956 * @inode: inode that need to be updated
1958 * Every time the inode is modified, the i_version field will be incremented.
1959 * The filesystem has to be mounted with i_version flag
1962 static inline void inode_inc_iversion(struct inode *inode)
1964 spin_lock(&inode->i_lock);
1966 spin_unlock(&inode->i_lock);
1969 enum file_time_flags {
1976 extern bool atime_needs_update(const struct path *, struct inode *);
1977 extern void touch_atime(const struct path *);
1978 static inline void file_accessed(struct file *file)
1980 if (!(file->f_flags & O_NOATIME))
1981 touch_atime(&file->f_path);
1984 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1985 int sync_inode_metadata(struct inode *inode, int wait);
1987 struct file_system_type {
1990 #define FS_REQUIRES_DEV 1
1991 #define FS_BINARY_MOUNTDATA 2
1992 #define FS_HAS_SUBTYPE 4
1993 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1994 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1995 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1996 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1997 struct dentry *(*mount) (struct file_system_type *, int,
1998 const char *, void *);
1999 void (*kill_sb) (struct super_block *);
2000 struct module *owner;
2001 struct file_system_type * next;
2002 struct hlist_head fs_supers;
2004 struct lock_class_key s_lock_key;
2005 struct lock_class_key s_umount_key;
2006 struct lock_class_key s_vfs_rename_key;
2007 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2009 struct lock_class_key i_lock_key;
2010 struct lock_class_key i_mutex_key;
2011 struct lock_class_key i_mutex_dir_key;
2014 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2016 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
2017 void *data, int (*fill_super)(struct super_block *, void *, int));
2018 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2019 int flags, const char *dev_name, void *data,
2020 int (*fill_super)(struct super_block *, void *, int));
2021 extern struct dentry *mount_single(struct file_system_type *fs_type,
2022 int flags, void *data,
2023 int (*fill_super)(struct super_block *, void *, int));
2024 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2025 int flags, void *data,
2026 int (*fill_super)(struct super_block *, void *, int));
2027 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2028 void generic_shutdown_super(struct super_block *sb);
2029 void kill_block_super(struct super_block *sb);
2030 void kill_anon_super(struct super_block *sb);
2031 void kill_litter_super(struct super_block *sb);
2032 void deactivate_super(struct super_block *sb);
2033 void deactivate_locked_super(struct super_block *sb);
2034 int set_anon_super(struct super_block *s, void *data);
2035 int get_anon_bdev(dev_t *);
2036 void free_anon_bdev(dev_t);
2037 struct super_block *sget(struct file_system_type *type,
2038 int (*test)(struct super_block *,void *),
2039 int (*set)(struct super_block *,void *),
2040 int flags, void *data);
2041 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2042 const struct super_operations *ops,
2043 const struct dentry_operations *dops,
2046 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2047 #define fops_get(fops) \
2048 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2049 #define fops_put(fops) \
2050 do { if (fops) module_put((fops)->owner); } while(0)
2052 * This one is to be used *ONLY* from ->open() instances.
2053 * fops must be non-NULL, pinned down *and* module dependencies
2054 * should be sufficient to pin the caller down as well.
2056 #define replace_fops(f, fops) \
2058 struct file *__file = (f); \
2059 fops_put(__file->f_op); \
2060 BUG_ON(!(__file->f_op = (fops))); \
2063 extern int register_filesystem(struct file_system_type *);
2064 extern int unregister_filesystem(struct file_system_type *);
2065 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2066 #define kern_mount(type) kern_mount_data(type, NULL)
2067 extern void kern_unmount(struct vfsmount *mnt);
2068 extern int may_umount_tree(struct vfsmount *);
2069 extern int may_umount(struct vfsmount *);
2070 extern long do_mount(const char *, const char __user *,
2071 const char *, unsigned long, void *);
2072 extern struct vfsmount *collect_mounts(struct path *);
2073 extern void drop_collected_mounts(struct vfsmount *);
2074 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2076 extern int vfs_statfs(struct path *, struct kstatfs *);
2077 extern int user_statfs(const char __user *, struct kstatfs *);
2078 extern int fd_statfs(int, struct kstatfs *);
2079 extern int vfs_ustat(dev_t, struct kstatfs *);
2080 extern int freeze_super(struct super_block *super);
2081 extern int thaw_super(struct super_block *super);
2082 extern bool our_mnt(struct vfsmount *mnt);
2084 extern int current_umask(void);
2086 extern void ihold(struct inode * inode);
2087 extern void iput(struct inode *);
2088 extern int generic_update_time(struct inode *, struct timespec *, int);
2091 extern struct kobject *fs_kobj;
2093 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2095 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2096 extern int locks_mandatory_locked(struct file *);
2097 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2100 * Candidates for mandatory locking have the setgid bit set
2101 * but no group execute bit - an otherwise meaningless combination.
2104 static inline int __mandatory_lock(struct inode *ino)
2106 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2110 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2111 * otherwise these will be advisory locks
2114 static inline int mandatory_lock(struct inode *ino)
2116 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2119 static inline int locks_verify_locked(struct file *file)
2121 if (mandatory_lock(file_inode(file)))
2122 return locks_mandatory_locked(file);
2126 static inline int locks_verify_truncate(struct inode *inode,
2130 if (!inode->i_flctx || !mandatory_lock(inode))
2133 if (size < inode->i_size) {
2134 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2137 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2142 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2144 static inline int locks_mandatory_locked(struct file *file)
2149 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2150 loff_t start, loff_t end, unsigned char type)
2155 static inline int __mandatory_lock(struct inode *inode)
2160 static inline int mandatory_lock(struct inode *inode)
2165 static inline int locks_verify_locked(struct file *file)
2170 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2176 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2179 #ifdef CONFIG_FILE_LOCKING
2180 static inline int break_lease(struct inode *inode, unsigned int mode)
2183 * Since this check is lockless, we must ensure that any refcounts
2184 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2185 * could end up racing with tasks trying to set a new lease on this
2189 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2190 return __break_lease(inode, mode, FL_LEASE);
2194 static inline int break_deleg(struct inode *inode, unsigned int mode)
2197 * Since this check is lockless, we must ensure that any refcounts
2198 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2199 * could end up racing with tasks trying to set a new lease on this
2203 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2204 return __break_lease(inode, mode, FL_DELEG);
2208 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2212 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2213 if (ret == -EWOULDBLOCK && delegated_inode) {
2214 *delegated_inode = inode;
2220 static inline int break_deleg_wait(struct inode **delegated_inode)
2224 ret = break_deleg(*delegated_inode, O_WRONLY);
2225 iput(*delegated_inode);
2226 *delegated_inode = NULL;
2230 static inline int break_layout(struct inode *inode, bool wait)
2233 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2234 return __break_lease(inode,
2235 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2240 #else /* !CONFIG_FILE_LOCKING */
2241 static inline int break_lease(struct inode *inode, unsigned int mode)
2246 static inline int break_deleg(struct inode *inode, unsigned int mode)
2251 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2256 static inline int break_deleg_wait(struct inode **delegated_inode)
2262 static inline int break_layout(struct inode *inode, bool wait)
2267 #endif /* CONFIG_FILE_LOCKING */
2272 const char *name; /* pointer to actual string */
2273 const __user char *uptr; /* original userland pointer */
2274 struct audit_names *aname;
2279 extern long vfs_truncate(struct path *, loff_t);
2280 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2282 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2284 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2286 extern struct file *file_open_name(struct filename *, int, umode_t);
2287 extern struct file *filp_open(const char *, int, umode_t);
2288 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2289 const char *, int, umode_t);
2290 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2291 extern int filp_close(struct file *, fl_owner_t id);
2293 extern struct filename *getname_flags(const char __user *, int, int *);
2294 extern struct filename *getname(const char __user *);
2295 extern struct filename *getname_kernel(const char *);
2296 extern void putname(struct filename *name);
2302 extern int finish_open(struct file *file, struct dentry *dentry,
2303 int (*open)(struct inode *, struct file *),
2305 extern int finish_no_open(struct file *file, struct dentry *dentry);
2309 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2312 extern void __init vfs_caches_init_early(void);
2313 extern void __init vfs_caches_init(void);
2315 extern struct kmem_cache *names_cachep;
2317 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2318 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2321 extern int register_blkdev(unsigned int, const char *);
2322 extern void unregister_blkdev(unsigned int, const char *);
2323 extern struct block_device *bdget(dev_t);
2324 extern struct block_device *bdgrab(struct block_device *bdev);
2325 extern void bd_set_size(struct block_device *, loff_t size);
2326 extern void bd_forget(struct inode *inode);
2327 extern void bdput(struct block_device *);
2328 extern void invalidate_bdev(struct block_device *);
2329 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2330 extern int sync_blockdev(struct block_device *bdev);
2331 extern void kill_bdev(struct block_device *);
2332 extern struct super_block *freeze_bdev(struct block_device *);
2333 extern void emergency_thaw_all(void);
2334 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2335 extern int fsync_bdev(struct block_device *);
2336 #ifdef CONFIG_FS_DAX
2337 extern bool blkdev_dax_capable(struct block_device *bdev);
2339 static inline bool blkdev_dax_capable(struct block_device *bdev)
2345 extern struct super_block *blockdev_superblock;
2347 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2349 return sb == blockdev_superblock;
2352 static inline void bd_forget(struct inode *inode) {}
2353 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2354 static inline void kill_bdev(struct block_device *bdev) {}
2355 static inline void invalidate_bdev(struct block_device *bdev) {}
2357 static inline struct super_block *freeze_bdev(struct block_device *sb)
2362 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2367 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2371 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2376 extern int sync_filesystem(struct super_block *);
2377 extern const struct file_operations def_blk_fops;
2378 extern const struct file_operations def_chr_fops;
2380 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2381 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2382 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2383 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2384 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2386 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2388 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2389 extern int __blkdev_reread_part(struct block_device *bdev);
2390 extern int blkdev_reread_part(struct block_device *bdev);
2393 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2394 extern void bd_unlink_disk_holder(struct block_device *bdev,
2395 struct gendisk *disk);
2397 static inline int bd_link_disk_holder(struct block_device *bdev,
2398 struct gendisk *disk)
2402 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2403 struct gendisk *disk)
2410 #define CHRDEV_MAJOR_HASH_SIZE 255
2411 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2412 extern int register_chrdev_region(dev_t, unsigned, const char *);
2413 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2414 unsigned int count, const char *name,
2415 const struct file_operations *fops);
2416 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2417 unsigned int count, const char *name);
2418 extern void unregister_chrdev_region(dev_t, unsigned);
2419 extern void chrdev_show(struct seq_file *,off_t);
2421 static inline int register_chrdev(unsigned int major, const char *name,
2422 const struct file_operations *fops)
2424 return __register_chrdev(major, 0, 256, name, fops);
2427 static inline void unregister_chrdev(unsigned int major, const char *name)
2429 __unregister_chrdev(major, 0, 256, name);
2432 /* fs/block_dev.c */
2433 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2434 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2437 #define BLKDEV_MAJOR_HASH_SIZE 255
2438 extern const char *__bdevname(dev_t, char *buffer);
2439 extern const char *bdevname(struct block_device *bdev, char *buffer);
2440 extern struct block_device *lookup_bdev(const char *);
2441 extern void blkdev_show(struct seq_file *,off_t);
2444 #define BLKDEV_MAJOR_HASH_SIZE 0
2447 extern void init_special_inode(struct inode *, umode_t, dev_t);
2449 /* Invalid inode operations -- fs/bad_inode.c */
2450 extern void make_bad_inode(struct inode *);
2451 extern bool is_bad_inode(struct inode *);
2455 * return READ, READA, or WRITE
2457 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2460 * return data direction, READ or WRITE
2462 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2464 extern void check_disk_size_change(struct gendisk *disk,
2465 struct block_device *bdev);
2466 extern int revalidate_disk(struct gendisk *);
2467 extern int check_disk_change(struct block_device *);
2468 extern int __invalidate_device(struct block_device *, bool);
2469 extern int invalidate_partition(struct gendisk *, int);
2471 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2472 pgoff_t start, pgoff_t end);
2474 static inline void invalidate_remote_inode(struct inode *inode)
2476 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2477 S_ISLNK(inode->i_mode))
2478 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2480 extern int invalidate_inode_pages2(struct address_space *mapping);
2481 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2482 pgoff_t start, pgoff_t end);
2483 extern int write_inode_now(struct inode *, int);
2484 extern int filemap_fdatawrite(struct address_space *);
2485 extern int filemap_flush(struct address_space *);
2486 extern int filemap_fdatawait(struct address_space *);
2487 extern void filemap_fdatawait_keep_errors(struct address_space *);
2488 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2490 extern int filemap_write_and_wait(struct address_space *mapping);
2491 extern int filemap_write_and_wait_range(struct address_space *mapping,
2492 loff_t lstart, loff_t lend);
2493 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2494 loff_t start, loff_t end, int sync_mode);
2495 extern int filemap_fdatawrite_range(struct address_space *mapping,
2496 loff_t start, loff_t end);
2498 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2500 extern int vfs_fsync(struct file *file, int datasync);
2501 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2503 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2505 return vfs_fsync_range(file, pos, pos + count - 1,
2506 (file->f_flags & __O_SYNC) ? 0 : 1);
2508 extern void emergency_sync(void);
2509 extern void emergency_remount(void);
2511 extern sector_t bmap(struct inode *, sector_t);
2513 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2514 extern int inode_permission(struct inode *, int);
2515 extern int __inode_permission(struct inode *, int);
2516 extern int generic_permission(struct inode *, int);
2517 extern int __check_sticky(struct inode *dir, struct inode *inode);
2519 static inline bool execute_ok(struct inode *inode)
2521 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2524 static inline void file_start_write(struct file *file)
2526 if (!S_ISREG(file_inode(file)->i_mode))
2528 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2531 static inline bool file_start_write_trylock(struct file *file)
2533 if (!S_ISREG(file_inode(file)->i_mode))
2535 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2538 static inline void file_end_write(struct file *file)
2540 if (!S_ISREG(file_inode(file)->i_mode))
2542 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2546 * get_write_access() gets write permission for a file.
2547 * put_write_access() releases this write permission.
2548 * This is used for regular files.
2549 * We cannot support write (and maybe mmap read-write shared) accesses and
2550 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2551 * can have the following values:
2552 * 0: no writers, no VM_DENYWRITE mappings
2553 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2554 * > 0: (i_writecount) users are writing to the file.
2556 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2557 * except for the cases where we don't hold i_writecount yet. Then we need to
2558 * use {get,deny}_write_access() - these functions check the sign and refuse
2559 * to do the change if sign is wrong.
2561 static inline int get_write_access(struct inode *inode)
2563 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2565 static inline int deny_write_access(struct file *file)
2567 struct inode *inode = file_inode(file);
2568 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2570 static inline void put_write_access(struct inode * inode)
2572 atomic_dec(&inode->i_writecount);
2574 static inline void allow_write_access(struct file *file)
2577 atomic_inc(&file_inode(file)->i_writecount);
2579 static inline bool inode_is_open_for_write(const struct inode *inode)
2581 return atomic_read(&inode->i_writecount) > 0;
2585 static inline void i_readcount_dec(struct inode *inode)
2587 BUG_ON(!atomic_read(&inode->i_readcount));
2588 atomic_dec(&inode->i_readcount);
2590 static inline void i_readcount_inc(struct inode *inode)
2592 atomic_inc(&inode->i_readcount);
2595 static inline void i_readcount_dec(struct inode *inode)
2599 static inline void i_readcount_inc(struct inode *inode)
2604 extern int do_pipe_flags(int *, int);
2606 enum kernel_read_file_id {
2607 READING_FIRMWARE = 1,
2609 READING_KEXEC_IMAGE,
2610 READING_KEXEC_INITRAMFS,
2615 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2616 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2617 enum kernel_read_file_id);
2618 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2619 enum kernel_read_file_id);
2620 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2621 enum kernel_read_file_id);
2622 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2623 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2624 extern struct file * open_exec(const char *);
2626 /* fs/dcache.c -- generic fs support functions */
2627 extern bool is_subdir(struct dentry *, struct dentry *);
2628 extern bool path_is_under(struct path *, struct path *);
2630 extern char *file_path(struct file *, char *, int);
2632 #include <linux/err.h>
2634 /* needed for stackable file system support */
2635 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2637 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2639 extern int inode_init_always(struct super_block *, struct inode *);
2640 extern void inode_init_once(struct inode *);
2641 extern void address_space_init_once(struct address_space *mapping);
2642 extern struct inode * igrab(struct inode *);
2643 extern ino_t iunique(struct super_block *, ino_t);
2644 extern int inode_needs_sync(struct inode *inode);
2645 extern int generic_delete_inode(struct inode *inode);
2646 static inline int generic_drop_inode(struct inode *inode)
2648 return !inode->i_nlink || inode_unhashed(inode);
2651 extern struct inode *ilookup5_nowait(struct super_block *sb,
2652 unsigned long hashval, int (*test)(struct inode *, void *),
2654 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2655 int (*test)(struct inode *, void *), void *data);
2656 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2658 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2659 extern struct inode * iget_locked(struct super_block *, unsigned long);
2660 extern struct inode *find_inode_nowait(struct super_block *,
2662 int (*match)(struct inode *,
2663 unsigned long, void *),
2665 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2666 extern int insert_inode_locked(struct inode *);
2667 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2668 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2670 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2672 extern void unlock_new_inode(struct inode *);
2673 extern unsigned int get_next_ino(void);
2675 extern void __iget(struct inode * inode);
2676 extern void iget_failed(struct inode *);
2677 extern void clear_inode(struct inode *);
2678 extern void __destroy_inode(struct inode *);
2679 extern struct inode *new_inode_pseudo(struct super_block *sb);
2680 extern struct inode *new_inode(struct super_block *sb);
2681 extern void free_inode_nonrcu(struct inode *inode);
2682 extern int should_remove_suid(struct dentry *);
2683 extern int file_remove_privs(struct file *);
2684 extern int dentry_needs_remove_privs(struct dentry *dentry);
2685 static inline int file_needs_remove_privs(struct file *file)
2687 return dentry_needs_remove_privs(file->f_path.dentry);
2690 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2691 static inline void insert_inode_hash(struct inode *inode)
2693 __insert_inode_hash(inode, inode->i_ino);
2696 extern void __remove_inode_hash(struct inode *);
2697 static inline void remove_inode_hash(struct inode *inode)
2699 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2700 __remove_inode_hash(inode);
2703 extern void inode_sb_list_add(struct inode *inode);
2706 extern blk_qc_t submit_bio(int, struct bio *);
2707 extern int bdev_read_only(struct block_device *);
2709 extern int set_blocksize(struct block_device *, int);
2710 extern int sb_set_blocksize(struct super_block *, int);
2711 extern int sb_min_blocksize(struct super_block *, int);
2713 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2714 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2715 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2716 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2717 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2718 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2719 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2720 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2722 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2723 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2725 /* fs/block_dev.c */
2726 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2727 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2728 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2730 extern void block_sync_page(struct page *page);
2733 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2734 struct pipe_inode_info *, size_t, unsigned int);
2735 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2736 struct pipe_inode_info *, size_t, unsigned int);
2737 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2738 struct file *, loff_t *, size_t, unsigned int);
2739 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2740 struct file *out, loff_t *, size_t len, unsigned int flags);
2741 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2742 loff_t *opos, size_t len, unsigned int flags);
2746 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2747 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2748 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2749 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2750 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2751 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2752 int whence, loff_t maxsize, loff_t eof);
2753 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2754 int whence, loff_t size);
2755 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2756 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2757 extern int generic_file_open(struct inode * inode, struct file * filp);
2758 extern int nonseekable_open(struct inode * inode, struct file * filp);
2761 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2762 loff_t file_offset);
2765 /* need locking between buffered and direct access */
2768 /* filesystem does not support filling holes */
2769 DIO_SKIP_HOLES = 0x02,
2771 /* filesystem can handle aio writes beyond i_size */
2772 DIO_ASYNC_EXTEND = 0x04,
2774 /* inode/fs/bdev does not need truncate protection */
2775 DIO_SKIP_DIO_COUNT = 0x08,
2778 void dio_end_io(struct bio *bio, int error);
2780 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2781 struct block_device *bdev, struct iov_iter *iter,
2782 loff_t offset, get_block_t get_block,
2783 dio_iodone_t end_io, dio_submit_t submit_io,
2786 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2787 struct inode *inode,
2788 struct iov_iter *iter, loff_t offset,
2789 get_block_t get_block)
2791 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2792 offset, get_block, NULL, NULL,
2793 DIO_LOCKING | DIO_SKIP_HOLES);
2797 void inode_dio_wait(struct inode *inode);
2800 * inode_dio_begin - signal start of a direct I/O requests
2801 * @inode: inode the direct I/O happens on
2803 * This is called once we've finished processing a direct I/O request,
2804 * and is used to wake up callers waiting for direct I/O to be quiesced.
2806 static inline void inode_dio_begin(struct inode *inode)
2808 atomic_inc(&inode->i_dio_count);
2812 * inode_dio_end - signal finish of a direct I/O requests
2813 * @inode: inode the direct I/O happens on
2815 * This is called once we've finished processing a direct I/O request,
2816 * and is used to wake up callers waiting for direct I/O to be quiesced.
2818 static inline void inode_dio_end(struct inode *inode)
2820 if (atomic_dec_and_test(&inode->i_dio_count))
2821 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2824 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2827 extern const struct file_operations generic_ro_fops;
2829 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2831 extern int readlink_copy(char __user *, int, const char *);
2832 extern int page_readlink(struct dentry *, char __user *, int);
2833 extern const char *page_get_link(struct dentry *, struct inode *,
2834 struct delayed_call *);
2835 extern void page_put_link(void *);
2836 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2838 extern int page_symlink(struct inode *inode, const char *symname, int len);
2839 extern const struct inode_operations page_symlink_inode_operations;
2840 extern void kfree_link(void *);
2841 extern int generic_readlink(struct dentry *, char __user *, int);
2842 extern void generic_fillattr(struct inode *, struct kstat *);
2843 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2844 extern int vfs_getattr(struct path *, struct kstat *);
2845 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2846 void inode_add_bytes(struct inode *inode, loff_t bytes);
2847 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2848 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2849 loff_t inode_get_bytes(struct inode *inode);
2850 void inode_set_bytes(struct inode *inode, loff_t bytes);
2851 const char *simple_get_link(struct dentry *, struct inode *,
2852 struct delayed_call *);
2853 extern const struct inode_operations simple_symlink_inode_operations;
2855 extern int iterate_dir(struct file *, struct dir_context *);
2857 extern int vfs_stat(const char __user *, struct kstat *);
2858 extern int vfs_lstat(const char __user *, struct kstat *);
2859 extern int vfs_fstat(unsigned int, struct kstat *);
2860 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2862 extern int __generic_block_fiemap(struct inode *inode,
2863 struct fiemap_extent_info *fieinfo,
2864 loff_t start, loff_t len,
2865 get_block_t *get_block);
2866 extern int generic_block_fiemap(struct inode *inode,
2867 struct fiemap_extent_info *fieinfo, u64 start,
2868 u64 len, get_block_t *get_block);
2870 extern void get_filesystem(struct file_system_type *fs);
2871 extern void put_filesystem(struct file_system_type *fs);
2872 extern struct file_system_type *get_fs_type(const char *name);
2873 extern struct super_block *get_super(struct block_device *);
2874 extern struct super_block *get_super_thawed(struct block_device *);
2875 extern struct super_block *get_active_super(struct block_device *bdev);
2876 extern void drop_super(struct super_block *sb);
2877 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2878 extern void iterate_supers_type(struct file_system_type *,
2879 void (*)(struct super_block *, void *), void *);
2881 extern int dcache_dir_open(struct inode *, struct file *);
2882 extern int dcache_dir_close(struct inode *, struct file *);
2883 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2884 extern int dcache_readdir(struct file *, struct dir_context *);
2885 extern int simple_setattr(struct dentry *, struct iattr *);
2886 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2887 extern int simple_statfs(struct dentry *, struct kstatfs *);
2888 extern int simple_open(struct inode *inode, struct file *file);
2889 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2890 extern int simple_unlink(struct inode *, struct dentry *);
2891 extern int simple_rmdir(struct inode *, struct dentry *);
2892 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2893 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2894 extern int simple_empty(struct dentry *);
2895 extern int simple_readpage(struct file *file, struct page *page);
2896 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2897 loff_t pos, unsigned len, unsigned flags,
2898 struct page **pagep, void **fsdata);
2899 extern int simple_write_end(struct file *file, struct address_space *mapping,
2900 loff_t pos, unsigned len, unsigned copied,
2901 struct page *page, void *fsdata);
2902 extern int always_delete_dentry(const struct dentry *);
2903 extern struct inode *alloc_anon_inode(struct super_block *);
2904 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2905 extern const struct dentry_operations simple_dentry_operations;
2907 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2908 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2909 extern const struct file_operations simple_dir_operations;
2910 extern const struct inode_operations simple_dir_inode_operations;
2911 extern void make_empty_dir_inode(struct inode *inode);
2912 extern bool is_empty_dir_inode(struct inode *inode);
2913 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2914 struct dentry *d_alloc_name(struct dentry *, const char *);
2915 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2916 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2917 extern void simple_release_fs(struct vfsmount **mount, int *count);
2919 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2920 loff_t *ppos, const void *from, size_t available);
2921 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2922 const void __user *from, size_t count);
2924 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2925 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2927 extern int generic_check_addressable(unsigned, u64);
2929 #ifdef CONFIG_MIGRATION
2930 extern int buffer_migrate_page(struct address_space *,
2931 struct page *, struct page *,
2934 #define buffer_migrate_page NULL
2937 extern int inode_change_ok(const struct inode *, struct iattr *);
2938 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2939 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2941 extern int file_update_time(struct file *file);
2943 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2944 extern void save_mount_options(struct super_block *sb, char *options);
2945 extern void replace_mount_options(struct super_block *sb, char *options);
2947 static inline bool io_is_direct(struct file *filp)
2949 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
2952 static inline int iocb_flags(struct file *file)
2955 if (file->f_flags & O_APPEND)
2957 if (io_is_direct(file))
2962 static inline ino_t parent_ino(struct dentry *dentry)
2967 * Don't strictly need d_lock here? If the parent ino could change
2968 * then surely we'd have a deeper race in the caller?
2970 spin_lock(&dentry->d_lock);
2971 res = dentry->d_parent->d_inode->i_ino;
2972 spin_unlock(&dentry->d_lock);
2976 /* Transaction based IO helpers */
2979 * An argresp is stored in an allocated page and holds the
2980 * size of the argument or response, along with its content
2982 struct simple_transaction_argresp {
2987 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2989 char *simple_transaction_get(struct file *file, const char __user *buf,
2991 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2992 size_t size, loff_t *pos);
2993 int simple_transaction_release(struct inode *inode, struct file *file);
2995 void simple_transaction_set(struct file *file, size_t n);
2998 * simple attribute files
3000 * These attributes behave similar to those in sysfs:
3002 * Writing to an attribute immediately sets a value, an open file can be
3003 * written to multiple times.
3005 * Reading from an attribute creates a buffer from the value that might get
3006 * read with multiple read calls. When the attribute has been read
3007 * completely, no further read calls are possible until the file is opened
3010 * All attributes contain a text representation of a numeric value
3011 * that are accessed with the get() and set() functions.
3013 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3014 static int __fops ## _open(struct inode *inode, struct file *file) \
3016 __simple_attr_check_format(__fmt, 0ull); \
3017 return simple_attr_open(inode, file, __get, __set, __fmt); \
3019 static const struct file_operations __fops = { \
3020 .owner = THIS_MODULE, \
3021 .open = __fops ## _open, \
3022 .release = simple_attr_release, \
3023 .read = simple_attr_read, \
3024 .write = simple_attr_write, \
3025 .llseek = generic_file_llseek, \
3028 static inline __printf(1, 2)
3029 void __simple_attr_check_format(const char *fmt, ...)
3031 /* don't do anything, just let the compiler check the arguments; */
3034 int simple_attr_open(struct inode *inode, struct file *file,
3035 int (*get)(void *, u64 *), int (*set)(void *, u64),
3037 int simple_attr_release(struct inode *inode, struct file *file);
3038 ssize_t simple_attr_read(struct file *file, char __user *buf,
3039 size_t len, loff_t *ppos);
3040 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3041 size_t len, loff_t *ppos);
3044 int proc_nr_files(struct ctl_table *table, int write,
3045 void __user *buffer, size_t *lenp, loff_t *ppos);
3046 int proc_nr_dentry(struct ctl_table *table, int write,
3047 void __user *buffer, size_t *lenp, loff_t *ppos);
3048 int proc_nr_inodes(struct ctl_table *table, int write,
3049 void __user *buffer, size_t *lenp, loff_t *ppos);
3050 int __init get_filesystem_list(char *buf);
3052 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3053 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3055 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3056 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3057 (flag & __FMODE_NONOTIFY)))
3059 static inline bool is_sxid(umode_t mode)
3061 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3064 static inline int check_sticky(struct inode *dir, struct inode *inode)
3066 if (!(dir->i_mode & S_ISVTX))
3069 return __check_sticky(dir, inode);
3072 static inline void inode_has_no_xattr(struct inode *inode)
3074 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3075 inode->i_flags |= S_NOSEC;
3078 static inline bool is_root_inode(struct inode *inode)
3080 return inode == inode->i_sb->s_root->d_inode;
3083 static inline bool dir_emit(struct dir_context *ctx,
3084 const char *name, int namelen,
3085 u64 ino, unsigned type)
3087 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3089 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3091 return ctx->actor(ctx, ".", 1, ctx->pos,
3092 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3094 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3096 return ctx->actor(ctx, "..", 2, ctx->pos,
3097 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3099 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3101 if (ctx->pos == 0) {
3102 if (!dir_emit_dot(file, ctx))
3106 if (ctx->pos == 1) {
3107 if (!dir_emit_dotdot(file, ctx))
3113 static inline bool dir_relax(struct inode *inode)
3115 inode_unlock(inode);
3117 return !IS_DEADDIR(inode);
3120 extern bool path_noexec(const struct path *path);
3121 extern void inode_nohighmem(struct inode *inode);
3123 #endif /* _LINUX_FS_H */