1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
41 #include <asm/byteorder.h>
42 #include <uapi/linux/fs.h>
44 struct backing_dev_info;
47 struct export_operations;
52 struct pipe_inode_info;
53 struct poll_table_struct;
55 struct vm_area_struct;
58 struct swap_info_struct;
60 struct workqueue_struct;
63 struct fscrypt_operations;
65 extern void __init inode_init(void);
66 extern void __init inode_init_early(void);
67 extern void __init files_init(void);
68 extern void __init files_maxfiles_init(void);
70 extern struct files_stat_struct files_stat;
71 extern unsigned long get_max_files(void);
72 extern unsigned int sysctl_nr_open;
73 extern struct inodes_stat_t inodes_stat;
74 extern int leases_enable, lease_break_time;
75 extern int sysctl_protected_symlinks;
76 extern int sysctl_protected_hardlinks;
77 extern int sysctl_protected_fifos;
78 extern int sysctl_protected_regular;
80 typedef __kernel_rwf_t rwf_t;
83 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
84 struct buffer_head *bh_result, int create);
85 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
86 ssize_t bytes, void *private);
88 #define MAY_EXEC 0x00000001
89 #define MAY_WRITE 0x00000002
90 #define MAY_READ 0x00000004
91 #define MAY_APPEND 0x00000008
92 #define MAY_ACCESS 0x00000010
93 #define MAY_OPEN 0x00000020
94 #define MAY_CHDIR 0x00000040
95 /* called from RCU mode, don't block */
96 #define MAY_NOT_BLOCK 0x00000080
99 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
100 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
103 /* file is open for reading */
104 #define FMODE_READ ((__force fmode_t)0x1)
105 /* file is open for writing */
106 #define FMODE_WRITE ((__force fmode_t)0x2)
107 /* file is seekable */
108 #define FMODE_LSEEK ((__force fmode_t)0x4)
109 /* file can be accessed using pread */
110 #define FMODE_PREAD ((__force fmode_t)0x8)
111 /* file can be accessed using pwrite */
112 #define FMODE_PWRITE ((__force fmode_t)0x10)
113 /* File is opened for execution with sys_execve / sys_uselib */
114 #define FMODE_EXEC ((__force fmode_t)0x20)
115 /* File is opened with O_NDELAY (only set for block devices) */
116 #define FMODE_NDELAY ((__force fmode_t)0x40)
117 /* File is opened with O_EXCL (only set for block devices) */
118 #define FMODE_EXCL ((__force fmode_t)0x80)
119 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
120 (specialy hack for floppy.c) */
121 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
122 /* 32bit hashes as llseek() offset (for directories) */
123 #define FMODE_32BITHASH ((__force fmode_t)0x200)
124 /* 64bit hashes as llseek() offset (for directories) */
125 #define FMODE_64BITHASH ((__force fmode_t)0x400)
128 * Don't update ctime and mtime.
130 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
131 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
133 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
135 /* Expect random access pattern */
136 #define FMODE_RANDOM ((__force fmode_t)0x1000)
138 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
139 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
141 /* File is opened with O_PATH; almost nothing can be done with it */
142 #define FMODE_PATH ((__force fmode_t)0x4000)
144 /* File needs atomic accesses to f_pos */
145 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
146 /* Write access to underlying fs */
147 #define FMODE_WRITER ((__force fmode_t)0x10000)
148 /* Has read method(s) */
149 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
150 /* Has write method(s) */
151 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
153 #define FMODE_OPENED ((__force fmode_t)0x80000)
154 #define FMODE_CREATED ((__force fmode_t)0x100000)
156 /* File was opened by fanotify and shouldn't generate fanotify events */
157 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
159 /* File is capable of returning -EAGAIN if I/O will block */
160 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
162 /* File does not contribute to nr_files count */
163 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
166 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
167 * that indicates that they should check the contents of the iovec are
168 * valid, but not check the memory that the iovec elements
171 #define CHECK_IOVEC_ONLY -1
174 * Attribute flags. These should be or-ed together to figure out what
177 #define ATTR_MODE (1 << 0)
178 #define ATTR_UID (1 << 1)
179 #define ATTR_GID (1 << 2)
180 #define ATTR_SIZE (1 << 3)
181 #define ATTR_ATIME (1 << 4)
182 #define ATTR_MTIME (1 << 5)
183 #define ATTR_CTIME (1 << 6)
184 #define ATTR_ATIME_SET (1 << 7)
185 #define ATTR_MTIME_SET (1 << 8)
186 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
187 #define ATTR_KILL_SUID (1 << 11)
188 #define ATTR_KILL_SGID (1 << 12)
189 #define ATTR_FILE (1 << 13)
190 #define ATTR_KILL_PRIV (1 << 14)
191 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
192 #define ATTR_TIMES_SET (1 << 16)
193 #define ATTR_TOUCH (1 << 17)
196 * Whiteout is represented by a char device. The following constants define the
197 * mode and device number to use.
199 #define WHITEOUT_MODE 0
200 #define WHITEOUT_DEV 0
203 * This is the Inode Attributes structure, used for notify_change(). It
204 * uses the above definitions as flags, to know which values have changed.
205 * Also, in this manner, a Filesystem can look at only the values it cares
206 * about. Basically, these are the attributes that the VFS layer can
207 * request to change from the FS layer.
209 * Derek Atkins <warlord@MIT.EDU> 94-10-20
212 unsigned int ia_valid;
217 struct timespec64 ia_atime;
218 struct timespec64 ia_mtime;
219 struct timespec64 ia_ctime;
222 * Not an attribute, but an auxiliary info for filesystems wanting to
223 * implement an ftruncate() like method. NOTE: filesystem should
224 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
226 struct file *ia_file;
230 * Includes for diskquotas.
232 #include <linux/quota.h>
235 * Maximum number of layers of fs stack. Needs to be limited to
236 * prevent kernel stack overflow
238 #define FILESYSTEM_MAX_STACK_DEPTH 2
241 * enum positive_aop_returns - aop return codes with specific semantics
243 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
244 * completed, that the page is still locked, and
245 * should be considered active. The VM uses this hint
246 * to return the page to the active list -- it won't
247 * be a candidate for writeback again in the near
248 * future. Other callers must be careful to unlock
249 * the page if they get this return. Returned by
252 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
253 * unlocked it and the page might have been truncated.
254 * The caller should back up to acquiring a new page and
255 * trying again. The aop will be taking reasonable
256 * precautions not to livelock. If the caller held a page
257 * reference, it should drop it before retrying. Returned
260 * address_space_operation functions return these large constants to indicate
261 * special semantics to the caller. These are much larger than the bytes in a
262 * page to allow for functions that return the number of bytes operated on in a
266 enum positive_aop_returns {
267 AOP_WRITEPAGE_ACTIVATE = 0x80000,
268 AOP_TRUNCATED_PAGE = 0x80001,
271 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
272 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
273 * helper code (eg buffer layer)
274 * to clear GFP_FS from alloc */
277 * oh the beauties of C type declarations.
280 struct address_space;
281 struct writeback_control;
284 * Write life time hint values.
285 * Stored in struct inode as u8.
288 WRITE_LIFE_NOT_SET = 0,
289 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
290 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
291 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
292 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
293 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
296 #define IOCB_EVENTFD (1 << 0)
297 #define IOCB_APPEND (1 << 1)
298 #define IOCB_DIRECT (1 << 2)
299 #define IOCB_HIPRI (1 << 3)
300 #define IOCB_DSYNC (1 << 4)
301 #define IOCB_SYNC (1 << 5)
302 #define IOCB_WRITE (1 << 6)
303 #define IOCB_NOWAIT (1 << 7)
306 struct file *ki_filp;
308 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
312 u16 ki_ioprio; /* See linux/ioprio.h */
313 } __randomize_layout;
315 static inline bool is_sync_kiocb(struct kiocb *kiocb)
317 return kiocb->ki_complete == NULL;
321 * "descriptor" for what we're up to with a read.
322 * This allows us to use the same read code yet
323 * have multiple different users of the data that
324 * we read from a file.
326 * The simplest case just copies the data to user
339 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
340 unsigned long, unsigned long);
342 struct address_space_operations {
343 int (*writepage)(struct page *page, struct writeback_control *wbc);
344 int (*readpage)(struct file *, struct page *);
346 /* Write back some dirty pages from this mapping. */
347 int (*writepages)(struct address_space *, struct writeback_control *);
349 /* Set a page dirty. Return true if this dirtied it */
350 int (*set_page_dirty)(struct page *page);
353 * Reads in the requested pages. Unlike ->readpage(), this is
354 * PURELY used for read-ahead!.
356 int (*readpages)(struct file *filp, struct address_space *mapping,
357 struct list_head *pages, unsigned nr_pages);
359 int (*write_begin)(struct file *, struct address_space *mapping,
360 loff_t pos, unsigned len, unsigned flags,
361 struct page **pagep, void **fsdata);
362 int (*write_end)(struct file *, struct address_space *mapping,
363 loff_t pos, unsigned len, unsigned copied,
364 struct page *page, void *fsdata);
366 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
367 sector_t (*bmap)(struct address_space *, sector_t);
368 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
369 int (*releasepage) (struct page *, gfp_t);
370 void (*freepage)(struct page *);
371 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
373 * migrate the contents of a page to the specified target. If
374 * migrate_mode is MIGRATE_ASYNC, it must not block.
376 int (*migratepage) (struct address_space *,
377 struct page *, struct page *, enum migrate_mode);
378 bool (*isolate_page)(struct page *, isolate_mode_t);
379 void (*putback_page)(struct page *);
380 int (*launder_page) (struct page *);
381 int (*is_partially_uptodate) (struct page *, unsigned long,
383 void (*is_dirty_writeback) (struct page *, bool *, bool *);
384 int (*error_remove_page)(struct address_space *, struct page *);
386 /* swapfile support */
387 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
389 void (*swap_deactivate)(struct file *file);
392 extern const struct address_space_operations empty_aops;
395 * pagecache_write_begin/pagecache_write_end must be used by general code
396 * to write into the pagecache.
398 int pagecache_write_begin(struct file *, struct address_space *mapping,
399 loff_t pos, unsigned len, unsigned flags,
400 struct page **pagep, void **fsdata);
402 int pagecache_write_end(struct file *, struct address_space *mapping,
403 loff_t pos, unsigned len, unsigned copied,
404 struct page *page, void *fsdata);
406 struct address_space {
407 struct inode *host; /* owner: inode, block_device */
408 struct radix_tree_root i_pages; /* cached pages */
409 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
410 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
411 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
412 /* Protected by the i_pages lock */
413 unsigned long nrpages; /* number of total pages */
414 /* number of shadow or DAX exceptional entries */
415 unsigned long nrexceptional;
416 pgoff_t writeback_index;/* writeback starts here */
417 const struct address_space_operations *a_ops; /* methods */
418 unsigned long flags; /* error bits */
419 spinlock_t private_lock; /* for use by the address_space */
420 gfp_t gfp_mask; /* implicit gfp mask for allocations */
421 struct list_head private_list; /* for use by the address_space */
422 void *private_data; /* ditto */
424 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
426 * On most architectures that alignment is already the case; but
427 * must be enforced here for CRIS, to let the least significant bit
428 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
430 struct request_queue;
432 struct block_device {
433 dev_t bd_dev; /* not a kdev_t - it's a search key */
435 struct inode * bd_inode; /* will die */
436 struct super_block * bd_super;
437 struct mutex bd_mutex; /* open/close mutex */
441 bool bd_write_holder;
443 struct list_head bd_holder_disks;
445 struct block_device * bd_contains;
446 unsigned bd_block_size;
448 struct hd_struct * bd_part;
449 /* number of times partitions within this device have been opened. */
450 unsigned bd_part_count;
452 struct gendisk * bd_disk;
453 struct request_queue * bd_queue;
454 struct backing_dev_info *bd_bdi;
455 struct list_head bd_list;
457 * Private data. You must have bd_claim'ed the block_device
458 * to use this. NOTE: bd_claim allows an owner to claim
459 * the same device multiple times, the owner must take special
460 * care to not mess up bd_private for that case.
462 unsigned long bd_private;
464 /* The counter of freeze processes */
465 int bd_fsfreeze_count;
466 /* Mutex for freeze */
467 struct mutex bd_fsfreeze_mutex;
468 } __randomize_layout;
471 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
474 #define PAGECACHE_TAG_DIRTY 0
475 #define PAGECACHE_TAG_WRITEBACK 1
476 #define PAGECACHE_TAG_TOWRITE 2
478 int mapping_tagged(struct address_space *mapping, int tag);
480 static inline void i_mmap_lock_write(struct address_space *mapping)
482 down_write(&mapping->i_mmap_rwsem);
485 static inline void i_mmap_unlock_write(struct address_space *mapping)
487 up_write(&mapping->i_mmap_rwsem);
490 static inline void i_mmap_lock_read(struct address_space *mapping)
492 down_read(&mapping->i_mmap_rwsem);
495 static inline void i_mmap_unlock_read(struct address_space *mapping)
497 up_read(&mapping->i_mmap_rwsem);
501 * Might pages of this file be mapped into userspace?
503 static inline int mapping_mapped(struct address_space *mapping)
505 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
509 * Might pages of this file have been modified in userspace?
510 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
511 * marks vma as VM_SHARED if it is shared, and the file was opened for
512 * writing i.e. vma may be mprotected writable even if now readonly.
514 * If i_mmap_writable is negative, no new writable mappings are allowed. You
515 * can only deny writable mappings, if none exists right now.
517 static inline int mapping_writably_mapped(struct address_space *mapping)
519 return atomic_read(&mapping->i_mmap_writable) > 0;
522 static inline int mapping_map_writable(struct address_space *mapping)
524 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
528 static inline void mapping_unmap_writable(struct address_space *mapping)
530 atomic_dec(&mapping->i_mmap_writable);
533 static inline int mapping_deny_writable(struct address_space *mapping)
535 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
539 static inline void mapping_allow_writable(struct address_space *mapping)
541 atomic_inc(&mapping->i_mmap_writable);
545 * Use sequence counter to get consistent i_size on 32-bit processors.
547 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
548 #include <linux/seqlock.h>
549 #define __NEED_I_SIZE_ORDERED
550 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
552 #define i_size_ordered_init(inode) do { } while (0)
556 #define ACL_NOT_CACHED ((void *)(-1))
557 #define ACL_DONT_CACHE ((void *)(-3))
559 static inline struct posix_acl *
560 uncached_acl_sentinel(struct task_struct *task)
562 return (void *)task + 1;
566 is_uncached_acl(struct posix_acl *acl)
568 return (long)acl & 1;
571 #define IOP_FASTPERM 0x0001
572 #define IOP_LOOKUP 0x0002
573 #define IOP_NOFOLLOW 0x0004
574 #define IOP_XATTR 0x0008
575 #define IOP_DEFAULT_READLINK 0x0010
577 struct fsnotify_mark_connector;
580 * Keep mostly read-only and often accessed (especially for
581 * the RCU path lookup and 'stat' data) fields at the beginning
582 * of the 'struct inode'
586 unsigned short i_opflags;
589 unsigned int i_flags;
591 #ifdef CONFIG_FS_POSIX_ACL
592 struct posix_acl *i_acl;
593 struct posix_acl *i_default_acl;
596 const struct inode_operations *i_op;
597 struct super_block *i_sb;
598 struct address_space *i_mapping;
600 #ifdef CONFIG_SECURITY
604 /* Stat data, not accessed from path walking */
607 * Filesystems may only read i_nlink directly. They shall use the
608 * following functions for modification:
610 * (set|clear|inc|drop)_nlink
611 * inode_(inc|dec)_link_count
614 const unsigned int i_nlink;
615 unsigned int __i_nlink;
619 struct timespec64 i_atime;
620 struct timespec64 i_mtime;
621 struct timespec64 i_ctime;
622 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
623 unsigned short i_bytes;
628 #ifdef __NEED_I_SIZE_ORDERED
629 seqcount_t i_size_seqcount;
633 unsigned long i_state;
634 struct rw_semaphore i_rwsem;
636 unsigned long dirtied_when; /* jiffies of first dirtying */
637 unsigned long dirtied_time_when;
639 struct hlist_node i_hash;
640 struct list_head i_io_list; /* backing dev IO list */
641 #ifdef CONFIG_CGROUP_WRITEBACK
642 struct bdi_writeback *i_wb; /* the associated cgroup wb */
644 /* foreign inode detection, see wbc_detach_inode() */
646 u16 i_wb_frn_avg_time;
647 u16 i_wb_frn_history;
649 struct list_head i_lru; /* inode LRU list */
650 struct list_head i_sb_list;
651 struct list_head i_wb_list; /* backing dev writeback list */
653 struct hlist_head i_dentry;
654 struct rcu_head i_rcu;
656 atomic64_t i_version;
658 atomic_t i_dio_count;
659 atomic_t i_writecount;
661 atomic_t i_readcount; /* struct files open RO */
663 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
664 struct file_lock_context *i_flctx;
665 struct address_space i_data;
666 struct list_head i_devices;
668 struct pipe_inode_info *i_pipe;
669 struct block_device *i_bdev;
677 #ifdef CONFIG_FSNOTIFY
678 __u32 i_fsnotify_mask; /* all events this inode cares about */
679 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
682 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
683 struct fscrypt_info *i_crypt_info;
686 void *i_private; /* fs or device private pointer */
687 } __randomize_layout;
689 static inline unsigned int i_blocksize(const struct inode *node)
691 return (1 << node->i_blkbits);
694 static inline int inode_unhashed(struct inode *inode)
696 return hlist_unhashed(&inode->i_hash);
700 * __mark_inode_dirty expects inodes to be hashed. Since we don't
701 * want special inodes in the fileset inode space, we make them
702 * appear hashed, but do not put on any lists. hlist_del()
703 * will work fine and require no locking.
705 static inline void inode_fake_hash(struct inode *inode)
707 hlist_add_fake(&inode->i_hash);
711 * inode->i_mutex nesting subclasses for the lock validator:
713 * 0: the object of the current VFS operation
717 * 4: second non-directory
718 * 5: second parent (when locking independent directories in rename)
720 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
721 * non-directories at once.
723 * The locking order between these classes is
724 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
726 enum inode_i_mutex_lock_class
736 static inline void inode_lock(struct inode *inode)
738 down_write(&inode->i_rwsem);
741 static inline void inode_unlock(struct inode *inode)
743 up_write(&inode->i_rwsem);
746 static inline void inode_lock_shared(struct inode *inode)
748 down_read(&inode->i_rwsem);
751 static inline void inode_unlock_shared(struct inode *inode)
753 up_read(&inode->i_rwsem);
756 static inline int inode_trylock(struct inode *inode)
758 return down_write_trylock(&inode->i_rwsem);
761 static inline int inode_trylock_shared(struct inode *inode)
763 return down_read_trylock(&inode->i_rwsem);
766 static inline int inode_is_locked(struct inode *inode)
768 return rwsem_is_locked(&inode->i_rwsem);
771 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
773 down_write_nested(&inode->i_rwsem, subclass);
776 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
778 down_read_nested(&inode->i_rwsem, subclass);
781 void lock_two_nondirectories(struct inode *, struct inode*);
782 void unlock_two_nondirectories(struct inode *, struct inode*);
785 * NOTE: in a 32bit arch with a preemptable kernel and
786 * an UP compile the i_size_read/write must be atomic
787 * with respect to the local cpu (unlike with preempt disabled),
788 * but they don't need to be atomic with respect to other cpus like in
789 * true SMP (so they need either to either locally disable irq around
790 * the read or for example on x86 they can be still implemented as a
791 * cmpxchg8b without the need of the lock prefix). For SMP compiles
792 * and 64bit archs it makes no difference if preempt is enabled or not.
794 static inline loff_t i_size_read(const struct inode *inode)
796 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
801 seq = read_seqcount_begin(&inode->i_size_seqcount);
802 i_size = inode->i_size;
803 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
805 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
809 i_size = inode->i_size;
813 return inode->i_size;
818 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
819 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
820 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
822 static inline void i_size_write(struct inode *inode, loff_t i_size)
824 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
826 write_seqcount_begin(&inode->i_size_seqcount);
827 inode->i_size = i_size;
828 write_seqcount_end(&inode->i_size_seqcount);
830 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
832 inode->i_size = i_size;
835 inode->i_size = i_size;
839 static inline unsigned iminor(const struct inode *inode)
841 return MINOR(inode->i_rdev);
844 static inline unsigned imajor(const struct inode *inode)
846 return MAJOR(inode->i_rdev);
849 extern struct block_device *I_BDEV(struct inode *inode);
852 rwlock_t lock; /* protects pid, uid, euid fields */
853 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
854 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
855 kuid_t uid, euid; /* uid/euid of process setting the owner */
856 int signum; /* posix.1b rt signal to be delivered on IO */
860 * Track a single file's readahead state
862 struct file_ra_state {
863 pgoff_t start; /* where readahead started */
864 unsigned int size; /* # of readahead pages */
865 unsigned int async_size; /* do asynchronous readahead when
866 there are only # of pages ahead */
868 unsigned int ra_pages; /* Maximum readahead window */
869 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
870 loff_t prev_pos; /* Cache last read() position */
874 * Check if @index falls in the readahead windows.
876 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
878 return (index >= ra->start &&
879 index < ra->start + ra->size);
884 struct llist_node fu_llist;
885 struct rcu_head fu_rcuhead;
888 struct inode *f_inode; /* cached value */
889 const struct file_operations *f_op;
892 * Protects f_ep_links, f_flags.
893 * Must not be taken from IRQ context.
896 enum rw_hint f_write_hint;
897 atomic_long_t f_count;
898 unsigned int f_flags;
900 struct mutex f_pos_lock;
902 struct fown_struct f_owner;
903 const struct cred *f_cred;
904 struct file_ra_state f_ra;
907 #ifdef CONFIG_SECURITY
910 /* needed for tty driver, and maybe others */
914 /* Used by fs/eventpoll.c to link all the hooks to this file */
915 struct list_head f_ep_links;
916 struct list_head f_tfile_llink;
917 #endif /* #ifdef CONFIG_EPOLL */
918 struct address_space *f_mapping;
921 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
926 /* file identifier */
927 unsigned char f_handle[0];
930 static inline struct file *get_file(struct file *f)
932 atomic_long_inc(&f->f_count);
935 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
936 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
937 #define file_count(x) atomic_long_read(&(x)->f_count)
939 #define MAX_NON_LFS ((1UL<<31) - 1)
941 /* Page cache limit. The filesystems should put that into their s_maxbytes
942 limits, otherwise bad things can happen in VM. */
943 #if BITS_PER_LONG==32
944 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
945 #elif BITS_PER_LONG==64
946 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
951 #define FL_DELEG 4 /* NFSv4 delegation */
952 #define FL_ACCESS 8 /* not trying to lock, just looking */
953 #define FL_EXISTS 16 /* when unlocking, test for existence */
954 #define FL_LEASE 32 /* lease held on this file */
955 #define FL_CLOSE 64 /* unlock on close */
956 #define FL_SLEEP 128 /* A blocking lock */
957 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
958 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
959 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
960 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
962 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
965 * Special return value from posix_lock_file() and vfs_lock_file() for
966 * asynchronous locking.
968 #define FILE_LOCK_DEFERRED 1
970 /* legacy typedef, should eventually be removed */
971 typedef void *fl_owner_t;
975 struct file_lock_operations {
976 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
977 void (*fl_release_private)(struct file_lock *);
980 struct lock_manager_operations {
981 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
982 unsigned long (*lm_owner_key)(struct file_lock *);
983 fl_owner_t (*lm_get_owner)(fl_owner_t);
984 void (*lm_put_owner)(fl_owner_t);
985 void (*lm_notify)(struct file_lock *); /* unblock callback */
986 int (*lm_grant)(struct file_lock *, int);
987 bool (*lm_break)(struct file_lock *);
988 int (*lm_change)(struct file_lock *, int, struct list_head *);
989 void (*lm_setup)(struct file_lock *, void **);
992 struct lock_manager {
993 struct list_head list;
995 * NFSv4 and up also want opens blocked during the grace period;
1002 void locks_start_grace(struct net *, struct lock_manager *);
1003 void locks_end_grace(struct lock_manager *);
1004 bool locks_in_grace(struct net *);
1005 bool opens_in_grace(struct net *);
1007 /* that will die - we need it for nfs_lock_info */
1008 #include <linux/nfs_fs_i.h>
1011 * struct file_lock represents a generic "file lock". It's used to represent
1012 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1013 * note that the same struct is used to represent both a request for a lock and
1014 * the lock itself, but the same object is never used for both.
1016 * FIXME: should we create a separate "struct lock_request" to help distinguish
1019 * The varous i_flctx lists are ordered by:
1022 * 2) lock range start
1025 * Obviously, the last two criteria only matter for POSIX locks.
1028 struct file_lock *fl_next; /* singly linked list for this inode */
1029 struct list_head fl_list; /* link into file_lock_context */
1030 struct hlist_node fl_link; /* node in global lists */
1031 struct list_head fl_block; /* circular list of blocked processes */
1032 fl_owner_t fl_owner;
1033 unsigned int fl_flags;
1034 unsigned char fl_type;
1035 unsigned int fl_pid;
1036 int fl_link_cpu; /* what cpu's list is this on? */
1037 wait_queue_head_t fl_wait;
1038 struct file *fl_file;
1042 struct fasync_struct * fl_fasync; /* for lease break notifications */
1043 /* for lease breaks: */
1044 unsigned long fl_break_time;
1045 unsigned long fl_downgrade_time;
1047 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1048 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1050 struct nfs_lock_info nfs_fl;
1051 struct nfs4_lock_info nfs4_fl;
1053 struct list_head link; /* link in AFS vnode's pending_locks list */
1054 int state; /* state of grant or error if -ve */
1057 } __randomize_layout;
1059 struct file_lock_context {
1060 spinlock_t flc_lock;
1061 struct list_head flc_flock;
1062 struct list_head flc_posix;
1063 struct list_head flc_lease;
1066 /* The following constant reflects the upper bound of the file/locking space */
1068 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1069 #define OFFSET_MAX INT_LIMIT(loff_t)
1070 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1073 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1075 #define locks_inode(f) file_inode(f)
1077 #ifdef CONFIG_FILE_LOCKING
1078 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1079 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1082 #if BITS_PER_LONG == 32
1083 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1084 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1088 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1089 extern int fcntl_getlease(struct file *filp);
1092 void locks_free_lock_context(struct inode *inode);
1093 void locks_free_lock(struct file_lock *fl);
1094 extern void locks_init_lock(struct file_lock *);
1095 extern struct file_lock * locks_alloc_lock(void);
1096 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1097 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1098 extern void locks_remove_posix(struct file *, fl_owner_t);
1099 extern void locks_remove_file(struct file *);
1100 extern void locks_release_private(struct file_lock *);
1101 extern void posix_test_lock(struct file *, struct file_lock *);
1102 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1103 extern int posix_unblock_lock(struct file_lock *);
1104 extern int vfs_test_lock(struct file *, struct file_lock *);
1105 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1106 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1107 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1108 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1109 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1110 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1111 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1112 extern int lease_modify(struct file_lock *, int, struct list_head *);
1113 struct files_struct;
1114 extern void show_fd_locks(struct seq_file *f,
1115 struct file *filp, struct files_struct *files);
1116 #else /* !CONFIG_FILE_LOCKING */
1117 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1118 struct flock __user *user)
1123 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1124 unsigned int cmd, struct flock __user *user)
1129 #if BITS_PER_LONG == 32
1130 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1131 struct flock64 __user *user)
1136 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1137 unsigned int cmd, struct flock64 __user *user)
1142 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1147 static inline int fcntl_getlease(struct file *filp)
1153 locks_free_lock_context(struct inode *inode)
1157 static inline void locks_init_lock(struct file_lock *fl)
1162 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1167 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1172 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1177 static inline void locks_remove_file(struct file *filp)
1182 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1187 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1188 struct file_lock *conflock)
1193 static inline int posix_unblock_lock(struct file_lock *waiter)
1198 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1203 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1204 struct file_lock *fl, struct file_lock *conf)
1209 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1214 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1219 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1224 static inline void lease_get_mtime(struct inode *inode,
1225 struct timespec64 *time)
1230 static inline int generic_setlease(struct file *filp, long arg,
1231 struct file_lock **flp, void **priv)
1236 static inline int vfs_setlease(struct file *filp, long arg,
1237 struct file_lock **lease, void **priv)
1242 static inline int lease_modify(struct file_lock *fl, int arg,
1243 struct list_head *dispose)
1248 struct files_struct;
1249 static inline void show_fd_locks(struct seq_file *f,
1250 struct file *filp, struct files_struct *files) {}
1251 #endif /* !CONFIG_FILE_LOCKING */
1253 static inline struct inode *file_inode(const struct file *f)
1258 static inline struct dentry *file_dentry(const struct file *file)
1260 return d_real(file->f_path.dentry, file_inode(file));
1263 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1265 return locks_lock_inode_wait(locks_inode(filp), fl);
1268 struct fasync_struct {
1272 struct fasync_struct *fa_next; /* singly linked list */
1273 struct file *fa_file;
1274 struct rcu_head fa_rcu;
1277 #define FASYNC_MAGIC 0x4601
1279 /* SMP safe fasync helpers: */
1280 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1281 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1282 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1283 extern struct fasync_struct *fasync_alloc(void);
1284 extern void fasync_free(struct fasync_struct *);
1286 /* can be called from interrupts */
1287 extern void kill_fasync(struct fasync_struct **, int, int);
1289 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1290 extern int f_setown(struct file *filp, unsigned long arg, int force);
1291 extern void f_delown(struct file *filp);
1292 extern pid_t f_getown(struct file *filp);
1293 extern int send_sigurg(struct fown_struct *fown);
1296 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1297 * represented in both.
1299 #define SB_RDONLY 1 /* Mount read-only */
1300 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1301 #define SB_NODEV 4 /* Disallow access to device special files */
1302 #define SB_NOEXEC 8 /* Disallow program execution */
1303 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1304 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1305 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1306 #define SB_NOATIME 1024 /* Do not update access times. */
1307 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1308 #define SB_SILENT 32768
1309 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1310 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1311 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1312 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1314 /* These sb flags are internal to the kernel */
1315 #define SB_SUBMOUNT (1<<26)
1316 #define SB_NOSEC (1<<28)
1317 #define SB_BORN (1<<29)
1318 #define SB_ACTIVE (1<<30)
1319 #define SB_NOUSER (1<<31)
1325 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1326 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1327 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1328 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1329 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1332 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1333 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1334 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1335 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1337 /* sb->s_iflags to limit user namespace mounts */
1338 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1339 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1340 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1342 /* Possible states of 'frozen' field */
1344 SB_UNFROZEN = 0, /* FS is unfrozen */
1345 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1346 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1347 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1348 * internal threads if needed) */
1349 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1352 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1355 int frozen; /* Is sb frozen? */
1356 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1357 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1360 struct super_block {
1361 struct list_head s_list; /* Keep this first */
1362 dev_t s_dev; /* search index; _not_ kdev_t */
1363 unsigned char s_blocksize_bits;
1364 unsigned long s_blocksize;
1365 loff_t s_maxbytes; /* Max file size */
1366 struct file_system_type *s_type;
1367 const struct super_operations *s_op;
1368 const struct dquot_operations *dq_op;
1369 const struct quotactl_ops *s_qcop;
1370 const struct export_operations *s_export_op;
1371 unsigned long s_flags;
1372 unsigned long s_iflags; /* internal SB_I_* flags */
1373 unsigned long s_magic;
1374 struct dentry *s_root;
1375 struct rw_semaphore s_umount;
1378 #ifdef CONFIG_SECURITY
1381 const struct xattr_handler **s_xattr;
1382 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1383 const struct fscrypt_operations *s_cop;
1385 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1386 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1387 struct block_device *s_bdev;
1388 struct backing_dev_info *s_bdi;
1389 struct mtd_info *s_mtd;
1390 struct hlist_node s_instances;
1391 unsigned int s_quota_types; /* Bitmask of supported quota types */
1392 struct quota_info s_dquot; /* Diskquota specific options */
1394 struct sb_writers s_writers;
1396 char s_id[32]; /* Informational name */
1397 uuid_t s_uuid; /* UUID */
1399 void *s_fs_info; /* Filesystem private info */
1400 unsigned int s_max_links;
1403 /* Granularity of c/m/atime in ns.
1404 Cannot be worse than a second */
1408 * The next field is for VFS *only*. No filesystems have any business
1409 * even looking at it. You had been warned.
1411 struct mutex s_vfs_rename_mutex; /* Kludge */
1414 * Filesystem subtype. If non-empty the filesystem type field
1415 * in /proc/mounts will be "type.subtype"
1419 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1422 * Saved pool identifier for cleancache (-1 means none)
1424 int cleancache_poolid;
1426 struct shrinker s_shrink; /* per-sb shrinker handle */
1428 /* Number of inodes with nlink == 0 but still referenced */
1429 atomic_long_t s_remove_count;
1431 /* Being remounted read-only */
1432 int s_readonly_remount;
1434 /* AIO completions deferred from interrupt context */
1435 struct workqueue_struct *s_dio_done_wq;
1436 struct hlist_head s_pins;
1439 * Owning user namespace and default context in which to
1440 * interpret filesystem uids, gids, quotas, device nodes,
1441 * xattrs and security labels.
1443 struct user_namespace *s_user_ns;
1446 * Keep the lru lists last in the structure so they always sit on their
1447 * own individual cachelines.
1449 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1450 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1451 struct rcu_head rcu;
1452 struct work_struct destroy_work;
1454 struct mutex s_sync_lock; /* sync serialisation lock */
1457 * Indicates how deep in a filesystem stack this SB is
1461 /* s_inode_list_lock protects s_inodes */
1462 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1463 struct list_head s_inodes; /* all inodes */
1465 spinlock_t s_inode_wblist_lock;
1466 struct list_head s_inodes_wb; /* writeback inodes */
1467 } __randomize_layout;
1469 /* Helper functions so that in most cases filesystems will
1470 * not need to deal directly with kuid_t and kgid_t and can
1471 * instead deal with the raw numeric values that are stored
1472 * in the filesystem.
1474 static inline uid_t i_uid_read(const struct inode *inode)
1476 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1479 static inline gid_t i_gid_read(const struct inode *inode)
1481 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1484 static inline void i_uid_write(struct inode *inode, uid_t uid)
1486 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1489 static inline void i_gid_write(struct inode *inode, gid_t gid)
1491 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1494 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1495 extern struct timespec64 current_time(struct inode *inode);
1498 * Snapshotting support.
1501 void __sb_end_write(struct super_block *sb, int level);
1502 int __sb_start_write(struct super_block *sb, int level, bool wait);
1504 #define __sb_writers_acquired(sb, lev) \
1505 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1506 #define __sb_writers_release(sb, lev) \
1507 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1510 * sb_end_write - drop write access to a superblock
1511 * @sb: the super we wrote to
1513 * Decrement number of writers to the filesystem. Wake up possible waiters
1514 * wanting to freeze the filesystem.
1516 static inline void sb_end_write(struct super_block *sb)
1518 __sb_end_write(sb, SB_FREEZE_WRITE);
1522 * sb_end_pagefault - drop write access to a superblock from a page fault
1523 * @sb: the super we wrote to
1525 * Decrement number of processes handling write page fault to the filesystem.
1526 * Wake up possible waiters wanting to freeze the filesystem.
1528 static inline void sb_end_pagefault(struct super_block *sb)
1530 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1534 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1535 * @sb: the super we wrote to
1537 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1538 * waiters wanting to freeze the filesystem.
1540 static inline void sb_end_intwrite(struct super_block *sb)
1542 __sb_end_write(sb, SB_FREEZE_FS);
1546 * sb_start_write - get write access to a superblock
1547 * @sb: the super we write to
1549 * When a process wants to write data or metadata to a file system (i.e. dirty
1550 * a page or an inode), it should embed the operation in a sb_start_write() -
1551 * sb_end_write() pair to get exclusion against file system freezing. This
1552 * function increments number of writers preventing freezing. If the file
1553 * system is already frozen, the function waits until the file system is
1556 * Since freeze protection behaves as a lock, users have to preserve
1557 * ordering of freeze protection and other filesystem locks. Generally,
1558 * freeze protection should be the outermost lock. In particular, we have:
1561 * -> i_mutex (write path, truncate, directory ops, ...)
1562 * -> s_umount (freeze_super, thaw_super)
1564 static inline void sb_start_write(struct super_block *sb)
1566 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1569 static inline int sb_start_write_trylock(struct super_block *sb)
1571 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1575 * sb_start_pagefault - get write access to a superblock from a page fault
1576 * @sb: the super we write to
1578 * When a process starts handling write page fault, it should embed the
1579 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1580 * exclusion against file system freezing. This is needed since the page fault
1581 * is going to dirty a page. This function increments number of running page
1582 * faults preventing freezing. If the file system is already frozen, the
1583 * function waits until the file system is thawed.
1585 * Since page fault freeze protection behaves as a lock, users have to preserve
1586 * ordering of freeze protection and other filesystem locks. It is advised to
1587 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1588 * handling code implies lock dependency:
1591 * -> sb_start_pagefault
1593 static inline void sb_start_pagefault(struct super_block *sb)
1595 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1599 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1600 * @sb: the super we write to
1602 * This is the third level of protection against filesystem freezing. It is
1603 * free for use by a filesystem. The only requirement is that it must rank
1604 * below sb_start_pagefault.
1606 * For example filesystem can call sb_start_intwrite() when starting a
1607 * transaction which somewhat eases handling of freezing for internal sources
1608 * of filesystem changes (internal fs threads, discarding preallocation on file
1611 static inline void sb_start_intwrite(struct super_block *sb)
1613 __sb_start_write(sb, SB_FREEZE_FS, true);
1616 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1618 return __sb_start_write(sb, SB_FREEZE_FS, false);
1622 extern bool inode_owner_or_capable(const struct inode *inode);
1625 * VFS helper functions..
1627 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1628 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1629 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1630 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1631 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1632 extern int vfs_rmdir(struct inode *, struct dentry *);
1633 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1634 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1635 extern int vfs_whiteout(struct inode *, struct dentry *);
1637 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1640 int vfs_mkobj(struct dentry *, umode_t,
1641 int (*f)(struct dentry *, umode_t, void *),
1644 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1647 * VFS file helper functions.
1649 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1651 extern bool may_open_dev(const struct path *path);
1653 * VFS FS_IOC_FIEMAP helper definitions.
1655 struct fiemap_extent_info {
1656 unsigned int fi_flags; /* Flags as passed from user */
1657 unsigned int fi_extents_mapped; /* Number of mapped extents */
1658 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1659 struct fiemap_extent __user *fi_extents_start; /* Start of
1660 fiemap_extent array */
1662 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1663 u64 phys, u64 len, u32 flags);
1664 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1669 * NOTE! These match bits 12..15 of stat.st_mode
1670 * (ie "(i_mode >> 12) & 15").
1672 #define DT_UNKNOWN 0
1683 * This is the "filldir" function type, used by readdir() to let
1684 * the kernel specify what kind of dirent layout it wants to have.
1685 * This allows the kernel to read directories into kernel space or
1686 * to have different dirent layouts depending on the binary type.
1689 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1692 struct dir_context {
1697 struct block_device_operations;
1699 /* These macros are for out of kernel modules to test that
1700 * the kernel supports the unlocked_ioctl and compat_ioctl
1701 * fields in struct file_operations. */
1702 #define HAVE_COMPAT_IOCTL 1
1703 #define HAVE_UNLOCKED_IOCTL 1
1706 * These flags let !MMU mmap() govern direct device mapping vs immediate
1707 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1709 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1710 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1711 * NOMMU_MAP_READ: Can be mapped for reading
1712 * NOMMU_MAP_WRITE: Can be mapped for writing
1713 * NOMMU_MAP_EXEC: Can be mapped for execution
1715 #define NOMMU_MAP_COPY 0x00000001
1716 #define NOMMU_MAP_DIRECT 0x00000008
1717 #define NOMMU_MAP_READ VM_MAYREAD
1718 #define NOMMU_MAP_WRITE VM_MAYWRITE
1719 #define NOMMU_MAP_EXEC VM_MAYEXEC
1721 #define NOMMU_VMFLAGS \
1722 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1725 * These flags control the behavior of the remap_file_range function pointer.
1726 * If it is called with len == 0 that means "remap to end of source file".
1727 * See Documentation/filesystems/vfs.txt for more details about this call.
1729 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1730 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1732 #define REMAP_FILE_DEDUP (1 << 0)
1733 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1736 * These flags signal that the caller is ok with altering various aspects of
1737 * the behavior of the remap operation. The changes must be made by the
1738 * implementation; the vfs remap helper functions can take advantage of them.
1739 * Flags in this category exist to preserve the quirky behavior of the hoisted
1740 * btrfs clone/dedupe ioctls.
1742 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1746 struct file_operations {
1747 struct module *owner;
1748 loff_t (*llseek) (struct file *, loff_t, int);
1749 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1750 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1751 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1752 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1753 int (*iterate) (struct file *, struct dir_context *);
1754 int (*iterate_shared) (struct file *, struct dir_context *);
1755 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1756 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1757 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1758 int (*mmap) (struct file *, struct vm_area_struct *);
1759 unsigned long mmap_supported_flags;
1760 int (*open) (struct inode *, struct file *);
1761 int (*flush) (struct file *, fl_owner_t id);
1762 int (*release) (struct inode *, struct file *);
1763 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1764 int (*fasync) (int, struct file *, int);
1765 int (*lock) (struct file *, int, struct file_lock *);
1766 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1767 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1768 int (*check_flags)(int);
1769 int (*flock) (struct file *, int, struct file_lock *);
1770 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1771 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1772 int (*setlease)(struct file *, long, struct file_lock **, void **);
1773 long (*fallocate)(struct file *file, int mode, loff_t offset,
1775 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1777 unsigned (*mmap_capabilities)(struct file *);
1779 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1780 loff_t, size_t, unsigned int);
1781 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1782 struct file *file_out, loff_t pos_out,
1783 loff_t len, unsigned int remap_flags);
1784 int (*fadvise)(struct file *, loff_t, loff_t, int);
1785 } __randomize_layout;
1787 struct inode_operations {
1788 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1789 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1790 int (*permission) (struct inode *, int);
1791 struct posix_acl * (*get_acl)(struct inode *, int);
1793 int (*readlink) (struct dentry *, char __user *,int);
1795 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1796 int (*link) (struct dentry *,struct inode *,struct dentry *);
1797 int (*unlink) (struct inode *,struct dentry *);
1798 int (*symlink) (struct inode *,struct dentry *,const char *);
1799 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1800 int (*rmdir) (struct inode *,struct dentry *);
1801 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1802 int (*rename) (struct inode *, struct dentry *,
1803 struct inode *, struct dentry *, unsigned int);
1804 int (*setattr) (struct dentry *, struct iattr *);
1805 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1806 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1807 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1809 int (*update_time)(struct inode *, struct timespec64 *, int);
1810 int (*atomic_open)(struct inode *, struct dentry *,
1811 struct file *, unsigned open_flag,
1812 umode_t create_mode);
1813 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1814 int (*set_acl)(struct inode *, struct posix_acl *, int);
1815 } ____cacheline_aligned;
1817 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1818 struct iov_iter *iter)
1820 return file->f_op->read_iter(kio, iter);
1823 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1824 struct iov_iter *iter)
1826 return file->f_op->write_iter(kio, iter);
1829 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1831 return file->f_op->mmap(file, vma);
1834 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1835 unsigned long nr_segs, unsigned long fast_segs,
1836 struct iovec *fast_pointer,
1837 struct iovec **ret_pointer);
1839 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1840 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1841 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1842 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1843 unsigned long, loff_t *, rwf_t);
1844 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1845 loff_t, size_t, unsigned int);
1846 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1847 struct file *file_out, loff_t pos_out,
1849 unsigned int remap_flags);
1850 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1851 struct file *file_out, loff_t pos_out,
1852 loff_t len, unsigned int remap_flags);
1853 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1854 struct file *file_out, loff_t pos_out,
1855 loff_t len, unsigned int remap_flags);
1856 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1857 struct inode *dest, loff_t destoff,
1858 loff_t len, bool *is_same);
1859 extern int vfs_dedupe_file_range(struct file *file,
1860 struct file_dedupe_range *same);
1861 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1862 struct file *dst_file, loff_t dst_pos,
1863 loff_t len, unsigned int remap_flags);
1866 struct super_operations {
1867 struct inode *(*alloc_inode)(struct super_block *sb);
1868 void (*destroy_inode)(struct inode *);
1870 void (*dirty_inode) (struct inode *, int flags);
1871 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1872 int (*drop_inode) (struct inode *);
1873 void (*evict_inode) (struct inode *);
1874 void (*put_super) (struct super_block *);
1875 int (*sync_fs)(struct super_block *sb, int wait);
1876 int (*freeze_super) (struct super_block *);
1877 int (*freeze_fs) (struct super_block *);
1878 int (*thaw_super) (struct super_block *);
1879 int (*unfreeze_fs) (struct super_block *);
1880 int (*statfs) (struct dentry *, struct kstatfs *);
1881 int (*remount_fs) (struct super_block *, int *, char *);
1882 void (*umount_begin) (struct super_block *);
1884 int (*show_options)(struct seq_file *, struct dentry *);
1885 int (*show_devname)(struct seq_file *, struct dentry *);
1886 int (*show_path)(struct seq_file *, struct dentry *);
1887 int (*show_stats)(struct seq_file *, struct dentry *);
1889 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1890 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1891 struct dquot **(*get_dquots)(struct inode *);
1893 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1894 long (*nr_cached_objects)(struct super_block *,
1895 struct shrink_control *);
1896 long (*free_cached_objects)(struct super_block *,
1897 struct shrink_control *);
1901 * Inode flags - they have no relation to superblock flags now
1903 #define S_SYNC 1 /* Writes are synced at once */
1904 #define S_NOATIME 2 /* Do not update access times */
1905 #define S_APPEND 4 /* Append-only file */
1906 #define S_IMMUTABLE 8 /* Immutable file */
1907 #define S_DEAD 16 /* removed, but still open directory */
1908 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1909 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1910 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1911 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1912 #define S_PRIVATE 512 /* Inode is fs-internal */
1913 #define S_IMA 1024 /* Inode has an associated IMA struct */
1914 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1915 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1916 #ifdef CONFIG_FS_DAX
1917 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1919 #define S_DAX 0 /* Make all the DAX code disappear */
1921 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1924 * Note that nosuid etc flags are inode-specific: setting some file-system
1925 * flags just means all the inodes inherit those flags by default. It might be
1926 * possible to override it selectively if you really wanted to with some
1927 * ioctl() that is not currently implemented.
1929 * Exception: SB_RDONLY is always applied to the entire file system.
1931 * Unfortunately, it is possible to change a filesystems flags with it mounted
1932 * with files in use. This means that all of the inodes will not have their
1933 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1934 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1936 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1938 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1939 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1940 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1941 ((inode)->i_flags & S_SYNC))
1942 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1943 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1944 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1945 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1946 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1948 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1949 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1950 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1951 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1953 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1954 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1955 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1956 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1957 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1958 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1959 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1960 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1961 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1963 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1964 (inode)->i_rdev == WHITEOUT_DEV)
1966 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1968 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1971 static inline enum rw_hint file_write_hint(struct file *file)
1973 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1974 return file->f_write_hint;
1976 return file_inode(file)->i_write_hint;
1979 static inline int iocb_flags(struct file *file);
1981 static inline u16 ki_hint_validate(enum rw_hint hint)
1983 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
1985 if (hint <= max_hint)
1990 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1992 *kiocb = (struct kiocb) {
1994 .ki_flags = iocb_flags(filp),
1995 .ki_hint = ki_hint_validate(file_write_hint(filp)),
1996 .ki_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, 0),
2001 * Inode state bits. Protected by inode->i_lock
2003 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2004 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2006 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2007 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2008 * various stages of removing an inode.
2010 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2012 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2013 * fdatasync(). i_atime is the usual cause.
2014 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2015 * these changes separately from I_DIRTY_SYNC so that we
2016 * don't have to write inode on fdatasync() when only
2017 * mtime has changed in it.
2018 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2019 * I_NEW Serves as both a mutex and completion notification.
2020 * New inodes set I_NEW. If two processes both create
2021 * the same inode, one of them will release its inode and
2022 * wait for I_NEW to be released before returning.
2023 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2024 * also cause waiting on I_NEW, without I_NEW actually
2025 * being set. find_inode() uses this to prevent returning
2026 * nearly-dead inodes.
2027 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2028 * is zero. I_FREEING must be set when I_WILL_FREE is
2030 * I_FREEING Set when inode is about to be freed but still has dirty
2031 * pages or buffers attached or the inode itself is still
2033 * I_CLEAR Added by clear_inode(). In this state the inode is
2034 * clean and can be destroyed. Inode keeps I_FREEING.
2036 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2037 * prohibited for many purposes. iget() must wait for
2038 * the inode to be completely released, then create it
2039 * anew. Other functions will just ignore such inodes,
2040 * if appropriate. I_NEW is used for waiting.
2042 * I_SYNC Writeback of inode is running. The bit is set during
2043 * data writeback, and cleared with a wakeup on the bit
2044 * address once it is done. The bit is also used to pin
2045 * the inode in memory for flusher thread.
2047 * I_REFERENCED Marks the inode as recently references on the LRU list.
2049 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2051 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2052 * synchronize competing switching instances and to tell
2053 * wb stat updates to grab the i_pages lock. See
2054 * inode_switch_wb_work_fn() for details.
2056 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2057 * and work dirs among overlayfs mounts.
2059 * I_CREATING New object's inode in the middle of setting up.
2061 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2063 #define I_DIRTY_SYNC (1 << 0)
2064 #define I_DIRTY_DATASYNC (1 << 1)
2065 #define I_DIRTY_PAGES (1 << 2)
2067 #define I_NEW (1 << __I_NEW)
2068 #define I_WILL_FREE (1 << 4)
2069 #define I_FREEING (1 << 5)
2070 #define I_CLEAR (1 << 6)
2072 #define I_SYNC (1 << __I_SYNC)
2073 #define I_REFERENCED (1 << 8)
2074 #define __I_DIO_WAKEUP 9
2075 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2076 #define I_LINKABLE (1 << 10)
2077 #define I_DIRTY_TIME (1 << 11)
2078 #define __I_DIRTY_TIME_EXPIRED 12
2079 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2080 #define I_WB_SWITCH (1 << 13)
2081 #define I_OVL_INUSE (1 << 14)
2082 #define I_CREATING (1 << 15)
2084 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2085 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2086 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2088 extern void __mark_inode_dirty(struct inode *, int);
2089 static inline void mark_inode_dirty(struct inode *inode)
2091 __mark_inode_dirty(inode, I_DIRTY);
2094 static inline void mark_inode_dirty_sync(struct inode *inode)
2096 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2099 extern void inc_nlink(struct inode *inode);
2100 extern void drop_nlink(struct inode *inode);
2101 extern void clear_nlink(struct inode *inode);
2102 extern void set_nlink(struct inode *inode, unsigned int nlink);
2104 static inline void inode_inc_link_count(struct inode *inode)
2107 mark_inode_dirty(inode);
2110 static inline void inode_dec_link_count(struct inode *inode)
2113 mark_inode_dirty(inode);
2116 enum file_time_flags {
2123 extern bool atime_needs_update(const struct path *, struct inode *);
2124 extern void touch_atime(const struct path *);
2125 static inline void file_accessed(struct file *file)
2127 if (!(file->f_flags & O_NOATIME))
2128 touch_atime(&file->f_path);
2131 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2132 int sync_inode_metadata(struct inode *inode, int wait);
2134 struct file_system_type {
2137 #define FS_REQUIRES_DEV 1
2138 #define FS_BINARY_MOUNTDATA 2
2139 #define FS_HAS_SUBTYPE 4
2140 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2141 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2142 struct dentry *(*mount) (struct file_system_type *, int,
2143 const char *, void *);
2144 void (*kill_sb) (struct super_block *);
2145 struct module *owner;
2146 struct file_system_type * next;
2147 struct hlist_head fs_supers;
2149 struct lock_class_key s_lock_key;
2150 struct lock_class_key s_umount_key;
2151 struct lock_class_key s_vfs_rename_key;
2152 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2154 struct lock_class_key i_lock_key;
2155 struct lock_class_key i_mutex_key;
2156 struct lock_class_key i_mutex_dir_key;
2159 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2161 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2162 int flags, void *data, void *ns, struct user_namespace *user_ns,
2163 int (*fill_super)(struct super_block *, void *, int));
2165 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2166 int flags, const char *dev_name, void *data,
2167 int (*fill_super)(struct super_block *, void *, int));
2169 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2170 int flags, const char *dev_name, void *data,
2171 int (*fill_super)(struct super_block *, void *, int))
2173 return ERR_PTR(-ENODEV);
2176 extern struct dentry *mount_single(struct file_system_type *fs_type,
2177 int flags, void *data,
2178 int (*fill_super)(struct super_block *, void *, int));
2179 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2180 int flags, void *data,
2181 int (*fill_super)(struct super_block *, void *, int));
2182 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2183 void generic_shutdown_super(struct super_block *sb);
2185 void kill_block_super(struct super_block *sb);
2187 static inline void kill_block_super(struct super_block *sb)
2192 void kill_anon_super(struct super_block *sb);
2193 void kill_litter_super(struct super_block *sb);
2194 void deactivate_super(struct super_block *sb);
2195 void deactivate_locked_super(struct super_block *sb);
2196 int set_anon_super(struct super_block *s, void *data);
2197 int get_anon_bdev(dev_t *);
2198 void free_anon_bdev(dev_t);
2199 struct super_block *sget_userns(struct file_system_type *type,
2200 int (*test)(struct super_block *,void *),
2201 int (*set)(struct super_block *,void *),
2202 int flags, struct user_namespace *user_ns,
2204 struct super_block *sget(struct file_system_type *type,
2205 int (*test)(struct super_block *,void *),
2206 int (*set)(struct super_block *,void *),
2207 int flags, void *data);
2208 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2209 const struct super_operations *ops,
2210 const struct xattr_handler **xattr,
2211 const struct dentry_operations *dops,
2214 static inline struct dentry *
2215 mount_pseudo(struct file_system_type *fs_type, char *name,
2216 const struct super_operations *ops,
2217 const struct dentry_operations *dops, unsigned long magic)
2219 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2222 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2223 #define fops_get(fops) \
2224 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2225 #define fops_put(fops) \
2226 do { if (fops) module_put((fops)->owner); } while(0)
2228 * This one is to be used *ONLY* from ->open() instances.
2229 * fops must be non-NULL, pinned down *and* module dependencies
2230 * should be sufficient to pin the caller down as well.
2232 #define replace_fops(f, fops) \
2234 struct file *__file = (f); \
2235 fops_put(__file->f_op); \
2236 BUG_ON(!(__file->f_op = (fops))); \
2239 extern int register_filesystem(struct file_system_type *);
2240 extern int unregister_filesystem(struct file_system_type *);
2241 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2242 #define kern_mount(type) kern_mount_data(type, NULL)
2243 extern void kern_unmount(struct vfsmount *mnt);
2244 extern int may_umount_tree(struct vfsmount *);
2245 extern int may_umount(struct vfsmount *);
2246 extern long do_mount(const char *, const char __user *,
2247 const char *, unsigned long, void *);
2248 extern struct vfsmount *collect_mounts(const struct path *);
2249 extern void drop_collected_mounts(struct vfsmount *);
2250 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2252 extern int vfs_statfs(const struct path *, struct kstatfs *);
2253 extern int user_statfs(const char __user *, struct kstatfs *);
2254 extern int fd_statfs(int, struct kstatfs *);
2255 extern int freeze_super(struct super_block *super);
2256 extern int thaw_super(struct super_block *super);
2257 extern bool our_mnt(struct vfsmount *mnt);
2258 extern __printf(2, 3)
2259 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2260 extern int super_setup_bdi(struct super_block *sb);
2262 extern int current_umask(void);
2264 extern void ihold(struct inode * inode);
2265 extern void iput(struct inode *);
2266 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2269 extern struct kobject *fs_kobj;
2271 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2273 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2274 extern int locks_mandatory_locked(struct file *);
2275 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2278 * Candidates for mandatory locking have the setgid bit set
2279 * but no group execute bit - an otherwise meaningless combination.
2282 static inline int __mandatory_lock(struct inode *ino)
2284 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2288 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2289 * otherwise these will be advisory locks
2292 static inline int mandatory_lock(struct inode *ino)
2294 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2297 static inline int locks_verify_locked(struct file *file)
2299 if (mandatory_lock(locks_inode(file)))
2300 return locks_mandatory_locked(file);
2304 static inline int locks_verify_truncate(struct inode *inode,
2308 if (!inode->i_flctx || !mandatory_lock(inode))
2311 if (size < inode->i_size) {
2312 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2315 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2320 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2322 static inline int locks_mandatory_locked(struct file *file)
2327 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2328 loff_t start, loff_t end, unsigned char type)
2333 static inline int __mandatory_lock(struct inode *inode)
2338 static inline int mandatory_lock(struct inode *inode)
2343 static inline int locks_verify_locked(struct file *file)
2348 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2354 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2357 #ifdef CONFIG_FILE_LOCKING
2358 static inline int break_lease(struct inode *inode, unsigned int mode)
2361 * Since this check is lockless, we must ensure that any refcounts
2362 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2363 * could end up racing with tasks trying to set a new lease on this
2367 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2368 return __break_lease(inode, mode, FL_LEASE);
2372 static inline int break_deleg(struct inode *inode, unsigned int mode)
2375 * Since this check is lockless, we must ensure that any refcounts
2376 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2377 * could end up racing with tasks trying to set a new lease on this
2381 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2382 return __break_lease(inode, mode, FL_DELEG);
2386 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2390 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2391 if (ret == -EWOULDBLOCK && delegated_inode) {
2392 *delegated_inode = inode;
2398 static inline int break_deleg_wait(struct inode **delegated_inode)
2402 ret = break_deleg(*delegated_inode, O_WRONLY);
2403 iput(*delegated_inode);
2404 *delegated_inode = NULL;
2408 static inline int break_layout(struct inode *inode, bool wait)
2411 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2412 return __break_lease(inode,
2413 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2418 #else /* !CONFIG_FILE_LOCKING */
2419 static inline int break_lease(struct inode *inode, unsigned int mode)
2424 static inline int break_deleg(struct inode *inode, unsigned int mode)
2429 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2434 static inline int break_deleg_wait(struct inode **delegated_inode)
2440 static inline int break_layout(struct inode *inode, bool wait)
2445 #endif /* CONFIG_FILE_LOCKING */
2450 const char *name; /* pointer to actual string */
2451 const __user char *uptr; /* original userland pointer */
2453 struct audit_names *aname;
2457 extern long vfs_truncate(const struct path *, loff_t);
2458 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2460 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2462 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2464 extern struct file *file_open_name(struct filename *, int, umode_t);
2465 extern struct file *filp_open(const char *, int, umode_t);
2466 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2467 const char *, int, umode_t);
2468 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2469 extern struct file * open_with_fake_path(const struct path *, int,
2470 struct inode*, const struct cred *);
2471 static inline struct file *file_clone_open(struct file *file)
2473 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2475 extern int filp_close(struct file *, fl_owner_t id);
2477 extern struct filename *getname_flags(const char __user *, int, int *);
2478 extern struct filename *getname(const char __user *);
2479 extern struct filename *getname_kernel(const char *);
2480 extern void putname(struct filename *name);
2482 extern int finish_open(struct file *file, struct dentry *dentry,
2483 int (*open)(struct inode *, struct file *));
2484 extern int finish_no_open(struct file *file, struct dentry *dentry);
2488 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2491 extern void __init vfs_caches_init_early(void);
2492 extern void __init vfs_caches_init(void);
2494 extern struct kmem_cache *names_cachep;
2496 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2497 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2500 extern int register_blkdev(unsigned int, const char *);
2501 extern void unregister_blkdev(unsigned int, const char *);
2502 extern void bdev_unhash_inode(dev_t dev);
2503 extern struct block_device *bdget(dev_t);
2504 extern struct block_device *bdgrab(struct block_device *bdev);
2505 extern void bd_set_size(struct block_device *, loff_t size);
2506 extern void bd_forget(struct inode *inode);
2507 extern void bdput(struct block_device *);
2508 extern void invalidate_bdev(struct block_device *);
2509 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2510 extern int sync_blockdev(struct block_device *bdev);
2511 extern void kill_bdev(struct block_device *);
2512 extern struct super_block *freeze_bdev(struct block_device *);
2513 extern void emergency_thaw_all(void);
2514 extern void emergency_thaw_bdev(struct super_block *sb);
2515 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2516 extern int fsync_bdev(struct block_device *);
2518 extern struct super_block *blockdev_superblock;
2520 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2522 return sb == blockdev_superblock;
2525 static inline void bd_forget(struct inode *inode) {}
2526 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2527 static inline void kill_bdev(struct block_device *bdev) {}
2528 static inline void invalidate_bdev(struct block_device *bdev) {}
2530 static inline struct super_block *freeze_bdev(struct block_device *sb)
2535 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2540 static inline int emergency_thaw_bdev(struct super_block *sb)
2545 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2549 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2554 extern int sync_filesystem(struct super_block *);
2555 extern const struct file_operations def_blk_fops;
2556 extern const struct file_operations def_chr_fops;
2558 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2559 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2560 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2561 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2562 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2564 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2566 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2567 extern int __blkdev_reread_part(struct block_device *bdev);
2568 extern int blkdev_reread_part(struct block_device *bdev);
2571 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2572 extern void bd_unlink_disk_holder(struct block_device *bdev,
2573 struct gendisk *disk);
2575 static inline int bd_link_disk_holder(struct block_device *bdev,
2576 struct gendisk *disk)
2580 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2581 struct gendisk *disk)
2588 #define CHRDEV_MAJOR_MAX 512
2589 /* Marks the bottom of the first segment of free char majors */
2590 #define CHRDEV_MAJOR_DYN_END 234
2591 /* Marks the top and bottom of the second segment of free char majors */
2592 #define CHRDEV_MAJOR_DYN_EXT_START 511
2593 #define CHRDEV_MAJOR_DYN_EXT_END 384
2595 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2596 extern int register_chrdev_region(dev_t, unsigned, const char *);
2597 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2598 unsigned int count, const char *name,
2599 const struct file_operations *fops);
2600 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2601 unsigned int count, const char *name);
2602 extern void unregister_chrdev_region(dev_t, unsigned);
2603 extern void chrdev_show(struct seq_file *,off_t);
2605 static inline int register_chrdev(unsigned int major, const char *name,
2606 const struct file_operations *fops)
2608 return __register_chrdev(major, 0, 256, name, fops);
2611 static inline void unregister_chrdev(unsigned int major, const char *name)
2613 __unregister_chrdev(major, 0, 256, name);
2616 /* fs/block_dev.c */
2617 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2618 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2621 #define BLKDEV_MAJOR_MAX 512
2622 extern const char *__bdevname(dev_t, char *buffer);
2623 extern const char *bdevname(struct block_device *bdev, char *buffer);
2624 extern struct block_device *lookup_bdev(const char *);
2625 extern void blkdev_show(struct seq_file *,off_t);
2628 #define BLKDEV_MAJOR_MAX 0
2631 extern void init_special_inode(struct inode *, umode_t, dev_t);
2633 /* Invalid inode operations -- fs/bad_inode.c */
2634 extern void make_bad_inode(struct inode *);
2635 extern bool is_bad_inode(struct inode *);
2638 extern void check_disk_size_change(struct gendisk *disk,
2639 struct block_device *bdev, bool verbose);
2640 extern int revalidate_disk(struct gendisk *);
2641 extern int check_disk_change(struct block_device *);
2642 extern int __invalidate_device(struct block_device *, bool);
2643 extern int invalidate_partition(struct gendisk *, int);
2645 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2646 pgoff_t start, pgoff_t end);
2648 static inline void invalidate_remote_inode(struct inode *inode)
2650 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2651 S_ISLNK(inode->i_mode))
2652 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2654 extern int invalidate_inode_pages2(struct address_space *mapping);
2655 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2656 pgoff_t start, pgoff_t end);
2657 extern int write_inode_now(struct inode *, int);
2658 extern int filemap_fdatawrite(struct address_space *);
2659 extern int filemap_flush(struct address_space *);
2660 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2661 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2664 static inline int filemap_fdatawait(struct address_space *mapping)
2666 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2669 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2671 extern int filemap_write_and_wait(struct address_space *mapping);
2672 extern int filemap_write_and_wait_range(struct address_space *mapping,
2673 loff_t lstart, loff_t lend);
2674 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2675 loff_t start, loff_t end, int sync_mode);
2676 extern int filemap_fdatawrite_range(struct address_space *mapping,
2677 loff_t start, loff_t end);
2678 extern int filemap_check_errors(struct address_space *mapping);
2679 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2681 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2683 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2684 extern int __must_check file_write_and_wait_range(struct file *file,
2685 loff_t start, loff_t end);
2687 static inline int file_write_and_wait(struct file *file)
2689 return file_write_and_wait_range(file, 0, LLONG_MAX);
2693 * filemap_set_wb_err - set a writeback error on an address_space
2694 * @mapping: mapping in which to set writeback error
2695 * @err: error to be set in mapping
2697 * When writeback fails in some way, we must record that error so that
2698 * userspace can be informed when fsync and the like are called. We endeavor
2699 * to report errors on any file that was open at the time of the error. Some
2700 * internal callers also need to know when writeback errors have occurred.
2702 * When a writeback error occurs, most filesystems will want to call
2703 * filemap_set_wb_err to record the error in the mapping so that it will be
2704 * automatically reported whenever fsync is called on the file.
2706 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2708 /* Fastpath for common case of no error */
2710 __filemap_set_wb_err(mapping, err);
2714 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2715 * @mapping: mapping to check for writeback errors
2716 * @since: previously-sampled errseq_t
2718 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2719 * the given value was sampled.
2721 * If it has then report the latest error set, otherwise return 0.
2723 static inline int filemap_check_wb_err(struct address_space *mapping,
2726 return errseq_check(&mapping->wb_err, since);
2730 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2731 * @mapping: mapping to be sampled
2733 * Writeback errors are always reported relative to a particular sample point
2734 * in the past. This function provides those sample points.
2736 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2738 return errseq_sample(&mapping->wb_err);
2741 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2743 extern int vfs_fsync(struct file *file, int datasync);
2746 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2747 * to already be updated for the write, and will return either the amount
2748 * of bytes passed in, or an error if syncing the file failed.
2750 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2752 if (iocb->ki_flags & IOCB_DSYNC) {
2753 int ret = vfs_fsync_range(iocb->ki_filp,
2754 iocb->ki_pos - count, iocb->ki_pos - 1,
2755 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2763 extern void emergency_sync(void);
2764 extern void emergency_remount(void);
2766 extern sector_t bmap(struct inode *, sector_t);
2768 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2769 extern int inode_permission(struct inode *, int);
2770 extern int generic_permission(struct inode *, int);
2771 extern int __check_sticky(struct inode *dir, struct inode *inode);
2773 static inline bool execute_ok(struct inode *inode)
2775 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2778 static inline void file_start_write(struct file *file)
2780 if (!S_ISREG(file_inode(file)->i_mode))
2782 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2785 static inline bool file_start_write_trylock(struct file *file)
2787 if (!S_ISREG(file_inode(file)->i_mode))
2789 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2792 static inline void file_end_write(struct file *file)
2794 if (!S_ISREG(file_inode(file)->i_mode))
2796 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2800 * get_write_access() gets write permission for a file.
2801 * put_write_access() releases this write permission.
2802 * This is used for regular files.
2803 * We cannot support write (and maybe mmap read-write shared) accesses and
2804 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2805 * can have the following values:
2806 * 0: no writers, no VM_DENYWRITE mappings
2807 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2808 * > 0: (i_writecount) users are writing to the file.
2810 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2811 * except for the cases where we don't hold i_writecount yet. Then we need to
2812 * use {get,deny}_write_access() - these functions check the sign and refuse
2813 * to do the change if sign is wrong.
2815 static inline int get_write_access(struct inode *inode)
2817 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2819 static inline int deny_write_access(struct file *file)
2821 struct inode *inode = file_inode(file);
2822 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2824 static inline void put_write_access(struct inode * inode)
2826 atomic_dec(&inode->i_writecount);
2828 static inline void allow_write_access(struct file *file)
2831 atomic_inc(&file_inode(file)->i_writecount);
2833 static inline bool inode_is_open_for_write(const struct inode *inode)
2835 return atomic_read(&inode->i_writecount) > 0;
2839 static inline void i_readcount_dec(struct inode *inode)
2841 BUG_ON(!atomic_read(&inode->i_readcount));
2842 atomic_dec(&inode->i_readcount);
2844 static inline void i_readcount_inc(struct inode *inode)
2846 atomic_inc(&inode->i_readcount);
2849 static inline void i_readcount_dec(struct inode *inode)
2853 static inline void i_readcount_inc(struct inode *inode)
2858 extern int do_pipe_flags(int *, int);
2860 #define __kernel_read_file_id(id) \
2861 id(UNKNOWN, unknown) \
2862 id(FIRMWARE, firmware) \
2863 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2864 id(MODULE, kernel-module) \
2865 id(KEXEC_IMAGE, kexec-image) \
2866 id(KEXEC_INITRAMFS, kexec-initramfs) \
2867 id(POLICY, security-policy) \
2868 id(X509_CERTIFICATE, x509-certificate) \
2871 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2872 #define __fid_stringify(dummy, str) #str,
2874 enum kernel_read_file_id {
2875 __kernel_read_file_id(__fid_enumify)
2878 static const char * const kernel_read_file_str[] = {
2879 __kernel_read_file_id(__fid_stringify)
2882 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2884 if ((unsigned)id >= READING_MAX_ID)
2885 return kernel_read_file_str[READING_UNKNOWN];
2887 return kernel_read_file_str[id];
2890 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2891 enum kernel_read_file_id);
2892 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2893 enum kernel_read_file_id);
2894 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2895 enum kernel_read_file_id);
2896 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2897 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2898 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2899 extern struct file * open_exec(const char *);
2901 /* fs/dcache.c -- generic fs support functions */
2902 extern bool is_subdir(struct dentry *, struct dentry *);
2903 extern bool path_is_under(const struct path *, const struct path *);
2905 extern char *file_path(struct file *, char *, int);
2907 #include <linux/err.h>
2909 /* needed for stackable file system support */
2910 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2912 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2914 extern int inode_init_always(struct super_block *, struct inode *);
2915 extern void inode_init_once(struct inode *);
2916 extern void address_space_init_once(struct address_space *mapping);
2917 extern struct inode * igrab(struct inode *);
2918 extern ino_t iunique(struct super_block *, ino_t);
2919 extern int inode_needs_sync(struct inode *inode);
2920 extern int generic_delete_inode(struct inode *inode);
2921 static inline int generic_drop_inode(struct inode *inode)
2923 return !inode->i_nlink || inode_unhashed(inode);
2926 extern struct inode *ilookup5_nowait(struct super_block *sb,
2927 unsigned long hashval, int (*test)(struct inode *, void *),
2929 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2930 int (*test)(struct inode *, void *), void *data);
2931 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2933 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2934 int (*test)(struct inode *, void *),
2935 int (*set)(struct inode *, void *),
2937 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2938 extern struct inode * iget_locked(struct super_block *, unsigned long);
2939 extern struct inode *find_inode_nowait(struct super_block *,
2941 int (*match)(struct inode *,
2942 unsigned long, void *),
2944 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2945 extern int insert_inode_locked(struct inode *);
2946 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2947 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2949 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2951 extern void unlock_new_inode(struct inode *);
2952 extern void discard_new_inode(struct inode *);
2953 extern unsigned int get_next_ino(void);
2954 extern void evict_inodes(struct super_block *sb);
2956 extern void __iget(struct inode * inode);
2957 extern void iget_failed(struct inode *);
2958 extern void clear_inode(struct inode *);
2959 extern void __destroy_inode(struct inode *);
2960 extern struct inode *new_inode_pseudo(struct super_block *sb);
2961 extern struct inode *new_inode(struct super_block *sb);
2962 extern void free_inode_nonrcu(struct inode *inode);
2963 extern int should_remove_suid(struct dentry *);
2964 extern int file_remove_privs(struct file *);
2966 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2967 static inline void insert_inode_hash(struct inode *inode)
2969 __insert_inode_hash(inode, inode->i_ino);
2972 extern void __remove_inode_hash(struct inode *);
2973 static inline void remove_inode_hash(struct inode *inode)
2975 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2976 __remove_inode_hash(inode);
2979 extern void inode_sb_list_add(struct inode *inode);
2982 extern int bdev_read_only(struct block_device *);
2984 extern int set_blocksize(struct block_device *, int);
2985 extern int sb_set_blocksize(struct super_block *, int);
2986 extern int sb_min_blocksize(struct super_block *, int);
2988 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2989 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2990 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2991 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
2992 struct file *file_out, loff_t pos_out,
2993 loff_t *count, unsigned int remap_flags);
2994 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2995 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2996 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2997 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2998 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3000 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3002 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3005 /* fs/block_dev.c */
3006 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3007 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3008 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3010 extern void block_sync_page(struct page *page);
3013 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3014 struct pipe_inode_info *, size_t, unsigned int);
3015 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3016 struct file *, loff_t *, size_t, unsigned int);
3017 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3018 struct file *out, loff_t *, size_t len, unsigned int flags);
3019 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3020 loff_t *opos, size_t len, unsigned int flags);
3024 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3025 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3026 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3027 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3028 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3029 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3030 int whence, loff_t maxsize, loff_t eof);
3031 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3032 int whence, loff_t size);
3033 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3034 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3035 extern int generic_file_open(struct inode * inode, struct file * filp);
3036 extern int nonseekable_open(struct inode * inode, struct file * filp);
3039 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3040 loff_t file_offset);
3043 /* need locking between buffered and direct access */
3046 /* filesystem does not support filling holes */
3047 DIO_SKIP_HOLES = 0x02,
3050 void dio_end_io(struct bio *bio);
3051 void dio_warn_stale_pagecache(struct file *filp);
3053 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3054 struct block_device *bdev, struct iov_iter *iter,
3055 get_block_t get_block,
3056 dio_iodone_t end_io, dio_submit_t submit_io,
3059 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3060 struct inode *inode,
3061 struct iov_iter *iter,
3062 get_block_t get_block)
3064 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3065 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3069 void inode_dio_wait(struct inode *inode);
3072 * inode_dio_begin - signal start of a direct I/O requests
3073 * @inode: inode the direct I/O happens on
3075 * This is called once we've finished processing a direct I/O request,
3076 * and is used to wake up callers waiting for direct I/O to be quiesced.
3078 static inline void inode_dio_begin(struct inode *inode)
3080 atomic_inc(&inode->i_dio_count);
3084 * inode_dio_end - signal finish of a direct I/O requests
3085 * @inode: inode the direct I/O happens on
3087 * This is called once we've finished processing a direct I/O request,
3088 * and is used to wake up callers waiting for direct I/O to be quiesced.
3090 static inline void inode_dio_end(struct inode *inode)
3092 if (atomic_dec_and_test(&inode->i_dio_count))
3093 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3096 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3099 extern const struct file_operations generic_ro_fops;
3101 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3103 extern int readlink_copy(char __user *, int, const char *);
3104 extern int page_readlink(struct dentry *, char __user *, int);
3105 extern const char *page_get_link(struct dentry *, struct inode *,
3106 struct delayed_call *);
3107 extern void page_put_link(void *);
3108 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3110 extern int page_symlink(struct inode *inode, const char *symname, int len);
3111 extern const struct inode_operations page_symlink_inode_operations;
3112 extern void kfree_link(void *);
3113 extern void generic_fillattr(struct inode *, struct kstat *);
3114 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3115 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3116 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3117 void inode_add_bytes(struct inode *inode, loff_t bytes);
3118 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3119 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3120 static inline loff_t __inode_get_bytes(struct inode *inode)
3122 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3124 loff_t inode_get_bytes(struct inode *inode);
3125 void inode_set_bytes(struct inode *inode, loff_t bytes);
3126 const char *simple_get_link(struct dentry *, struct inode *,
3127 struct delayed_call *);
3128 extern const struct inode_operations simple_symlink_inode_operations;
3130 extern int iterate_dir(struct file *, struct dir_context *);
3132 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3133 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3135 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3137 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3138 stat, STATX_BASIC_STATS);
3140 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3142 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3143 stat, STATX_BASIC_STATS);
3145 static inline int vfs_fstatat(int dfd, const char __user *filename,
3146 struct kstat *stat, int flags)
3148 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3149 stat, STATX_BASIC_STATS);
3151 static inline int vfs_fstat(int fd, struct kstat *stat)
3153 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3157 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3158 extern int vfs_readlink(struct dentry *, char __user *, int);
3160 extern int __generic_block_fiemap(struct inode *inode,
3161 struct fiemap_extent_info *fieinfo,
3162 loff_t start, loff_t len,
3163 get_block_t *get_block);
3164 extern int generic_block_fiemap(struct inode *inode,
3165 struct fiemap_extent_info *fieinfo, u64 start,
3166 u64 len, get_block_t *get_block);
3168 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3169 extern void put_filesystem(struct file_system_type *fs);
3170 extern struct file_system_type *get_fs_type(const char *name);
3171 extern struct super_block *get_super(struct block_device *);
3172 extern struct super_block *get_super_thawed(struct block_device *);
3173 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3174 extern struct super_block *get_active_super(struct block_device *bdev);
3175 extern void drop_super(struct super_block *sb);
3176 extern void drop_super_exclusive(struct super_block *sb);
3177 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3178 extern void iterate_supers_type(struct file_system_type *,
3179 void (*)(struct super_block *, void *), void *);
3181 extern int dcache_dir_open(struct inode *, struct file *);
3182 extern int dcache_dir_close(struct inode *, struct file *);
3183 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3184 extern int dcache_readdir(struct file *, struct dir_context *);
3185 extern int simple_setattr(struct dentry *, struct iattr *);
3186 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3187 extern int simple_statfs(struct dentry *, struct kstatfs *);
3188 extern int simple_open(struct inode *inode, struct file *file);
3189 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3190 extern int simple_unlink(struct inode *, struct dentry *);
3191 extern int simple_rmdir(struct inode *, struct dentry *);
3192 extern int simple_rename(struct inode *, struct dentry *,
3193 struct inode *, struct dentry *, unsigned int);
3194 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3195 extern int noop_set_page_dirty(struct page *page);
3196 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3197 unsigned int length);
3198 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3199 extern int simple_empty(struct dentry *);
3200 extern int simple_readpage(struct file *file, struct page *page);
3201 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3202 loff_t pos, unsigned len, unsigned flags,
3203 struct page **pagep, void **fsdata);
3204 extern int simple_write_end(struct file *file, struct address_space *mapping,
3205 loff_t pos, unsigned len, unsigned copied,
3206 struct page *page, void *fsdata);
3207 extern int always_delete_dentry(const struct dentry *);
3208 extern struct inode *alloc_anon_inode(struct super_block *);
3209 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3210 extern const struct dentry_operations simple_dentry_operations;
3212 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3213 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3214 extern const struct file_operations simple_dir_operations;
3215 extern const struct inode_operations simple_dir_inode_operations;
3216 extern void make_empty_dir_inode(struct inode *inode);
3217 extern bool is_empty_dir_inode(struct inode *inode);
3218 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3219 struct dentry *d_alloc_name(struct dentry *, const char *);
3220 extern int simple_fill_super(struct super_block *, unsigned long,
3221 const struct tree_descr *);
3222 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3223 extern void simple_release_fs(struct vfsmount **mount, int *count);
3225 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3226 loff_t *ppos, const void *from, size_t available);
3227 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3228 const void __user *from, size_t count);
3230 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3231 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3233 extern int generic_check_addressable(unsigned, u64);
3235 #ifdef CONFIG_MIGRATION
3236 extern int buffer_migrate_page(struct address_space *,
3237 struct page *, struct page *,
3240 #define buffer_migrate_page NULL
3243 extern int setattr_prepare(struct dentry *, struct iattr *);
3244 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3245 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3247 extern int file_update_time(struct file *file);
3249 static inline bool io_is_direct(struct file *filp)
3251 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3254 static inline bool vma_is_dax(struct vm_area_struct *vma)
3256 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3259 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3261 struct inode *inode;
3265 if (!vma_is_dax(vma))
3267 inode = file_inode(vma->vm_file);
3268 if (S_ISCHR(inode->i_mode))
3269 return false; /* device-dax */
3273 static inline int iocb_flags(struct file *file)
3276 if (file->f_flags & O_APPEND)
3278 if (io_is_direct(file))
3280 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3282 if (file->f_flags & __O_SYNC)
3287 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3289 if (unlikely(flags & ~RWF_SUPPORTED))
3292 if (flags & RWF_NOWAIT) {
3293 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3295 ki->ki_flags |= IOCB_NOWAIT;
3297 if (flags & RWF_HIPRI)
3298 ki->ki_flags |= IOCB_HIPRI;
3299 if (flags & RWF_DSYNC)
3300 ki->ki_flags |= IOCB_DSYNC;
3301 if (flags & RWF_SYNC)
3302 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3303 if (flags & RWF_APPEND)
3304 ki->ki_flags |= IOCB_APPEND;
3308 static inline ino_t parent_ino(struct dentry *dentry)
3313 * Don't strictly need d_lock here? If the parent ino could change
3314 * then surely we'd have a deeper race in the caller?
3316 spin_lock(&dentry->d_lock);
3317 res = dentry->d_parent->d_inode->i_ino;
3318 spin_unlock(&dentry->d_lock);
3322 /* Transaction based IO helpers */
3325 * An argresp is stored in an allocated page and holds the
3326 * size of the argument or response, along with its content
3328 struct simple_transaction_argresp {
3333 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3335 char *simple_transaction_get(struct file *file, const char __user *buf,
3337 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3338 size_t size, loff_t *pos);
3339 int simple_transaction_release(struct inode *inode, struct file *file);
3341 void simple_transaction_set(struct file *file, size_t n);
3344 * simple attribute files
3346 * These attributes behave similar to those in sysfs:
3348 * Writing to an attribute immediately sets a value, an open file can be
3349 * written to multiple times.
3351 * Reading from an attribute creates a buffer from the value that might get
3352 * read with multiple read calls. When the attribute has been read
3353 * completely, no further read calls are possible until the file is opened
3356 * All attributes contain a text representation of a numeric value
3357 * that are accessed with the get() and set() functions.
3359 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3360 static int __fops ## _open(struct inode *inode, struct file *file) \
3362 __simple_attr_check_format(__fmt, 0ull); \
3363 return simple_attr_open(inode, file, __get, __set, __fmt); \
3365 static const struct file_operations __fops = { \
3366 .owner = THIS_MODULE, \
3367 .open = __fops ## _open, \
3368 .release = simple_attr_release, \
3369 .read = simple_attr_read, \
3370 .write = simple_attr_write, \
3371 .llseek = generic_file_llseek, \
3374 static inline __printf(1, 2)
3375 void __simple_attr_check_format(const char *fmt, ...)
3377 /* don't do anything, just let the compiler check the arguments; */
3380 int simple_attr_open(struct inode *inode, struct file *file,
3381 int (*get)(void *, u64 *), int (*set)(void *, u64),
3383 int simple_attr_release(struct inode *inode, struct file *file);
3384 ssize_t simple_attr_read(struct file *file, char __user *buf,
3385 size_t len, loff_t *ppos);
3386 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3387 size_t len, loff_t *ppos);
3390 int proc_nr_files(struct ctl_table *table, int write,
3391 void __user *buffer, size_t *lenp, loff_t *ppos);
3392 int proc_nr_dentry(struct ctl_table *table, int write,
3393 void __user *buffer, size_t *lenp, loff_t *ppos);
3394 int proc_nr_inodes(struct ctl_table *table, int write,
3395 void __user *buffer, size_t *lenp, loff_t *ppos);
3396 int __init get_filesystem_list(char *buf);
3398 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3399 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3401 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3402 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3403 (flag & __FMODE_NONOTIFY)))
3405 static inline bool is_sxid(umode_t mode)
3407 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3410 static inline int check_sticky(struct inode *dir, struct inode *inode)
3412 if (!(dir->i_mode & S_ISVTX))
3415 return __check_sticky(dir, inode);
3418 static inline void inode_has_no_xattr(struct inode *inode)
3420 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3421 inode->i_flags |= S_NOSEC;
3424 static inline bool is_root_inode(struct inode *inode)
3426 return inode == inode->i_sb->s_root->d_inode;
3429 static inline bool dir_emit(struct dir_context *ctx,
3430 const char *name, int namelen,
3431 u64 ino, unsigned type)
3433 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3435 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3437 return ctx->actor(ctx, ".", 1, ctx->pos,
3438 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3440 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3442 return ctx->actor(ctx, "..", 2, ctx->pos,
3443 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3445 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3447 if (ctx->pos == 0) {
3448 if (!dir_emit_dot(file, ctx))
3452 if (ctx->pos == 1) {
3453 if (!dir_emit_dotdot(file, ctx))
3459 static inline bool dir_relax(struct inode *inode)
3461 inode_unlock(inode);
3463 return !IS_DEADDIR(inode);
3466 static inline bool dir_relax_shared(struct inode *inode)
3468 inode_unlock_shared(inode);
3469 inode_lock_shared(inode);
3470 return !IS_DEADDIR(inode);
3473 extern bool path_noexec(const struct path *path);
3474 extern void inode_nohighmem(struct inode *inode);
3477 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3480 #endif /* _LINUX_FS_H */