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/mm_types.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
33 #include <linux/workqueue.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/delayed_call.h>
37 #include <asm/byteorder.h>
38 #include <uapi/linux/fs.h>
40 struct backing_dev_info;
42 struct export_operations;
47 struct pipe_inode_info;
48 struct poll_table_struct;
50 struct vm_area_struct;
53 struct swap_info_struct;
55 struct workqueue_struct;
58 struct fscrypt_operations;
60 extern void __init inode_init(void);
61 extern void __init inode_init_early(void);
62 extern void __init files_init(void);
63 extern void __init files_maxfiles_init(void);
65 extern struct files_stat_struct files_stat;
66 extern unsigned long get_max_files(void);
67 extern unsigned int sysctl_nr_open;
68 extern struct inodes_stat_t inodes_stat;
69 extern int leases_enable, lease_break_time;
70 extern int sysctl_protected_symlinks;
71 extern int sysctl_protected_hardlinks;
72 extern int sysctl_protected_fifos;
73 extern int sysctl_protected_regular;
76 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
77 struct buffer_head *bh_result, int create);
78 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
79 ssize_t bytes, void *private);
81 #define MAY_EXEC 0x00000001
82 #define MAY_WRITE 0x00000002
83 #define MAY_READ 0x00000004
84 #define MAY_APPEND 0x00000008
85 #define MAY_ACCESS 0x00000010
86 #define MAY_OPEN 0x00000020
87 #define MAY_CHDIR 0x00000040
88 /* called from RCU mode, don't block */
89 #define MAY_NOT_BLOCK 0x00000080
92 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
93 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
96 /* file is open for reading */
97 #define FMODE_READ ((__force fmode_t)0x1)
98 /* file is open for writing */
99 #define FMODE_WRITE ((__force fmode_t)0x2)
100 /* file is seekable */
101 #define FMODE_LSEEK ((__force fmode_t)0x4)
102 /* file can be accessed using pread */
103 #define FMODE_PREAD ((__force fmode_t)0x8)
104 /* file can be accessed using pwrite */
105 #define FMODE_PWRITE ((__force fmode_t)0x10)
106 /* File is opened for execution with sys_execve / sys_uselib */
107 #define FMODE_EXEC ((__force fmode_t)0x20)
108 /* File is opened with O_NDELAY (only set for block devices) */
109 #define FMODE_NDELAY ((__force fmode_t)0x40)
110 /* File is opened with O_EXCL (only set for block devices) */
111 #define FMODE_EXCL ((__force fmode_t)0x80)
112 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
113 (specialy hack for floppy.c) */
114 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
115 /* 32bit hashes as llseek() offset (for directories) */
116 #define FMODE_32BITHASH ((__force fmode_t)0x200)
117 /* 64bit hashes as llseek() offset (for directories) */
118 #define FMODE_64BITHASH ((__force fmode_t)0x400)
121 * Don't update ctime and mtime.
123 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
124 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
126 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
128 /* Expect random access pattern */
129 #define FMODE_RANDOM ((__force fmode_t)0x1000)
131 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
132 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
134 /* File is opened with O_PATH; almost nothing can be done with it */
135 #define FMODE_PATH ((__force fmode_t)0x4000)
137 /* File needs atomic accesses to f_pos */
138 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
139 /* Write access to underlying fs */
140 #define FMODE_WRITER ((__force fmode_t)0x10000)
141 /* Has read method(s) */
142 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
143 /* Has write method(s) */
144 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
146 /* File is stream-like */
147 #define FMODE_STREAM ((__force fmode_t)0x200000)
149 /* File was opened by fanotify and shouldn't generate fanotify events */
150 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
153 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
154 * that indicates that they should check the contents of the iovec are
155 * valid, but not check the memory that the iovec elements
158 #define CHECK_IOVEC_ONLY -1
161 * The below are the various read and write flags that we support. Some of
162 * them include behavioral modifiers that send information down to the
163 * block layer and IO scheduler. They should be used along with a req_op.
166 * The block layer uses device plugging to defer IO a little bit, in
167 * the hope that we will see more IO very shortly. This increases
168 * coalescing of adjacent IO and thus reduces the number of IOs we
169 * have to send to the device. It also allows for better queuing,
170 * if the IO isn't mergeable. If the caller is going to be waiting
171 * for the IO, then he must ensure that the device is unplugged so
172 * that the IO is dispatched to the driver.
174 * All IO is handled async in Linux. This is fine for background
175 * writes, but for reads or writes that someone waits for completion
176 * on, we want to notify the block layer and IO scheduler so that they
177 * know about it. That allows them to make better scheduling
178 * decisions. So when the below references 'sync' and 'async', it
179 * is referencing this priority hint.
181 * With that in mind, the available types are:
183 * READ A normal read operation. Device will be plugged.
184 * READ_SYNC A synchronous read. Device is not plugged, caller can
185 * immediately wait on this read without caring about
187 * WRITE A normal async write. Device will be plugged.
188 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
189 * the hint that someone will be waiting on this IO
190 * shortly. The write equivalent of READ_SYNC.
191 * WRITE_ODIRECT Special case write for O_DIRECT only.
192 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
193 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
194 * non-volatile media on completion.
195 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
196 * by a cache flush and data is guaranteed to be on
197 * non-volatile media on completion.
200 #define RW_MASK REQ_OP_WRITE
202 #define READ REQ_OP_READ
203 #define WRITE REQ_OP_WRITE
205 #define READ_SYNC REQ_SYNC
206 #define WRITE_SYNC (REQ_SYNC | REQ_NOIDLE)
207 #define WRITE_ODIRECT REQ_SYNC
208 #define WRITE_FLUSH (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH)
209 #define WRITE_FUA (REQ_SYNC | REQ_NOIDLE | REQ_FUA)
210 #define WRITE_FLUSH_FUA (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH | REQ_FUA)
213 * Attribute flags. These should be or-ed together to figure out what
216 #define ATTR_MODE (1 << 0)
217 #define ATTR_UID (1 << 1)
218 #define ATTR_GID (1 << 2)
219 #define ATTR_SIZE (1 << 3)
220 #define ATTR_ATIME (1 << 4)
221 #define ATTR_MTIME (1 << 5)
222 #define ATTR_CTIME (1 << 6)
223 #define ATTR_ATIME_SET (1 << 7)
224 #define ATTR_MTIME_SET (1 << 8)
225 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
226 #define ATTR_ATTR_FLAG (1 << 10)
227 #define ATTR_KILL_SUID (1 << 11)
228 #define ATTR_KILL_SGID (1 << 12)
229 #define ATTR_FILE (1 << 13)
230 #define ATTR_KILL_PRIV (1 << 14)
231 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
232 #define ATTR_TIMES_SET (1 << 16)
233 #define ATTR_TOUCH (1 << 17)
236 * Whiteout is represented by a char device. The following constants define the
237 * mode and device number to use.
239 #define WHITEOUT_MODE 0
240 #define WHITEOUT_DEV 0
243 * This is the Inode Attributes structure, used for notify_change(). It
244 * uses the above definitions as flags, to know which values have changed.
245 * Also, in this manner, a Filesystem can look at only the values it cares
246 * about. Basically, these are the attributes that the VFS layer can
247 * request to change from the FS layer.
249 * Derek Atkins <warlord@MIT.EDU> 94-10-20
252 unsigned int ia_valid;
257 struct timespec ia_atime;
258 struct timespec ia_mtime;
259 struct timespec ia_ctime;
262 * Not an attribute, but an auxiliary info for filesystems wanting to
263 * implement an ftruncate() like method. NOTE: filesystem should
264 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
266 struct file *ia_file;
270 * Includes for diskquotas.
272 #include <linux/quota.h>
275 * Maximum number of layers of fs stack. Needs to be limited to
276 * prevent kernel stack overflow
278 #define FILESYSTEM_MAX_STACK_DEPTH 2
281 * enum positive_aop_returns - aop return codes with specific semantics
283 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
284 * completed, that the page is still locked, and
285 * should be considered active. The VM uses this hint
286 * to return the page to the active list -- it won't
287 * be a candidate for writeback again in the near
288 * future. Other callers must be careful to unlock
289 * the page if they get this return. Returned by
292 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
293 * unlocked it and the page might have been truncated.
294 * The caller should back up to acquiring a new page and
295 * trying again. The aop will be taking reasonable
296 * precautions not to livelock. If the caller held a page
297 * reference, it should drop it before retrying. Returned
300 * address_space_operation functions return these large constants to indicate
301 * special semantics to the caller. These are much larger than the bytes in a
302 * page to allow for functions that return the number of bytes operated on in a
306 enum positive_aop_returns {
307 AOP_WRITEPAGE_ACTIVATE = 0x80000,
308 AOP_TRUNCATED_PAGE = 0x80001,
311 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
312 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
313 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
314 * helper code (eg buffer layer)
315 * to clear GFP_FS from alloc */
318 * oh the beauties of C type declarations.
321 struct address_space;
322 struct writeback_control;
324 #define IOCB_EVENTFD (1 << 0)
325 #define IOCB_APPEND (1 << 1)
326 #define IOCB_DIRECT (1 << 2)
327 #define IOCB_HIPRI (1 << 3)
328 #define IOCB_DSYNC (1 << 4)
329 #define IOCB_SYNC (1 << 5)
330 #define IOCB_WRITE (1 << 6)
333 struct file *ki_filp;
335 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
340 static inline bool is_sync_kiocb(struct kiocb *kiocb)
342 return kiocb->ki_complete == NULL;
345 static inline int iocb_flags(struct file *file);
347 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
349 *kiocb = (struct kiocb) {
351 .ki_flags = iocb_flags(filp),
356 * "descriptor" for what we're up to with a read.
357 * This allows us to use the same read code yet
358 * have multiple different users of the data that
359 * we read from a file.
361 * The simplest case just copies the data to user
374 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
375 unsigned long, unsigned long);
377 struct address_space_operations {
378 int (*writepage)(struct page *page, struct writeback_control *wbc);
379 int (*readpage)(struct file *, struct page *);
381 /* Write back some dirty pages from this mapping. */
382 int (*writepages)(struct address_space *, struct writeback_control *);
384 /* Set a page dirty. Return true if this dirtied it */
385 int (*set_page_dirty)(struct page *page);
387 int (*readpages)(struct file *filp, struct address_space *mapping,
388 struct list_head *pages, unsigned nr_pages);
390 int (*write_begin)(struct file *, struct address_space *mapping,
391 loff_t pos, unsigned len, unsigned flags,
392 struct page **pagep, void **fsdata);
393 int (*write_end)(struct file *, struct address_space *mapping,
394 loff_t pos, unsigned len, unsigned copied,
395 struct page *page, void *fsdata);
397 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
398 sector_t (*bmap)(struct address_space *, sector_t);
399 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
400 int (*releasepage) (struct page *, gfp_t);
401 void (*freepage)(struct page *);
402 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
404 * migrate the contents of a page to the specified target. If
405 * migrate_mode is MIGRATE_ASYNC, it must not block.
407 int (*migratepage) (struct address_space *,
408 struct page *, struct page *, enum migrate_mode);
409 bool (*isolate_page)(struct page *, isolate_mode_t);
410 void (*putback_page)(struct page *);
411 int (*launder_page) (struct page *);
412 int (*is_partially_uptodate) (struct page *, unsigned long,
414 void (*is_dirty_writeback) (struct page *, bool *, bool *);
415 int (*error_remove_page)(struct address_space *, struct page *);
417 /* swapfile support */
418 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
420 void (*swap_deactivate)(struct file *file);
423 extern const struct address_space_operations empty_aops;
426 * pagecache_write_begin/pagecache_write_end must be used by general code
427 * to write into the pagecache.
429 int pagecache_write_begin(struct file *, struct address_space *mapping,
430 loff_t pos, unsigned len, unsigned flags,
431 struct page **pagep, void **fsdata);
433 int pagecache_write_end(struct file *, struct address_space *mapping,
434 loff_t pos, unsigned len, unsigned copied,
435 struct page *page, void *fsdata);
437 struct address_space {
438 struct inode *host; /* owner: inode, block_device */
439 struct radix_tree_root page_tree; /* radix tree of all pages */
440 spinlock_t tree_lock; /* and lock protecting it */
441 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
442 struct rb_root i_mmap; /* tree of private and shared mappings */
443 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
444 /* Protected by tree_lock together with the radix tree */
445 unsigned long nrpages; /* number of total pages */
446 /* number of shadow or DAX exceptional entries */
447 unsigned long nrexceptional;
448 pgoff_t writeback_index;/* writeback starts here */
449 const struct address_space_operations *a_ops; /* methods */
450 unsigned long flags; /* error bits */
451 spinlock_t private_lock; /* for use by the address_space */
452 gfp_t gfp_mask; /* implicit gfp mask for allocations */
453 struct list_head private_list; /* ditto */
454 void *private_data; /* ditto */
455 } __attribute__((aligned(sizeof(long))));
457 * On most architectures that alignment is already the case; but
458 * must be enforced here for CRIS, to let the least significant bit
459 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
461 struct request_queue;
463 struct block_device {
464 dev_t bd_dev; /* not a kdev_t - it's a search key */
466 struct inode * bd_inode; /* will die */
467 struct super_block * bd_super;
468 struct mutex bd_mutex; /* open/close mutex */
472 bool bd_write_holder;
474 struct list_head bd_holder_disks;
476 struct block_device * bd_contains;
477 unsigned bd_block_size;
478 struct hd_struct * bd_part;
479 /* number of times partitions within this device have been opened. */
480 unsigned bd_part_count;
482 struct gendisk * bd_disk;
483 struct request_queue * bd_queue;
484 struct list_head bd_list;
486 * Private data. You must have bd_claim'ed the block_device
487 * to use this. NOTE: bd_claim allows an owner to claim
488 * the same device multiple times, the owner must take special
489 * care to not mess up bd_private for that case.
491 unsigned long bd_private;
493 /* The counter of freeze processes */
494 int bd_fsfreeze_count;
495 /* Mutex for freeze */
496 struct mutex bd_fsfreeze_mutex;
500 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
503 #define PAGECACHE_TAG_DIRTY 0
504 #define PAGECACHE_TAG_WRITEBACK 1
505 #define PAGECACHE_TAG_TOWRITE 2
507 int mapping_tagged(struct address_space *mapping, int tag);
509 static inline void i_mmap_lock_write(struct address_space *mapping)
511 down_write(&mapping->i_mmap_rwsem);
514 static inline void i_mmap_unlock_write(struct address_space *mapping)
516 up_write(&mapping->i_mmap_rwsem);
519 static inline void i_mmap_lock_read(struct address_space *mapping)
521 down_read(&mapping->i_mmap_rwsem);
524 static inline void i_mmap_unlock_read(struct address_space *mapping)
526 up_read(&mapping->i_mmap_rwsem);
530 * Might pages of this file be mapped into userspace?
532 static inline int mapping_mapped(struct address_space *mapping)
534 return !RB_EMPTY_ROOT(&mapping->i_mmap);
538 * Might pages of this file have been modified in userspace?
539 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
540 * marks vma as VM_SHARED if it is shared, and the file was opened for
541 * writing i.e. vma may be mprotected writable even if now readonly.
543 * If i_mmap_writable is negative, no new writable mappings are allowed. You
544 * can only deny writable mappings, if none exists right now.
546 static inline int mapping_writably_mapped(struct address_space *mapping)
548 return atomic_read(&mapping->i_mmap_writable) > 0;
551 static inline int mapping_map_writable(struct address_space *mapping)
553 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
557 static inline void mapping_unmap_writable(struct address_space *mapping)
559 atomic_dec(&mapping->i_mmap_writable);
562 static inline int mapping_deny_writable(struct address_space *mapping)
564 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
568 static inline void mapping_allow_writable(struct address_space *mapping)
570 atomic_inc(&mapping->i_mmap_writable);
574 * Use sequence counter to get consistent i_size on 32-bit processors.
576 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
577 #include <linux/seqlock.h>
578 #define __NEED_I_SIZE_ORDERED
579 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
581 #define i_size_ordered_init(inode) do { } while (0)
585 #define ACL_NOT_CACHED ((void *)(-1))
586 #define ACL_DONT_CACHE ((void *)(-3))
588 static inline struct posix_acl *
589 uncached_acl_sentinel(struct task_struct *task)
591 return (void *)task + 1;
595 is_uncached_acl(struct posix_acl *acl)
597 return (long)acl & 1;
600 #define IOP_FASTPERM 0x0001
601 #define IOP_LOOKUP 0x0002
602 #define IOP_NOFOLLOW 0x0004
603 #define IOP_XATTR 0x0008
606 * Keep mostly read-only and often accessed (especially for
607 * the RCU path lookup and 'stat' data) fields at the beginning
608 * of the 'struct inode'
612 unsigned short i_opflags;
615 unsigned int i_flags;
617 #ifdef CONFIG_FS_POSIX_ACL
618 struct posix_acl *i_acl;
619 struct posix_acl *i_default_acl;
622 const struct inode_operations *i_op;
623 struct super_block *i_sb;
624 struct address_space *i_mapping;
626 #ifdef CONFIG_SECURITY
630 /* Stat data, not accessed from path walking */
633 * Filesystems may only read i_nlink directly. They shall use the
634 * following functions for modification:
636 * (set|clear|inc|drop)_nlink
637 * inode_(inc|dec)_link_count
640 const unsigned int i_nlink;
641 unsigned int __i_nlink;
645 struct timespec i_atime;
646 struct timespec i_mtime;
647 struct timespec i_ctime;
648 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
649 unsigned short i_bytes;
650 unsigned int i_blkbits;
653 #ifdef __NEED_I_SIZE_ORDERED
654 seqcount_t i_size_seqcount;
658 unsigned long i_state;
659 struct rw_semaphore i_rwsem;
661 unsigned long dirtied_when; /* jiffies of first dirtying */
662 unsigned long dirtied_time_when;
664 struct hlist_node i_hash;
665 struct list_head i_io_list; /* backing dev IO list */
666 #ifdef CONFIG_CGROUP_WRITEBACK
667 struct bdi_writeback *i_wb; /* the associated cgroup wb */
669 /* foreign inode detection, see wbc_detach_inode() */
671 u16 i_wb_frn_avg_time;
672 u16 i_wb_frn_history;
674 struct list_head i_lru; /* inode LRU list */
675 struct list_head i_sb_list;
676 struct list_head i_wb_list; /* backing dev writeback list */
678 struct hlist_head i_dentry;
679 struct rcu_head i_rcu;
682 atomic64_t i_sequence; /* see futex */
684 atomic_t i_dio_count;
685 atomic_t i_writecount;
687 atomic_t i_readcount; /* struct files open RO */
689 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
690 struct file_lock_context *i_flctx;
691 struct address_space i_data;
692 struct list_head i_devices;
694 struct pipe_inode_info *i_pipe;
695 struct block_device *i_bdev;
703 #ifdef CONFIG_FSNOTIFY
704 __u32 i_fsnotify_mask; /* all events this inode cares about */
705 struct hlist_head i_fsnotify_marks;
708 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
709 struct fscrypt_info *i_crypt_info;
712 void *i_private; /* fs or device private pointer */
715 static inline unsigned int i_blocksize(const struct inode *node)
717 return (1 << node->i_blkbits);
720 static inline int inode_unhashed(struct inode *inode)
722 return hlist_unhashed(&inode->i_hash);
726 * inode->i_mutex nesting subclasses for the lock validator:
728 * 0: the object of the current VFS operation
732 * 4: second non-directory
733 * 5: second parent (when locking independent directories in rename)
735 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
736 * non-directories at once.
738 * The locking order between these classes is
739 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
741 enum inode_i_mutex_lock_class
751 static inline void inode_lock(struct inode *inode)
753 down_write(&inode->i_rwsem);
756 static inline void inode_unlock(struct inode *inode)
758 up_write(&inode->i_rwsem);
761 static inline void inode_lock_shared(struct inode *inode)
763 down_read(&inode->i_rwsem);
766 static inline void inode_unlock_shared(struct inode *inode)
768 up_read(&inode->i_rwsem);
771 static inline int inode_trylock(struct inode *inode)
773 return down_write_trylock(&inode->i_rwsem);
776 static inline int inode_trylock_shared(struct inode *inode)
778 return down_read_trylock(&inode->i_rwsem);
781 static inline int inode_is_locked(struct inode *inode)
783 return rwsem_is_locked(&inode->i_rwsem);
786 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
788 down_write_nested(&inode->i_rwsem, subclass);
791 void lock_two_nondirectories(struct inode *, struct inode*);
792 void unlock_two_nondirectories(struct inode *, struct inode*);
795 * NOTE: in a 32bit arch with a preemptable kernel and
796 * an UP compile the i_size_read/write must be atomic
797 * with respect to the local cpu (unlike with preempt disabled),
798 * but they don't need to be atomic with respect to other cpus like in
799 * true SMP (so they need either to either locally disable irq around
800 * the read or for example on x86 they can be still implemented as a
801 * cmpxchg8b without the need of the lock prefix). For SMP compiles
802 * and 64bit archs it makes no difference if preempt is enabled or not.
804 static inline loff_t i_size_read(const struct inode *inode)
806 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
811 seq = read_seqcount_begin(&inode->i_size_seqcount);
812 i_size = inode->i_size;
813 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
815 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
819 i_size = inode->i_size;
823 return inode->i_size;
828 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
829 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
830 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
832 static inline void i_size_write(struct inode *inode, loff_t i_size)
834 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
836 write_seqcount_begin(&inode->i_size_seqcount);
837 inode->i_size = i_size;
838 write_seqcount_end(&inode->i_size_seqcount);
840 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
842 inode->i_size = i_size;
845 inode->i_size = i_size;
849 static inline unsigned iminor(const struct inode *inode)
851 return MINOR(inode->i_rdev);
854 static inline unsigned imajor(const struct inode *inode)
856 return MAJOR(inode->i_rdev);
859 extern struct block_device *I_BDEV(struct inode *inode);
862 rwlock_t lock; /* protects pid, uid, euid fields */
863 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
864 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
865 kuid_t uid, euid; /* uid/euid of process setting the owner */
866 int signum; /* posix.1b rt signal to be delivered on IO */
870 * Track a single file's readahead state
872 struct file_ra_state {
873 pgoff_t start; /* where readahead started */
874 unsigned int size; /* # of readahead pages */
875 unsigned int async_size; /* do asynchronous readahead when
876 there are only # of pages ahead */
878 unsigned int ra_pages; /* Maximum readahead window */
879 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
880 loff_t prev_pos; /* Cache last read() position */
884 * Check if @index falls in the readahead windows.
886 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
888 return (index >= ra->start &&
889 index < ra->start + ra->size);
894 struct llist_node fu_llist;
895 struct rcu_head fu_rcuhead;
898 struct inode *f_inode; /* cached value */
899 const struct file_operations *f_op;
902 * Protects f_ep_links, f_flags.
903 * Must not be taken from IRQ context.
906 atomic_long_t f_count;
907 unsigned int f_flags;
909 struct mutex f_pos_lock;
911 struct fown_struct f_owner;
912 const struct cred *f_cred;
913 struct file_ra_state f_ra;
916 #ifdef CONFIG_SECURITY
919 /* needed for tty driver, and maybe others */
923 /* Used by fs/eventpoll.c to link all the hooks to this file */
924 struct list_head f_ep_links;
925 struct list_head f_tfile_llink;
926 #endif /* #ifdef CONFIG_EPOLL */
927 struct address_space *f_mapping;
928 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
933 /* file identifier */
934 unsigned char f_handle[];
937 static inline struct file *get_file(struct file *f)
939 atomic_long_inc(&f->f_count);
942 #define get_file_rcu_many(x, cnt) \
943 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
944 #define get_file_rcu(x) get_file_rcu_many((x), 1)
945 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
946 #define file_count(x) atomic_long_read(&(x)->f_count)
948 #define MAX_NON_LFS ((1UL<<31) - 1)
950 /* Page cache limit. The filesystems should put that into their s_maxbytes
951 limits, otherwise bad things can happen in VM. */
952 #if BITS_PER_LONG==32
953 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
954 #elif BITS_PER_LONG==64
955 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
960 #define FL_DELEG 4 /* NFSv4 delegation */
961 #define FL_ACCESS 8 /* not trying to lock, just looking */
962 #define FL_EXISTS 16 /* when unlocking, test for existence */
963 #define FL_LEASE 32 /* lease held on this file */
964 #define FL_CLOSE 64 /* unlock on close */
965 #define FL_SLEEP 128 /* A blocking lock */
966 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
967 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
968 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
969 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
972 * Special return value from posix_lock_file() and vfs_lock_file() for
973 * asynchronous locking.
975 #define FILE_LOCK_DEFERRED 1
977 /* legacy typedef, should eventually be removed */
978 typedef void *fl_owner_t;
982 struct file_lock_operations {
983 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
984 void (*fl_release_private)(struct file_lock *);
987 struct lock_manager_operations {
988 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
989 unsigned long (*lm_owner_key)(struct file_lock *);
990 fl_owner_t (*lm_get_owner)(fl_owner_t);
991 void (*lm_put_owner)(fl_owner_t);
992 void (*lm_notify)(struct file_lock *); /* unblock callback */
993 int (*lm_grant)(struct file_lock *, int);
994 bool (*lm_break)(struct file_lock *);
995 int (*lm_change)(struct file_lock *, int, struct list_head *);
996 void (*lm_setup)(struct file_lock *, void **);
999 struct lock_manager {
1000 struct list_head list;
1002 * NFSv4 and up also want opens blocked during the grace period;
1009 void locks_start_grace(struct net *, struct lock_manager *);
1010 void locks_end_grace(struct lock_manager *);
1011 int locks_in_grace(struct net *);
1012 int opens_in_grace(struct net *);
1014 /* that will die - we need it for nfs_lock_info */
1015 #include <linux/nfs_fs_i.h>
1018 * struct file_lock represents a generic "file lock". It's used to represent
1019 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1020 * note that the same struct is used to represent both a request for a lock and
1021 * the lock itself, but the same object is never used for both.
1023 * FIXME: should we create a separate "struct lock_request" to help distinguish
1026 * The varous i_flctx lists are ordered by:
1029 * 2) lock range start
1032 * Obviously, the last two criteria only matter for POSIX locks.
1035 struct file_lock *fl_next; /* singly linked list for this inode */
1036 struct list_head fl_list; /* link into file_lock_context */
1037 struct hlist_node fl_link; /* node in global lists */
1038 struct list_head fl_block; /* circular list of blocked processes */
1039 fl_owner_t fl_owner;
1040 unsigned int fl_flags;
1041 unsigned char fl_type;
1042 unsigned int fl_pid;
1043 int fl_link_cpu; /* what cpu's list is this on? */
1044 struct pid *fl_nspid;
1045 wait_queue_head_t fl_wait;
1046 struct file *fl_file;
1050 struct fasync_struct * fl_fasync; /* for lease break notifications */
1051 /* for lease breaks: */
1052 unsigned long fl_break_time;
1053 unsigned long fl_downgrade_time;
1055 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1056 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1058 struct nfs_lock_info nfs_fl;
1059 struct nfs4_lock_info nfs4_fl;
1061 struct list_head link; /* link in AFS vnode's pending_locks list */
1062 int state; /* state of grant or error if -ve */
1067 struct file_lock_context {
1068 spinlock_t flc_lock;
1069 struct list_head flc_flock;
1070 struct list_head flc_posix;
1071 struct list_head flc_lease;
1074 /* The following constant reflects the upper bound of the file/locking space */
1076 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1077 #define OFFSET_MAX INT_LIMIT(loff_t)
1078 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1081 #include <linux/fcntl.h>
1083 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1086 * Return the inode to use for locking
1088 * For overlayfs this should be the overlay inode, not the real inode returned
1089 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1092 static inline struct inode *locks_inode(const struct file *f)
1094 return f->f_path.dentry->d_inode;
1097 #ifdef CONFIG_FILE_LOCKING
1098 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1099 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1100 struct flock __user *);
1102 #if BITS_PER_LONG == 32
1103 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1104 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1105 struct flock64 __user *);
1108 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1109 extern int fcntl_getlease(struct file *filp);
1112 void locks_free_lock_context(struct inode *inode);
1113 void locks_free_lock(struct file_lock *fl);
1114 extern void locks_init_lock(struct file_lock *);
1115 extern struct file_lock * locks_alloc_lock(void);
1116 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1117 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1118 extern void locks_remove_posix(struct file *, fl_owner_t);
1119 extern void locks_remove_file(struct file *);
1120 extern void locks_release_private(struct file_lock *);
1121 extern void posix_test_lock(struct file *, struct file_lock *);
1122 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1123 extern int posix_unblock_lock(struct file_lock *);
1124 extern int vfs_test_lock(struct file *, struct file_lock *);
1125 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1126 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1127 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1128 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1129 extern void lease_get_mtime(struct inode *, struct timespec *time);
1130 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1131 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1132 extern int lease_modify(struct file_lock *, int, struct list_head *);
1133 struct files_struct;
1134 extern void show_fd_locks(struct seq_file *f,
1135 struct file *filp, struct files_struct *files);
1136 #else /* !CONFIG_FILE_LOCKING */
1137 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1138 struct flock __user *user)
1143 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1144 unsigned int cmd, struct flock __user *user)
1149 #if BITS_PER_LONG == 32
1150 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1151 struct flock64 __user *user)
1156 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1157 unsigned int cmd, struct flock64 __user *user)
1162 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1167 static inline int fcntl_getlease(struct file *filp)
1173 locks_free_lock_context(struct inode *inode)
1177 static inline void locks_init_lock(struct file_lock *fl)
1182 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1187 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1192 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1197 static inline void locks_remove_file(struct file *filp)
1202 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1207 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1208 struct file_lock *conflock)
1213 static inline int posix_unblock_lock(struct file_lock *waiter)
1218 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1223 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1224 struct file_lock *fl, struct file_lock *conf)
1229 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1234 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1239 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1244 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1249 static inline int generic_setlease(struct file *filp, long arg,
1250 struct file_lock **flp, void **priv)
1255 static inline int vfs_setlease(struct file *filp, long arg,
1256 struct file_lock **lease, void **priv)
1261 static inline int lease_modify(struct file_lock *fl, int arg,
1262 struct list_head *dispose)
1267 struct files_struct;
1268 static inline void show_fd_locks(struct seq_file *f,
1269 struct file *filp, struct files_struct *files) {}
1270 #endif /* !CONFIG_FILE_LOCKING */
1272 static inline struct inode *file_inode(const struct file *f)
1277 static inline struct dentry *file_dentry(const struct file *file)
1279 return d_real(file->f_path.dentry, file_inode(file), 0);
1282 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1284 return locks_lock_inode_wait(locks_inode(filp), fl);
1287 struct fasync_struct {
1291 struct fasync_struct *fa_next; /* singly linked list */
1292 struct file *fa_file;
1293 struct rcu_head fa_rcu;
1296 #define FASYNC_MAGIC 0x4601
1298 /* SMP safe fasync helpers: */
1299 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1300 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1301 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1302 extern struct fasync_struct *fasync_alloc(void);
1303 extern void fasync_free(struct fasync_struct *);
1305 /* can be called from interrupts */
1306 extern void kill_fasync(struct fasync_struct **, int, int);
1308 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1309 extern void f_setown(struct file *filp, unsigned long arg, int force);
1310 extern void f_delown(struct file *filp);
1311 extern pid_t f_getown(struct file *filp);
1312 extern int send_sigurg(struct fown_struct *fown);
1320 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1321 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1322 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1323 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1324 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1327 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1328 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1329 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1330 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1332 /* sb->s_iflags to limit user namespace mounts */
1333 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1335 /* Possible states of 'frozen' field */
1337 SB_UNFROZEN = 0, /* FS is unfrozen */
1338 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1339 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1340 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1341 * internal threads if needed) */
1342 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1345 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1348 int frozen; /* Is sb frozen? */
1349 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1350 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1353 struct super_block {
1354 struct list_head s_list; /* Keep this first */
1355 dev_t s_dev; /* search index; _not_ kdev_t */
1356 unsigned char s_blocksize_bits;
1357 unsigned long s_blocksize;
1358 loff_t s_maxbytes; /* Max file size */
1359 struct file_system_type *s_type;
1360 const struct super_operations *s_op;
1361 const struct dquot_operations *dq_op;
1362 const struct quotactl_ops *s_qcop;
1363 const struct export_operations *s_export_op;
1364 unsigned long s_flags;
1365 unsigned long s_iflags; /* internal SB_I_* flags */
1366 unsigned long s_magic;
1367 struct dentry *s_root;
1368 struct rw_semaphore s_umount;
1371 #ifdef CONFIG_SECURITY
1374 const struct xattr_handler **s_xattr;
1376 const struct fscrypt_operations *s_cop;
1378 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1379 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1380 struct block_device *s_bdev;
1381 struct backing_dev_info *s_bdi;
1382 struct mtd_info *s_mtd;
1383 struct hlist_node s_instances;
1384 unsigned int s_quota_types; /* Bitmask of supported quota types */
1385 struct quota_info s_dquot; /* Diskquota specific options */
1387 struct sb_writers s_writers;
1389 char s_id[32]; /* Informational name */
1390 u8 s_uuid[16]; /* UUID */
1392 void *s_fs_info; /* Filesystem private info */
1393 unsigned int s_max_links;
1396 /* Granularity of c/m/atime in ns.
1397 Cannot be worse than a second */
1401 * The next field is for VFS *only*. No filesystems have any business
1402 * even looking at it. You had been warned.
1404 struct mutex s_vfs_rename_mutex; /* Kludge */
1407 * Filesystem subtype. If non-empty the filesystem type field
1408 * in /proc/mounts will be "type.subtype"
1413 * Saved mount options for lazy filesystems using
1414 * generic_show_options()
1416 char __rcu *s_options;
1417 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1420 * Saved pool identifier for cleancache (-1 means none)
1422 int cleancache_poolid;
1424 struct shrinker s_shrink; /* per-sb shrinker handle */
1426 /* Number of inodes with nlink == 0 but still referenced */
1427 atomic_long_t s_remove_count;
1429 /* Being remounted read-only */
1430 int s_readonly_remount;
1432 /* AIO completions deferred from interrupt context */
1433 struct workqueue_struct *s_dio_done_wq;
1434 struct hlist_head s_pins;
1437 * Owning user namespace and default context in which to
1438 * interpret filesystem uids, gids, quotas, device nodes,
1439 * xattrs and security labels.
1441 struct user_namespace *s_user_ns;
1444 * Keep the lru lists last in the structure so they always sit on their
1445 * own individual cachelines.
1447 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1448 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1449 struct rcu_head rcu;
1450 struct work_struct destroy_work;
1452 struct mutex s_sync_lock; /* sync serialisation lock */
1455 * Indicates how deep in a filesystem stack this SB is
1459 /* s_inode_list_lock protects s_inodes */
1460 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1461 struct list_head s_inodes; /* all inodes */
1463 spinlock_t s_inode_wblist_lock;
1464 struct list_head s_inodes_wb; /* writeback inodes */
1467 /* Helper functions so that in most cases filesystems will
1468 * not need to deal directly with kuid_t and kgid_t and can
1469 * instead deal with the raw numeric values that are stored
1470 * in the filesystem.
1472 static inline uid_t i_uid_read(const struct inode *inode)
1474 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1477 static inline gid_t i_gid_read(const struct inode *inode)
1479 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1482 static inline void i_uid_write(struct inode *inode, uid_t uid)
1484 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1487 static inline void i_gid_write(struct inode *inode, gid_t gid)
1489 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1492 extern struct timespec current_fs_time(struct super_block *sb);
1493 extern struct timespec current_time(struct inode *inode);
1496 * Snapshotting support.
1499 void __sb_end_write(struct super_block *sb, int level);
1500 int __sb_start_write(struct super_block *sb, int level, bool wait);
1502 #define __sb_writers_acquired(sb, lev) \
1503 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1504 #define __sb_writers_release(sb, lev) \
1505 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1508 * sb_end_write - drop write access to a superblock
1509 * @sb: the super we wrote to
1511 * Decrement number of writers to the filesystem. Wake up possible waiters
1512 * wanting to freeze the filesystem.
1514 static inline void sb_end_write(struct super_block *sb)
1516 __sb_end_write(sb, SB_FREEZE_WRITE);
1520 * sb_end_pagefault - drop write access to a superblock from a page fault
1521 * @sb: the super we wrote to
1523 * Decrement number of processes handling write page fault to the filesystem.
1524 * Wake up possible waiters wanting to freeze the filesystem.
1526 static inline void sb_end_pagefault(struct super_block *sb)
1528 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1532 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1533 * @sb: the super we wrote to
1535 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1536 * waiters wanting to freeze the filesystem.
1538 static inline void sb_end_intwrite(struct super_block *sb)
1540 __sb_end_write(sb, SB_FREEZE_FS);
1544 * sb_start_write - get write access to a superblock
1545 * @sb: the super we write to
1547 * When a process wants to write data or metadata to a file system (i.e. dirty
1548 * a page or an inode), it should embed the operation in a sb_start_write() -
1549 * sb_end_write() pair to get exclusion against file system freezing. This
1550 * function increments number of writers preventing freezing. If the file
1551 * system is already frozen, the function waits until the file system is
1554 * Since freeze protection behaves as a lock, users have to preserve
1555 * ordering of freeze protection and other filesystem locks. Generally,
1556 * freeze protection should be the outermost lock. In particular, we have:
1559 * -> i_mutex (write path, truncate, directory ops, ...)
1560 * -> s_umount (freeze_super, thaw_super)
1562 static inline void sb_start_write(struct super_block *sb)
1564 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1567 static inline int sb_start_write_trylock(struct super_block *sb)
1569 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1573 * sb_start_pagefault - get write access to a superblock from a page fault
1574 * @sb: the super we write to
1576 * When a process starts handling write page fault, it should embed the
1577 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1578 * exclusion against file system freezing. This is needed since the page fault
1579 * is going to dirty a page. This function increments number of running page
1580 * faults preventing freezing. If the file system is already frozen, the
1581 * function waits until the file system is thawed.
1583 * Since page fault freeze protection behaves as a lock, users have to preserve
1584 * ordering of freeze protection and other filesystem locks. It is advised to
1585 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1586 * handling code implies lock dependency:
1589 * -> sb_start_pagefault
1591 static inline void sb_start_pagefault(struct super_block *sb)
1593 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1597 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1598 * @sb: the super we write to
1600 * This is the third level of protection against filesystem freezing. It is
1601 * free for use by a filesystem. The only requirement is that it must rank
1602 * below sb_start_pagefault.
1604 * For example filesystem can call sb_start_intwrite() when starting a
1605 * transaction which somewhat eases handling of freezing for internal sources
1606 * of filesystem changes (internal fs threads, discarding preallocation on file
1609 static inline void sb_start_intwrite(struct super_block *sb)
1611 __sb_start_write(sb, SB_FREEZE_FS, true);
1615 extern bool inode_owner_or_capable(const struct inode *inode);
1618 * VFS helper functions..
1620 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1621 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1622 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1623 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1624 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1625 extern int vfs_rmdir(struct inode *, struct dentry *);
1626 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1627 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1628 extern int vfs_whiteout(struct inode *, struct dentry *);
1631 * VFS file helper functions.
1633 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1635 extern bool may_open_dev(const struct path *path);
1637 * VFS FS_IOC_FIEMAP helper definitions.
1639 struct fiemap_extent_info {
1640 unsigned int fi_flags; /* Flags as passed from user */
1641 unsigned int fi_extents_mapped; /* Number of mapped extents */
1642 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1643 struct fiemap_extent __user *fi_extents_start; /* Start of
1644 fiemap_extent array */
1646 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1647 u64 phys, u64 len, u32 flags);
1648 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1653 * NOTE! These match bits 12..15 of stat.st_mode
1654 * (ie "(i_mode >> 12) & 15").
1656 #define DT_UNKNOWN 0
1667 * This is the "filldir" function type, used by readdir() to let
1668 * the kernel specify what kind of dirent layout it wants to have.
1669 * This allows the kernel to read directories into kernel space or
1670 * to have different dirent layouts depending on the binary type.
1673 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1676 struct dir_context {
1677 const filldir_t actor;
1681 struct block_device_operations;
1683 /* These macros are for out of kernel modules to test that
1684 * the kernel supports the unlocked_ioctl and compat_ioctl
1685 * fields in struct file_operations. */
1686 #define HAVE_COMPAT_IOCTL 1
1687 #define HAVE_UNLOCKED_IOCTL 1
1690 * These flags let !MMU mmap() govern direct device mapping vs immediate
1691 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1693 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1694 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1695 * NOMMU_MAP_READ: Can be mapped for reading
1696 * NOMMU_MAP_WRITE: Can be mapped for writing
1697 * NOMMU_MAP_EXEC: Can be mapped for execution
1699 #define NOMMU_MAP_COPY 0x00000001
1700 #define NOMMU_MAP_DIRECT 0x00000008
1701 #define NOMMU_MAP_READ VM_MAYREAD
1702 #define NOMMU_MAP_WRITE VM_MAYWRITE
1703 #define NOMMU_MAP_EXEC VM_MAYEXEC
1705 #define NOMMU_VMFLAGS \
1706 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1711 struct file_operations {
1712 struct module *owner;
1713 loff_t (*llseek) (struct file *, loff_t, int);
1714 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1715 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1716 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1717 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1718 int (*iterate) (struct file *, struct dir_context *);
1719 int (*iterate_shared) (struct file *, struct dir_context *);
1720 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1721 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1722 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1723 int (*mmap) (struct file *, struct vm_area_struct *);
1724 int (*open) (struct inode *, struct file *);
1725 int (*flush) (struct file *, fl_owner_t id);
1726 int (*release) (struct inode *, struct file *);
1727 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1728 int (*fasync) (int, struct file *, int);
1729 int (*lock) (struct file *, int, struct file_lock *);
1730 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1731 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1732 int (*check_flags)(int);
1733 int (*flock) (struct file *, int, struct file_lock *);
1734 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1735 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1736 int (*setlease)(struct file *, long, struct file_lock **, void **);
1737 long (*fallocate)(struct file *file, int mode, loff_t offset,
1739 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1741 unsigned (*mmap_capabilities)(struct file *);
1743 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1744 loff_t, size_t, unsigned int);
1745 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1747 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1751 struct inode_operations {
1752 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1753 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1754 int (*permission) (struct inode *, int);
1755 struct posix_acl * (*get_acl)(struct inode *, int);
1757 int (*readlink) (struct dentry *, char __user *,int);
1759 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1760 int (*link) (struct dentry *,struct inode *,struct dentry *);
1761 int (*unlink) (struct inode *,struct dentry *);
1762 int (*symlink) (struct inode *,struct dentry *,const char *);
1763 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1764 int (*rmdir) (struct inode *,struct dentry *);
1765 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1766 int (*rename) (struct inode *, struct dentry *,
1767 struct inode *, struct dentry *, unsigned int);
1768 int (*setattr) (struct dentry *, struct iattr *);
1769 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1770 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1771 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1773 int (*update_time)(struct inode *, struct timespec *, int);
1774 int (*atomic_open)(struct inode *, struct dentry *,
1775 struct file *, unsigned open_flag,
1776 umode_t create_mode, int *opened);
1777 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1778 int (*set_acl)(struct inode *, struct posix_acl *, int);
1779 } ____cacheline_aligned;
1781 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1782 unsigned long nr_segs, unsigned long fast_segs,
1783 struct iovec *fast_pointer,
1784 struct iovec **ret_pointer);
1786 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1787 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1788 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1789 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1790 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1791 unsigned long, loff_t *, int);
1792 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1793 unsigned long, loff_t *, int);
1794 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1795 loff_t, size_t, unsigned int);
1796 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1797 struct file *file_out, loff_t pos_out, u64 len);
1798 extern int vfs_dedupe_file_range(struct file *file,
1799 struct file_dedupe_range *same);
1801 struct super_operations {
1802 struct inode *(*alloc_inode)(struct super_block *sb);
1803 void (*destroy_inode)(struct inode *);
1805 void (*dirty_inode) (struct inode *, int flags);
1806 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1807 int (*drop_inode) (struct inode *);
1808 void (*evict_inode) (struct inode *);
1809 void (*put_super) (struct super_block *);
1810 int (*sync_fs)(struct super_block *sb, int wait);
1811 int (*freeze_super) (struct super_block *);
1812 int (*freeze_fs) (struct super_block *);
1813 int (*thaw_super) (struct super_block *);
1814 int (*unfreeze_fs) (struct super_block *);
1815 int (*statfs) (struct dentry *, struct kstatfs *);
1816 int (*remount_fs) (struct super_block *, int *, char *);
1817 void (*umount_begin) (struct super_block *);
1819 int (*show_options)(struct seq_file *, struct dentry *);
1820 int (*show_devname)(struct seq_file *, struct dentry *);
1821 int (*show_path)(struct seq_file *, struct dentry *);
1822 int (*show_stats)(struct seq_file *, struct dentry *);
1824 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1825 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1826 struct dquot **(*get_dquots)(struct inode *);
1828 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1829 long (*nr_cached_objects)(struct super_block *,
1830 struct shrink_control *);
1831 long (*free_cached_objects)(struct super_block *,
1832 struct shrink_control *);
1836 * Inode flags - they have no relation to superblock flags now
1838 #define S_SYNC 1 /* Writes are synced at once */
1839 #define S_NOATIME 2 /* Do not update access times */
1840 #define S_APPEND 4 /* Append-only file */
1841 #define S_IMMUTABLE 8 /* Immutable file */
1842 #define S_DEAD 16 /* removed, but still open directory */
1843 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1844 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1845 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1846 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1847 #define S_PRIVATE 512 /* Inode is fs-internal */
1848 #define S_IMA 1024 /* Inode has an associated IMA struct */
1849 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1850 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1851 #ifdef CONFIG_FS_DAX
1852 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1854 #define S_DAX 0 /* Make all the DAX code disappear */
1858 * Note that nosuid etc flags are inode-specific: setting some file-system
1859 * flags just means all the inodes inherit those flags by default. It might be
1860 * possible to override it selectively if you really wanted to with some
1861 * ioctl() that is not currently implemented.
1863 * Exception: MS_RDONLY is always applied to the entire file system.
1865 * Unfortunately, it is possible to change a filesystems flags with it mounted
1866 * with files in use. This means that all of the inodes will not have their
1867 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1868 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1870 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1872 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1873 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1874 ((inode)->i_flags & S_SYNC))
1875 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1876 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1877 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1878 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1879 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1881 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1882 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1883 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1884 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1886 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1887 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1888 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1889 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1890 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1891 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1892 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1893 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1895 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1896 (inode)->i_rdev == WHITEOUT_DEV)
1898 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1900 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1904 * Inode state bits. Protected by inode->i_lock
1906 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1907 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1909 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1910 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1911 * various stages of removing an inode.
1913 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1915 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1916 * fdatasync(). i_atime is the usual cause.
1917 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1918 * these changes separately from I_DIRTY_SYNC so that we
1919 * don't have to write inode on fdatasync() when only
1920 * mtime has changed in it.
1921 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1922 * I_NEW Serves as both a mutex and completion notification.
1923 * New inodes set I_NEW. If two processes both create
1924 * the same inode, one of them will release its inode and
1925 * wait for I_NEW to be released before returning.
1926 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1927 * also cause waiting on I_NEW, without I_NEW actually
1928 * being set. find_inode() uses this to prevent returning
1929 * nearly-dead inodes.
1930 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1931 * is zero. I_FREEING must be set when I_WILL_FREE is
1933 * I_FREEING Set when inode is about to be freed but still has dirty
1934 * pages or buffers attached or the inode itself is still
1936 * I_CLEAR Added by clear_inode(). In this state the inode is
1937 * clean and can be destroyed. Inode keeps I_FREEING.
1939 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1940 * prohibited for many purposes. iget() must wait for
1941 * the inode to be completely released, then create it
1942 * anew. Other functions will just ignore such inodes,
1943 * if appropriate. I_NEW is used for waiting.
1945 * I_SYNC Writeback of inode is running. The bit is set during
1946 * data writeback, and cleared with a wakeup on the bit
1947 * address once it is done. The bit is also used to pin
1948 * the inode in memory for flusher thread.
1950 * I_REFERENCED Marks the inode as recently references on the LRU list.
1952 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1954 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1955 * synchronize competing switching instances and to tell
1956 * wb stat updates to grab mapping->tree_lock. See
1957 * inode_switch_wb_work_fn() for details.
1959 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
1960 * Used to detect that mark_inode_dirty() should not move
1961 * inode between dirty lists.
1963 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1965 #define I_DIRTY_SYNC (1 << 0)
1966 #define I_DIRTY_DATASYNC (1 << 1)
1967 #define I_DIRTY_PAGES (1 << 2)
1969 #define I_NEW (1 << __I_NEW)
1970 #define I_WILL_FREE (1 << 4)
1971 #define I_FREEING (1 << 5)
1972 #define I_CLEAR (1 << 6)
1974 #define I_SYNC (1 << __I_SYNC)
1975 #define I_REFERENCED (1 << 8)
1976 #define __I_DIO_WAKEUP 9
1977 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1978 #define I_LINKABLE (1 << 10)
1979 #define I_DIRTY_TIME (1 << 11)
1980 #define I_DIRTY_TIME_EXPIRED (1 << 12)
1981 #define I_WB_SWITCH (1 << 13)
1982 #define I_SYNC_QUEUED (1 << 17)
1984 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1985 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1987 extern void __mark_inode_dirty(struct inode *, int);
1988 static inline void mark_inode_dirty(struct inode *inode)
1990 __mark_inode_dirty(inode, I_DIRTY);
1993 static inline void mark_inode_dirty_sync(struct inode *inode)
1995 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1998 extern void inc_nlink(struct inode *inode);
1999 extern void drop_nlink(struct inode *inode);
2000 extern void clear_nlink(struct inode *inode);
2001 extern void set_nlink(struct inode *inode, unsigned int nlink);
2003 static inline void inode_inc_link_count(struct inode *inode)
2006 mark_inode_dirty(inode);
2009 static inline void inode_dec_link_count(struct inode *inode)
2012 mark_inode_dirty(inode);
2016 * inode_inc_iversion - increments i_version
2017 * @inode: inode that need to be updated
2019 * Every time the inode is modified, the i_version field will be incremented.
2020 * The filesystem has to be mounted with i_version flag
2023 static inline void inode_inc_iversion(struct inode *inode)
2025 spin_lock(&inode->i_lock);
2027 spin_unlock(&inode->i_lock);
2030 enum file_time_flags {
2037 extern void touch_atime(const struct path *);
2038 static inline void file_accessed(struct file *file)
2040 if (!(file->f_flags & O_NOATIME))
2041 touch_atime(&file->f_path);
2044 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2045 int sync_inode_metadata(struct inode *inode, int wait);
2047 struct file_system_type {
2050 #define FS_REQUIRES_DEV 1
2051 #define FS_BINARY_MOUNTDATA 2
2052 #define FS_HAS_SUBTYPE 4
2053 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2054 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2055 struct dentry *(*mount) (struct file_system_type *, int,
2056 const char *, void *);
2057 void (*kill_sb) (struct super_block *);
2058 struct module *owner;
2059 struct file_system_type * next;
2060 struct hlist_head fs_supers;
2062 struct lock_class_key s_lock_key;
2063 struct lock_class_key s_umount_key;
2064 struct lock_class_key s_vfs_rename_key;
2065 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2067 struct lock_class_key i_lock_key;
2068 struct lock_class_key i_mutex_key;
2069 struct lock_class_key i_mutex_dir_key;
2072 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2074 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2075 int flags, void *data, void *ns, struct user_namespace *user_ns,
2076 int (*fill_super)(struct super_block *, void *, int));
2077 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2078 int flags, const char *dev_name, void *data,
2079 int (*fill_super)(struct super_block *, void *, int));
2080 extern struct dentry *mount_single(struct file_system_type *fs_type,
2081 int flags, void *data,
2082 int (*fill_super)(struct super_block *, void *, int));
2083 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2084 int flags, void *data,
2085 int (*fill_super)(struct super_block *, void *, int));
2086 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2087 void generic_shutdown_super(struct super_block *sb);
2088 void kill_block_super(struct super_block *sb);
2089 void kill_anon_super(struct super_block *sb);
2090 void kill_litter_super(struct super_block *sb);
2091 void deactivate_super(struct super_block *sb);
2092 void deactivate_locked_super(struct super_block *sb);
2093 int set_anon_super(struct super_block *s, void *data);
2094 int get_anon_bdev(dev_t *);
2095 void free_anon_bdev(dev_t);
2096 struct super_block *sget_userns(struct file_system_type *type,
2097 int (*test)(struct super_block *,void *),
2098 int (*set)(struct super_block *,void *),
2099 int flags, struct user_namespace *user_ns,
2101 struct super_block *sget(struct file_system_type *type,
2102 int (*test)(struct super_block *,void *),
2103 int (*set)(struct super_block *,void *),
2104 int flags, void *data);
2105 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2106 const struct super_operations *ops,
2107 const struct xattr_handler **xattr,
2108 const struct dentry_operations *dops,
2111 static inline struct dentry *
2112 mount_pseudo(struct file_system_type *fs_type, char *name,
2113 const struct super_operations *ops,
2114 const struct dentry_operations *dops, unsigned long magic)
2116 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2119 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2120 #define fops_get(fops) \
2121 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2122 #define fops_put(fops) \
2123 do { if (fops) module_put((fops)->owner); } while(0)
2125 * This one is to be used *ONLY* from ->open() instances.
2126 * fops must be non-NULL, pinned down *and* module dependencies
2127 * should be sufficient to pin the caller down as well.
2129 #define replace_fops(f, fops) \
2131 struct file *__file = (f); \
2132 fops_put(__file->f_op); \
2133 BUG_ON(!(__file->f_op = (fops))); \
2136 extern int register_filesystem(struct file_system_type *);
2137 extern int unregister_filesystem(struct file_system_type *);
2138 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2139 #define kern_mount(type) kern_mount_data(type, NULL)
2140 extern void kern_unmount(struct vfsmount *mnt);
2141 extern int may_umount_tree(struct vfsmount *);
2142 extern int may_umount(struct vfsmount *);
2143 extern long do_mount(const char *, const char __user *,
2144 const char *, unsigned long, void *);
2145 extern struct vfsmount *collect_mounts(struct path *);
2146 extern void drop_collected_mounts(struct vfsmount *);
2147 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2149 extern int vfs_statfs(struct path *, struct kstatfs *);
2150 extern int user_statfs(const char __user *, struct kstatfs *);
2151 extern int fd_statfs(int, struct kstatfs *);
2152 extern int vfs_ustat(dev_t, struct kstatfs *);
2153 extern int freeze_super(struct super_block *super);
2154 extern int thaw_super(struct super_block *super);
2155 extern bool our_mnt(struct vfsmount *mnt);
2157 extern int current_umask(void);
2159 extern void ihold(struct inode * inode);
2160 extern void iput(struct inode *);
2161 extern int generic_update_time(struct inode *, struct timespec *, int);
2164 extern struct kobject *fs_kobj;
2166 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2168 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2169 extern int locks_mandatory_locked(struct file *);
2170 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2173 * Candidates for mandatory locking have the setgid bit set
2174 * but no group execute bit - an otherwise meaningless combination.
2177 static inline int __mandatory_lock(struct inode *ino)
2179 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2183 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2184 * otherwise these will be advisory locks
2187 static inline int mandatory_lock(struct inode *ino)
2189 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2192 static inline int locks_verify_locked(struct file *file)
2194 if (mandatory_lock(locks_inode(file)))
2195 return locks_mandatory_locked(file);
2199 static inline int locks_verify_truncate(struct inode *inode,
2203 if (!inode->i_flctx || !mandatory_lock(inode))
2206 if (size < inode->i_size) {
2207 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2210 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2215 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2217 static inline int locks_mandatory_locked(struct file *file)
2222 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2223 loff_t start, loff_t end, unsigned char type)
2228 static inline int __mandatory_lock(struct inode *inode)
2233 static inline int mandatory_lock(struct inode *inode)
2238 static inline int locks_verify_locked(struct file *file)
2243 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2249 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2252 #ifdef CONFIG_FILE_LOCKING
2253 static inline int break_lease(struct inode *inode, unsigned int mode)
2256 * Since this check is lockless, we must ensure that any refcounts
2257 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2258 * could end up racing with tasks trying to set a new lease on this
2262 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2263 return __break_lease(inode, mode, FL_LEASE);
2267 static inline int break_deleg(struct inode *inode, unsigned int mode)
2270 * Since this check is lockless, we must ensure that any refcounts
2271 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2272 * could end up racing with tasks trying to set a new lease on this
2276 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2277 return __break_lease(inode, mode, FL_DELEG);
2281 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2285 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2286 if (ret == -EWOULDBLOCK && delegated_inode) {
2287 *delegated_inode = inode;
2293 static inline int break_deleg_wait(struct inode **delegated_inode)
2297 ret = break_deleg(*delegated_inode, O_WRONLY);
2298 iput(*delegated_inode);
2299 *delegated_inode = NULL;
2303 static inline int break_layout(struct inode *inode, bool wait)
2306 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2307 return __break_lease(inode,
2308 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2313 #else /* !CONFIG_FILE_LOCKING */
2314 static inline int break_lease(struct inode *inode, unsigned int mode)
2319 static inline int break_deleg(struct inode *inode, unsigned int mode)
2324 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2329 static inline int break_deleg_wait(struct inode **delegated_inode)
2335 static inline int break_layout(struct inode *inode, bool wait)
2340 #endif /* CONFIG_FILE_LOCKING */
2345 const char *name; /* pointer to actual string */
2346 const __user char *uptr; /* original userland pointer */
2347 struct audit_names *aname;
2352 extern long vfs_truncate(const struct path *, loff_t);
2353 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2355 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2357 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2359 extern struct file *file_open_name(struct filename *, int, umode_t);
2360 extern struct file *filp_open(const char *, int, umode_t);
2361 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2362 const char *, int, umode_t);
2363 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2364 extern int filp_close(struct file *, fl_owner_t id);
2366 extern struct filename *getname_flags(const char __user *, int, int *);
2367 extern struct filename *getname(const char __user *);
2368 extern struct filename *getname_kernel(const char *);
2369 extern void putname(struct filename *name);
2375 extern int finish_open(struct file *file, struct dentry *dentry,
2376 int (*open)(struct inode *, struct file *),
2378 extern int finish_no_open(struct file *file, struct dentry *dentry);
2382 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2385 extern void __init vfs_caches_init_early(void);
2386 extern void __init vfs_caches_init(void);
2388 extern struct kmem_cache *names_cachep;
2390 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2391 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2394 extern int register_blkdev(unsigned int, const char *);
2395 extern void unregister_blkdev(unsigned int, const char *);
2396 extern struct block_device *bdget(dev_t);
2397 extern struct block_device *bdgrab(struct block_device *bdev);
2398 extern void bd_set_size(struct block_device *, loff_t size);
2399 extern void bd_forget(struct inode *inode);
2400 extern void bdput(struct block_device *);
2401 extern void invalidate_bdev(struct block_device *);
2402 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2403 extern int sync_blockdev(struct block_device *bdev);
2404 extern void kill_bdev(struct block_device *);
2405 extern struct super_block *freeze_bdev(struct block_device *);
2406 extern void emergency_thaw_all(void);
2407 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2408 extern int fsync_bdev(struct block_device *);
2410 extern struct super_block *blockdev_superblock;
2412 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2414 return sb == blockdev_superblock;
2417 static inline void bd_forget(struct inode *inode) {}
2418 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2419 static inline void kill_bdev(struct block_device *bdev) {}
2420 static inline void invalidate_bdev(struct block_device *bdev) {}
2422 static inline struct super_block *freeze_bdev(struct block_device *sb)
2427 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2432 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2436 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2441 extern int sync_filesystem(struct super_block *);
2442 extern const struct file_operations def_blk_fops;
2443 extern const struct file_operations def_chr_fops;
2445 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2446 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2447 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2448 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2449 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2451 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2453 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2454 extern int __blkdev_reread_part(struct block_device *bdev);
2455 extern int blkdev_reread_part(struct block_device *bdev);
2458 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2459 extern void bd_unlink_disk_holder(struct block_device *bdev,
2460 struct gendisk *disk);
2462 static inline int bd_link_disk_holder(struct block_device *bdev,
2463 struct gendisk *disk)
2467 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2468 struct gendisk *disk)
2475 #define CHRDEV_MAJOR_HASH_SIZE 255
2476 /* Marks the bottom of the first segment of free char majors */
2477 #define CHRDEV_MAJOR_DYN_END 234
2478 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2479 extern int register_chrdev_region(dev_t, unsigned, const char *);
2480 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2481 unsigned int count, const char *name,
2482 const struct file_operations *fops);
2483 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2484 unsigned int count, const char *name);
2485 extern void unregister_chrdev_region(dev_t, unsigned);
2486 extern void chrdev_show(struct seq_file *,off_t);
2488 static inline int register_chrdev(unsigned int major, const char *name,
2489 const struct file_operations *fops)
2491 return __register_chrdev(major, 0, 256, name, fops);
2494 static inline void unregister_chrdev(unsigned int major, const char *name)
2496 __unregister_chrdev(major, 0, 256, name);
2499 /* fs/block_dev.c */
2500 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2501 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2504 #define BLKDEV_MAJOR_HASH_SIZE 255
2505 extern const char *__bdevname(dev_t, char *buffer);
2506 extern const char *bdevname(struct block_device *bdev, char *buffer);
2507 extern struct block_device *lookup_bdev(const char *);
2508 extern void blkdev_show(struct seq_file *,off_t);
2511 #define BLKDEV_MAJOR_HASH_SIZE 0
2514 extern void init_special_inode(struct inode *, umode_t, dev_t);
2516 /* Invalid inode operations -- fs/bad_inode.c */
2517 extern void make_bad_inode(struct inode *);
2518 extern bool is_bad_inode(struct inode *);
2521 static inline bool op_is_write(unsigned int op)
2523 return op == REQ_OP_READ ? false : true;
2527 * return data direction, READ or WRITE
2529 static inline int bio_data_dir(struct bio *bio)
2531 return op_is_write(bio_op(bio)) ? WRITE : READ;
2534 extern void check_disk_size_change(struct gendisk *disk,
2535 struct block_device *bdev);
2536 extern int revalidate_disk(struct gendisk *);
2537 extern int check_disk_change(struct block_device *);
2538 extern int __invalidate_device(struct block_device *, bool);
2539 extern int invalidate_partition(struct gendisk *, int);
2541 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2542 pgoff_t start, pgoff_t end);
2544 static inline void invalidate_remote_inode(struct inode *inode)
2546 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2547 S_ISLNK(inode->i_mode))
2548 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2550 extern int invalidate_inode_pages2(struct address_space *mapping);
2551 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2552 pgoff_t start, pgoff_t end);
2553 extern int write_inode_now(struct inode *, int);
2554 extern int filemap_fdatawrite(struct address_space *);
2555 extern int filemap_flush(struct address_space *);
2556 extern int filemap_fdatawait(struct address_space *);
2557 extern void filemap_fdatawait_keep_errors(struct address_space *);
2558 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2560 extern int filemap_write_and_wait(struct address_space *mapping);
2561 extern int filemap_write_and_wait_range(struct address_space *mapping,
2562 loff_t lstart, loff_t lend);
2563 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2564 loff_t start, loff_t end, int sync_mode);
2565 extern int filemap_fdatawrite_range(struct address_space *mapping,
2566 loff_t start, loff_t end);
2567 extern int filemap_check_errors(struct address_space *mapping);
2569 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2571 extern int vfs_fsync(struct file *file, int datasync);
2574 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2575 * to already be updated for the write, and will return either the amount
2576 * of bytes passed in, or an error if syncing the file failed.
2578 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2580 if (iocb->ki_flags & IOCB_DSYNC) {
2581 int ret = vfs_fsync_range(iocb->ki_filp,
2582 iocb->ki_pos - count, iocb->ki_pos - 1,
2583 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2591 extern void emergency_sync(void);
2592 extern void emergency_remount(void);
2594 extern sector_t bmap(struct inode *, sector_t);
2596 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2597 extern int inode_permission(struct inode *, int);
2598 extern int __inode_permission(struct inode *, int);
2599 extern int generic_permission(struct inode *, int);
2600 extern int __check_sticky(struct inode *dir, struct inode *inode);
2602 static inline bool execute_ok(struct inode *inode)
2604 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2607 static inline void file_start_write(struct file *file)
2609 if (!S_ISREG(file_inode(file)->i_mode))
2611 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2614 static inline bool file_start_write_trylock(struct file *file)
2616 if (!S_ISREG(file_inode(file)->i_mode))
2618 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2621 static inline void file_end_write(struct file *file)
2623 if (!S_ISREG(file_inode(file)->i_mode))
2625 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2629 * get_write_access() gets write permission for a file.
2630 * put_write_access() releases this write permission.
2631 * This is used for regular files.
2632 * We cannot support write (and maybe mmap read-write shared) accesses and
2633 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2634 * can have the following values:
2635 * 0: no writers, no VM_DENYWRITE mappings
2636 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2637 * > 0: (i_writecount) users are writing to the file.
2639 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2640 * except for the cases where we don't hold i_writecount yet. Then we need to
2641 * use {get,deny}_write_access() - these functions check the sign and refuse
2642 * to do the change if sign is wrong.
2644 static inline int get_write_access(struct inode *inode)
2646 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2648 static inline int deny_write_access(struct file *file)
2650 struct inode *inode = file_inode(file);
2651 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2653 static inline void put_write_access(struct inode * inode)
2655 atomic_dec(&inode->i_writecount);
2657 static inline void allow_write_access(struct file *file)
2660 atomic_inc(&file_inode(file)->i_writecount);
2662 static inline bool inode_is_open_for_write(const struct inode *inode)
2664 return atomic_read(&inode->i_writecount) > 0;
2668 static inline void i_readcount_dec(struct inode *inode)
2670 BUG_ON(!atomic_read(&inode->i_readcount));
2671 atomic_dec(&inode->i_readcount);
2673 static inline void i_readcount_inc(struct inode *inode)
2675 atomic_inc(&inode->i_readcount);
2678 static inline void i_readcount_dec(struct inode *inode)
2682 static inline void i_readcount_inc(struct inode *inode)
2687 extern int do_pipe_flags(int *, int);
2689 #define __kernel_read_file_id(id) \
2690 id(UNKNOWN, unknown) \
2691 id(FIRMWARE, firmware) \
2692 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2693 id(MODULE, kernel-module) \
2694 id(KEXEC_IMAGE, kexec-image) \
2695 id(KEXEC_INITRAMFS, kexec-initramfs) \
2696 id(POLICY, security-policy) \
2699 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2700 #define __fid_stringify(dummy, str) #str,
2702 enum kernel_read_file_id {
2703 __kernel_read_file_id(__fid_enumify)
2706 static const char * const kernel_read_file_str[] = {
2707 __kernel_read_file_id(__fid_stringify)
2710 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2712 if (id < 0 || id >= READING_MAX_ID)
2713 return kernel_read_file_str[READING_UNKNOWN];
2715 return kernel_read_file_str[id];
2718 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2719 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2720 enum kernel_read_file_id);
2721 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2722 enum kernel_read_file_id);
2723 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2724 enum kernel_read_file_id);
2725 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2726 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2727 extern struct file * open_exec(const char *);
2729 /* fs/dcache.c -- generic fs support functions */
2730 extern bool is_subdir(struct dentry *, struct dentry *);
2731 extern bool path_is_under(struct path *, struct path *);
2733 extern char *file_path(struct file *, char *, int);
2735 #include <linux/err.h>
2737 /* needed for stackable file system support */
2738 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2740 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2742 extern int inode_init_always(struct super_block *, struct inode *);
2743 extern void inode_init_once(struct inode *);
2744 extern void address_space_init_once(struct address_space *mapping);
2745 extern struct inode * igrab(struct inode *);
2746 extern ino_t iunique(struct super_block *, ino_t);
2747 extern int inode_needs_sync(struct inode *inode);
2748 extern int generic_delete_inode(struct inode *inode);
2749 static inline int generic_drop_inode(struct inode *inode)
2751 return !inode->i_nlink || inode_unhashed(inode);
2754 extern struct inode *ilookup5_nowait(struct super_block *sb,
2755 unsigned long hashval, int (*test)(struct inode *, void *),
2757 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2758 int (*test)(struct inode *, void *), void *data);
2759 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2761 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2762 extern struct inode * iget_locked(struct super_block *, unsigned long);
2763 extern struct inode *find_inode_nowait(struct super_block *,
2765 int (*match)(struct inode *,
2766 unsigned long, void *),
2768 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2769 extern int insert_inode_locked(struct inode *);
2770 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2771 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2773 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2775 extern void unlock_new_inode(struct inode *);
2776 extern unsigned int get_next_ino(void);
2777 extern void evict_inodes(struct super_block *sb);
2779 extern void __iget(struct inode * inode);
2780 extern void iget_failed(struct inode *);
2781 extern void clear_inode(struct inode *);
2782 extern void __destroy_inode(struct inode *);
2783 extern struct inode *new_inode_pseudo(struct super_block *sb);
2784 extern struct inode *new_inode(struct super_block *sb);
2785 extern void free_inode_nonrcu(struct inode *inode);
2786 extern int should_remove_suid(struct dentry *);
2787 extern int file_remove_privs(struct file *);
2789 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2790 static inline void insert_inode_hash(struct inode *inode)
2792 __insert_inode_hash(inode, inode->i_ino);
2795 extern void __remove_inode_hash(struct inode *);
2796 static inline void remove_inode_hash(struct inode *inode)
2798 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2799 __remove_inode_hash(inode);
2802 extern void inode_sb_list_add(struct inode *inode);
2805 extern blk_qc_t submit_bio(struct bio *);
2806 extern int bdev_read_only(struct block_device *);
2808 extern int set_blocksize(struct block_device *, int);
2809 extern int sb_set_blocksize(struct super_block *, int);
2810 extern int sb_min_blocksize(struct super_block *, int);
2812 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2813 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2814 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2815 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2816 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2817 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2818 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2819 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2821 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2822 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2824 /* fs/block_dev.c */
2825 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2826 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2827 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2829 extern void block_sync_page(struct page *page);
2832 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2833 struct pipe_inode_info *, size_t, unsigned int);
2834 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2835 struct file *, loff_t *, size_t, unsigned int);
2836 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2837 struct file *out, loff_t *, size_t len, unsigned int flags);
2838 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2839 loff_t *opos, size_t len, unsigned int flags);
2843 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2844 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2845 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2846 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2847 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2848 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2849 int whence, loff_t maxsize, loff_t eof);
2850 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2851 int whence, loff_t size);
2852 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2853 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2854 extern int generic_file_open(struct inode * inode, struct file * filp);
2855 extern int nonseekable_open(struct inode * inode, struct file * filp);
2856 extern int stream_open(struct inode * inode, struct file * filp);
2859 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2860 loff_t file_offset);
2863 /* need locking between buffered and direct access */
2866 /* filesystem does not support filling holes */
2867 DIO_SKIP_HOLES = 0x02,
2869 /* filesystem can handle aio writes beyond i_size */
2870 DIO_ASYNC_EXTEND = 0x04,
2872 /* inode/fs/bdev does not need truncate protection */
2873 DIO_SKIP_DIO_COUNT = 0x08,
2876 void dio_end_io(struct bio *bio, int error);
2878 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2879 struct block_device *bdev, struct iov_iter *iter,
2880 get_block_t get_block,
2881 dio_iodone_t end_io, dio_submit_t submit_io,
2884 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2885 struct inode *inode,
2886 struct iov_iter *iter,
2887 get_block_t get_block)
2889 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2890 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2894 void inode_dio_wait(struct inode *inode);
2897 * inode_dio_begin - signal start of a direct I/O requests
2898 * @inode: inode the direct I/O happens on
2900 * This is called once we've finished processing a direct I/O request,
2901 * and is used to wake up callers waiting for direct I/O to be quiesced.
2903 static inline void inode_dio_begin(struct inode *inode)
2905 atomic_inc(&inode->i_dio_count);
2909 * inode_dio_end - signal finish of a direct I/O requests
2910 * @inode: inode the direct I/O happens on
2912 * This is called once we've finished processing a direct I/O request,
2913 * and is used to wake up callers waiting for direct I/O to be quiesced.
2915 static inline void inode_dio_end(struct inode *inode)
2917 if (atomic_dec_and_test(&inode->i_dio_count))
2918 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2921 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2924 extern const struct file_operations generic_ro_fops;
2926 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2928 extern int readlink_copy(char __user *, int, const char *);
2929 extern int page_readlink(struct dentry *, char __user *, int);
2930 extern const char *page_get_link(struct dentry *, struct inode *,
2931 struct delayed_call *);
2932 extern void page_put_link(void *);
2933 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2935 extern int page_symlink(struct inode *inode, const char *symname, int len);
2936 extern const struct inode_operations page_symlink_inode_operations;
2937 extern void kfree_link(void *);
2938 extern int generic_readlink(struct dentry *, char __user *, int);
2939 extern void generic_fillattr(struct inode *, struct kstat *);
2940 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2941 extern int vfs_getattr(struct path *, struct kstat *);
2942 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2943 void inode_add_bytes(struct inode *inode, loff_t bytes);
2944 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2945 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2946 loff_t inode_get_bytes(struct inode *inode);
2947 void inode_set_bytes(struct inode *inode, loff_t bytes);
2948 const char *simple_get_link(struct dentry *, struct inode *,
2949 struct delayed_call *);
2950 extern const struct inode_operations simple_symlink_inode_operations;
2952 extern int iterate_dir(struct file *, struct dir_context *);
2954 extern int vfs_stat(const char __user *, struct kstat *);
2955 extern int vfs_lstat(const char __user *, struct kstat *);
2956 extern int vfs_fstat(unsigned int, struct kstat *);
2957 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2958 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
2960 extern int __generic_block_fiemap(struct inode *inode,
2961 struct fiemap_extent_info *fieinfo,
2962 loff_t start, loff_t len,
2963 get_block_t *get_block);
2964 extern int generic_block_fiemap(struct inode *inode,
2965 struct fiemap_extent_info *fieinfo, u64 start,
2966 u64 len, get_block_t *get_block);
2968 extern void get_filesystem(struct file_system_type *fs);
2969 extern void put_filesystem(struct file_system_type *fs);
2970 extern struct file_system_type *get_fs_type(const char *name);
2971 extern struct super_block *get_super(struct block_device *);
2972 extern struct super_block *get_super_thawed(struct block_device *);
2973 extern struct super_block *get_active_super(struct block_device *bdev);
2974 extern void drop_super(struct super_block *sb);
2975 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2976 extern void iterate_supers_type(struct file_system_type *,
2977 void (*)(struct super_block *, void *), void *);
2979 extern int dcache_dir_open(struct inode *, struct file *);
2980 extern int dcache_dir_close(struct inode *, struct file *);
2981 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2982 extern int dcache_readdir(struct file *, struct dir_context *);
2983 extern int simple_setattr(struct dentry *, struct iattr *);
2984 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2985 extern int simple_statfs(struct dentry *, struct kstatfs *);
2986 extern int simple_open(struct inode *inode, struct file *file);
2987 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2988 extern int simple_unlink(struct inode *, struct dentry *);
2989 extern int simple_rmdir(struct inode *, struct dentry *);
2990 extern int simple_rename(struct inode *, struct dentry *,
2991 struct inode *, struct dentry *, unsigned int);
2992 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2993 extern int simple_empty(struct dentry *);
2994 extern int simple_readpage(struct file *file, struct page *page);
2995 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2996 loff_t pos, unsigned len, unsigned flags,
2997 struct page **pagep, void **fsdata);
2998 extern int simple_write_end(struct file *file, struct address_space *mapping,
2999 loff_t pos, unsigned len, unsigned copied,
3000 struct page *page, void *fsdata);
3001 extern int always_delete_dentry(const struct dentry *);
3002 extern struct inode *alloc_anon_inode(struct super_block *);
3003 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3004 extern const struct dentry_operations simple_dentry_operations;
3006 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3007 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3008 extern const struct file_operations simple_dir_operations;
3009 extern const struct inode_operations simple_dir_inode_operations;
3010 extern void make_empty_dir_inode(struct inode *inode);
3011 extern bool is_empty_dir_inode(struct inode *inode);
3012 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
3013 struct dentry *d_alloc_name(struct dentry *, const char *);
3014 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
3015 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3016 extern void simple_release_fs(struct vfsmount **mount, int *count);
3018 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3019 loff_t *ppos, const void *from, size_t available);
3020 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3021 const void __user *from, size_t count);
3023 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3024 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3026 extern int generic_check_addressable(unsigned, u64);
3028 #ifdef CONFIG_MIGRATION
3029 extern int buffer_migrate_page(struct address_space *,
3030 struct page *, struct page *,
3033 #define buffer_migrate_page NULL
3036 extern int setattr_prepare(struct dentry *, struct iattr *);
3037 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3038 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3040 extern int file_update_time(struct file *file);
3042 extern int generic_show_options(struct seq_file *m, struct dentry *root);
3043 extern void save_mount_options(struct super_block *sb, char *options);
3044 extern void replace_mount_options(struct super_block *sb, char *options);
3046 static inline bool io_is_direct(struct file *filp)
3048 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3051 static inline bool vma_is_dax(struct vm_area_struct *vma)
3053 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3056 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3058 struct inode *inode;
3062 if (!vma_is_dax(vma))
3064 inode = file_inode(vma->vm_file);
3065 if (S_ISCHR(inode->i_mode))
3066 return false; /* device-dax */
3070 static inline int iocb_flags(struct file *file)
3073 if (file->f_flags & O_APPEND)
3075 if (io_is_direct(file))
3077 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3079 if (file->f_flags & __O_SYNC)
3084 static inline ino_t parent_ino(struct dentry *dentry)
3089 * Don't strictly need d_lock here? If the parent ino could change
3090 * then surely we'd have a deeper race in the caller?
3092 spin_lock(&dentry->d_lock);
3093 res = dentry->d_parent->d_inode->i_ino;
3094 spin_unlock(&dentry->d_lock);
3098 /* Transaction based IO helpers */
3101 * An argresp is stored in an allocated page and holds the
3102 * size of the argument or response, along with its content
3104 struct simple_transaction_argresp {
3109 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3111 char *simple_transaction_get(struct file *file, const char __user *buf,
3113 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3114 size_t size, loff_t *pos);
3115 int simple_transaction_release(struct inode *inode, struct file *file);
3117 void simple_transaction_set(struct file *file, size_t n);
3120 * simple attribute files
3122 * These attributes behave similar to those in sysfs:
3124 * Writing to an attribute immediately sets a value, an open file can be
3125 * written to multiple times.
3127 * Reading from an attribute creates a buffer from the value that might get
3128 * read with multiple read calls. When the attribute has been read
3129 * completely, no further read calls are possible until the file is opened
3132 * All attributes contain a text representation of a numeric value
3133 * that are accessed with the get() and set() functions.
3135 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3136 static int __fops ## _open(struct inode *inode, struct file *file) \
3138 __simple_attr_check_format(__fmt, 0ull); \
3139 return simple_attr_open(inode, file, __get, __set, __fmt); \
3141 static const struct file_operations __fops = { \
3142 .owner = THIS_MODULE, \
3143 .open = __fops ## _open, \
3144 .release = simple_attr_release, \
3145 .read = simple_attr_read, \
3146 .write = simple_attr_write, \
3147 .llseek = generic_file_llseek, \
3150 static inline __printf(1, 2)
3151 void __simple_attr_check_format(const char *fmt, ...)
3153 /* don't do anything, just let the compiler check the arguments; */
3156 int simple_attr_open(struct inode *inode, struct file *file,
3157 int (*get)(void *, u64 *), int (*set)(void *, u64),
3159 int simple_attr_release(struct inode *inode, struct file *file);
3160 ssize_t simple_attr_read(struct file *file, char __user *buf,
3161 size_t len, loff_t *ppos);
3162 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3163 size_t len, loff_t *ppos);
3166 int proc_nr_files(struct ctl_table *table, int write,
3167 void __user *buffer, size_t *lenp, loff_t *ppos);
3168 int proc_nr_dentry(struct ctl_table *table, int write,
3169 void __user *buffer, size_t *lenp, loff_t *ppos);
3170 int proc_nr_inodes(struct ctl_table *table, int write,
3171 void __user *buffer, size_t *lenp, loff_t *ppos);
3172 int __init get_filesystem_list(char *buf);
3174 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3175 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3177 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3178 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3179 (flag & __FMODE_NONOTIFY)))
3181 static inline bool is_sxid(umode_t mode)
3183 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3186 static inline int check_sticky(struct inode *dir, struct inode *inode)
3188 if (!(dir->i_mode & S_ISVTX))
3191 return __check_sticky(dir, inode);
3194 static inline void inode_has_no_xattr(struct inode *inode)
3196 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3197 inode->i_flags |= S_NOSEC;
3200 static inline bool is_root_inode(struct inode *inode)
3202 return inode == inode->i_sb->s_root->d_inode;
3205 static inline bool dir_emit(struct dir_context *ctx,
3206 const char *name, int namelen,
3207 u64 ino, unsigned type)
3209 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3211 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3213 return ctx->actor(ctx, ".", 1, ctx->pos,
3214 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3216 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3218 return ctx->actor(ctx, "..", 2, ctx->pos,
3219 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3221 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3223 if (ctx->pos == 0) {
3224 if (!dir_emit_dot(file, ctx))
3228 if (ctx->pos == 1) {
3229 if (!dir_emit_dotdot(file, ctx))
3235 static inline bool dir_relax(struct inode *inode)
3237 inode_unlock(inode);
3239 return !IS_DEADDIR(inode);
3242 static inline bool dir_relax_shared(struct inode *inode)
3244 inode_unlock_shared(inode);
3245 inode_lock_shared(inode);
3246 return !IS_DEADDIR(inode);
3249 extern bool path_noexec(const struct path *path);
3250 extern void inode_nohighmem(struct inode *inode);
3252 #endif /* _LINUX_FS_H */