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 is stream-like */
157 #define FMODE_STREAM ((__force fmode_t)0x200000)
159 /* File was opened by fanotify and shouldn't generate fanotify events */
160 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
162 /* File is capable of returning -EAGAIN if I/O will block */
163 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
165 /* File does not contribute to nr_files count */
166 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
169 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
170 * that indicates that they should check the contents of the iovec are
171 * valid, but not check the memory that the iovec elements
174 #define CHECK_IOVEC_ONLY -1
177 * Attribute flags. These should be or-ed together to figure out what
180 #define ATTR_MODE (1 << 0)
181 #define ATTR_UID (1 << 1)
182 #define ATTR_GID (1 << 2)
183 #define ATTR_SIZE (1 << 3)
184 #define ATTR_ATIME (1 << 4)
185 #define ATTR_MTIME (1 << 5)
186 #define ATTR_CTIME (1 << 6)
187 #define ATTR_ATIME_SET (1 << 7)
188 #define ATTR_MTIME_SET (1 << 8)
189 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
190 #define ATTR_KILL_SUID (1 << 11)
191 #define ATTR_KILL_SGID (1 << 12)
192 #define ATTR_FILE (1 << 13)
193 #define ATTR_KILL_PRIV (1 << 14)
194 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
195 #define ATTR_TIMES_SET (1 << 16)
196 #define ATTR_TOUCH (1 << 17)
199 * Whiteout is represented by a char device. The following constants define the
200 * mode and device number to use.
202 #define WHITEOUT_MODE 0
203 #define WHITEOUT_DEV 0
206 * This is the Inode Attributes structure, used for notify_change(). It
207 * uses the above definitions as flags, to know which values have changed.
208 * Also, in this manner, a Filesystem can look at only the values it cares
209 * about. Basically, these are the attributes that the VFS layer can
210 * request to change from the FS layer.
212 * Derek Atkins <warlord@MIT.EDU> 94-10-20
215 unsigned int ia_valid;
220 struct timespec64 ia_atime;
221 struct timespec64 ia_mtime;
222 struct timespec64 ia_ctime;
225 * Not an attribute, but an auxiliary info for filesystems wanting to
226 * implement an ftruncate() like method. NOTE: filesystem should
227 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
229 struct file *ia_file;
233 * Includes for diskquotas.
235 #include <linux/quota.h>
238 * Maximum number of layers of fs stack. Needs to be limited to
239 * prevent kernel stack overflow
241 #define FILESYSTEM_MAX_STACK_DEPTH 2
244 * enum positive_aop_returns - aop return codes with specific semantics
246 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
247 * completed, that the page is still locked, and
248 * should be considered active. The VM uses this hint
249 * to return the page to the active list -- it won't
250 * be a candidate for writeback again in the near
251 * future. Other callers must be careful to unlock
252 * the page if they get this return. Returned by
255 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
256 * unlocked it and the page might have been truncated.
257 * The caller should back up to acquiring a new page and
258 * trying again. The aop will be taking reasonable
259 * precautions not to livelock. If the caller held a page
260 * reference, it should drop it before retrying. Returned
263 * address_space_operation functions return these large constants to indicate
264 * special semantics to the caller. These are much larger than the bytes in a
265 * page to allow for functions that return the number of bytes operated on in a
269 enum positive_aop_returns {
270 AOP_WRITEPAGE_ACTIVATE = 0x80000,
271 AOP_TRUNCATED_PAGE = 0x80001,
274 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
275 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
276 * helper code (eg buffer layer)
277 * to clear GFP_FS from alloc */
280 * oh the beauties of C type declarations.
283 struct address_space;
284 struct writeback_control;
287 * Write life time hint values.
288 * Stored in struct inode as u8.
291 WRITE_LIFE_NOT_SET = 0,
292 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
293 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
294 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
295 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
296 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
299 #define IOCB_EVENTFD (1 << 0)
300 #define IOCB_APPEND (1 << 1)
301 #define IOCB_DIRECT (1 << 2)
302 #define IOCB_HIPRI (1 << 3)
303 #define IOCB_DSYNC (1 << 4)
304 #define IOCB_SYNC (1 << 5)
305 #define IOCB_WRITE (1 << 6)
306 #define IOCB_NOWAIT (1 << 7)
309 struct file *ki_filp;
311 /* The 'ki_filp' pointer is shared in a union for aio */
312 randomized_struct_fields_start
315 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
319 u16 ki_ioprio; /* See linux/ioprio.h */
321 randomized_struct_fields_end
324 static inline bool is_sync_kiocb(struct kiocb *kiocb)
326 return kiocb->ki_complete == NULL;
330 * "descriptor" for what we're up to with a read.
331 * This allows us to use the same read code yet
332 * have multiple different users of the data that
333 * we read from a file.
335 * The simplest case just copies the data to user
348 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
349 unsigned long, unsigned long);
351 struct address_space_operations {
352 int (*writepage)(struct page *page, struct writeback_control *wbc);
353 int (*readpage)(struct file *, struct page *);
355 /* Write back some dirty pages from this mapping. */
356 int (*writepages)(struct address_space *, struct writeback_control *);
358 /* Set a page dirty. Return true if this dirtied it */
359 int (*set_page_dirty)(struct page *page);
362 * Reads in the requested pages. Unlike ->readpage(), this is
363 * PURELY used for read-ahead!.
365 int (*readpages)(struct file *filp, struct address_space *mapping,
366 struct list_head *pages, unsigned nr_pages);
368 int (*write_begin)(struct file *, struct address_space *mapping,
369 loff_t pos, unsigned len, unsigned flags,
370 struct page **pagep, void **fsdata);
371 int (*write_end)(struct file *, struct address_space *mapping,
372 loff_t pos, unsigned len, unsigned copied,
373 struct page *page, void *fsdata);
375 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
376 sector_t (*bmap)(struct address_space *, sector_t);
377 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
378 int (*releasepage) (struct page *, gfp_t);
379 void (*freepage)(struct page *);
380 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
382 * migrate the contents of a page to the specified target. If
383 * migrate_mode is MIGRATE_ASYNC, it must not block.
385 int (*migratepage) (struct address_space *,
386 struct page *, struct page *, enum migrate_mode);
387 bool (*isolate_page)(struct page *, isolate_mode_t);
388 void (*putback_page)(struct page *);
389 int (*launder_page) (struct page *);
390 int (*is_partially_uptodate) (struct page *, unsigned long,
392 void (*is_dirty_writeback) (struct page *, bool *, bool *);
393 int (*error_remove_page)(struct address_space *, struct page *);
395 /* swapfile support */
396 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
398 void (*swap_deactivate)(struct file *file);
401 extern const struct address_space_operations empty_aops;
404 * pagecache_write_begin/pagecache_write_end must be used by general code
405 * to write into the pagecache.
407 int pagecache_write_begin(struct file *, struct address_space *mapping,
408 loff_t pos, unsigned len, unsigned flags,
409 struct page **pagep, void **fsdata);
411 int pagecache_write_end(struct file *, struct address_space *mapping,
412 loff_t pos, unsigned len, unsigned copied,
413 struct page *page, void *fsdata);
415 struct address_space {
416 struct inode *host; /* owner: inode, block_device */
417 struct radix_tree_root i_pages; /* cached pages */
418 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
419 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
420 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
421 /* Protected by the i_pages lock */
422 unsigned long nrpages; /* number of total pages */
423 /* number of shadow or DAX exceptional entries */
424 unsigned long nrexceptional;
425 pgoff_t writeback_index;/* writeback starts here */
426 const struct address_space_operations *a_ops; /* methods */
427 unsigned long flags; /* error bits */
428 spinlock_t private_lock; /* for use by the address_space */
429 gfp_t gfp_mask; /* implicit gfp mask for allocations */
430 struct list_head private_list; /* for use by the address_space */
431 void *private_data; /* ditto */
433 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
435 * On most architectures that alignment is already the case; but
436 * must be enforced here for CRIS, to let the least significant bit
437 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
439 struct request_queue;
441 struct block_device {
442 dev_t bd_dev; /* not a kdev_t - it's a search key */
444 struct inode * bd_inode; /* will die */
445 struct super_block * bd_super;
446 struct mutex bd_mutex; /* open/close mutex */
450 bool bd_write_holder;
452 struct list_head bd_holder_disks;
454 struct block_device * bd_contains;
455 unsigned bd_block_size;
457 struct hd_struct * bd_part;
458 /* number of times partitions within this device have been opened. */
459 unsigned bd_part_count;
461 struct gendisk * bd_disk;
462 struct request_queue * bd_queue;
463 struct backing_dev_info *bd_bdi;
464 struct list_head bd_list;
466 * Private data. You must have bd_claim'ed the block_device
467 * to use this. NOTE: bd_claim allows an owner to claim
468 * the same device multiple times, the owner must take special
469 * care to not mess up bd_private for that case.
471 unsigned long bd_private;
473 /* The counter of freeze processes */
474 int bd_fsfreeze_count;
475 /* Mutex for freeze */
476 struct mutex bd_fsfreeze_mutex;
477 } __randomize_layout;
480 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
483 #define PAGECACHE_TAG_DIRTY 0
484 #define PAGECACHE_TAG_WRITEBACK 1
485 #define PAGECACHE_TAG_TOWRITE 2
487 int mapping_tagged(struct address_space *mapping, int tag);
489 static inline void i_mmap_lock_write(struct address_space *mapping)
491 down_write(&mapping->i_mmap_rwsem);
494 static inline void i_mmap_unlock_write(struct address_space *mapping)
496 up_write(&mapping->i_mmap_rwsem);
499 static inline void i_mmap_lock_read(struct address_space *mapping)
501 down_read(&mapping->i_mmap_rwsem);
504 static inline void i_mmap_unlock_read(struct address_space *mapping)
506 up_read(&mapping->i_mmap_rwsem);
510 * Might pages of this file be mapped into userspace?
512 static inline int mapping_mapped(struct address_space *mapping)
514 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
518 * Might pages of this file have been modified in userspace?
519 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
520 * marks vma as VM_SHARED if it is shared, and the file was opened for
521 * writing i.e. vma may be mprotected writable even if now readonly.
523 * If i_mmap_writable is negative, no new writable mappings are allowed. You
524 * can only deny writable mappings, if none exists right now.
526 static inline int mapping_writably_mapped(struct address_space *mapping)
528 return atomic_read(&mapping->i_mmap_writable) > 0;
531 static inline int mapping_map_writable(struct address_space *mapping)
533 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
537 static inline void mapping_unmap_writable(struct address_space *mapping)
539 atomic_dec(&mapping->i_mmap_writable);
542 static inline int mapping_deny_writable(struct address_space *mapping)
544 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
548 static inline void mapping_allow_writable(struct address_space *mapping)
550 atomic_inc(&mapping->i_mmap_writable);
554 * Use sequence counter to get consistent i_size on 32-bit processors.
556 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
557 #include <linux/seqlock.h>
558 #define __NEED_I_SIZE_ORDERED
559 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
561 #define i_size_ordered_init(inode) do { } while (0)
565 #define ACL_NOT_CACHED ((void *)(-1))
566 #define ACL_DONT_CACHE ((void *)(-3))
568 static inline struct posix_acl *
569 uncached_acl_sentinel(struct task_struct *task)
571 return (void *)task + 1;
575 is_uncached_acl(struct posix_acl *acl)
577 return (long)acl & 1;
580 #define IOP_FASTPERM 0x0001
581 #define IOP_LOOKUP 0x0002
582 #define IOP_NOFOLLOW 0x0004
583 #define IOP_XATTR 0x0008
584 #define IOP_DEFAULT_READLINK 0x0010
586 struct fsnotify_mark_connector;
589 * Keep mostly read-only and often accessed (especially for
590 * the RCU path lookup and 'stat' data) fields at the beginning
591 * of the 'struct inode'
595 unsigned short i_opflags;
598 unsigned int i_flags;
600 #ifdef CONFIG_FS_POSIX_ACL
601 struct posix_acl *i_acl;
602 struct posix_acl *i_default_acl;
605 const struct inode_operations *i_op;
606 struct super_block *i_sb;
607 struct address_space *i_mapping;
609 #ifdef CONFIG_SECURITY
613 /* Stat data, not accessed from path walking */
616 * Filesystems may only read i_nlink directly. They shall use the
617 * following functions for modification:
619 * (set|clear|inc|drop)_nlink
620 * inode_(inc|dec)_link_count
623 const unsigned int i_nlink;
624 unsigned int __i_nlink;
628 struct timespec64 i_atime;
629 struct timespec64 i_mtime;
630 struct timespec64 i_ctime;
631 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
632 unsigned short i_bytes;
637 #ifdef __NEED_I_SIZE_ORDERED
638 seqcount_t i_size_seqcount;
642 unsigned long i_state;
643 struct rw_semaphore i_rwsem;
645 unsigned long dirtied_when; /* jiffies of first dirtying */
646 unsigned long dirtied_time_when;
648 struct hlist_node i_hash;
649 struct list_head i_io_list; /* backing dev IO list */
650 #ifdef CONFIG_CGROUP_WRITEBACK
651 struct bdi_writeback *i_wb; /* the associated cgroup wb */
653 /* foreign inode detection, see wbc_detach_inode() */
655 u16 i_wb_frn_avg_time;
656 u16 i_wb_frn_history;
658 struct list_head i_lru; /* inode LRU list */
659 struct list_head i_sb_list;
660 struct list_head i_wb_list; /* backing dev writeback list */
662 struct hlist_head i_dentry;
663 struct rcu_head i_rcu;
665 atomic64_t i_version;
666 atomic64_t i_sequence; /* see futex */
668 atomic_t i_dio_count;
669 atomic_t i_writecount;
671 atomic_t i_readcount; /* struct files open RO */
673 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
674 struct file_lock_context *i_flctx;
675 struct address_space i_data;
676 struct list_head i_devices;
678 struct pipe_inode_info *i_pipe;
679 struct block_device *i_bdev;
687 #ifdef CONFIG_FSNOTIFY
688 __u32 i_fsnotify_mask; /* all events this inode cares about */
689 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
692 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
693 struct fscrypt_info *i_crypt_info;
696 void *i_private; /* fs or device private pointer */
697 } __randomize_layout;
699 static inline unsigned int i_blocksize(const struct inode *node)
701 return (1 << node->i_blkbits);
704 static inline int inode_unhashed(struct inode *inode)
706 return hlist_unhashed(&inode->i_hash);
710 * __mark_inode_dirty expects inodes to be hashed. Since we don't
711 * want special inodes in the fileset inode space, we make them
712 * appear hashed, but do not put on any lists. hlist_del()
713 * will work fine and require no locking.
715 static inline void inode_fake_hash(struct inode *inode)
717 hlist_add_fake(&inode->i_hash);
721 * inode->i_mutex nesting subclasses for the lock validator:
723 * 0: the object of the current VFS operation
727 * 4: second non-directory
728 * 5: second parent (when locking independent directories in rename)
730 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
731 * non-directories at once.
733 * The locking order between these classes is
734 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
736 enum inode_i_mutex_lock_class
746 static inline void inode_lock(struct inode *inode)
748 down_write(&inode->i_rwsem);
751 static inline void inode_unlock(struct inode *inode)
753 up_write(&inode->i_rwsem);
756 static inline void inode_lock_shared(struct inode *inode)
758 down_read(&inode->i_rwsem);
761 static inline void inode_unlock_shared(struct inode *inode)
763 up_read(&inode->i_rwsem);
766 static inline int inode_trylock(struct inode *inode)
768 return down_write_trylock(&inode->i_rwsem);
771 static inline int inode_trylock_shared(struct inode *inode)
773 return down_read_trylock(&inode->i_rwsem);
776 static inline int inode_is_locked(struct inode *inode)
778 return rwsem_is_locked(&inode->i_rwsem);
781 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
783 down_write_nested(&inode->i_rwsem, subclass);
786 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
788 down_read_nested(&inode->i_rwsem, subclass);
791 void lock_two_nondirectories(struct inode *, struct inode*);
792 void unlock_two_nondirectories(struct inode *, struct inode*);
795 * NOTE: in a 32bit arch with a preemptable kernel and
796 * an UP compile the i_size_read/write must be atomic
797 * with respect to the local cpu (unlike with preempt disabled),
798 * but they don't need to be atomic with respect to other cpus like in
799 * true SMP (so they need either to either locally disable irq around
800 * the read or for example on x86 they can be still implemented as a
801 * cmpxchg8b without the need of the lock prefix). For SMP compiles
802 * and 64bit archs it makes no difference if preempt is enabled or not.
804 static inline loff_t i_size_read(const struct inode *inode)
806 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
811 seq = read_seqcount_begin(&inode->i_size_seqcount);
812 i_size = inode->i_size;
813 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
815 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
819 i_size = inode->i_size;
823 return inode->i_size;
828 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
829 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
830 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
832 static inline void i_size_write(struct inode *inode, loff_t i_size)
834 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
836 write_seqcount_begin(&inode->i_size_seqcount);
837 inode->i_size = i_size;
838 write_seqcount_end(&inode->i_size_seqcount);
840 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
842 inode->i_size = i_size;
845 inode->i_size = i_size;
849 static inline unsigned iminor(const struct inode *inode)
851 return MINOR(inode->i_rdev);
854 static inline unsigned imajor(const struct inode *inode)
856 return MAJOR(inode->i_rdev);
859 extern struct block_device *I_BDEV(struct inode *inode);
862 rwlock_t lock; /* protects pid, uid, euid fields */
863 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
864 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
865 kuid_t uid, euid; /* uid/euid of process setting the owner */
866 int signum; /* posix.1b rt signal to be delivered on IO */
870 * Track a single file's readahead state
872 struct file_ra_state {
873 pgoff_t start; /* where readahead started */
874 unsigned int size; /* # of readahead pages */
875 unsigned int async_size; /* do asynchronous readahead when
876 there are only # of pages ahead */
878 unsigned int ra_pages; /* Maximum readahead window */
879 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
880 loff_t prev_pos; /* Cache last read() position */
884 * Check if @index falls in the readahead windows.
886 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
888 return (index >= ra->start &&
889 index < ra->start + ra->size);
894 struct llist_node fu_llist;
895 struct rcu_head fu_rcuhead;
898 struct inode *f_inode; /* cached value */
899 const struct file_operations *f_op;
902 * Protects f_ep_links, f_flags.
903 * Must not be taken from IRQ context.
906 enum rw_hint f_write_hint;
907 atomic_long_t f_count;
908 unsigned int f_flags;
910 struct mutex f_pos_lock;
912 struct fown_struct f_owner;
913 const struct cred *f_cred;
914 struct file_ra_state f_ra;
917 #ifdef CONFIG_SECURITY
920 /* needed for tty driver, and maybe others */
924 /* Used by fs/eventpoll.c to link all the hooks to this file */
925 struct list_head f_ep_links;
926 struct list_head f_tfile_llink;
927 #endif /* #ifdef CONFIG_EPOLL */
928 struct address_space *f_mapping;
931 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
936 /* file identifier */
937 unsigned char f_handle[];
940 static inline struct file *get_file(struct file *f)
942 atomic_long_inc(&f->f_count);
945 #define get_file_rcu_many(x, cnt) \
946 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
947 #define get_file_rcu(x) get_file_rcu_many((x), 1)
948 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
949 #define file_count(x) atomic_long_read(&(x)->f_count)
951 #define MAX_NON_LFS ((1UL<<31) - 1)
953 /* Page cache limit. The filesystems should put that into their s_maxbytes
954 limits, otherwise bad things can happen in VM. */
955 #if BITS_PER_LONG==32
956 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
957 #elif BITS_PER_LONG==64
958 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
963 #define FL_DELEG 4 /* NFSv4 delegation */
964 #define FL_ACCESS 8 /* not trying to lock, just looking */
965 #define FL_EXISTS 16 /* when unlocking, test for existence */
966 #define FL_LEASE 32 /* lease held on this file */
967 #define FL_CLOSE 64 /* unlock on close */
968 #define FL_SLEEP 128 /* A blocking lock */
969 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
970 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
971 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
972 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
974 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
977 * Special return value from posix_lock_file() and vfs_lock_file() for
978 * asynchronous locking.
980 #define FILE_LOCK_DEFERRED 1
982 /* legacy typedef, should eventually be removed */
983 typedef void *fl_owner_t;
987 struct file_lock_operations {
988 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
989 void (*fl_release_private)(struct file_lock *);
992 struct lock_manager_operations {
993 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
994 unsigned long (*lm_owner_key)(struct file_lock *);
995 fl_owner_t (*lm_get_owner)(fl_owner_t);
996 void (*lm_put_owner)(fl_owner_t);
997 void (*lm_notify)(struct file_lock *); /* unblock callback */
998 int (*lm_grant)(struct file_lock *, int);
999 bool (*lm_break)(struct file_lock *);
1000 int (*lm_change)(struct file_lock *, int, struct list_head *);
1001 void (*lm_setup)(struct file_lock *, void **);
1004 struct lock_manager {
1005 struct list_head list;
1007 * NFSv4 and up also want opens blocked during the grace period;
1014 void locks_start_grace(struct net *, struct lock_manager *);
1015 void locks_end_grace(struct lock_manager *);
1016 bool locks_in_grace(struct net *);
1017 bool opens_in_grace(struct net *);
1019 /* that will die - we need it for nfs_lock_info */
1020 #include <linux/nfs_fs_i.h>
1023 * struct file_lock represents a generic "file lock". It's used to represent
1024 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1025 * note that the same struct is used to represent both a request for a lock and
1026 * the lock itself, but the same object is never used for both.
1028 * FIXME: should we create a separate "struct lock_request" to help distinguish
1031 * The varous i_flctx lists are ordered by:
1034 * 2) lock range start
1037 * Obviously, the last two criteria only matter for POSIX locks.
1040 struct file_lock *fl_next; /* singly linked list for this inode */
1041 struct list_head fl_list; /* link into file_lock_context */
1042 struct hlist_node fl_link; /* node in global lists */
1043 struct list_head fl_block; /* circular list of blocked processes */
1044 fl_owner_t fl_owner;
1045 unsigned int fl_flags;
1046 unsigned char fl_type;
1047 unsigned int fl_pid;
1048 int fl_link_cpu; /* what cpu's list is this on? */
1049 wait_queue_head_t fl_wait;
1050 struct file *fl_file;
1054 struct fasync_struct * fl_fasync; /* for lease break notifications */
1055 /* for lease breaks: */
1056 unsigned long fl_break_time;
1057 unsigned long fl_downgrade_time;
1059 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1060 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1062 struct nfs_lock_info nfs_fl;
1063 struct nfs4_lock_info nfs4_fl;
1065 struct list_head link; /* link in AFS vnode's pending_locks list */
1066 int state; /* state of grant or error if -ve */
1069 } __randomize_layout;
1071 struct file_lock_context {
1072 spinlock_t flc_lock;
1073 struct list_head flc_flock;
1074 struct list_head flc_posix;
1075 struct list_head flc_lease;
1078 /* The following constant reflects the upper bound of the file/locking space */
1080 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1081 #define OFFSET_MAX INT_LIMIT(loff_t)
1082 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1085 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1087 #define locks_inode(f) file_inode(f)
1089 #ifdef CONFIG_FILE_LOCKING
1090 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1091 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1094 #if BITS_PER_LONG == 32
1095 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1096 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1100 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1101 extern int fcntl_getlease(struct file *filp);
1104 void locks_free_lock_context(struct inode *inode);
1105 void locks_free_lock(struct file_lock *fl);
1106 extern void locks_init_lock(struct file_lock *);
1107 extern struct file_lock * locks_alloc_lock(void);
1108 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1109 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1110 extern void locks_remove_posix(struct file *, fl_owner_t);
1111 extern void locks_remove_file(struct file *);
1112 extern void locks_release_private(struct file_lock *);
1113 extern void posix_test_lock(struct file *, struct file_lock *);
1114 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1115 extern int posix_unblock_lock(struct file_lock *);
1116 extern int vfs_test_lock(struct file *, struct file_lock *);
1117 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1118 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1119 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1120 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1121 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1122 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1123 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1124 extern int lease_modify(struct file_lock *, int, struct list_head *);
1125 struct files_struct;
1126 extern void show_fd_locks(struct seq_file *f,
1127 struct file *filp, struct files_struct *files);
1128 #else /* !CONFIG_FILE_LOCKING */
1129 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1130 struct flock __user *user)
1135 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1136 unsigned int cmd, struct flock __user *user)
1141 #if BITS_PER_LONG == 32
1142 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1143 struct flock64 __user *user)
1148 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1149 unsigned int cmd, struct flock64 __user *user)
1154 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1159 static inline int fcntl_getlease(struct file *filp)
1165 locks_free_lock_context(struct inode *inode)
1169 static inline void locks_init_lock(struct file_lock *fl)
1174 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1179 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1184 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1189 static inline void locks_remove_file(struct file *filp)
1194 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1199 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1200 struct file_lock *conflock)
1205 static inline int posix_unblock_lock(struct file_lock *waiter)
1210 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1215 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1216 struct file_lock *fl, struct file_lock *conf)
1221 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1226 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1231 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1236 static inline void lease_get_mtime(struct inode *inode,
1237 struct timespec64 *time)
1242 static inline int generic_setlease(struct file *filp, long arg,
1243 struct file_lock **flp, void **priv)
1248 static inline int vfs_setlease(struct file *filp, long arg,
1249 struct file_lock **lease, void **priv)
1254 static inline int lease_modify(struct file_lock *fl, int arg,
1255 struct list_head *dispose)
1260 struct files_struct;
1261 static inline void show_fd_locks(struct seq_file *f,
1262 struct file *filp, struct files_struct *files) {}
1263 #endif /* !CONFIG_FILE_LOCKING */
1265 static inline struct inode *file_inode(const struct file *f)
1270 static inline struct dentry *file_dentry(const struct file *file)
1272 return d_real(file->f_path.dentry, file_inode(file));
1275 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1277 return locks_lock_inode_wait(locks_inode(filp), fl);
1280 struct fasync_struct {
1284 struct fasync_struct *fa_next; /* singly linked list */
1285 struct file *fa_file;
1286 struct rcu_head fa_rcu;
1289 #define FASYNC_MAGIC 0x4601
1291 /* SMP safe fasync helpers: */
1292 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1293 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1294 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1295 extern struct fasync_struct *fasync_alloc(void);
1296 extern void fasync_free(struct fasync_struct *);
1298 /* can be called from interrupts */
1299 extern void kill_fasync(struct fasync_struct **, int, int);
1301 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1302 extern int f_setown(struct file *filp, unsigned long arg, int force);
1303 extern void f_delown(struct file *filp);
1304 extern pid_t f_getown(struct file *filp);
1305 extern int send_sigurg(struct fown_struct *fown);
1308 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1309 * represented in both.
1311 #define SB_RDONLY 1 /* Mount read-only */
1312 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1313 #define SB_NODEV 4 /* Disallow access to device special files */
1314 #define SB_NOEXEC 8 /* Disallow program execution */
1315 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1316 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1317 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1318 #define SB_NOATIME 1024 /* Do not update access times. */
1319 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1320 #define SB_SILENT 32768
1321 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1322 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1323 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1324 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1326 /* These sb flags are internal to the kernel */
1327 #define SB_SUBMOUNT (1<<26)
1328 #define SB_NOSEC (1<<28)
1329 #define SB_BORN (1<<29)
1330 #define SB_ACTIVE (1<<30)
1331 #define SB_NOUSER (1<<31)
1337 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1338 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1339 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1340 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1341 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1344 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1345 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1346 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1347 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1349 /* sb->s_iflags to limit user namespace mounts */
1350 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1351 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1352 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1354 /* Possible states of 'frozen' field */
1356 SB_UNFROZEN = 0, /* FS is unfrozen */
1357 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1358 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1359 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1360 * internal threads if needed) */
1361 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1364 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1367 int frozen; /* Is sb frozen? */
1368 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1369 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1372 struct super_block {
1373 struct list_head s_list; /* Keep this first */
1374 dev_t s_dev; /* search index; _not_ kdev_t */
1375 unsigned char s_blocksize_bits;
1376 unsigned long s_blocksize;
1377 loff_t s_maxbytes; /* Max file size */
1378 struct file_system_type *s_type;
1379 const struct super_operations *s_op;
1380 const struct dquot_operations *dq_op;
1381 const struct quotactl_ops *s_qcop;
1382 const struct export_operations *s_export_op;
1383 unsigned long s_flags;
1384 unsigned long s_iflags; /* internal SB_I_* flags */
1385 unsigned long s_magic;
1386 struct dentry *s_root;
1387 struct rw_semaphore s_umount;
1390 #ifdef CONFIG_SECURITY
1393 const struct xattr_handler **s_xattr;
1394 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1395 const struct fscrypt_operations *s_cop;
1397 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1398 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1399 struct block_device *s_bdev;
1400 struct backing_dev_info *s_bdi;
1401 struct mtd_info *s_mtd;
1402 struct hlist_node s_instances;
1403 unsigned int s_quota_types; /* Bitmask of supported quota types */
1404 struct quota_info s_dquot; /* Diskquota specific options */
1406 struct sb_writers s_writers;
1408 char s_id[32]; /* Informational name */
1409 uuid_t s_uuid; /* UUID */
1411 void *s_fs_info; /* Filesystem private info */
1412 unsigned int s_max_links;
1415 /* Granularity of c/m/atime in ns.
1416 Cannot be worse than a second */
1420 * The next field is for VFS *only*. No filesystems have any business
1421 * even looking at it. You had been warned.
1423 struct mutex s_vfs_rename_mutex; /* Kludge */
1426 * Filesystem subtype. If non-empty the filesystem type field
1427 * in /proc/mounts will be "type.subtype"
1431 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1434 * Saved pool identifier for cleancache (-1 means none)
1436 int cleancache_poolid;
1438 struct shrinker s_shrink; /* per-sb shrinker handle */
1440 /* Number of inodes with nlink == 0 but still referenced */
1441 atomic_long_t s_remove_count;
1443 /* Pending fsnotify inode refs */
1444 atomic_long_t s_fsnotify_inode_refs;
1446 /* Being remounted read-only */
1447 int s_readonly_remount;
1449 /* AIO completions deferred from interrupt context */
1450 struct workqueue_struct *s_dio_done_wq;
1451 struct hlist_head s_pins;
1454 * Owning user namespace and default context in which to
1455 * interpret filesystem uids, gids, quotas, device nodes,
1456 * xattrs and security labels.
1458 struct user_namespace *s_user_ns;
1461 * Keep the lru lists last in the structure so they always sit on their
1462 * own individual cachelines.
1464 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1465 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1466 struct rcu_head rcu;
1467 struct work_struct destroy_work;
1469 struct mutex s_sync_lock; /* sync serialisation lock */
1472 * Indicates how deep in a filesystem stack this SB is
1476 /* s_inode_list_lock protects s_inodes */
1477 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1478 struct list_head s_inodes; /* all inodes */
1480 spinlock_t s_inode_wblist_lock;
1481 struct list_head s_inodes_wb; /* writeback inodes */
1482 } __randomize_layout;
1484 /* Helper functions so that in most cases filesystems will
1485 * not need to deal directly with kuid_t and kgid_t and can
1486 * instead deal with the raw numeric values that are stored
1487 * in the filesystem.
1489 static inline uid_t i_uid_read(const struct inode *inode)
1491 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1494 static inline gid_t i_gid_read(const struct inode *inode)
1496 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1499 static inline void i_uid_write(struct inode *inode, uid_t uid)
1501 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1504 static inline void i_gid_write(struct inode *inode, gid_t gid)
1506 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1509 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1510 extern struct timespec64 current_time(struct inode *inode);
1513 * Snapshotting support.
1516 void __sb_end_write(struct super_block *sb, int level);
1517 int __sb_start_write(struct super_block *sb, int level, bool wait);
1519 #define __sb_writers_acquired(sb, lev) \
1520 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1521 #define __sb_writers_release(sb, lev) \
1522 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1525 * sb_end_write - drop write access to a superblock
1526 * @sb: the super we wrote to
1528 * Decrement number of writers to the filesystem. Wake up possible waiters
1529 * wanting to freeze the filesystem.
1531 static inline void sb_end_write(struct super_block *sb)
1533 __sb_end_write(sb, SB_FREEZE_WRITE);
1537 * sb_end_pagefault - drop write access to a superblock from a page fault
1538 * @sb: the super we wrote to
1540 * Decrement number of processes handling write page fault to the filesystem.
1541 * Wake up possible waiters wanting to freeze the filesystem.
1543 static inline void sb_end_pagefault(struct super_block *sb)
1545 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1549 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1550 * @sb: the super we wrote to
1552 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1553 * waiters wanting to freeze the filesystem.
1555 static inline void sb_end_intwrite(struct super_block *sb)
1557 __sb_end_write(sb, SB_FREEZE_FS);
1561 * sb_start_write - get write access to a superblock
1562 * @sb: the super we write to
1564 * When a process wants to write data or metadata to a file system (i.e. dirty
1565 * a page or an inode), it should embed the operation in a sb_start_write() -
1566 * sb_end_write() pair to get exclusion against file system freezing. This
1567 * function increments number of writers preventing freezing. If the file
1568 * system is already frozen, the function waits until the file system is
1571 * Since freeze protection behaves as a lock, users have to preserve
1572 * ordering of freeze protection and other filesystem locks. Generally,
1573 * freeze protection should be the outermost lock. In particular, we have:
1576 * -> i_mutex (write path, truncate, directory ops, ...)
1577 * -> s_umount (freeze_super, thaw_super)
1579 static inline void sb_start_write(struct super_block *sb)
1581 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1584 static inline int sb_start_write_trylock(struct super_block *sb)
1586 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1590 * sb_start_pagefault - get write access to a superblock from a page fault
1591 * @sb: the super we write to
1593 * When a process starts handling write page fault, it should embed the
1594 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1595 * exclusion against file system freezing. This is needed since the page fault
1596 * is going to dirty a page. This function increments number of running page
1597 * faults preventing freezing. If the file system is already frozen, the
1598 * function waits until the file system is thawed.
1600 * Since page fault freeze protection behaves as a lock, users have to preserve
1601 * ordering of freeze protection and other filesystem locks. It is advised to
1602 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1603 * handling code implies lock dependency:
1606 * -> sb_start_pagefault
1608 static inline void sb_start_pagefault(struct super_block *sb)
1610 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1614 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1615 * @sb: the super we write to
1617 * This is the third level of protection against filesystem freezing. It is
1618 * free for use by a filesystem. The only requirement is that it must rank
1619 * below sb_start_pagefault.
1621 * For example filesystem can call sb_start_intwrite() when starting a
1622 * transaction which somewhat eases handling of freezing for internal sources
1623 * of filesystem changes (internal fs threads, discarding preallocation on file
1626 static inline void sb_start_intwrite(struct super_block *sb)
1628 __sb_start_write(sb, SB_FREEZE_FS, true);
1631 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1633 return __sb_start_write(sb, SB_FREEZE_FS, false);
1637 extern bool inode_owner_or_capable(const struct inode *inode);
1640 * VFS helper functions..
1642 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1643 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1644 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1645 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1646 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1647 extern int vfs_rmdir(struct inode *, struct dentry *);
1648 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1649 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1650 extern int vfs_whiteout(struct inode *, struct dentry *);
1652 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1655 int vfs_mkobj(struct dentry *, umode_t,
1656 int (*f)(struct dentry *, umode_t, void *),
1659 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1662 * VFS file helper functions.
1664 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1666 extern bool may_open_dev(const struct path *path);
1668 * VFS FS_IOC_FIEMAP helper definitions.
1670 struct fiemap_extent_info {
1671 unsigned int fi_flags; /* Flags as passed from user */
1672 unsigned int fi_extents_mapped; /* Number of mapped extents */
1673 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1674 struct fiemap_extent __user *fi_extents_start; /* Start of
1675 fiemap_extent array */
1677 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1678 u64 phys, u64 len, u32 flags);
1679 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1684 * NOTE! These match bits 12..15 of stat.st_mode
1685 * (ie "(i_mode >> 12) & 15").
1687 #define DT_UNKNOWN 0
1698 * This is the "filldir" function type, used by readdir() to let
1699 * the kernel specify what kind of dirent layout it wants to have.
1700 * This allows the kernel to read directories into kernel space or
1701 * to have different dirent layouts depending on the binary type.
1704 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1707 struct dir_context {
1712 struct block_device_operations;
1714 /* These macros are for out of kernel modules to test that
1715 * the kernel supports the unlocked_ioctl and compat_ioctl
1716 * fields in struct file_operations. */
1717 #define HAVE_COMPAT_IOCTL 1
1718 #define HAVE_UNLOCKED_IOCTL 1
1721 * These flags let !MMU mmap() govern direct device mapping vs immediate
1722 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1724 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1725 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1726 * NOMMU_MAP_READ: Can be mapped for reading
1727 * NOMMU_MAP_WRITE: Can be mapped for writing
1728 * NOMMU_MAP_EXEC: Can be mapped for execution
1730 #define NOMMU_MAP_COPY 0x00000001
1731 #define NOMMU_MAP_DIRECT 0x00000008
1732 #define NOMMU_MAP_READ VM_MAYREAD
1733 #define NOMMU_MAP_WRITE VM_MAYWRITE
1734 #define NOMMU_MAP_EXEC VM_MAYEXEC
1736 #define NOMMU_VMFLAGS \
1737 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1742 struct file_operations {
1743 struct module *owner;
1744 loff_t (*llseek) (struct file *, loff_t, int);
1745 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1746 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1747 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1748 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1749 int (*iterate) (struct file *, struct dir_context *);
1750 int (*iterate_shared) (struct file *, struct dir_context *);
1751 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1752 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1753 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1754 int (*mmap) (struct file *, struct vm_area_struct *);
1755 unsigned long mmap_supported_flags;
1756 int (*open) (struct inode *, struct file *);
1757 int (*flush) (struct file *, fl_owner_t id);
1758 int (*release) (struct inode *, struct file *);
1759 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1760 int (*fasync) (int, struct file *, int);
1761 int (*lock) (struct file *, int, struct file_lock *);
1762 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1763 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1764 int (*check_flags)(int);
1765 int (*flock) (struct file *, int, struct file_lock *);
1766 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1767 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1768 int (*setlease)(struct file *, long, struct file_lock **, void **);
1769 long (*fallocate)(struct file *file, int mode, loff_t offset,
1771 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1773 unsigned (*mmap_capabilities)(struct file *);
1775 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1776 loff_t, size_t, unsigned int);
1777 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1779 int (*dedupe_file_range)(struct file *, loff_t, struct file *, loff_t,
1781 int (*fadvise)(struct file *, loff_t, loff_t, int);
1782 } __randomize_layout;
1784 struct inode_operations {
1785 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1786 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1787 int (*permission) (struct inode *, int);
1788 struct posix_acl * (*get_acl)(struct inode *, int);
1790 int (*readlink) (struct dentry *, char __user *,int);
1792 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1793 int (*link) (struct dentry *,struct inode *,struct dentry *);
1794 int (*unlink) (struct inode *,struct dentry *);
1795 int (*symlink) (struct inode *,struct dentry *,const char *);
1796 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1797 int (*rmdir) (struct inode *,struct dentry *);
1798 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1799 int (*rename) (struct inode *, struct dentry *,
1800 struct inode *, struct dentry *, unsigned int);
1801 int (*setattr) (struct dentry *, struct iattr *);
1802 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1803 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1804 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1806 int (*update_time)(struct inode *, struct timespec64 *, int);
1807 int (*atomic_open)(struct inode *, struct dentry *,
1808 struct file *, unsigned open_flag,
1809 umode_t create_mode);
1810 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1811 int (*set_acl)(struct inode *, struct posix_acl *, int);
1812 } ____cacheline_aligned;
1814 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1815 struct iov_iter *iter)
1817 return file->f_op->read_iter(kio, iter);
1820 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1821 struct iov_iter *iter)
1823 return file->f_op->write_iter(kio, iter);
1826 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1828 return file->f_op->mmap(file, vma);
1831 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1832 unsigned long nr_segs, unsigned long fast_segs,
1833 struct iovec *fast_pointer,
1834 struct iovec **ret_pointer);
1836 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1837 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1838 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1839 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1840 unsigned long, loff_t *, rwf_t);
1841 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1842 loff_t, size_t, unsigned int);
1843 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1844 struct inode *inode_out, loff_t pos_out,
1845 u64 *len, bool is_dedupe);
1846 extern int do_clone_file_range(struct file *file_in, loff_t pos_in,
1847 struct file *file_out, loff_t pos_out, u64 len);
1848 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1849 struct file *file_out, loff_t pos_out, u64 len);
1850 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1851 struct inode *dest, loff_t destoff,
1852 loff_t len, bool *is_same);
1853 extern int vfs_dedupe_file_range(struct file *file,
1854 struct file_dedupe_range *same);
1855 extern int vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1856 struct file *dst_file, loff_t dst_pos,
1860 struct super_operations {
1861 struct inode *(*alloc_inode)(struct super_block *sb);
1862 void (*destroy_inode)(struct inode *);
1864 void (*dirty_inode) (struct inode *, int flags);
1865 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1866 int (*drop_inode) (struct inode *);
1867 void (*evict_inode) (struct inode *);
1868 void (*put_super) (struct super_block *);
1869 int (*sync_fs)(struct super_block *sb, int wait);
1870 int (*freeze_super) (struct super_block *);
1871 int (*freeze_fs) (struct super_block *);
1872 int (*thaw_super) (struct super_block *);
1873 int (*unfreeze_fs) (struct super_block *);
1874 int (*statfs) (struct dentry *, struct kstatfs *);
1875 int (*remount_fs) (struct super_block *, int *, char *);
1876 void (*umount_begin) (struct super_block *);
1878 int (*show_options)(struct seq_file *, struct dentry *);
1879 int (*show_devname)(struct seq_file *, struct dentry *);
1880 int (*show_path)(struct seq_file *, struct dentry *);
1881 int (*show_stats)(struct seq_file *, struct dentry *);
1883 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1884 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1885 struct dquot **(*get_dquots)(struct inode *);
1887 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1888 long (*nr_cached_objects)(struct super_block *,
1889 struct shrink_control *);
1890 long (*free_cached_objects)(struct super_block *,
1891 struct shrink_control *);
1895 * Inode flags - they have no relation to superblock flags now
1897 #define S_SYNC 1 /* Writes are synced at once */
1898 #define S_NOATIME 2 /* Do not update access times */
1899 #define S_APPEND 4 /* Append-only file */
1900 #define S_IMMUTABLE 8 /* Immutable file */
1901 #define S_DEAD 16 /* removed, but still open directory */
1902 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1903 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1904 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1905 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1906 #define S_PRIVATE 512 /* Inode is fs-internal */
1907 #define S_IMA 1024 /* Inode has an associated IMA struct */
1908 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1909 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1910 #ifdef CONFIG_FS_DAX
1911 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1913 #define S_DAX 0 /* Make all the DAX code disappear */
1915 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1918 * Note that nosuid etc flags are inode-specific: setting some file-system
1919 * flags just means all the inodes inherit those flags by default. It might be
1920 * possible to override it selectively if you really wanted to with some
1921 * ioctl() that is not currently implemented.
1923 * Exception: SB_RDONLY is always applied to the entire file system.
1925 * Unfortunately, it is possible to change a filesystems flags with it mounted
1926 * with files in use. This means that all of the inodes will not have their
1927 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1928 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1930 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1932 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1933 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1934 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1935 ((inode)->i_flags & S_SYNC))
1936 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1937 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1938 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1939 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1940 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1942 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1943 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1944 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1945 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1947 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1948 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1949 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1950 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1951 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1952 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1953 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1954 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1955 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1957 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1958 (inode)->i_rdev == WHITEOUT_DEV)
1960 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1962 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1965 static inline enum rw_hint file_write_hint(struct file *file)
1967 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1968 return file->f_write_hint;
1970 return file_inode(file)->i_write_hint;
1973 static inline int iocb_flags(struct file *file);
1975 static inline u16 ki_hint_validate(enum rw_hint hint)
1977 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
1979 if (hint <= max_hint)
1984 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1986 *kiocb = (struct kiocb) {
1988 .ki_flags = iocb_flags(filp),
1989 .ki_hint = ki_hint_validate(file_write_hint(filp)),
1990 .ki_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, 0),
1995 * Inode state bits. Protected by inode->i_lock
1997 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1998 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2000 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2001 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2002 * various stages of removing an inode.
2004 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2006 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2007 * fdatasync(). i_atime is the usual cause.
2008 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2009 * these changes separately from I_DIRTY_SYNC so that we
2010 * don't have to write inode on fdatasync() when only
2011 * mtime has changed in it.
2012 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2013 * I_NEW Serves as both a mutex and completion notification.
2014 * New inodes set I_NEW. If two processes both create
2015 * the same inode, one of them will release its inode and
2016 * wait for I_NEW to be released before returning.
2017 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2018 * also cause waiting on I_NEW, without I_NEW actually
2019 * being set. find_inode() uses this to prevent returning
2020 * nearly-dead inodes.
2021 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2022 * is zero. I_FREEING must be set when I_WILL_FREE is
2024 * I_FREEING Set when inode is about to be freed but still has dirty
2025 * pages or buffers attached or the inode itself is still
2027 * I_CLEAR Added by clear_inode(). In this state the inode is
2028 * clean and can be destroyed. Inode keeps I_FREEING.
2030 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2031 * prohibited for many purposes. iget() must wait for
2032 * the inode to be completely released, then create it
2033 * anew. Other functions will just ignore such inodes,
2034 * if appropriate. I_NEW is used for waiting.
2036 * I_SYNC Writeback of inode is running. The bit is set during
2037 * data writeback, and cleared with a wakeup on the bit
2038 * address once it is done. The bit is also used to pin
2039 * the inode in memory for flusher thread.
2041 * I_REFERENCED Marks the inode as recently references on the LRU list.
2043 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2045 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2046 * synchronize competing switching instances and to tell
2047 * wb stat updates to grab the i_pages lock. See
2048 * inode_switch_wb_work_fn() for details.
2050 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2051 * and work dirs among overlayfs mounts.
2053 * I_CREATING New object's inode in the middle of setting up.
2055 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2056 * Used to detect that mark_inode_dirty() should not move
2057 * inode between dirty lists.
2059 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2061 #define I_DIRTY_SYNC (1 << 0)
2062 #define I_DIRTY_DATASYNC (1 << 1)
2063 #define I_DIRTY_PAGES (1 << 2)
2065 #define I_NEW (1 << __I_NEW)
2066 #define I_WILL_FREE (1 << 4)
2067 #define I_FREEING (1 << 5)
2068 #define I_CLEAR (1 << 6)
2070 #define I_SYNC (1 << __I_SYNC)
2071 #define I_REFERENCED (1 << 8)
2072 #define __I_DIO_WAKEUP 9
2073 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2074 #define I_LINKABLE (1 << 10)
2075 #define I_DIRTY_TIME (1 << 11)
2076 #define I_WB_SWITCH (1 << 13)
2077 #define I_OVL_INUSE (1 << 14)
2078 #define I_CREATING (1 << 15)
2079 #define I_SYNC_QUEUED (1 << 17)
2081 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2082 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2083 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2085 extern void __mark_inode_dirty(struct inode *, int);
2086 static inline void mark_inode_dirty(struct inode *inode)
2088 __mark_inode_dirty(inode, I_DIRTY);
2091 static inline void mark_inode_dirty_sync(struct inode *inode)
2093 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2096 extern void inc_nlink(struct inode *inode);
2097 extern void drop_nlink(struct inode *inode);
2098 extern void clear_nlink(struct inode *inode);
2099 extern void set_nlink(struct inode *inode, unsigned int nlink);
2101 static inline void inode_inc_link_count(struct inode *inode)
2104 mark_inode_dirty(inode);
2107 static inline void inode_dec_link_count(struct inode *inode)
2110 mark_inode_dirty(inode);
2113 enum file_time_flags {
2120 extern bool atime_needs_update(const struct path *, struct inode *);
2121 extern void touch_atime(const struct path *);
2122 static inline void file_accessed(struct file *file)
2124 if (!(file->f_flags & O_NOATIME))
2125 touch_atime(&file->f_path);
2128 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2129 int sync_inode_metadata(struct inode *inode, int wait);
2131 struct file_system_type {
2134 #define FS_REQUIRES_DEV 1
2135 #define FS_BINARY_MOUNTDATA 2
2136 #define FS_HAS_SUBTYPE 4
2137 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2138 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2139 struct dentry *(*mount) (struct file_system_type *, int,
2140 const char *, void *);
2141 void (*kill_sb) (struct super_block *);
2142 struct module *owner;
2143 struct file_system_type * next;
2144 struct hlist_head fs_supers;
2146 struct lock_class_key s_lock_key;
2147 struct lock_class_key s_umount_key;
2148 struct lock_class_key s_vfs_rename_key;
2149 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2151 struct lock_class_key i_lock_key;
2152 struct lock_class_key i_mutex_key;
2153 struct lock_class_key i_mutex_dir_key;
2156 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2158 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2159 int flags, void *data, void *ns, struct user_namespace *user_ns,
2160 int (*fill_super)(struct super_block *, void *, int));
2162 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2163 int flags, const char *dev_name, void *data,
2164 int (*fill_super)(struct super_block *, void *, int));
2166 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2167 int flags, const char *dev_name, void *data,
2168 int (*fill_super)(struct super_block *, void *, int))
2170 return ERR_PTR(-ENODEV);
2173 extern struct dentry *mount_single(struct file_system_type *fs_type,
2174 int flags, void *data,
2175 int (*fill_super)(struct super_block *, void *, int));
2176 extern struct dentry *mount_nodev(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_subtree(struct vfsmount *mnt, const char *path);
2180 void generic_shutdown_super(struct super_block *sb);
2182 void kill_block_super(struct super_block *sb);
2184 static inline void kill_block_super(struct super_block *sb)
2189 void kill_anon_super(struct super_block *sb);
2190 void kill_litter_super(struct super_block *sb);
2191 void deactivate_super(struct super_block *sb);
2192 void deactivate_locked_super(struct super_block *sb);
2193 int set_anon_super(struct super_block *s, void *data);
2194 int get_anon_bdev(dev_t *);
2195 void free_anon_bdev(dev_t);
2196 struct super_block *sget_userns(struct file_system_type *type,
2197 int (*test)(struct super_block *,void *),
2198 int (*set)(struct super_block *,void *),
2199 int flags, struct user_namespace *user_ns,
2201 struct super_block *sget(struct file_system_type *type,
2202 int (*test)(struct super_block *,void *),
2203 int (*set)(struct super_block *,void *),
2204 int flags, void *data);
2205 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2206 const struct super_operations *ops,
2207 const struct xattr_handler **xattr,
2208 const struct dentry_operations *dops,
2211 static inline struct dentry *
2212 mount_pseudo(struct file_system_type *fs_type, char *name,
2213 const struct super_operations *ops,
2214 const struct dentry_operations *dops, unsigned long magic)
2216 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2219 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2220 #define fops_get(fops) \
2221 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2222 #define fops_put(fops) \
2223 do { if (fops) module_put((fops)->owner); } while(0)
2225 * This one is to be used *ONLY* from ->open() instances.
2226 * fops must be non-NULL, pinned down *and* module dependencies
2227 * should be sufficient to pin the caller down as well.
2229 #define replace_fops(f, fops) \
2231 struct file *__file = (f); \
2232 fops_put(__file->f_op); \
2233 BUG_ON(!(__file->f_op = (fops))); \
2236 extern int register_filesystem(struct file_system_type *);
2237 extern int unregister_filesystem(struct file_system_type *);
2238 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2239 #define kern_mount(type) kern_mount_data(type, NULL)
2240 extern void kern_unmount(struct vfsmount *mnt);
2241 extern int may_umount_tree(struct vfsmount *);
2242 extern int may_umount(struct vfsmount *);
2243 extern long do_mount(const char *, const char __user *,
2244 const char *, unsigned long, void *);
2245 extern struct vfsmount *collect_mounts(const struct path *);
2246 extern void drop_collected_mounts(struct vfsmount *);
2247 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2249 extern int vfs_statfs(const struct path *, struct kstatfs *);
2250 extern int user_statfs(const char __user *, struct kstatfs *);
2251 extern int fd_statfs(int, struct kstatfs *);
2252 extern int freeze_super(struct super_block *super);
2253 extern int thaw_super(struct super_block *super);
2254 extern bool our_mnt(struct vfsmount *mnt);
2255 extern __printf(2, 3)
2256 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2257 extern int super_setup_bdi(struct super_block *sb);
2259 extern int current_umask(void);
2261 extern void ihold(struct inode * inode);
2262 extern void iput(struct inode *);
2263 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2266 extern struct kobject *fs_kobj;
2268 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2270 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2271 extern int locks_mandatory_locked(struct file *);
2272 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2275 * Candidates for mandatory locking have the setgid bit set
2276 * but no group execute bit - an otherwise meaningless combination.
2279 static inline int __mandatory_lock(struct inode *ino)
2281 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2285 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2286 * otherwise these will be advisory locks
2289 static inline int mandatory_lock(struct inode *ino)
2291 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2294 static inline int locks_verify_locked(struct file *file)
2296 if (mandatory_lock(locks_inode(file)))
2297 return locks_mandatory_locked(file);
2301 static inline int locks_verify_truncate(struct inode *inode,
2305 if (!inode->i_flctx || !mandatory_lock(inode))
2308 if (size < inode->i_size) {
2309 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2312 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2317 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2319 static inline int locks_mandatory_locked(struct file *file)
2324 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2325 loff_t start, loff_t end, unsigned char type)
2330 static inline int __mandatory_lock(struct inode *inode)
2335 static inline int mandatory_lock(struct inode *inode)
2340 static inline int locks_verify_locked(struct file *file)
2345 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2351 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2354 #ifdef CONFIG_FILE_LOCKING
2355 static inline int break_lease(struct inode *inode, unsigned int mode)
2358 * Since this check is lockless, we must ensure that any refcounts
2359 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2360 * could end up racing with tasks trying to set a new lease on this
2364 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2365 return __break_lease(inode, mode, FL_LEASE);
2369 static inline int break_deleg(struct inode *inode, unsigned int mode)
2372 * Since this check is lockless, we must ensure that any refcounts
2373 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2374 * could end up racing with tasks trying to set a new lease on this
2378 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2379 return __break_lease(inode, mode, FL_DELEG);
2383 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2387 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2388 if (ret == -EWOULDBLOCK && delegated_inode) {
2389 *delegated_inode = inode;
2395 static inline int break_deleg_wait(struct inode **delegated_inode)
2399 ret = break_deleg(*delegated_inode, O_WRONLY);
2400 iput(*delegated_inode);
2401 *delegated_inode = NULL;
2405 static inline int break_layout(struct inode *inode, bool wait)
2408 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2409 return __break_lease(inode,
2410 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2415 #else /* !CONFIG_FILE_LOCKING */
2416 static inline int break_lease(struct inode *inode, unsigned int mode)
2421 static inline int break_deleg(struct inode *inode, unsigned int mode)
2426 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2431 static inline int break_deleg_wait(struct inode **delegated_inode)
2437 static inline int break_layout(struct inode *inode, bool wait)
2442 #endif /* CONFIG_FILE_LOCKING */
2447 const char *name; /* pointer to actual string */
2448 const __user char *uptr; /* original userland pointer */
2450 struct audit_names *aname;
2454 extern long vfs_truncate(const struct path *, loff_t);
2455 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2457 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2459 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2461 extern struct file *file_open_name(struct filename *, int, umode_t);
2462 extern struct file *filp_open(const char *, int, umode_t);
2463 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2464 const char *, int, umode_t);
2465 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2466 extern struct file * open_with_fake_path(const struct path *, int,
2467 struct inode*, const struct cred *);
2468 static inline struct file *file_clone_open(struct file *file)
2470 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2472 extern int filp_close(struct file *, fl_owner_t id);
2474 extern struct filename *getname_flags(const char __user *, int, int *);
2475 extern struct filename *getname(const char __user *);
2476 extern struct filename *getname_kernel(const char *);
2477 extern void putname(struct filename *name);
2479 extern int finish_open(struct file *file, struct dentry *dentry,
2480 int (*open)(struct inode *, struct file *));
2481 extern int finish_no_open(struct file *file, struct dentry *dentry);
2485 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2488 extern void __init vfs_caches_init_early(void);
2489 extern void __init vfs_caches_init(void);
2491 extern struct kmem_cache *names_cachep;
2493 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2494 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2497 extern int register_blkdev(unsigned int, const char *);
2498 extern void unregister_blkdev(unsigned int, const char *);
2499 extern void bdev_unhash_inode(dev_t dev);
2500 extern struct block_device *bdget(dev_t);
2501 extern struct block_device *bdgrab(struct block_device *bdev);
2502 extern void bd_set_size(struct block_device *, loff_t size);
2503 extern void bd_forget(struct inode *inode);
2504 extern void bdput(struct block_device *);
2505 extern void invalidate_bdev(struct block_device *);
2506 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2507 extern int sync_blockdev(struct block_device *bdev);
2508 extern void kill_bdev(struct block_device *);
2509 extern struct super_block *freeze_bdev(struct block_device *);
2510 extern void emergency_thaw_all(void);
2511 extern void emergency_thaw_bdev(struct super_block *sb);
2512 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2513 extern int fsync_bdev(struct block_device *);
2515 extern struct super_block *blockdev_superblock;
2517 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2519 return sb == blockdev_superblock;
2522 static inline void bd_forget(struct inode *inode) {}
2523 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2524 static inline void kill_bdev(struct block_device *bdev) {}
2525 static inline void invalidate_bdev(struct block_device *bdev) {}
2527 static inline struct super_block *freeze_bdev(struct block_device *sb)
2532 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2537 static inline int emergency_thaw_bdev(struct super_block *sb)
2542 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2546 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2551 extern int sync_filesystem(struct super_block *);
2552 extern const struct file_operations def_blk_fops;
2553 extern const struct file_operations def_chr_fops;
2555 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2556 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2557 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2558 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2559 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2561 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2563 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2564 extern int __blkdev_reread_part(struct block_device *bdev);
2565 extern int blkdev_reread_part(struct block_device *bdev);
2568 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2569 extern void bd_unlink_disk_holder(struct block_device *bdev,
2570 struct gendisk *disk);
2572 static inline int bd_link_disk_holder(struct block_device *bdev,
2573 struct gendisk *disk)
2577 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2578 struct gendisk *disk)
2585 #define CHRDEV_MAJOR_MAX 512
2586 /* Marks the bottom of the first segment of free char majors */
2587 #define CHRDEV_MAJOR_DYN_END 234
2588 /* Marks the top and bottom of the second segment of free char majors */
2589 #define CHRDEV_MAJOR_DYN_EXT_START 511
2590 #define CHRDEV_MAJOR_DYN_EXT_END 384
2592 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2593 extern int register_chrdev_region(dev_t, unsigned, const char *);
2594 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2595 unsigned int count, const char *name,
2596 const struct file_operations *fops);
2597 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2598 unsigned int count, const char *name);
2599 extern void unregister_chrdev_region(dev_t, unsigned);
2600 extern void chrdev_show(struct seq_file *,off_t);
2602 static inline int register_chrdev(unsigned int major, const char *name,
2603 const struct file_operations *fops)
2605 return __register_chrdev(major, 0, 256, name, fops);
2608 static inline void unregister_chrdev(unsigned int major, const char *name)
2610 __unregister_chrdev(major, 0, 256, name);
2613 /* fs/block_dev.c */
2614 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2615 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2618 #define BLKDEV_MAJOR_MAX 512
2619 extern const char *__bdevname(dev_t, char *buffer);
2620 extern const char *bdevname(struct block_device *bdev, char *buffer);
2621 extern struct block_device *lookup_bdev(const char *);
2622 extern void blkdev_show(struct seq_file *,off_t);
2625 #define BLKDEV_MAJOR_MAX 0
2628 extern void init_special_inode(struct inode *, umode_t, dev_t);
2630 /* Invalid inode operations -- fs/bad_inode.c */
2631 extern void make_bad_inode(struct inode *);
2632 extern bool is_bad_inode(struct inode *);
2635 extern void check_disk_size_change(struct gendisk *disk,
2636 struct block_device *bdev, bool verbose);
2637 extern int revalidate_disk(struct gendisk *);
2638 extern int check_disk_change(struct block_device *);
2639 extern int __invalidate_device(struct block_device *, bool);
2640 extern int invalidate_partition(struct gendisk *, int);
2642 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2643 pgoff_t start, pgoff_t end);
2645 static inline void invalidate_remote_inode(struct inode *inode)
2647 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2648 S_ISLNK(inode->i_mode))
2649 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2651 extern int invalidate_inode_pages2(struct address_space *mapping);
2652 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2653 pgoff_t start, pgoff_t end);
2654 extern int write_inode_now(struct inode *, int);
2655 extern int filemap_fdatawrite(struct address_space *);
2656 extern int filemap_flush(struct address_space *);
2657 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2658 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2660 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2661 loff_t start_byte, loff_t end_byte);
2663 static inline int filemap_fdatawait(struct address_space *mapping)
2665 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2668 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2670 extern int filemap_write_and_wait(struct address_space *mapping);
2671 extern int filemap_write_and_wait_range(struct address_space *mapping,
2672 loff_t lstart, loff_t lend);
2673 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2674 loff_t start, loff_t end, int sync_mode);
2675 extern int filemap_fdatawrite_range(struct address_space *mapping,
2676 loff_t start, loff_t end);
2677 extern int filemap_check_errors(struct address_space *mapping);
2678 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2680 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2682 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2683 extern int __must_check file_write_and_wait_range(struct file *file,
2684 loff_t start, loff_t end);
2686 static inline int file_write_and_wait(struct file *file)
2688 return file_write_and_wait_range(file, 0, LLONG_MAX);
2692 * filemap_set_wb_err - set a writeback error on an address_space
2693 * @mapping: mapping in which to set writeback error
2694 * @err: error to be set in mapping
2696 * When writeback fails in some way, we must record that error so that
2697 * userspace can be informed when fsync and the like are called. We endeavor
2698 * to report errors on any file that was open at the time of the error. Some
2699 * internal callers also need to know when writeback errors have occurred.
2701 * When a writeback error occurs, most filesystems will want to call
2702 * filemap_set_wb_err to record the error in the mapping so that it will be
2703 * automatically reported whenever fsync is called on the file.
2705 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2707 /* Fastpath for common case of no error */
2709 __filemap_set_wb_err(mapping, err);
2713 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2714 * @mapping: mapping to check for writeback errors
2715 * @since: previously-sampled errseq_t
2717 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2718 * the given value was sampled.
2720 * If it has then report the latest error set, otherwise return 0.
2722 static inline int filemap_check_wb_err(struct address_space *mapping,
2725 return errseq_check(&mapping->wb_err, since);
2729 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2730 * @mapping: mapping to be sampled
2732 * Writeback errors are always reported relative to a particular sample point
2733 * in the past. This function provides those sample points.
2735 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2737 return errseq_sample(&mapping->wb_err);
2740 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2742 extern int vfs_fsync(struct file *file, int datasync);
2745 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2746 * to already be updated for the write, and will return either the amount
2747 * of bytes passed in, or an error if syncing the file failed.
2749 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2751 if (iocb->ki_flags & IOCB_DSYNC) {
2752 int ret = vfs_fsync_range(iocb->ki_filp,
2753 iocb->ki_pos - count, iocb->ki_pos - 1,
2754 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2762 extern void emergency_sync(void);
2763 extern void emergency_remount(void);
2765 extern sector_t bmap(struct inode *, sector_t);
2767 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2768 extern int inode_permission(struct inode *, int);
2769 extern int generic_permission(struct inode *, int);
2770 extern int __check_sticky(struct inode *dir, struct inode *inode);
2772 static inline bool execute_ok(struct inode *inode)
2774 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2777 static inline void file_start_write(struct file *file)
2779 if (!S_ISREG(file_inode(file)->i_mode))
2781 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2784 static inline bool file_start_write_trylock(struct file *file)
2786 if (!S_ISREG(file_inode(file)->i_mode))
2788 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2791 static inline void file_end_write(struct file *file)
2793 if (!S_ISREG(file_inode(file)->i_mode))
2795 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2799 * get_write_access() gets write permission for a file.
2800 * put_write_access() releases this write permission.
2801 * This is used for regular files.
2802 * We cannot support write (and maybe mmap read-write shared) accesses and
2803 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2804 * can have the following values:
2805 * 0: no writers, no VM_DENYWRITE mappings
2806 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2807 * > 0: (i_writecount) users are writing to the file.
2809 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2810 * except for the cases where we don't hold i_writecount yet. Then we need to
2811 * use {get,deny}_write_access() - these functions check the sign and refuse
2812 * to do the change if sign is wrong.
2814 static inline int get_write_access(struct inode *inode)
2816 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2818 static inline int deny_write_access(struct file *file)
2820 struct inode *inode = file_inode(file);
2821 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2823 static inline void put_write_access(struct inode * inode)
2825 atomic_dec(&inode->i_writecount);
2827 static inline void allow_write_access(struct file *file)
2830 atomic_inc(&file_inode(file)->i_writecount);
2832 static inline bool inode_is_open_for_write(const struct inode *inode)
2834 return atomic_read(&inode->i_writecount) > 0;
2838 static inline void i_readcount_dec(struct inode *inode)
2840 BUG_ON(!atomic_read(&inode->i_readcount));
2841 atomic_dec(&inode->i_readcount);
2843 static inline void i_readcount_inc(struct inode *inode)
2845 atomic_inc(&inode->i_readcount);
2848 static inline void i_readcount_dec(struct inode *inode)
2852 static inline void i_readcount_inc(struct inode *inode)
2857 extern int do_pipe_flags(int *, int);
2859 #define __kernel_read_file_id(id) \
2860 id(UNKNOWN, unknown) \
2861 id(FIRMWARE, firmware) \
2862 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2863 id(MODULE, kernel-module) \
2864 id(KEXEC_IMAGE, kexec-image) \
2865 id(KEXEC_INITRAMFS, kexec-initramfs) \
2866 id(POLICY, security-policy) \
2867 id(X509_CERTIFICATE, x509-certificate) \
2870 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2871 #define __fid_stringify(dummy, str) #str,
2873 enum kernel_read_file_id {
2874 __kernel_read_file_id(__fid_enumify)
2877 static const char * const kernel_read_file_str[] = {
2878 __kernel_read_file_id(__fid_stringify)
2881 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2883 if ((unsigned)id >= READING_MAX_ID)
2884 return kernel_read_file_str[READING_UNKNOWN];
2886 return kernel_read_file_str[id];
2889 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2890 enum kernel_read_file_id);
2891 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2892 enum kernel_read_file_id);
2893 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2894 enum kernel_read_file_id);
2895 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2896 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2897 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2898 extern struct file * open_exec(const char *);
2900 /* fs/dcache.c -- generic fs support functions */
2901 extern bool is_subdir(struct dentry *, struct dentry *);
2902 extern bool path_is_under(const struct path *, const struct path *);
2904 extern char *file_path(struct file *, char *, int);
2906 #include <linux/err.h>
2908 /* needed for stackable file system support */
2909 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2911 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2913 extern int inode_init_always(struct super_block *, struct inode *);
2914 extern void inode_init_once(struct inode *);
2915 extern void address_space_init_once(struct address_space *mapping);
2916 extern struct inode * igrab(struct inode *);
2917 extern ino_t iunique(struct super_block *, ino_t);
2918 extern int inode_needs_sync(struct inode *inode);
2919 extern int generic_delete_inode(struct inode *inode);
2920 static inline int generic_drop_inode(struct inode *inode)
2922 return !inode->i_nlink || inode_unhashed(inode);
2925 extern struct inode *ilookup5_nowait(struct super_block *sb,
2926 unsigned long hashval, int (*test)(struct inode *, void *),
2928 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2929 int (*test)(struct inode *, void *), void *data);
2930 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2932 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2933 int (*test)(struct inode *, void *),
2934 int (*set)(struct inode *, void *),
2936 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2937 extern struct inode * iget_locked(struct super_block *, unsigned long);
2938 extern struct inode *find_inode_nowait(struct super_block *,
2940 int (*match)(struct inode *,
2941 unsigned long, void *),
2943 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2944 extern int insert_inode_locked(struct inode *);
2945 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2946 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2948 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2950 extern void unlock_new_inode(struct inode *);
2951 extern void discard_new_inode(struct inode *);
2952 extern unsigned int get_next_ino(void);
2953 extern void evict_inodes(struct super_block *sb);
2955 extern void __iget(struct inode * inode);
2956 extern void iget_failed(struct inode *);
2957 extern void clear_inode(struct inode *);
2958 extern void __destroy_inode(struct inode *);
2959 extern struct inode *new_inode_pseudo(struct super_block *sb);
2960 extern struct inode *new_inode(struct super_block *sb);
2961 extern void free_inode_nonrcu(struct inode *inode);
2962 extern int should_remove_suid(struct dentry *);
2963 extern int file_remove_privs(struct file *);
2965 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2966 static inline void insert_inode_hash(struct inode *inode)
2968 __insert_inode_hash(inode, inode->i_ino);
2971 extern void __remove_inode_hash(struct inode *);
2972 static inline void remove_inode_hash(struct inode *inode)
2974 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2975 __remove_inode_hash(inode);
2978 extern void inode_sb_list_add(struct inode *inode);
2981 extern int bdev_read_only(struct block_device *);
2983 extern int set_blocksize(struct block_device *, int);
2984 extern int sb_set_blocksize(struct super_block *, int);
2985 extern int sb_min_blocksize(struct super_block *, int);
2987 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2988 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2989 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2990 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2991 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2992 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2993 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2994 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2996 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2998 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3001 /* fs/block_dev.c */
3002 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3003 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3004 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3006 extern void block_sync_page(struct page *page);
3009 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3010 struct pipe_inode_info *, size_t, unsigned int);
3011 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3012 struct file *, loff_t *, size_t, unsigned int);
3013 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3014 struct file *out, loff_t *, size_t len, unsigned int flags);
3015 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3016 loff_t *opos, size_t len, unsigned int flags);
3020 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3021 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3022 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3023 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3024 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3025 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3026 int whence, loff_t maxsize, loff_t eof);
3027 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3028 int whence, loff_t size);
3029 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3030 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3031 extern int generic_file_open(struct inode * inode, struct file * filp);
3032 extern int nonseekable_open(struct inode * inode, struct file * filp);
3033 extern int stream_open(struct inode * inode, struct file * filp);
3036 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3037 loff_t file_offset);
3040 /* need locking between buffered and direct access */
3043 /* filesystem does not support filling holes */
3044 DIO_SKIP_HOLES = 0x02,
3047 void dio_end_io(struct bio *bio);
3048 void dio_warn_stale_pagecache(struct file *filp);
3050 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3051 struct block_device *bdev, struct iov_iter *iter,
3052 get_block_t get_block,
3053 dio_iodone_t end_io, dio_submit_t submit_io,
3056 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3057 struct inode *inode,
3058 struct iov_iter *iter,
3059 get_block_t get_block)
3061 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3062 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3066 void inode_dio_wait(struct inode *inode);
3069 * inode_dio_begin - signal start of a direct I/O requests
3070 * @inode: inode the direct I/O happens on
3072 * This is called once we've finished processing a direct I/O request,
3073 * and is used to wake up callers waiting for direct I/O to be quiesced.
3075 static inline void inode_dio_begin(struct inode *inode)
3077 atomic_inc(&inode->i_dio_count);
3081 * inode_dio_end - signal finish of a direct I/O requests
3082 * @inode: inode the direct I/O happens on
3084 * This is called once we've finished processing a direct I/O request,
3085 * and is used to wake up callers waiting for direct I/O to be quiesced.
3087 static inline void inode_dio_end(struct inode *inode)
3089 if (atomic_dec_and_test(&inode->i_dio_count))
3090 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3093 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3096 extern const struct file_operations generic_ro_fops;
3098 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3100 extern int readlink_copy(char __user *, int, const char *);
3101 extern int page_readlink(struct dentry *, char __user *, int);
3102 extern const char *page_get_link(struct dentry *, struct inode *,
3103 struct delayed_call *);
3104 extern void page_put_link(void *);
3105 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3107 extern int page_symlink(struct inode *inode, const char *symname, int len);
3108 extern const struct inode_operations page_symlink_inode_operations;
3109 extern void kfree_link(void *);
3110 extern void generic_fillattr(struct inode *, struct kstat *);
3111 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3112 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3113 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3114 void inode_add_bytes(struct inode *inode, loff_t bytes);
3115 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3116 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3117 static inline loff_t __inode_get_bytes(struct inode *inode)
3119 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3121 loff_t inode_get_bytes(struct inode *inode);
3122 void inode_set_bytes(struct inode *inode, loff_t bytes);
3123 const char *simple_get_link(struct dentry *, struct inode *,
3124 struct delayed_call *);
3125 extern const struct inode_operations simple_symlink_inode_operations;
3127 extern int iterate_dir(struct file *, struct dir_context *);
3129 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3130 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3132 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3134 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3135 stat, STATX_BASIC_STATS);
3137 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3139 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3140 stat, STATX_BASIC_STATS);
3142 static inline int vfs_fstatat(int dfd, const char __user *filename,
3143 struct kstat *stat, int flags)
3145 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3146 stat, STATX_BASIC_STATS);
3148 static inline int vfs_fstat(int fd, struct kstat *stat)
3150 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3154 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3155 extern int vfs_readlink(struct dentry *, char __user *, int);
3157 extern int __generic_block_fiemap(struct inode *inode,
3158 struct fiemap_extent_info *fieinfo,
3159 loff_t start, loff_t len,
3160 get_block_t *get_block);
3161 extern int generic_block_fiemap(struct inode *inode,
3162 struct fiemap_extent_info *fieinfo, u64 start,
3163 u64 len, get_block_t *get_block);
3165 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3166 extern void put_filesystem(struct file_system_type *fs);
3167 extern struct file_system_type *get_fs_type(const char *name);
3168 extern struct super_block *get_super(struct block_device *);
3169 extern struct super_block *get_super_thawed(struct block_device *);
3170 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3171 extern struct super_block *get_active_super(struct block_device *bdev);
3172 extern void drop_super(struct super_block *sb);
3173 extern void drop_super_exclusive(struct super_block *sb);
3174 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3175 extern void iterate_supers_type(struct file_system_type *,
3176 void (*)(struct super_block *, void *), void *);
3178 extern int dcache_dir_open(struct inode *, struct file *);
3179 extern int dcache_dir_close(struct inode *, struct file *);
3180 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3181 extern int dcache_readdir(struct file *, struct dir_context *);
3182 extern int simple_setattr(struct dentry *, struct iattr *);
3183 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3184 extern int simple_statfs(struct dentry *, struct kstatfs *);
3185 extern int simple_open(struct inode *inode, struct file *file);
3186 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3187 extern int simple_unlink(struct inode *, struct dentry *);
3188 extern int simple_rmdir(struct inode *, struct dentry *);
3189 extern int simple_rename(struct inode *, struct dentry *,
3190 struct inode *, struct dentry *, unsigned int);
3191 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3192 extern int noop_set_page_dirty(struct page *page);
3193 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3194 unsigned int length);
3195 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3196 extern int simple_empty(struct dentry *);
3197 extern int simple_readpage(struct file *file, struct page *page);
3198 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3199 loff_t pos, unsigned len, unsigned flags,
3200 struct page **pagep, void **fsdata);
3201 extern int simple_write_end(struct file *file, struct address_space *mapping,
3202 loff_t pos, unsigned len, unsigned copied,
3203 struct page *page, void *fsdata);
3204 extern int always_delete_dentry(const struct dentry *);
3205 extern struct inode *alloc_anon_inode(struct super_block *);
3206 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3207 extern const struct dentry_operations simple_dentry_operations;
3209 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3210 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3211 extern const struct file_operations simple_dir_operations;
3212 extern const struct inode_operations simple_dir_inode_operations;
3213 extern void make_empty_dir_inode(struct inode *inode);
3214 extern bool is_empty_dir_inode(struct inode *inode);
3215 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3216 struct dentry *d_alloc_name(struct dentry *, const char *);
3217 extern int simple_fill_super(struct super_block *, unsigned long,
3218 const struct tree_descr *);
3219 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3220 extern void simple_release_fs(struct vfsmount **mount, int *count);
3222 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3223 loff_t *ppos, const void *from, size_t available);
3224 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3225 const void __user *from, size_t count);
3227 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3228 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3230 extern int generic_check_addressable(unsigned, u64);
3232 #ifdef CONFIG_MIGRATION
3233 extern int buffer_migrate_page(struct address_space *,
3234 struct page *, struct page *,
3237 #define buffer_migrate_page NULL
3240 extern int setattr_prepare(struct dentry *, struct iattr *);
3241 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3242 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3244 extern int file_update_time(struct file *file);
3246 static inline bool io_is_direct(struct file *filp)
3248 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3251 static inline bool vma_is_dax(struct vm_area_struct *vma)
3253 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3256 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3258 struct inode *inode;
3262 if (!vma_is_dax(vma))
3264 inode = file_inode(vma->vm_file);
3265 if (S_ISCHR(inode->i_mode))
3266 return false; /* device-dax */
3270 static inline int iocb_flags(struct file *file)
3273 if (file->f_flags & O_APPEND)
3275 if (io_is_direct(file))
3277 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3279 if (file->f_flags & __O_SYNC)
3284 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3286 if (unlikely(flags & ~RWF_SUPPORTED))
3289 if (flags & RWF_NOWAIT) {
3290 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3292 ki->ki_flags |= IOCB_NOWAIT;
3294 if (flags & RWF_HIPRI)
3295 ki->ki_flags |= IOCB_HIPRI;
3296 if (flags & RWF_DSYNC)
3297 ki->ki_flags |= IOCB_DSYNC;
3298 if (flags & RWF_SYNC)
3299 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3300 if (flags & RWF_APPEND)
3301 ki->ki_flags |= IOCB_APPEND;
3305 static inline ino_t parent_ino(struct dentry *dentry)
3310 * Don't strictly need d_lock here? If the parent ino could change
3311 * then surely we'd have a deeper race in the caller?
3313 spin_lock(&dentry->d_lock);
3314 res = dentry->d_parent->d_inode->i_ino;
3315 spin_unlock(&dentry->d_lock);
3319 /* Transaction based IO helpers */
3322 * An argresp is stored in an allocated page and holds the
3323 * size of the argument or response, along with its content
3325 struct simple_transaction_argresp {
3330 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3332 char *simple_transaction_get(struct file *file, const char __user *buf,
3334 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3335 size_t size, loff_t *pos);
3336 int simple_transaction_release(struct inode *inode, struct file *file);
3338 void simple_transaction_set(struct file *file, size_t n);
3341 * simple attribute files
3343 * These attributes behave similar to those in sysfs:
3345 * Writing to an attribute immediately sets a value, an open file can be
3346 * written to multiple times.
3348 * Reading from an attribute creates a buffer from the value that might get
3349 * read with multiple read calls. When the attribute has been read
3350 * completely, no further read calls are possible until the file is opened
3353 * All attributes contain a text representation of a numeric value
3354 * that are accessed with the get() and set() functions.
3356 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3357 static int __fops ## _open(struct inode *inode, struct file *file) \
3359 __simple_attr_check_format(__fmt, 0ull); \
3360 return simple_attr_open(inode, file, __get, __set, __fmt); \
3362 static const struct file_operations __fops = { \
3363 .owner = THIS_MODULE, \
3364 .open = __fops ## _open, \
3365 .release = simple_attr_release, \
3366 .read = simple_attr_read, \
3367 .write = simple_attr_write, \
3368 .llseek = generic_file_llseek, \
3371 static inline __printf(1, 2)
3372 void __simple_attr_check_format(const char *fmt, ...)
3374 /* don't do anything, just let the compiler check the arguments; */
3377 int simple_attr_open(struct inode *inode, struct file *file,
3378 int (*get)(void *, u64 *), int (*set)(void *, u64),
3380 int simple_attr_release(struct inode *inode, struct file *file);
3381 ssize_t simple_attr_read(struct file *file, char __user *buf,
3382 size_t len, loff_t *ppos);
3383 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3384 size_t len, loff_t *ppos);
3387 int proc_nr_files(struct ctl_table *table, int write,
3388 void __user *buffer, size_t *lenp, loff_t *ppos);
3389 int proc_nr_dentry(struct ctl_table *table, int write,
3390 void __user *buffer, size_t *lenp, loff_t *ppos);
3391 int proc_nr_inodes(struct ctl_table *table, int write,
3392 void __user *buffer, size_t *lenp, loff_t *ppos);
3393 int __init get_filesystem_list(char *buf);
3395 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3396 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3398 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3399 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3400 (flag & __FMODE_NONOTIFY)))
3402 static inline bool is_sxid(umode_t mode)
3404 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3407 static inline int check_sticky(struct inode *dir, struct inode *inode)
3409 if (!(dir->i_mode & S_ISVTX))
3412 return __check_sticky(dir, inode);
3415 static inline void inode_has_no_xattr(struct inode *inode)
3417 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3418 inode->i_flags |= S_NOSEC;
3421 static inline bool is_root_inode(struct inode *inode)
3423 return inode == inode->i_sb->s_root->d_inode;
3426 static inline bool dir_emit(struct dir_context *ctx,
3427 const char *name, int namelen,
3428 u64 ino, unsigned type)
3430 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3432 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3434 return ctx->actor(ctx, ".", 1, ctx->pos,
3435 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3437 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3439 return ctx->actor(ctx, "..", 2, ctx->pos,
3440 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3442 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3444 if (ctx->pos == 0) {
3445 if (!dir_emit_dot(file, ctx))
3449 if (ctx->pos == 1) {
3450 if (!dir_emit_dotdot(file, ctx))
3456 static inline bool dir_relax(struct inode *inode)
3458 inode_unlock(inode);
3460 return !IS_DEADDIR(inode);
3463 static inline bool dir_relax_shared(struct inode *inode)
3465 inode_unlock_shared(inode);
3466 inode_lock_shared(inode);
3467 return !IS_DEADDIR(inode);
3470 extern bool path_noexec(const struct path *path);
3471 extern void inode_nohighmem(struct inode *inode);
3474 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3477 #endif /* _LINUX_FS_H */