Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1/* SPDX-License-Identifier: GPL-2.0 */
2#ifndef _LINUX_FS_H
3#define _LINUX_FS_H
4
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/rbtree.h>
17#include <linux/init.h>
18#include <linux/pid.h>
19#include <linux/bug.h>
20#include <linux/mutex.h>
21#include <linux/rwsem.h>
22#include <linux/mm_types.h>
23#include <linux/capability.h>
24#include <linux/semaphore.h>
25#include <linux/fcntl.h>
26#include <linux/fiemap.h>
27#include <linux/rculist_bl.h>
28#include <linux/atomic.h>
29#include <linux/shrinker.h>
30#include <linux/migrate_mode.h>
31#include <linux/uidgid.h>
32#include <linux/lockdep.h>
33#include <linux/percpu-rwsem.h>
34#include <linux/workqueue.h>
35#include <linux/delayed_call.h>
36#include <linux/uuid.h>
37#include <linux/errseq.h>
38
39#include <asm/byteorder.h>
40#include <uapi/linux/fs.h>
41
42struct backing_dev_info;
43struct bdi_writeback;
44struct bio;
45struct export_operations;
46struct hd_geometry;
47struct iovec;
48struct kiocb;
49struct kobject;
50struct pipe_inode_info;
51struct poll_table_struct;
52struct kstatfs;
53struct vm_area_struct;
54struct vfsmount;
55struct cred;
56struct swap_info_struct;
57struct seq_file;
58struct workqueue_struct;
59struct iov_iter;
60struct fscrypt_info;
61struct fscrypt_operations;
62
63extern void __init inode_init(void);
64extern void __init inode_init_early(void);
65extern void __init files_init(void);
66extern void __init files_maxfiles_init(void);
67
68extern struct files_stat_struct files_stat;
69extern unsigned long get_max_files(void);
70extern unsigned int sysctl_nr_open;
71extern struct inodes_stat_t inodes_stat;
72extern int leases_enable, lease_break_time;
73extern int sysctl_protected_symlinks;
74extern int sysctl_protected_hardlinks;
75
76typedef __kernel_rwf_t rwf_t;
77
78struct buffer_head;
79typedef int (get_block_t)(struct inode *inode, sector_t iblock,
80 struct buffer_head *bh_result, int create);
81typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
82 ssize_t bytes, void *private);
83
84#define MAY_EXEC 0x00000001
85#define MAY_WRITE 0x00000002
86#define MAY_READ 0x00000004
87#define MAY_APPEND 0x00000008
88#define MAY_ACCESS 0x00000010
89#define MAY_OPEN 0x00000020
90#define MAY_CHDIR 0x00000040
91/* called from RCU mode, don't block */
92#define MAY_NOT_BLOCK 0x00000080
93
94/*
95 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
96 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
97 */
98
99/* file is open for reading */
100#define FMODE_READ ((__force fmode_t)0x1)
101/* file is open for writing */
102#define FMODE_WRITE ((__force fmode_t)0x2)
103/* file is seekable */
104#define FMODE_LSEEK ((__force fmode_t)0x4)
105/* file can be accessed using pread */
106#define FMODE_PREAD ((__force fmode_t)0x8)
107/* file can be accessed using pwrite */
108#define FMODE_PWRITE ((__force fmode_t)0x10)
109/* File is opened for execution with sys_execve / sys_uselib */
110#define FMODE_EXEC ((__force fmode_t)0x20)
111/* File is opened with O_NDELAY (only set for block devices) */
112#define FMODE_NDELAY ((__force fmode_t)0x40)
113/* File is opened with O_EXCL (only set for block devices) */
114#define FMODE_EXCL ((__force fmode_t)0x80)
115/* File is opened using open(.., 3, ..) and is writeable only for ioctls
116 (specialy hack for floppy.c) */
117#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
118/* 32bit hashes as llseek() offset (for directories) */
119#define FMODE_32BITHASH ((__force fmode_t)0x200)
120/* 64bit hashes as llseek() offset (for directories) */
121#define FMODE_64BITHASH ((__force fmode_t)0x400)
122
123/*
124 * Don't update ctime and mtime.
125 *
126 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
127 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
128 */
129#define FMODE_NOCMTIME ((__force fmode_t)0x800)
130
131/* Expect random access pattern */
132#define FMODE_RANDOM ((__force fmode_t)0x1000)
133
134/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
135#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
136
137/* File is opened with O_PATH; almost nothing can be done with it */
138#define FMODE_PATH ((__force fmode_t)0x4000)
139
140/* File needs atomic accesses to f_pos */
141#define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
142/* Write access to underlying fs */
143#define FMODE_WRITER ((__force fmode_t)0x10000)
144/* Has read method(s) */
145#define FMODE_CAN_READ ((__force fmode_t)0x20000)
146/* Has write method(s) */
147#define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
148
149/* File was opened by fanotify and shouldn't generate fanotify events */
150#define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
151
152/* File is capable of returning -EAGAIN if I/O will block */
153#define FMODE_NOWAIT ((__force fmode_t)0x8000000)
154
155/*
156 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
157 * that indicates that they should check the contents of the iovec are
158 * valid, but not check the memory that the iovec elements
159 * points too.
160 */
161#define CHECK_IOVEC_ONLY -1
162
163/*
164 * Attribute flags. These should be or-ed together to figure out what
165 * has been changed!
166 */
167#define ATTR_MODE (1 << 0)
168#define ATTR_UID (1 << 1)
169#define ATTR_GID (1 << 2)
170#define ATTR_SIZE (1 << 3)
171#define ATTR_ATIME (1 << 4)
172#define ATTR_MTIME (1 << 5)
173#define ATTR_CTIME (1 << 6)
174#define ATTR_ATIME_SET (1 << 7)
175#define ATTR_MTIME_SET (1 << 8)
176#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
177#define ATTR_ATTR_FLAG (1 << 10)
178#define ATTR_KILL_SUID (1 << 11)
179#define ATTR_KILL_SGID (1 << 12)
180#define ATTR_FILE (1 << 13)
181#define ATTR_KILL_PRIV (1 << 14)
182#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
183#define ATTR_TIMES_SET (1 << 16)
184#define ATTR_TOUCH (1 << 17)
185
186/*
187 * Whiteout is represented by a char device. The following constants define the
188 * mode and device number to use.
189 */
190#define WHITEOUT_MODE 0
191#define WHITEOUT_DEV 0
192
193/*
194 * This is the Inode Attributes structure, used for notify_change(). It
195 * uses the above definitions as flags, to know which values have changed.
196 * Also, in this manner, a Filesystem can look at only the values it cares
197 * about. Basically, these are the attributes that the VFS layer can
198 * request to change from the FS layer.
199 *
200 * Derek Atkins <warlord@MIT.EDU> 94-10-20
201 */
202struct iattr {
203 unsigned int ia_valid;
204 umode_t ia_mode;
205 kuid_t ia_uid;
206 kgid_t ia_gid;
207 loff_t ia_size;
208 struct timespec ia_atime;
209 struct timespec ia_mtime;
210 struct timespec ia_ctime;
211
212 /*
213 * Not an attribute, but an auxiliary info for filesystems wanting to
214 * implement an ftruncate() like method. NOTE: filesystem should
215 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
216 */
217 struct file *ia_file;
218};
219
220/*
221 * Includes for diskquotas.
222 */
223#include <linux/quota.h>
224
225/*
226 * Maximum number of layers of fs stack. Needs to be limited to
227 * prevent kernel stack overflow
228 */
229#define FILESYSTEM_MAX_STACK_DEPTH 2
230
231/**
232 * enum positive_aop_returns - aop return codes with specific semantics
233 *
234 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
235 * completed, that the page is still locked, and
236 * should be considered active. The VM uses this hint
237 * to return the page to the active list -- it won't
238 * be a candidate for writeback again in the near
239 * future. Other callers must be careful to unlock
240 * the page if they get this return. Returned by
241 * writepage();
242 *
243 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
244 * unlocked it and the page might have been truncated.
245 * The caller should back up to acquiring a new page and
246 * trying again. The aop will be taking reasonable
247 * precautions not to livelock. If the caller held a page
248 * reference, it should drop it before retrying. Returned
249 * by readpage().
250 *
251 * address_space_operation functions return these large constants to indicate
252 * special semantics to the caller. These are much larger than the bytes in a
253 * page to allow for functions that return the number of bytes operated on in a
254 * given page.
255 */
256
257enum positive_aop_returns {
258 AOP_WRITEPAGE_ACTIVATE = 0x80000,
259 AOP_TRUNCATED_PAGE = 0x80001,
260};
261
262#define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
263#define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
264 * helper code (eg buffer layer)
265 * to clear GFP_FS from alloc */
266
267/*
268 * oh the beauties of C type declarations.
269 */
270struct page;
271struct address_space;
272struct writeback_control;
273
274/*
275 * Write life time hint values.
276 */
277enum rw_hint {
278 WRITE_LIFE_NOT_SET = 0,
279 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
280 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
281 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
282 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
283 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
284};
285
286#define IOCB_EVENTFD (1 << 0)
287#define IOCB_APPEND (1 << 1)
288#define IOCB_DIRECT (1 << 2)
289#define IOCB_HIPRI (1 << 3)
290#define IOCB_DSYNC (1 << 4)
291#define IOCB_SYNC (1 << 5)
292#define IOCB_WRITE (1 << 6)
293#define IOCB_NOWAIT (1 << 7)
294
295struct kiocb {
296 struct file *ki_filp;
297 loff_t ki_pos;
298 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
299 void *private;
300 int ki_flags;
301 enum rw_hint ki_hint;
302} __randomize_layout;
303
304static inline bool is_sync_kiocb(struct kiocb *kiocb)
305{
306 return kiocb->ki_complete == NULL;
307}
308
309/*
310 * "descriptor" for what we're up to with a read.
311 * This allows us to use the same read code yet
312 * have multiple different users of the data that
313 * we read from a file.
314 *
315 * The simplest case just copies the data to user
316 * mode.
317 */
318typedef struct {
319 size_t written;
320 size_t count;
321 union {
322 char __user *buf;
323 void *data;
324 } arg;
325 int error;
326} read_descriptor_t;
327
328typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
329 unsigned long, unsigned long);
330
331struct address_space_operations {
332 int (*writepage)(struct page *page, struct writeback_control *wbc);
333 int (*readpage)(struct file *, struct page *);
334
335 /* Write back some dirty pages from this mapping. */
336 int (*writepages)(struct address_space *, struct writeback_control *);
337
338 /* Set a page dirty. Return true if this dirtied it */
339 int (*set_page_dirty)(struct page *page);
340
341 int (*readpages)(struct file *filp, struct address_space *mapping,
342 struct list_head *pages, unsigned nr_pages);
343
344 int (*write_begin)(struct file *, struct address_space *mapping,
345 loff_t pos, unsigned len, unsigned flags,
346 struct page **pagep, void **fsdata);
347 int (*write_end)(struct file *, struct address_space *mapping,
348 loff_t pos, unsigned len, unsigned copied,
349 struct page *page, void *fsdata);
350
351 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
352 sector_t (*bmap)(struct address_space *, sector_t);
353 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
354 int (*releasepage) (struct page *, gfp_t);
355 void (*freepage)(struct page *);
356 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
357 /*
358 * migrate the contents of a page to the specified target. If
359 * migrate_mode is MIGRATE_ASYNC, it must not block.
360 */
361 int (*migratepage) (struct address_space *,
362 struct page *, struct page *, enum migrate_mode);
363 bool (*isolate_page)(struct page *, isolate_mode_t);
364 void (*putback_page)(struct page *);
365 int (*launder_page) (struct page *);
366 int (*is_partially_uptodate) (struct page *, unsigned long,
367 unsigned long);
368 void (*is_dirty_writeback) (struct page *, bool *, bool *);
369 int (*error_remove_page)(struct address_space *, struct page *);
370
371 /* swapfile support */
372 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
373 sector_t *span);
374 void (*swap_deactivate)(struct file *file);
375};
376
377extern const struct address_space_operations empty_aops;
378
379/*
380 * pagecache_write_begin/pagecache_write_end must be used by general code
381 * to write into the pagecache.
382 */
383int pagecache_write_begin(struct file *, struct address_space *mapping,
384 loff_t pos, unsigned len, unsigned flags,
385 struct page **pagep, void **fsdata);
386
387int pagecache_write_end(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned copied,
389 struct page *page, void *fsdata);
390
391struct address_space {
392 struct inode *host; /* owner: inode, block_device */
393 struct radix_tree_root page_tree; /* radix tree of all pages */
394 spinlock_t tree_lock; /* and lock protecting it */
395 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
396 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
397 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
398 /* Protected by tree_lock together with the radix tree */
399 unsigned long nrpages; /* number of total pages */
400 /* number of shadow or DAX exceptional entries */
401 unsigned long nrexceptional;
402 pgoff_t writeback_index;/* writeback starts here */
403 const struct address_space_operations *a_ops; /* methods */
404 unsigned long flags; /* error bits */
405 spinlock_t private_lock; /* for use by the address_space */
406 gfp_t gfp_mask; /* implicit gfp mask for allocations */
407 struct list_head private_list; /* for use by the address_space */
408 void *private_data; /* ditto */
409 errseq_t wb_err;
410} __attribute__((aligned(sizeof(long)))) __randomize_layout;
411 /*
412 * On most architectures that alignment is already the case; but
413 * must be enforced here for CRIS, to let the least significant bit
414 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
415 */
416struct request_queue;
417
418struct block_device {
419 dev_t bd_dev; /* not a kdev_t - it's a search key */
420 int bd_openers;
421 struct inode * bd_inode; /* will die */
422 struct super_block * bd_super;
423 struct mutex bd_mutex; /* open/close mutex */
424 void * bd_claiming;
425 void * bd_holder;
426 int bd_holders;
427 bool bd_write_holder;
428#ifdef CONFIG_SYSFS
429 struct list_head bd_holder_disks;
430#endif
431 struct block_device * bd_contains;
432 unsigned bd_block_size;
433 u8 bd_partno;
434 struct hd_struct * bd_part;
435 /* number of times partitions within this device have been opened. */
436 unsigned bd_part_count;
437 int bd_invalidated;
438 struct gendisk * bd_disk;
439 struct request_queue * bd_queue;
440 struct backing_dev_info *bd_bdi;
441 struct list_head bd_list;
442 /*
443 * Private data. You must have bd_claim'ed the block_device
444 * to use this. NOTE: bd_claim allows an owner to claim
445 * the same device multiple times, the owner must take special
446 * care to not mess up bd_private for that case.
447 */
448 unsigned long bd_private;
449
450 /* The counter of freeze processes */
451 int bd_fsfreeze_count;
452 /* Mutex for freeze */
453 struct mutex bd_fsfreeze_mutex;
454} __randomize_layout;
455
456/*
457 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
458 * radix trees
459 */
460#define PAGECACHE_TAG_DIRTY 0
461#define PAGECACHE_TAG_WRITEBACK 1
462#define PAGECACHE_TAG_TOWRITE 2
463
464int mapping_tagged(struct address_space *mapping, int tag);
465
466static inline void i_mmap_lock_write(struct address_space *mapping)
467{
468 down_write(&mapping->i_mmap_rwsem);
469}
470
471static inline void i_mmap_unlock_write(struct address_space *mapping)
472{
473 up_write(&mapping->i_mmap_rwsem);
474}
475
476static inline void i_mmap_lock_read(struct address_space *mapping)
477{
478 down_read(&mapping->i_mmap_rwsem);
479}
480
481static inline void i_mmap_unlock_read(struct address_space *mapping)
482{
483 up_read(&mapping->i_mmap_rwsem);
484}
485
486/*
487 * Might pages of this file be mapped into userspace?
488 */
489static inline int mapping_mapped(struct address_space *mapping)
490{
491 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
492}
493
494/*
495 * Might pages of this file have been modified in userspace?
496 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
497 * marks vma as VM_SHARED if it is shared, and the file was opened for
498 * writing i.e. vma may be mprotected writable even if now readonly.
499 *
500 * If i_mmap_writable is negative, no new writable mappings are allowed. You
501 * can only deny writable mappings, if none exists right now.
502 */
503static inline int mapping_writably_mapped(struct address_space *mapping)
504{
505 return atomic_read(&mapping->i_mmap_writable) > 0;
506}
507
508static inline int mapping_map_writable(struct address_space *mapping)
509{
510 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
511 0 : -EPERM;
512}
513
514static inline void mapping_unmap_writable(struct address_space *mapping)
515{
516 atomic_dec(&mapping->i_mmap_writable);
517}
518
519static inline int mapping_deny_writable(struct address_space *mapping)
520{
521 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
522 0 : -EBUSY;
523}
524
525static inline void mapping_allow_writable(struct address_space *mapping)
526{
527 atomic_inc(&mapping->i_mmap_writable);
528}
529
530/*
531 * Use sequence counter to get consistent i_size on 32-bit processors.
532 */
533#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
534#include <linux/seqlock.h>
535#define __NEED_I_SIZE_ORDERED
536#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
537#else
538#define i_size_ordered_init(inode) do { } while (0)
539#endif
540
541struct posix_acl;
542#define ACL_NOT_CACHED ((void *)(-1))
543#define ACL_DONT_CACHE ((void *)(-3))
544
545static inline struct posix_acl *
546uncached_acl_sentinel(struct task_struct *task)
547{
548 return (void *)task + 1;
549}
550
551static inline bool
552is_uncached_acl(struct posix_acl *acl)
553{
554 return (long)acl & 1;
555}
556
557#define IOP_FASTPERM 0x0001
558#define IOP_LOOKUP 0x0002
559#define IOP_NOFOLLOW 0x0004
560#define IOP_XATTR 0x0008
561#define IOP_DEFAULT_READLINK 0x0010
562
563struct fsnotify_mark_connector;
564
565/*
566 * Keep mostly read-only and often accessed (especially for
567 * the RCU path lookup and 'stat' data) fields at the beginning
568 * of the 'struct inode'
569 */
570struct inode {
571 umode_t i_mode;
572 unsigned short i_opflags;
573 kuid_t i_uid;
574 kgid_t i_gid;
575 unsigned int i_flags;
576
577#ifdef CONFIG_FS_POSIX_ACL
578 struct posix_acl *i_acl;
579 struct posix_acl *i_default_acl;
580#endif
581
582 const struct inode_operations *i_op;
583 struct super_block *i_sb;
584 struct address_space *i_mapping;
585
586#ifdef CONFIG_SECURITY
587 void *i_security;
588#endif
589
590 /* Stat data, not accessed from path walking */
591 unsigned long i_ino;
592 /*
593 * Filesystems may only read i_nlink directly. They shall use the
594 * following functions for modification:
595 *
596 * (set|clear|inc|drop)_nlink
597 * inode_(inc|dec)_link_count
598 */
599 union {
600 const unsigned int i_nlink;
601 unsigned int __i_nlink;
602 };
603 dev_t i_rdev;
604 loff_t i_size;
605 struct timespec i_atime;
606 struct timespec i_mtime;
607 struct timespec i_ctime;
608 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
609 unsigned short i_bytes;
610 unsigned int i_blkbits;
611 enum rw_hint i_write_hint;
612 blkcnt_t i_blocks;
613
614#ifdef __NEED_I_SIZE_ORDERED
615 seqcount_t i_size_seqcount;
616#endif
617
618 /* Misc */
619 unsigned long i_state;
620 struct rw_semaphore i_rwsem;
621
622 unsigned long dirtied_when; /* jiffies of first dirtying */
623 unsigned long dirtied_time_when;
624
625 struct hlist_node i_hash;
626 struct list_head i_io_list; /* backing dev IO list */
627#ifdef CONFIG_CGROUP_WRITEBACK
628 struct bdi_writeback *i_wb; /* the associated cgroup wb */
629
630 /* foreign inode detection, see wbc_detach_inode() */
631 int i_wb_frn_winner;
632 u16 i_wb_frn_avg_time;
633 u16 i_wb_frn_history;
634#endif
635 struct list_head i_lru; /* inode LRU list */
636 struct list_head i_sb_list;
637 struct list_head i_wb_list; /* backing dev writeback list */
638 union {
639 struct hlist_head i_dentry;
640 struct rcu_head i_rcu;
641 };
642 u64 i_version;
643 atomic_t i_count;
644 atomic_t i_dio_count;
645 atomic_t i_writecount;
646#ifdef CONFIG_IMA
647 atomic_t i_readcount; /* struct files open RO */
648#endif
649 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
650 struct file_lock_context *i_flctx;
651 struct address_space i_data;
652 struct list_head i_devices;
653 union {
654 struct pipe_inode_info *i_pipe;
655 struct block_device *i_bdev;
656 struct cdev *i_cdev;
657 char *i_link;
658 unsigned i_dir_seq;
659 };
660
661 __u32 i_generation;
662
663#ifdef CONFIG_FSNOTIFY
664 __u32 i_fsnotify_mask; /* all events this inode cares about */
665 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
666#endif
667
668#if IS_ENABLED(CONFIG_FS_ENCRYPTION)
669 struct fscrypt_info *i_crypt_info;
670#endif
671
672 void *i_private; /* fs or device private pointer */
673} __randomize_layout;
674
675static inline unsigned int i_blocksize(const struct inode *node)
676{
677 return (1 << node->i_blkbits);
678}
679
680static inline int inode_unhashed(struct inode *inode)
681{
682 return hlist_unhashed(&inode->i_hash);
683}
684
685/*
686 * inode->i_mutex nesting subclasses for the lock validator:
687 *
688 * 0: the object of the current VFS operation
689 * 1: parent
690 * 2: child/target
691 * 3: xattr
692 * 4: second non-directory
693 * 5: second parent (when locking independent directories in rename)
694 *
695 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
696 * non-directories at once.
697 *
698 * The locking order between these classes is
699 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
700 */
701enum inode_i_mutex_lock_class
702{
703 I_MUTEX_NORMAL,
704 I_MUTEX_PARENT,
705 I_MUTEX_CHILD,
706 I_MUTEX_XATTR,
707 I_MUTEX_NONDIR2,
708 I_MUTEX_PARENT2,
709};
710
711static inline void inode_lock(struct inode *inode)
712{
713 down_write(&inode->i_rwsem);
714}
715
716static inline void inode_unlock(struct inode *inode)
717{
718 up_write(&inode->i_rwsem);
719}
720
721static inline void inode_lock_shared(struct inode *inode)
722{
723 down_read(&inode->i_rwsem);
724}
725
726static inline void inode_unlock_shared(struct inode *inode)
727{
728 up_read(&inode->i_rwsem);
729}
730
731static inline int inode_trylock(struct inode *inode)
732{
733 return down_write_trylock(&inode->i_rwsem);
734}
735
736static inline int inode_trylock_shared(struct inode *inode)
737{
738 return down_read_trylock(&inode->i_rwsem);
739}
740
741static inline int inode_is_locked(struct inode *inode)
742{
743 return rwsem_is_locked(&inode->i_rwsem);
744}
745
746static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
747{
748 down_write_nested(&inode->i_rwsem, subclass);
749}
750
751void lock_two_nondirectories(struct inode *, struct inode*);
752void unlock_two_nondirectories(struct inode *, struct inode*);
753
754/*
755 * NOTE: in a 32bit arch with a preemptable kernel and
756 * an UP compile the i_size_read/write must be atomic
757 * with respect to the local cpu (unlike with preempt disabled),
758 * but they don't need to be atomic with respect to other cpus like in
759 * true SMP (so they need either to either locally disable irq around
760 * the read or for example on x86 they can be still implemented as a
761 * cmpxchg8b without the need of the lock prefix). For SMP compiles
762 * and 64bit archs it makes no difference if preempt is enabled or not.
763 */
764static inline loff_t i_size_read(const struct inode *inode)
765{
766#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
767 loff_t i_size;
768 unsigned int seq;
769
770 do {
771 seq = read_seqcount_begin(&inode->i_size_seqcount);
772 i_size = inode->i_size;
773 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
774 return i_size;
775#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
776 loff_t i_size;
777
778 preempt_disable();
779 i_size = inode->i_size;
780 preempt_enable();
781 return i_size;
782#else
783 return inode->i_size;
784#endif
785}
786
787/*
788 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
789 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
790 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
791 */
792static inline void i_size_write(struct inode *inode, loff_t i_size)
793{
794#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
795 preempt_disable();
796 write_seqcount_begin(&inode->i_size_seqcount);
797 inode->i_size = i_size;
798 write_seqcount_end(&inode->i_size_seqcount);
799 preempt_enable();
800#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
801 preempt_disable();
802 inode->i_size = i_size;
803 preempt_enable();
804#else
805 inode->i_size = i_size;
806#endif
807}
808
809static inline unsigned iminor(const struct inode *inode)
810{
811 return MINOR(inode->i_rdev);
812}
813
814static inline unsigned imajor(const struct inode *inode)
815{
816 return MAJOR(inode->i_rdev);
817}
818
819extern struct block_device *I_BDEV(struct inode *inode);
820
821struct fown_struct {
822 rwlock_t lock; /* protects pid, uid, euid fields */
823 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
824 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
825 kuid_t uid, euid; /* uid/euid of process setting the owner */
826 int signum; /* posix.1b rt signal to be delivered on IO */
827};
828
829/*
830 * Track a single file's readahead state
831 */
832struct file_ra_state {
833 pgoff_t start; /* where readahead started */
834 unsigned int size; /* # of readahead pages */
835 unsigned int async_size; /* do asynchronous readahead when
836 there are only # of pages ahead */
837
838 unsigned int ra_pages; /* Maximum readahead window */
839 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
840 loff_t prev_pos; /* Cache last read() position */
841};
842
843/*
844 * Check if @index falls in the readahead windows.
845 */
846static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
847{
848 return (index >= ra->start &&
849 index < ra->start + ra->size);
850}
851
852struct file {
853 union {
854 struct llist_node fu_llist;
855 struct rcu_head fu_rcuhead;
856 } f_u;
857 struct path f_path;
858 struct inode *f_inode; /* cached value */
859 const struct file_operations *f_op;
860
861 /*
862 * Protects f_ep_links, f_flags.
863 * Must not be taken from IRQ context.
864 */
865 spinlock_t f_lock;
866 enum rw_hint f_write_hint;
867 atomic_long_t f_count;
868 unsigned int f_flags;
869 fmode_t f_mode;
870 struct mutex f_pos_lock;
871 loff_t f_pos;
872 struct fown_struct f_owner;
873 const struct cred *f_cred;
874 struct file_ra_state f_ra;
875
876 u64 f_version;
877#ifdef CONFIG_SECURITY
878 void *f_security;
879#endif
880 /* needed for tty driver, and maybe others */
881 void *private_data;
882
883#ifdef CONFIG_EPOLL
884 /* Used by fs/eventpoll.c to link all the hooks to this file */
885 struct list_head f_ep_links;
886 struct list_head f_tfile_llink;
887#endif /* #ifdef CONFIG_EPOLL */
888 struct address_space *f_mapping;
889 errseq_t f_wb_err;
890} __randomize_layout
891 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
892
893struct file_handle {
894 __u32 handle_bytes;
895 int handle_type;
896 /* file identifier */
897 unsigned char f_handle[0];
898};
899
900static inline struct file *get_file(struct file *f)
901{
902 atomic_long_inc(&f->f_count);
903 return f;
904}
905#define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
906#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
907#define file_count(x) atomic_long_read(&(x)->f_count)
908
909#define MAX_NON_LFS ((1UL<<31) - 1)
910
911/* Page cache limit. The filesystems should put that into their s_maxbytes
912 limits, otherwise bad things can happen in VM. */
913#if BITS_PER_LONG==32
914#define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
915#elif BITS_PER_LONG==64
916#define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
917#endif
918
919#define FL_POSIX 1
920#define FL_FLOCK 2
921#define FL_DELEG 4 /* NFSv4 delegation */
922#define FL_ACCESS 8 /* not trying to lock, just looking */
923#define FL_EXISTS 16 /* when unlocking, test for existence */
924#define FL_LEASE 32 /* lease held on this file */
925#define FL_CLOSE 64 /* unlock on close */
926#define FL_SLEEP 128 /* A blocking lock */
927#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
928#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
929#define FL_OFDLCK 1024 /* lock is "owned" by struct file */
930#define FL_LAYOUT 2048 /* outstanding pNFS layout */
931
932#define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
933
934/*
935 * Special return value from posix_lock_file() and vfs_lock_file() for
936 * asynchronous locking.
937 */
938#define FILE_LOCK_DEFERRED 1
939
940/* legacy typedef, should eventually be removed */
941typedef void *fl_owner_t;
942
943struct file_lock;
944
945struct file_lock_operations {
946 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
947 void (*fl_release_private)(struct file_lock *);
948};
949
950struct lock_manager_operations {
951 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
952 unsigned long (*lm_owner_key)(struct file_lock *);
953 fl_owner_t (*lm_get_owner)(fl_owner_t);
954 void (*lm_put_owner)(fl_owner_t);
955 void (*lm_notify)(struct file_lock *); /* unblock callback */
956 int (*lm_grant)(struct file_lock *, int);
957 bool (*lm_break)(struct file_lock *);
958 int (*lm_change)(struct file_lock *, int, struct list_head *);
959 void (*lm_setup)(struct file_lock *, void **);
960};
961
962struct lock_manager {
963 struct list_head list;
964 /*
965 * NFSv4 and up also want opens blocked during the grace period;
966 * NLM doesn't care:
967 */
968 bool block_opens;
969};
970
971struct net;
972void locks_start_grace(struct net *, struct lock_manager *);
973void locks_end_grace(struct lock_manager *);
974int locks_in_grace(struct net *);
975int opens_in_grace(struct net *);
976
977/* that will die - we need it for nfs_lock_info */
978#include <linux/nfs_fs_i.h>
979
980/*
981 * struct file_lock represents a generic "file lock". It's used to represent
982 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
983 * note that the same struct is used to represent both a request for a lock and
984 * the lock itself, but the same object is never used for both.
985 *
986 * FIXME: should we create a separate "struct lock_request" to help distinguish
987 * these two uses?
988 *
989 * The varous i_flctx lists are ordered by:
990 *
991 * 1) lock owner
992 * 2) lock range start
993 * 3) lock range end
994 *
995 * Obviously, the last two criteria only matter for POSIX locks.
996 */
997struct file_lock {
998 struct file_lock *fl_next; /* singly linked list for this inode */
999 struct list_head fl_list; /* link into file_lock_context */
1000 struct hlist_node fl_link; /* node in global lists */
1001 struct list_head fl_block; /* circular list of blocked processes */
1002 fl_owner_t fl_owner;
1003 unsigned int fl_flags;
1004 unsigned char fl_type;
1005 unsigned int fl_pid;
1006 int fl_link_cpu; /* what cpu's list is this on? */
1007 wait_queue_head_t fl_wait;
1008 struct file *fl_file;
1009 loff_t fl_start;
1010 loff_t fl_end;
1011
1012 struct fasync_struct * fl_fasync; /* for lease break notifications */
1013 /* for lease breaks: */
1014 unsigned long fl_break_time;
1015 unsigned long fl_downgrade_time;
1016
1017 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1018 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1019 union {
1020 struct nfs_lock_info nfs_fl;
1021 struct nfs4_lock_info nfs4_fl;
1022 struct {
1023 struct list_head link; /* link in AFS vnode's pending_locks list */
1024 int state; /* state of grant or error if -ve */
1025 } afs;
1026 } fl_u;
1027} __randomize_layout;
1028
1029struct file_lock_context {
1030 spinlock_t flc_lock;
1031 struct list_head flc_flock;
1032 struct list_head flc_posix;
1033 struct list_head flc_lease;
1034};
1035
1036/* The following constant reflects the upper bound of the file/locking space */
1037#ifndef OFFSET_MAX
1038#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1039#define OFFSET_MAX INT_LIMIT(loff_t)
1040#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1041#endif
1042
1043extern void send_sigio(struct fown_struct *fown, int fd, int band);
1044
1045/*
1046 * Return the inode to use for locking
1047 *
1048 * For overlayfs this should be the overlay inode, not the real inode returned
1049 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1050 * equal.
1051 */
1052static inline struct inode *locks_inode(const struct file *f)
1053{
1054 return f->f_path.dentry->d_inode;
1055}
1056
1057#ifdef CONFIG_FILE_LOCKING
1058extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1059extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1060 struct flock *);
1061
1062#if BITS_PER_LONG == 32
1063extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1064extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1065 struct flock64 *);
1066#endif
1067
1068extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1069extern int fcntl_getlease(struct file *filp);
1070
1071/* fs/locks.c */
1072void locks_free_lock_context(struct inode *inode);
1073void locks_free_lock(struct file_lock *fl);
1074extern void locks_init_lock(struct file_lock *);
1075extern struct file_lock * locks_alloc_lock(void);
1076extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1077extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1078extern void locks_remove_posix(struct file *, fl_owner_t);
1079extern void locks_remove_file(struct file *);
1080extern void locks_release_private(struct file_lock *);
1081extern void posix_test_lock(struct file *, struct file_lock *);
1082extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1083extern int posix_unblock_lock(struct file_lock *);
1084extern int vfs_test_lock(struct file *, struct file_lock *);
1085extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1086extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1087extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1088extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1089extern void lease_get_mtime(struct inode *, struct timespec *time);
1090extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1091extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1092extern int lease_modify(struct file_lock *, int, struct list_head *);
1093struct files_struct;
1094extern void show_fd_locks(struct seq_file *f,
1095 struct file *filp, struct files_struct *files);
1096#else /* !CONFIG_FILE_LOCKING */
1097static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1098 struct flock __user *user)
1099{
1100 return -EINVAL;
1101}
1102
1103static inline int fcntl_setlk(unsigned int fd, struct file *file,
1104 unsigned int cmd, struct flock __user *user)
1105{
1106 return -EACCES;
1107}
1108
1109#if BITS_PER_LONG == 32
1110static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1111 struct flock64 __user *user)
1112{
1113 return -EINVAL;
1114}
1115
1116static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1117 unsigned int cmd, struct flock64 __user *user)
1118{
1119 return -EACCES;
1120}
1121#endif
1122static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1123{
1124 return -EINVAL;
1125}
1126
1127static inline int fcntl_getlease(struct file *filp)
1128{
1129 return F_UNLCK;
1130}
1131
1132static inline void
1133locks_free_lock_context(struct inode *inode)
1134{
1135}
1136
1137static inline void locks_init_lock(struct file_lock *fl)
1138{
1139 return;
1140}
1141
1142static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1143{
1144 return;
1145}
1146
1147static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1148{
1149 return;
1150}
1151
1152static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1153{
1154 return;
1155}
1156
1157static inline void locks_remove_file(struct file *filp)
1158{
1159 return;
1160}
1161
1162static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1163{
1164 return;
1165}
1166
1167static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1168 struct file_lock *conflock)
1169{
1170 return -ENOLCK;
1171}
1172
1173static inline int posix_unblock_lock(struct file_lock *waiter)
1174{
1175 return -ENOENT;
1176}
1177
1178static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1179{
1180 return 0;
1181}
1182
1183static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1184 struct file_lock *fl, struct file_lock *conf)
1185{
1186 return -ENOLCK;
1187}
1188
1189static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1190{
1191 return 0;
1192}
1193
1194static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1195{
1196 return -ENOLCK;
1197}
1198
1199static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1200{
1201 return 0;
1202}
1203
1204static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1205{
1206 return;
1207}
1208
1209static inline int generic_setlease(struct file *filp, long arg,
1210 struct file_lock **flp, void **priv)
1211{
1212 return -EINVAL;
1213}
1214
1215static inline int vfs_setlease(struct file *filp, long arg,
1216 struct file_lock **lease, void **priv)
1217{
1218 return -EINVAL;
1219}
1220
1221static inline int lease_modify(struct file_lock *fl, int arg,
1222 struct list_head *dispose)
1223{
1224 return -EINVAL;
1225}
1226
1227struct files_struct;
1228static inline void show_fd_locks(struct seq_file *f,
1229 struct file *filp, struct files_struct *files) {}
1230#endif /* !CONFIG_FILE_LOCKING */
1231
1232static inline struct inode *file_inode(const struct file *f)
1233{
1234 return f->f_inode;
1235}
1236
1237static inline struct dentry *file_dentry(const struct file *file)
1238{
1239 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
1240}
1241
1242static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1243{
1244 return locks_lock_inode_wait(locks_inode(filp), fl);
1245}
1246
1247struct fasync_struct {
1248 spinlock_t fa_lock;
1249 int magic;
1250 int fa_fd;
1251 struct fasync_struct *fa_next; /* singly linked list */
1252 struct file *fa_file;
1253 struct rcu_head fa_rcu;
1254};
1255
1256#define FASYNC_MAGIC 0x4601
1257
1258/* SMP safe fasync helpers: */
1259extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1260extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1261extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1262extern struct fasync_struct *fasync_alloc(void);
1263extern void fasync_free(struct fasync_struct *);
1264
1265/* can be called from interrupts */
1266extern void kill_fasync(struct fasync_struct **, int, int);
1267
1268extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1269extern int f_setown(struct file *filp, unsigned long arg, int force);
1270extern void f_delown(struct file *filp);
1271extern pid_t f_getown(struct file *filp);
1272extern int send_sigurg(struct fown_struct *fown);
1273
1274/*
1275 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1276 * represented in both.
1277 */
1278#define SB_RDONLY 1 /* Mount read-only */
1279#define SB_NOSUID 2 /* Ignore suid and sgid bits */
1280#define SB_NODEV 4 /* Disallow access to device special files */
1281#define SB_NOEXEC 8 /* Disallow program execution */
1282#define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1283#define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1284#define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1285#define SB_NOATIME 1024 /* Do not update access times. */
1286#define SB_NODIRATIME 2048 /* Do not update directory access times */
1287#define SB_SILENT 32768
1288#define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1289#define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1290#define SB_I_VERSION (1<<23) /* Update inode I_version field */
1291#define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1292
1293/* These sb flags are internal to the kernel */
1294#define SB_SUBMOUNT (1<<26)
1295#define SB_NOREMOTELOCK (1<<27)
1296#define SB_NOSEC (1<<28)
1297#define SB_BORN (1<<29)
1298#define SB_ACTIVE (1<<30)
1299#define SB_NOUSER (1<<31)
1300
1301/*
1302 * Umount options
1303 */
1304
1305#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1306#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1307#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1308#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1309#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1310
1311/* sb->s_iflags */
1312#define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1313#define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1314#define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1315
1316/* sb->s_iflags to limit user namespace mounts */
1317#define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1318
1319/* Possible states of 'frozen' field */
1320enum {
1321 SB_UNFROZEN = 0, /* FS is unfrozen */
1322 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1323 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1324 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1325 * internal threads if needed) */
1326 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1327};
1328
1329#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1330
1331struct sb_writers {
1332 int frozen; /* Is sb frozen? */
1333 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1334 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1335};
1336
1337struct super_block {
1338 struct list_head s_list; /* Keep this first */
1339 dev_t s_dev; /* search index; _not_ kdev_t */
1340 unsigned char s_blocksize_bits;
1341 unsigned long s_blocksize;
1342 loff_t s_maxbytes; /* Max file size */
1343 struct file_system_type *s_type;
1344 const struct super_operations *s_op;
1345 const struct dquot_operations *dq_op;
1346 const struct quotactl_ops *s_qcop;
1347 const struct export_operations *s_export_op;
1348 unsigned long s_flags;
1349 unsigned long s_iflags; /* internal SB_I_* flags */
1350 unsigned long s_magic;
1351 struct dentry *s_root;
1352 struct rw_semaphore s_umount;
1353 int s_count;
1354 atomic_t s_active;
1355#ifdef CONFIG_SECURITY
1356 void *s_security;
1357#endif
1358 const struct xattr_handler **s_xattr;
1359
1360 const struct fscrypt_operations *s_cop;
1361
1362 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1363 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1364 struct block_device *s_bdev;
1365 struct backing_dev_info *s_bdi;
1366 struct mtd_info *s_mtd;
1367 struct hlist_node s_instances;
1368 unsigned int s_quota_types; /* Bitmask of supported quota types */
1369 struct quota_info s_dquot; /* Diskquota specific options */
1370
1371 struct sb_writers s_writers;
1372
1373 char s_id[32]; /* Informational name */
1374 uuid_t s_uuid; /* UUID */
1375
1376 void *s_fs_info; /* Filesystem private info */
1377 unsigned int s_max_links;
1378 fmode_t s_mode;
1379
1380 /* Granularity of c/m/atime in ns.
1381 Cannot be worse than a second */
1382 u32 s_time_gran;
1383
1384 /*
1385 * The next field is for VFS *only*. No filesystems have any business
1386 * even looking at it. You had been warned.
1387 */
1388 struct mutex s_vfs_rename_mutex; /* Kludge */
1389
1390 /*
1391 * Filesystem subtype. If non-empty the filesystem type field
1392 * in /proc/mounts will be "type.subtype"
1393 */
1394 char *s_subtype;
1395
1396 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1397
1398 /*
1399 * Saved pool identifier for cleancache (-1 means none)
1400 */
1401 int cleancache_poolid;
1402
1403 struct shrinker s_shrink; /* per-sb shrinker handle */
1404
1405 /* Number of inodes with nlink == 0 but still referenced */
1406 atomic_long_t s_remove_count;
1407
1408 /* Being remounted read-only */
1409 int s_readonly_remount;
1410
1411 /* AIO completions deferred from interrupt context */
1412 struct workqueue_struct *s_dio_done_wq;
1413 struct hlist_head s_pins;
1414
1415 /*
1416 * Owning user namespace and default context in which to
1417 * interpret filesystem uids, gids, quotas, device nodes,
1418 * xattrs and security labels.
1419 */
1420 struct user_namespace *s_user_ns;
1421
1422 /*
1423 * Keep the lru lists last in the structure so they always sit on their
1424 * own individual cachelines.
1425 */
1426 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1427 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1428 struct rcu_head rcu;
1429 struct work_struct destroy_work;
1430
1431 struct mutex s_sync_lock; /* sync serialisation lock */
1432
1433 /*
1434 * Indicates how deep in a filesystem stack this SB is
1435 */
1436 int s_stack_depth;
1437
1438 /* s_inode_list_lock protects s_inodes */
1439 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1440 struct list_head s_inodes; /* all inodes */
1441
1442 spinlock_t s_inode_wblist_lock;
1443 struct list_head s_inodes_wb; /* writeback inodes */
1444} __randomize_layout;
1445
1446/* Helper functions so that in most cases filesystems will
1447 * not need to deal directly with kuid_t and kgid_t and can
1448 * instead deal with the raw numeric values that are stored
1449 * in the filesystem.
1450 */
1451static inline uid_t i_uid_read(const struct inode *inode)
1452{
1453 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1454}
1455
1456static inline gid_t i_gid_read(const struct inode *inode)
1457{
1458 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1459}
1460
1461static inline void i_uid_write(struct inode *inode, uid_t uid)
1462{
1463 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1464}
1465
1466static inline void i_gid_write(struct inode *inode, gid_t gid)
1467{
1468 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1469}
1470
1471extern struct timespec current_time(struct inode *inode);
1472
1473/*
1474 * Snapshotting support.
1475 */
1476
1477void __sb_end_write(struct super_block *sb, int level);
1478int __sb_start_write(struct super_block *sb, int level, bool wait);
1479
1480#define __sb_writers_acquired(sb, lev) \
1481 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1482#define __sb_writers_release(sb, lev) \
1483 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1484
1485/**
1486 * sb_end_write - drop write access to a superblock
1487 * @sb: the super we wrote to
1488 *
1489 * Decrement number of writers to the filesystem. Wake up possible waiters
1490 * wanting to freeze the filesystem.
1491 */
1492static inline void sb_end_write(struct super_block *sb)
1493{
1494 __sb_end_write(sb, SB_FREEZE_WRITE);
1495}
1496
1497/**
1498 * sb_end_pagefault - drop write access to a superblock from a page fault
1499 * @sb: the super we wrote to
1500 *
1501 * Decrement number of processes handling write page fault to the filesystem.
1502 * Wake up possible waiters wanting to freeze the filesystem.
1503 */
1504static inline void sb_end_pagefault(struct super_block *sb)
1505{
1506 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1507}
1508
1509/**
1510 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1511 * @sb: the super we wrote to
1512 *
1513 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1514 * waiters wanting to freeze the filesystem.
1515 */
1516static inline void sb_end_intwrite(struct super_block *sb)
1517{
1518 __sb_end_write(sb, SB_FREEZE_FS);
1519}
1520
1521/**
1522 * sb_start_write - get write access to a superblock
1523 * @sb: the super we write to
1524 *
1525 * When a process wants to write data or metadata to a file system (i.e. dirty
1526 * a page or an inode), it should embed the operation in a sb_start_write() -
1527 * sb_end_write() pair to get exclusion against file system freezing. This
1528 * function increments number of writers preventing freezing. If the file
1529 * system is already frozen, the function waits until the file system is
1530 * thawed.
1531 *
1532 * Since freeze protection behaves as a lock, users have to preserve
1533 * ordering of freeze protection and other filesystem locks. Generally,
1534 * freeze protection should be the outermost lock. In particular, we have:
1535 *
1536 * sb_start_write
1537 * -> i_mutex (write path, truncate, directory ops, ...)
1538 * -> s_umount (freeze_super, thaw_super)
1539 */
1540static inline void sb_start_write(struct super_block *sb)
1541{
1542 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1543}
1544
1545static inline int sb_start_write_trylock(struct super_block *sb)
1546{
1547 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1548}
1549
1550/**
1551 * sb_start_pagefault - get write access to a superblock from a page fault
1552 * @sb: the super we write to
1553 *
1554 * When a process starts handling write page fault, it should embed the
1555 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1556 * exclusion against file system freezing. This is needed since the page fault
1557 * is going to dirty a page. This function increments number of running page
1558 * faults preventing freezing. If the file system is already frozen, the
1559 * function waits until the file system is thawed.
1560 *
1561 * Since page fault freeze protection behaves as a lock, users have to preserve
1562 * ordering of freeze protection and other filesystem locks. It is advised to
1563 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1564 * handling code implies lock dependency:
1565 *
1566 * mmap_sem
1567 * -> sb_start_pagefault
1568 */
1569static inline void sb_start_pagefault(struct super_block *sb)
1570{
1571 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1572}
1573
1574/*
1575 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1576 * @sb: the super we write to
1577 *
1578 * This is the third level of protection against filesystem freezing. It is
1579 * free for use by a filesystem. The only requirement is that it must rank
1580 * below sb_start_pagefault.
1581 *
1582 * For example filesystem can call sb_start_intwrite() when starting a
1583 * transaction which somewhat eases handling of freezing for internal sources
1584 * of filesystem changes (internal fs threads, discarding preallocation on file
1585 * close, etc.).
1586 */
1587static inline void sb_start_intwrite(struct super_block *sb)
1588{
1589 __sb_start_write(sb, SB_FREEZE_FS, true);
1590}
1591
1592
1593extern bool inode_owner_or_capable(const struct inode *inode);
1594
1595/*
1596 * VFS helper functions..
1597 */
1598extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1599extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1600extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1601extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1602extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1603extern int vfs_rmdir(struct inode *, struct dentry *);
1604extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1605extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1606extern int vfs_whiteout(struct inode *, struct dentry *);
1607
1608extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1609 int open_flag);
1610
1611/*
1612 * VFS file helper functions.
1613 */
1614extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1615 umode_t mode);
1616extern bool may_open_dev(const struct path *path);
1617/*
1618 * VFS FS_IOC_FIEMAP helper definitions.
1619 */
1620struct fiemap_extent_info {
1621 unsigned int fi_flags; /* Flags as passed from user */
1622 unsigned int fi_extents_mapped; /* Number of mapped extents */
1623 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1624 struct fiemap_extent __user *fi_extents_start; /* Start of
1625 fiemap_extent array */
1626};
1627int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1628 u64 phys, u64 len, u32 flags);
1629int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1630
1631/*
1632 * File types
1633 *
1634 * NOTE! These match bits 12..15 of stat.st_mode
1635 * (ie "(i_mode >> 12) & 15").
1636 */
1637#define DT_UNKNOWN 0
1638#define DT_FIFO 1
1639#define DT_CHR 2
1640#define DT_DIR 4
1641#define DT_BLK 6
1642#define DT_REG 8
1643#define DT_LNK 10
1644#define DT_SOCK 12
1645#define DT_WHT 14
1646
1647/*
1648 * This is the "filldir" function type, used by readdir() to let
1649 * the kernel specify what kind of dirent layout it wants to have.
1650 * This allows the kernel to read directories into kernel space or
1651 * to have different dirent layouts depending on the binary type.
1652 */
1653struct dir_context;
1654typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1655 unsigned);
1656
1657struct dir_context {
1658 const filldir_t actor;
1659 loff_t pos;
1660};
1661
1662struct block_device_operations;
1663
1664/* These macros are for out of kernel modules to test that
1665 * the kernel supports the unlocked_ioctl and compat_ioctl
1666 * fields in struct file_operations. */
1667#define HAVE_COMPAT_IOCTL 1
1668#define HAVE_UNLOCKED_IOCTL 1
1669
1670/*
1671 * These flags let !MMU mmap() govern direct device mapping vs immediate
1672 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1673 *
1674 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1675 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1676 * NOMMU_MAP_READ: Can be mapped for reading
1677 * NOMMU_MAP_WRITE: Can be mapped for writing
1678 * NOMMU_MAP_EXEC: Can be mapped for execution
1679 */
1680#define NOMMU_MAP_COPY 0x00000001
1681#define NOMMU_MAP_DIRECT 0x00000008
1682#define NOMMU_MAP_READ VM_MAYREAD
1683#define NOMMU_MAP_WRITE VM_MAYWRITE
1684#define NOMMU_MAP_EXEC VM_MAYEXEC
1685
1686#define NOMMU_VMFLAGS \
1687 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1688
1689
1690struct iov_iter;
1691
1692struct file_operations {
1693 struct module *owner;
1694 loff_t (*llseek) (struct file *, loff_t, int);
1695 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1696 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1697 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1698 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1699 int (*iterate) (struct file *, struct dir_context *);
1700 int (*iterate_shared) (struct file *, struct dir_context *);
1701 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1702 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1703 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1704 int (*mmap) (struct file *, struct vm_area_struct *);
1705 int (*open) (struct inode *, struct file *);
1706 int (*flush) (struct file *, fl_owner_t id);
1707 int (*release) (struct inode *, struct file *);
1708 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1709 int (*fasync) (int, struct file *, int);
1710 int (*lock) (struct file *, int, struct file_lock *);
1711 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1712 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1713 int (*check_flags)(int);
1714 int (*flock) (struct file *, int, struct file_lock *);
1715 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1716 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1717 int (*setlease)(struct file *, long, struct file_lock **, void **);
1718 long (*fallocate)(struct file *file, int mode, loff_t offset,
1719 loff_t len);
1720 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1721#ifndef CONFIG_MMU
1722 unsigned (*mmap_capabilities)(struct file *);
1723#endif
1724 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1725 loff_t, size_t, unsigned int);
1726 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1727 u64);
1728 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1729 u64);
1730} __randomize_layout;
1731
1732struct inode_operations {
1733 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1734 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1735 int (*permission) (struct inode *, int);
1736 struct posix_acl * (*get_acl)(struct inode *, int);
1737
1738 int (*readlink) (struct dentry *, char __user *,int);
1739
1740 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1741 int (*link) (struct dentry *,struct inode *,struct dentry *);
1742 int (*unlink) (struct inode *,struct dentry *);
1743 int (*symlink) (struct inode *,struct dentry *,const char *);
1744 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1745 int (*rmdir) (struct inode *,struct dentry *);
1746 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1747 int (*rename) (struct inode *, struct dentry *,
1748 struct inode *, struct dentry *, unsigned int);
1749 int (*setattr) (struct dentry *, struct iattr *);
1750 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1751 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1752 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1753 u64 len);
1754 int (*update_time)(struct inode *, struct timespec *, int);
1755 int (*atomic_open)(struct inode *, struct dentry *,
1756 struct file *, unsigned open_flag,
1757 umode_t create_mode, int *opened);
1758 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1759 int (*set_acl)(struct inode *, struct posix_acl *, int);
1760} ____cacheline_aligned;
1761
1762static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1763 struct iov_iter *iter)
1764{
1765 return file->f_op->read_iter(kio, iter);
1766}
1767
1768static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1769 struct iov_iter *iter)
1770{
1771 return file->f_op->write_iter(kio, iter);
1772}
1773
1774static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1775{
1776 return file->f_op->mmap(file, vma);
1777}
1778
1779ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1780 unsigned long nr_segs, unsigned long fast_segs,
1781 struct iovec *fast_pointer,
1782 struct iovec **ret_pointer);
1783
1784extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1785extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1786extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1787extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1788 unsigned long, loff_t *, rwf_t);
1789extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1790 loff_t, size_t, unsigned int);
1791extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1792 struct inode *inode_out, loff_t pos_out,
1793 u64 *len, bool is_dedupe);
1794extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1795 struct file *file_out, loff_t pos_out, u64 len);
1796extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1797 struct inode *dest, loff_t destoff,
1798 loff_t len, bool *is_same);
1799extern int vfs_dedupe_file_range(struct file *file,
1800 struct file_dedupe_range *same);
1801
1802struct super_operations {
1803 struct inode *(*alloc_inode)(struct super_block *sb);
1804 void (*destroy_inode)(struct inode *);
1805
1806 void (*dirty_inode) (struct inode *, int flags);
1807 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1808 int (*drop_inode) (struct inode *);
1809 void (*evict_inode) (struct inode *);
1810 void (*put_super) (struct super_block *);
1811 int (*sync_fs)(struct super_block *sb, int wait);
1812 int (*freeze_super) (struct super_block *);
1813 int (*freeze_fs) (struct super_block *);
1814 int (*thaw_super) (struct super_block *);
1815 int (*unfreeze_fs) (struct super_block *);
1816 int (*statfs) (struct dentry *, struct kstatfs *);
1817 int (*remount_fs) (struct super_block *, int *, char *);
1818 void (*umount_begin) (struct super_block *);
1819
1820 int (*show_options)(struct seq_file *, struct dentry *);
1821 int (*show_devname)(struct seq_file *, struct dentry *);
1822 int (*show_path)(struct seq_file *, struct dentry *);
1823 int (*show_stats)(struct seq_file *, struct dentry *);
1824#ifdef CONFIG_QUOTA
1825 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1826 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1827 struct dquot **(*get_dquots)(struct inode *);
1828#endif
1829 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1830 long (*nr_cached_objects)(struct super_block *,
1831 struct shrink_control *);
1832 long (*free_cached_objects)(struct super_block *,
1833 struct shrink_control *);
1834};
1835
1836/*
1837 * Inode flags - they have no relation to superblock flags now
1838 */
1839#define S_SYNC 1 /* Writes are synced at once */
1840#define S_NOATIME 2 /* Do not update access times */
1841#define S_APPEND 4 /* Append-only file */
1842#define S_IMMUTABLE 8 /* Immutable file */
1843#define S_DEAD 16 /* removed, but still open directory */
1844#define S_NOQUOTA 32 /* Inode is not counted to quota */
1845#define S_DIRSYNC 64 /* Directory modifications are synchronous */
1846#define S_NOCMTIME 128 /* Do not update file c/mtime */
1847#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1848#define S_PRIVATE 512 /* Inode is fs-internal */
1849#define S_IMA 1024 /* Inode has an associated IMA struct */
1850#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1851#define S_NOSEC 4096 /* no suid or xattr security attributes */
1852#ifdef CONFIG_FS_DAX
1853#define S_DAX 8192 /* Direct Access, avoiding the page cache */
1854#else
1855#define S_DAX 0 /* Make all the DAX code disappear */
1856#endif
1857
1858/*
1859 * Note that nosuid etc flags are inode-specific: setting some file-system
1860 * flags just means all the inodes inherit those flags by default. It might be
1861 * possible to override it selectively if you really wanted to with some
1862 * ioctl() that is not currently implemented.
1863 *
1864 * Exception: SB_RDONLY is always applied to the entire file system.
1865 *
1866 * Unfortunately, it is possible to change a filesystems flags with it mounted
1867 * with files in use. This means that all of the inodes will not have their
1868 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1869 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1870 */
1871#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1872
1873static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & MS_RDONLY; }
1874#define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1875#define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1876 ((inode)->i_flags & S_SYNC))
1877#define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1878 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1879#define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1880#define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1881#define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1882
1883#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1884#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1885#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1886#define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1887
1888#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1889#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1890#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1891#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1892#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1893#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1894#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1895#define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1896
1897#define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1898 (inode)->i_rdev == WHITEOUT_DEV)
1899
1900static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1901{
1902 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1903}
1904
1905static inline enum rw_hint file_write_hint(struct file *file)
1906{
1907 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1908 return file->f_write_hint;
1909
1910 return file_inode(file)->i_write_hint;
1911}
1912
1913static inline int iocb_flags(struct file *file);
1914
1915static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1916{
1917 *kiocb = (struct kiocb) {
1918 .ki_filp = filp,
1919 .ki_flags = iocb_flags(filp),
1920 .ki_hint = file_write_hint(filp),
1921 };
1922}
1923
1924/*
1925 * Inode state bits. Protected by inode->i_lock
1926 *
1927 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1928 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1929 *
1930 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1931 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1932 * various stages of removing an inode.
1933 *
1934 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1935 *
1936 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1937 * fdatasync(). i_atime is the usual cause.
1938 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1939 * these changes separately from I_DIRTY_SYNC so that we
1940 * don't have to write inode on fdatasync() when only
1941 * mtime has changed in it.
1942 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1943 * I_NEW Serves as both a mutex and completion notification.
1944 * New inodes set I_NEW. If two processes both create
1945 * the same inode, one of them will release its inode and
1946 * wait for I_NEW to be released before returning.
1947 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1948 * also cause waiting on I_NEW, without I_NEW actually
1949 * being set. find_inode() uses this to prevent returning
1950 * nearly-dead inodes.
1951 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1952 * is zero. I_FREEING must be set when I_WILL_FREE is
1953 * cleared.
1954 * I_FREEING Set when inode is about to be freed but still has dirty
1955 * pages or buffers attached or the inode itself is still
1956 * dirty.
1957 * I_CLEAR Added by clear_inode(). In this state the inode is
1958 * clean and can be destroyed. Inode keeps I_FREEING.
1959 *
1960 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1961 * prohibited for many purposes. iget() must wait for
1962 * the inode to be completely released, then create it
1963 * anew. Other functions will just ignore such inodes,
1964 * if appropriate. I_NEW is used for waiting.
1965 *
1966 * I_SYNC Writeback of inode is running. The bit is set during
1967 * data writeback, and cleared with a wakeup on the bit
1968 * address once it is done. The bit is also used to pin
1969 * the inode in memory for flusher thread.
1970 *
1971 * I_REFERENCED Marks the inode as recently references on the LRU list.
1972 *
1973 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1974 *
1975 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1976 * synchronize competing switching instances and to tell
1977 * wb stat updates to grab mapping->tree_lock. See
1978 * inode_switch_wb_work_fn() for details.
1979 *
1980 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1981 * and work dirs among overlayfs mounts.
1982 *
1983 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1984 */
1985#define I_DIRTY_SYNC (1 << 0)
1986#define I_DIRTY_DATASYNC (1 << 1)
1987#define I_DIRTY_PAGES (1 << 2)
1988#define __I_NEW 3
1989#define I_NEW (1 << __I_NEW)
1990#define I_WILL_FREE (1 << 4)
1991#define I_FREEING (1 << 5)
1992#define I_CLEAR (1 << 6)
1993#define __I_SYNC 7
1994#define I_SYNC (1 << __I_SYNC)
1995#define I_REFERENCED (1 << 8)
1996#define __I_DIO_WAKEUP 9
1997#define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1998#define I_LINKABLE (1 << 10)
1999#define I_DIRTY_TIME (1 << 11)
2000#define __I_DIRTY_TIME_EXPIRED 12
2001#define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2002#define I_WB_SWITCH (1 << 13)
2003#define I_OVL_INUSE (1 << 14)
2004
2005#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
2006#define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2007
2008extern void __mark_inode_dirty(struct inode *, int);
2009static inline void mark_inode_dirty(struct inode *inode)
2010{
2011 __mark_inode_dirty(inode, I_DIRTY);
2012}
2013
2014static inline void mark_inode_dirty_sync(struct inode *inode)
2015{
2016 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2017}
2018
2019extern void inc_nlink(struct inode *inode);
2020extern void drop_nlink(struct inode *inode);
2021extern void clear_nlink(struct inode *inode);
2022extern void set_nlink(struct inode *inode, unsigned int nlink);
2023
2024static inline void inode_inc_link_count(struct inode *inode)
2025{
2026 inc_nlink(inode);
2027 mark_inode_dirty(inode);
2028}
2029
2030static inline void inode_dec_link_count(struct inode *inode)
2031{
2032 drop_nlink(inode);
2033 mark_inode_dirty(inode);
2034}
2035
2036/**
2037 * inode_inc_iversion - increments i_version
2038 * @inode: inode that need to be updated
2039 *
2040 * Every time the inode is modified, the i_version field will be incremented.
2041 * The filesystem has to be mounted with i_version flag
2042 */
2043
2044static inline void inode_inc_iversion(struct inode *inode)
2045{
2046 spin_lock(&inode->i_lock);
2047 inode->i_version++;
2048 spin_unlock(&inode->i_lock);
2049}
2050
2051enum file_time_flags {
2052 S_ATIME = 1,
2053 S_MTIME = 2,
2054 S_CTIME = 4,
2055 S_VERSION = 8,
2056};
2057
2058extern void touch_atime(const struct path *);
2059static inline void file_accessed(struct file *file)
2060{
2061 if (!(file->f_flags & O_NOATIME))
2062 touch_atime(&file->f_path);
2063}
2064
2065int sync_inode(struct inode *inode, struct writeback_control *wbc);
2066int sync_inode_metadata(struct inode *inode, int wait);
2067
2068struct file_system_type {
2069 const char *name;
2070 int fs_flags;
2071#define FS_REQUIRES_DEV 1
2072#define FS_BINARY_MOUNTDATA 2
2073#define FS_HAS_SUBTYPE 4
2074#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2075#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2076 struct dentry *(*mount) (struct file_system_type *, int,
2077 const char *, void *);
2078 void (*kill_sb) (struct super_block *);
2079 struct module *owner;
2080 struct file_system_type * next;
2081 struct hlist_head fs_supers;
2082
2083 struct lock_class_key s_lock_key;
2084 struct lock_class_key s_umount_key;
2085 struct lock_class_key s_vfs_rename_key;
2086 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2087
2088 struct lock_class_key i_lock_key;
2089 struct lock_class_key i_mutex_key;
2090 struct lock_class_key i_mutex_dir_key;
2091};
2092
2093#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2094
2095extern struct dentry *mount_ns(struct file_system_type *fs_type,
2096 int flags, void *data, void *ns, struct user_namespace *user_ns,
2097 int (*fill_super)(struct super_block *, void *, int));
2098extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2099 int flags, const char *dev_name, void *data,
2100 int (*fill_super)(struct super_block *, void *, int));
2101extern struct dentry *mount_single(struct file_system_type *fs_type,
2102 int flags, void *data,
2103 int (*fill_super)(struct super_block *, void *, int));
2104extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2105 int flags, void *data,
2106 int (*fill_super)(struct super_block *, void *, int));
2107extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2108void generic_shutdown_super(struct super_block *sb);
2109void kill_block_super(struct super_block *sb);
2110void kill_anon_super(struct super_block *sb);
2111void kill_litter_super(struct super_block *sb);
2112void deactivate_super(struct super_block *sb);
2113void deactivate_locked_super(struct super_block *sb);
2114int set_anon_super(struct super_block *s, void *data);
2115int get_anon_bdev(dev_t *);
2116void free_anon_bdev(dev_t);
2117struct super_block *sget_userns(struct file_system_type *type,
2118 int (*test)(struct super_block *,void *),
2119 int (*set)(struct super_block *,void *),
2120 int flags, struct user_namespace *user_ns,
2121 void *data);
2122struct super_block *sget(struct file_system_type *type,
2123 int (*test)(struct super_block *,void *),
2124 int (*set)(struct super_block *,void *),
2125 int flags, void *data);
2126extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2127 const struct super_operations *ops,
2128 const struct xattr_handler **xattr,
2129 const struct dentry_operations *dops,
2130 unsigned long);
2131
2132static inline struct dentry *
2133mount_pseudo(struct file_system_type *fs_type, char *name,
2134 const struct super_operations *ops,
2135 const struct dentry_operations *dops, unsigned long magic)
2136{
2137 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2138}
2139
2140/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2141#define fops_get(fops) \
2142 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2143#define fops_put(fops) \
2144 do { if (fops) module_put((fops)->owner); } while(0)
2145/*
2146 * This one is to be used *ONLY* from ->open() instances.
2147 * fops must be non-NULL, pinned down *and* module dependencies
2148 * should be sufficient to pin the caller down as well.
2149 */
2150#define replace_fops(f, fops) \
2151 do { \
2152 struct file *__file = (f); \
2153 fops_put(__file->f_op); \
2154 BUG_ON(!(__file->f_op = (fops))); \
2155 } while(0)
2156
2157extern int register_filesystem(struct file_system_type *);
2158extern int unregister_filesystem(struct file_system_type *);
2159extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2160#define kern_mount(type) kern_mount_data(type, NULL)
2161extern void kern_unmount(struct vfsmount *mnt);
2162extern int may_umount_tree(struct vfsmount *);
2163extern int may_umount(struct vfsmount *);
2164extern long do_mount(const char *, const char __user *,
2165 const char *, unsigned long, void *);
2166extern struct vfsmount *collect_mounts(const struct path *);
2167extern void drop_collected_mounts(struct vfsmount *);
2168extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2169 struct vfsmount *);
2170extern int vfs_statfs(const struct path *, struct kstatfs *);
2171extern int user_statfs(const char __user *, struct kstatfs *);
2172extern int fd_statfs(int, struct kstatfs *);
2173extern int vfs_ustat(dev_t, struct kstatfs *);
2174extern int freeze_super(struct super_block *super);
2175extern int thaw_super(struct super_block *super);
2176extern bool our_mnt(struct vfsmount *mnt);
2177extern __printf(2, 3)
2178int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2179extern int super_setup_bdi(struct super_block *sb);
2180
2181extern int current_umask(void);
2182
2183extern void ihold(struct inode * inode);
2184extern void iput(struct inode *);
2185extern int generic_update_time(struct inode *, struct timespec *, int);
2186
2187/* /sys/fs */
2188extern struct kobject *fs_kobj;
2189
2190#define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2191
2192#ifdef CONFIG_MANDATORY_FILE_LOCKING
2193extern int locks_mandatory_locked(struct file *);
2194extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2195
2196/*
2197 * Candidates for mandatory locking have the setgid bit set
2198 * but no group execute bit - an otherwise meaningless combination.
2199 */
2200
2201static inline int __mandatory_lock(struct inode *ino)
2202{
2203 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2204}
2205
2206/*
2207 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2208 * otherwise these will be advisory locks
2209 */
2210
2211static inline int mandatory_lock(struct inode *ino)
2212{
2213 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2214}
2215
2216static inline int locks_verify_locked(struct file *file)
2217{
2218 if (mandatory_lock(locks_inode(file)))
2219 return locks_mandatory_locked(file);
2220 return 0;
2221}
2222
2223static inline int locks_verify_truncate(struct inode *inode,
2224 struct file *f,
2225 loff_t size)
2226{
2227 if (!inode->i_flctx || !mandatory_lock(inode))
2228 return 0;
2229
2230 if (size < inode->i_size) {
2231 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2232 F_WRLCK);
2233 } else {
2234 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2235 F_WRLCK);
2236 }
2237}
2238
2239#else /* !CONFIG_MANDATORY_FILE_LOCKING */
2240
2241static inline int locks_mandatory_locked(struct file *file)
2242{
2243 return 0;
2244}
2245
2246static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2247 loff_t start, loff_t end, unsigned char type)
2248{
2249 return 0;
2250}
2251
2252static inline int __mandatory_lock(struct inode *inode)
2253{
2254 return 0;
2255}
2256
2257static inline int mandatory_lock(struct inode *inode)
2258{
2259 return 0;
2260}
2261
2262static inline int locks_verify_locked(struct file *file)
2263{
2264 return 0;
2265}
2266
2267static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2268 size_t size)
2269{
2270 return 0;
2271}
2272
2273#endif /* CONFIG_MANDATORY_FILE_LOCKING */
2274
2275
2276#ifdef CONFIG_FILE_LOCKING
2277static inline int break_lease(struct inode *inode, unsigned int mode)
2278{
2279 /*
2280 * Since this check is lockless, we must ensure that any refcounts
2281 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2282 * could end up racing with tasks trying to set a new lease on this
2283 * file.
2284 */
2285 smp_mb();
2286 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2287 return __break_lease(inode, mode, FL_LEASE);
2288 return 0;
2289}
2290
2291static inline int break_deleg(struct inode *inode, unsigned int mode)
2292{
2293 /*
2294 * Since this check is lockless, we must ensure that any refcounts
2295 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2296 * could end up racing with tasks trying to set a new lease on this
2297 * file.
2298 */
2299 smp_mb();
2300 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2301 return __break_lease(inode, mode, FL_DELEG);
2302 return 0;
2303}
2304
2305static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2306{
2307 int ret;
2308
2309 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2310 if (ret == -EWOULDBLOCK && delegated_inode) {
2311 *delegated_inode = inode;
2312 ihold(inode);
2313 }
2314 return ret;
2315}
2316
2317static inline int break_deleg_wait(struct inode **delegated_inode)
2318{
2319 int ret;
2320
2321 ret = break_deleg(*delegated_inode, O_WRONLY);
2322 iput(*delegated_inode);
2323 *delegated_inode = NULL;
2324 return ret;
2325}
2326
2327static inline int break_layout(struct inode *inode, bool wait)
2328{
2329 smp_mb();
2330 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2331 return __break_lease(inode,
2332 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2333 FL_LAYOUT);
2334 return 0;
2335}
2336
2337#else /* !CONFIG_FILE_LOCKING */
2338static inline int break_lease(struct inode *inode, unsigned int mode)
2339{
2340 return 0;
2341}
2342
2343static inline int break_deleg(struct inode *inode, unsigned int mode)
2344{
2345 return 0;
2346}
2347
2348static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2349{
2350 return 0;
2351}
2352
2353static inline int break_deleg_wait(struct inode **delegated_inode)
2354{
2355 BUG();
2356 return 0;
2357}
2358
2359static inline int break_layout(struct inode *inode, bool wait)
2360{
2361 return 0;
2362}
2363
2364#endif /* CONFIG_FILE_LOCKING */
2365
2366/* fs/open.c */
2367struct audit_names;
2368struct filename {
2369 const char *name; /* pointer to actual string */
2370 const __user char *uptr; /* original userland pointer */
2371 struct audit_names *aname;
2372 int refcnt;
2373 const char iname[];
2374};
2375
2376extern long vfs_truncate(const struct path *, loff_t);
2377extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2378 struct file *filp);
2379extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2380 loff_t len);
2381extern long do_sys_open(int dfd, const char __user *filename, int flags,
2382 umode_t mode);
2383extern struct file *file_open_name(struct filename *, int, umode_t);
2384extern struct file *filp_open(const char *, int, umode_t);
2385extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2386 const char *, int, umode_t);
2387extern struct file * dentry_open(const struct path *, int, const struct cred *);
2388extern int filp_close(struct file *, fl_owner_t id);
2389
2390extern struct filename *getname_flags(const char __user *, int, int *);
2391extern struct filename *getname(const char __user *);
2392extern struct filename *getname_kernel(const char *);
2393extern void putname(struct filename *name);
2394
2395enum {
2396 FILE_CREATED = 1,
2397 FILE_OPENED = 2
2398};
2399extern int finish_open(struct file *file, struct dentry *dentry,
2400 int (*open)(struct inode *, struct file *),
2401 int *opened);
2402extern int finish_no_open(struct file *file, struct dentry *dentry);
2403
2404/* fs/ioctl.c */
2405
2406extern int ioctl_preallocate(struct file *filp, void __user *argp);
2407
2408/* fs/dcache.c */
2409extern void __init vfs_caches_init_early(void);
2410extern void __init vfs_caches_init(void);
2411
2412extern struct kmem_cache *names_cachep;
2413
2414#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2415#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2416
2417#ifdef CONFIG_BLOCK
2418extern int register_blkdev(unsigned int, const char *);
2419extern void unregister_blkdev(unsigned int, const char *);
2420extern void bdev_unhash_inode(dev_t dev);
2421extern struct block_device *bdget(dev_t);
2422extern struct block_device *bdgrab(struct block_device *bdev);
2423extern void bd_set_size(struct block_device *, loff_t size);
2424extern void bd_forget(struct inode *inode);
2425extern void bdput(struct block_device *);
2426extern void invalidate_bdev(struct block_device *);
2427extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2428extern int sync_blockdev(struct block_device *bdev);
2429extern void kill_bdev(struct block_device *);
2430extern struct super_block *freeze_bdev(struct block_device *);
2431extern void emergency_thaw_all(void);
2432extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2433extern int fsync_bdev(struct block_device *);
2434
2435extern struct super_block *blockdev_superblock;
2436
2437static inline bool sb_is_blkdev_sb(struct super_block *sb)
2438{
2439 return sb == blockdev_superblock;
2440}
2441#else
2442static inline void bd_forget(struct inode *inode) {}
2443static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2444static inline void kill_bdev(struct block_device *bdev) {}
2445static inline void invalidate_bdev(struct block_device *bdev) {}
2446
2447static inline struct super_block *freeze_bdev(struct block_device *sb)
2448{
2449 return NULL;
2450}
2451
2452static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2453{
2454 return 0;
2455}
2456
2457static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2458{
2459}
2460
2461static inline bool sb_is_blkdev_sb(struct super_block *sb)
2462{
2463 return false;
2464}
2465#endif
2466extern int sync_filesystem(struct super_block *);
2467extern const struct file_operations def_blk_fops;
2468extern const struct file_operations def_chr_fops;
2469#ifdef CONFIG_BLOCK
2470extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2471extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2472extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2473extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2474extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2475 void *holder);
2476extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2477 void *holder);
2478extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2479extern int __blkdev_reread_part(struct block_device *bdev);
2480extern int blkdev_reread_part(struct block_device *bdev);
2481
2482#ifdef CONFIG_SYSFS
2483extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2484extern void bd_unlink_disk_holder(struct block_device *bdev,
2485 struct gendisk *disk);
2486#else
2487static inline int bd_link_disk_holder(struct block_device *bdev,
2488 struct gendisk *disk)
2489{
2490 return 0;
2491}
2492static inline void bd_unlink_disk_holder(struct block_device *bdev,
2493 struct gendisk *disk)
2494{
2495}
2496#endif
2497#endif
2498
2499/* fs/char_dev.c */
2500#define CHRDEV_MAJOR_MAX 512
2501/* Marks the bottom of the first segment of free char majors */
2502#define CHRDEV_MAJOR_DYN_END 234
2503/* Marks the top and bottom of the second segment of free char majors */
2504#define CHRDEV_MAJOR_DYN_EXT_START 511
2505#define CHRDEV_MAJOR_DYN_EXT_END 384
2506
2507extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2508extern int register_chrdev_region(dev_t, unsigned, const char *);
2509extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2510 unsigned int count, const char *name,
2511 const struct file_operations *fops);
2512extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2513 unsigned int count, const char *name);
2514extern void unregister_chrdev_region(dev_t, unsigned);
2515extern void chrdev_show(struct seq_file *,off_t);
2516
2517static inline int register_chrdev(unsigned int major, const char *name,
2518 const struct file_operations *fops)
2519{
2520 return __register_chrdev(major, 0, 256, name, fops);
2521}
2522
2523static inline void unregister_chrdev(unsigned int major, const char *name)
2524{
2525 __unregister_chrdev(major, 0, 256, name);
2526}
2527
2528/* fs/block_dev.c */
2529#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2530#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2531
2532#ifdef CONFIG_BLOCK
2533#define BLKDEV_MAJOR_MAX 512
2534extern const char *__bdevname(dev_t, char *buffer);
2535extern const char *bdevname(struct block_device *bdev, char *buffer);
2536extern struct block_device *lookup_bdev(const char *);
2537extern void blkdev_show(struct seq_file *,off_t);
2538
2539#else
2540#define BLKDEV_MAJOR_MAX 0
2541#endif
2542
2543extern void init_special_inode(struct inode *, umode_t, dev_t);
2544
2545/* Invalid inode operations -- fs/bad_inode.c */
2546extern void make_bad_inode(struct inode *);
2547extern bool is_bad_inode(struct inode *);
2548
2549#ifdef CONFIG_BLOCK
2550extern void check_disk_size_change(struct gendisk *disk,
2551 struct block_device *bdev);
2552extern int revalidate_disk(struct gendisk *);
2553extern int check_disk_change(struct block_device *);
2554extern int __invalidate_device(struct block_device *, bool);
2555extern int invalidate_partition(struct gendisk *, int);
2556#endif
2557unsigned long invalidate_mapping_pages(struct address_space *mapping,
2558 pgoff_t start, pgoff_t end);
2559
2560static inline void invalidate_remote_inode(struct inode *inode)
2561{
2562 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2563 S_ISLNK(inode->i_mode))
2564 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2565}
2566extern int invalidate_inode_pages2(struct address_space *mapping);
2567extern int invalidate_inode_pages2_range(struct address_space *mapping,
2568 pgoff_t start, pgoff_t end);
2569extern int write_inode_now(struct inode *, int);
2570extern int filemap_fdatawrite(struct address_space *);
2571extern int filemap_flush(struct address_space *);
2572extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2573extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2574 loff_t lend);
2575
2576static inline int filemap_fdatawait(struct address_space *mapping)
2577{
2578 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2579}
2580
2581extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2582 loff_t lend);
2583extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2584 loff_t lend);
2585extern int filemap_write_and_wait(struct address_space *mapping);
2586extern int filemap_write_and_wait_range(struct address_space *mapping,
2587 loff_t lstart, loff_t lend);
2588extern int __filemap_fdatawrite_range(struct address_space *mapping,
2589 loff_t start, loff_t end, int sync_mode);
2590extern int filemap_fdatawrite_range(struct address_space *mapping,
2591 loff_t start, loff_t end);
2592extern int filemap_check_errors(struct address_space *mapping);
2593extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2594
2595extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2596 loff_t lend);
2597extern int __must_check file_check_and_advance_wb_err(struct file *file);
2598extern int __must_check file_write_and_wait_range(struct file *file,
2599 loff_t start, loff_t end);
2600
2601static inline int file_write_and_wait(struct file *file)
2602{
2603 return file_write_and_wait_range(file, 0, LLONG_MAX);
2604}
2605
2606/**
2607 * filemap_set_wb_err - set a writeback error on an address_space
2608 * @mapping: mapping in which to set writeback error
2609 * @err: error to be set in mapping
2610 *
2611 * When writeback fails in some way, we must record that error so that
2612 * userspace can be informed when fsync and the like are called. We endeavor
2613 * to report errors on any file that was open at the time of the error. Some
2614 * internal callers also need to know when writeback errors have occurred.
2615 *
2616 * When a writeback error occurs, most filesystems will want to call
2617 * filemap_set_wb_err to record the error in the mapping so that it will be
2618 * automatically reported whenever fsync is called on the file.
2619 */
2620static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2621{
2622 /* Fastpath for common case of no error */
2623 if (unlikely(err))
2624 __filemap_set_wb_err(mapping, err);
2625}
2626
2627/**
2628 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2629 * @mapping: mapping to check for writeback errors
2630 * @since: previously-sampled errseq_t
2631 *
2632 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2633 * the given value was sampled.
2634 *
2635 * If it has then report the latest error set, otherwise return 0.
2636 */
2637static inline int filemap_check_wb_err(struct address_space *mapping,
2638 errseq_t since)
2639{
2640 return errseq_check(&mapping->wb_err, since);
2641}
2642
2643/**
2644 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2645 * @mapping: mapping to be sampled
2646 *
2647 * Writeback errors are always reported relative to a particular sample point
2648 * in the past. This function provides those sample points.
2649 */
2650static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2651{
2652 return errseq_sample(&mapping->wb_err);
2653}
2654
2655extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2656 int datasync);
2657extern int vfs_fsync(struct file *file, int datasync);
2658
2659/*
2660 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2661 * to already be updated for the write, and will return either the amount
2662 * of bytes passed in, or an error if syncing the file failed.
2663 */
2664static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2665{
2666 if (iocb->ki_flags & IOCB_DSYNC) {
2667 int ret = vfs_fsync_range(iocb->ki_filp,
2668 iocb->ki_pos - count, iocb->ki_pos - 1,
2669 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2670 if (ret)
2671 return ret;
2672 }
2673
2674 return count;
2675}
2676
2677extern void emergency_sync(void);
2678extern void emergency_remount(void);
2679#ifdef CONFIG_BLOCK
2680extern sector_t bmap(struct inode *, sector_t);
2681#endif
2682extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2683extern int inode_permission(struct inode *, int);
2684extern int __inode_permission(struct inode *, int);
2685extern int generic_permission(struct inode *, int);
2686extern int __check_sticky(struct inode *dir, struct inode *inode);
2687
2688static inline bool execute_ok(struct inode *inode)
2689{
2690 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2691}
2692
2693static inline void file_start_write(struct file *file)
2694{
2695 if (!S_ISREG(file_inode(file)->i_mode))
2696 return;
2697 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2698}
2699
2700static inline bool file_start_write_trylock(struct file *file)
2701{
2702 if (!S_ISREG(file_inode(file)->i_mode))
2703 return true;
2704 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2705}
2706
2707static inline void file_end_write(struct file *file)
2708{
2709 if (!S_ISREG(file_inode(file)->i_mode))
2710 return;
2711 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2712}
2713
2714static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2715 struct file *file_out, loff_t pos_out,
2716 u64 len)
2717{
2718 int ret;
2719
2720 file_start_write(file_out);
2721 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2722 file_end_write(file_out);
2723
2724 return ret;
2725}
2726
2727/*
2728 * get_write_access() gets write permission for a file.
2729 * put_write_access() releases this write permission.
2730 * This is used for regular files.
2731 * We cannot support write (and maybe mmap read-write shared) accesses and
2732 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2733 * can have the following values:
2734 * 0: no writers, no VM_DENYWRITE mappings
2735 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2736 * > 0: (i_writecount) users are writing to the file.
2737 *
2738 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2739 * except for the cases where we don't hold i_writecount yet. Then we need to
2740 * use {get,deny}_write_access() - these functions check the sign and refuse
2741 * to do the change if sign is wrong.
2742 */
2743static inline int get_write_access(struct inode *inode)
2744{
2745 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2746}
2747static inline int deny_write_access(struct file *file)
2748{
2749 struct inode *inode = file_inode(file);
2750 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2751}
2752static inline void put_write_access(struct inode * inode)
2753{
2754 atomic_dec(&inode->i_writecount);
2755}
2756static inline void allow_write_access(struct file *file)
2757{
2758 if (file)
2759 atomic_inc(&file_inode(file)->i_writecount);
2760}
2761static inline bool inode_is_open_for_write(const struct inode *inode)
2762{
2763 return atomic_read(&inode->i_writecount) > 0;
2764}
2765
2766#ifdef CONFIG_IMA
2767static inline void i_readcount_dec(struct inode *inode)
2768{
2769 BUG_ON(!atomic_read(&inode->i_readcount));
2770 atomic_dec(&inode->i_readcount);
2771}
2772static inline void i_readcount_inc(struct inode *inode)
2773{
2774 atomic_inc(&inode->i_readcount);
2775}
2776#else
2777static inline void i_readcount_dec(struct inode *inode)
2778{
2779 return;
2780}
2781static inline void i_readcount_inc(struct inode *inode)
2782{
2783 return;
2784}
2785#endif
2786extern int do_pipe_flags(int *, int);
2787
2788#define __kernel_read_file_id(id) \
2789 id(UNKNOWN, unknown) \
2790 id(FIRMWARE, firmware) \
2791 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2792 id(MODULE, kernel-module) \
2793 id(KEXEC_IMAGE, kexec-image) \
2794 id(KEXEC_INITRAMFS, kexec-initramfs) \
2795 id(POLICY, security-policy) \
2796 id(MAX_ID, )
2797
2798#define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2799#define __fid_stringify(dummy, str) #str,
2800
2801enum kernel_read_file_id {
2802 __kernel_read_file_id(__fid_enumify)
2803};
2804
2805static const char * const kernel_read_file_str[] = {
2806 __kernel_read_file_id(__fid_stringify)
2807};
2808
2809static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2810{
2811 if ((unsigned)id >= READING_MAX_ID)
2812 return kernel_read_file_str[READING_UNKNOWN];
2813
2814 return kernel_read_file_str[id];
2815}
2816
2817extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2818 enum kernel_read_file_id);
2819extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2820 enum kernel_read_file_id);
2821extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2822 enum kernel_read_file_id);
2823extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2824extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2825extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2826extern struct file * open_exec(const char *);
2827
2828/* fs/dcache.c -- generic fs support functions */
2829extern bool is_subdir(struct dentry *, struct dentry *);
2830extern bool path_is_under(const struct path *, const struct path *);
2831
2832extern char *file_path(struct file *, char *, int);
2833
2834#include <linux/err.h>
2835
2836/* needed for stackable file system support */
2837extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2838
2839extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2840
2841extern int inode_init_always(struct super_block *, struct inode *);
2842extern void inode_init_once(struct inode *);
2843extern void address_space_init_once(struct address_space *mapping);
2844extern struct inode * igrab(struct inode *);
2845extern ino_t iunique(struct super_block *, ino_t);
2846extern int inode_needs_sync(struct inode *inode);
2847extern int generic_delete_inode(struct inode *inode);
2848static inline int generic_drop_inode(struct inode *inode)
2849{
2850 return !inode->i_nlink || inode_unhashed(inode);
2851}
2852
2853extern struct inode *ilookup5_nowait(struct super_block *sb,
2854 unsigned long hashval, int (*test)(struct inode *, void *),
2855 void *data);
2856extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2857 int (*test)(struct inode *, void *), void *data);
2858extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2859
2860extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2861extern struct inode * iget_locked(struct super_block *, unsigned long);
2862extern struct inode *find_inode_nowait(struct super_block *,
2863 unsigned long,
2864 int (*match)(struct inode *,
2865 unsigned long, void *),
2866 void *data);
2867extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2868extern int insert_inode_locked(struct inode *);
2869#ifdef CONFIG_DEBUG_LOCK_ALLOC
2870extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2871#else
2872static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2873#endif
2874extern void unlock_new_inode(struct inode *);
2875extern unsigned int get_next_ino(void);
2876extern void evict_inodes(struct super_block *sb);
2877
2878extern void __iget(struct inode * inode);
2879extern void iget_failed(struct inode *);
2880extern void clear_inode(struct inode *);
2881extern void __destroy_inode(struct inode *);
2882extern struct inode *new_inode_pseudo(struct super_block *sb);
2883extern struct inode *new_inode(struct super_block *sb);
2884extern void free_inode_nonrcu(struct inode *inode);
2885extern int should_remove_suid(struct dentry *);
2886extern int file_remove_privs(struct file *);
2887
2888extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2889static inline void insert_inode_hash(struct inode *inode)
2890{
2891 __insert_inode_hash(inode, inode->i_ino);
2892}
2893
2894extern void __remove_inode_hash(struct inode *);
2895static inline void remove_inode_hash(struct inode *inode)
2896{
2897 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2898 __remove_inode_hash(inode);
2899}
2900
2901extern void inode_sb_list_add(struct inode *inode);
2902
2903#ifdef CONFIG_BLOCK
2904extern int bdev_read_only(struct block_device *);
2905#endif
2906extern int set_blocksize(struct block_device *, int);
2907extern int sb_set_blocksize(struct super_block *, int);
2908extern int sb_min_blocksize(struct super_block *, int);
2909
2910extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2911extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2912extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2913extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2914extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2915extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2916extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2917extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2918
2919ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2920 rwf_t flags);
2921ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2922 rwf_t flags);
2923
2924/* fs/block_dev.c */
2925extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2926extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2927extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2928 int datasync);
2929extern void block_sync_page(struct page *page);
2930
2931/* fs/splice.c */
2932extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2933 struct pipe_inode_info *, size_t, unsigned int);
2934extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2935 struct file *, loff_t *, size_t, unsigned int);
2936extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2937 struct file *out, loff_t *, size_t len, unsigned int flags);
2938extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2939 loff_t *opos, size_t len, unsigned int flags);
2940
2941
2942extern void
2943file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2944extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2945extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2946extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2947extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2948extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2949 int whence, loff_t maxsize, loff_t eof);
2950extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2951 int whence, loff_t size);
2952extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2953extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2954extern int generic_file_open(struct inode * inode, struct file * filp);
2955extern int nonseekable_open(struct inode * inode, struct file * filp);
2956
2957#ifdef CONFIG_BLOCK
2958typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2959 loff_t file_offset);
2960
2961enum {
2962 /* need locking between buffered and direct access */
2963 DIO_LOCKING = 0x01,
2964
2965 /* filesystem does not support filling holes */
2966 DIO_SKIP_HOLES = 0x02,
2967
2968 /* filesystem can handle aio writes beyond i_size */
2969 DIO_ASYNC_EXTEND = 0x04,
2970
2971 /* inode/fs/bdev does not need truncate protection */
2972 DIO_SKIP_DIO_COUNT = 0x08,
2973};
2974
2975void dio_end_io(struct bio *bio);
2976
2977ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2978 struct block_device *bdev, struct iov_iter *iter,
2979 get_block_t get_block,
2980 dio_iodone_t end_io, dio_submit_t submit_io,
2981 int flags);
2982
2983static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2984 struct inode *inode,
2985 struct iov_iter *iter,
2986 get_block_t get_block)
2987{
2988 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2989 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2990}
2991#endif
2992
2993void inode_dio_wait(struct inode *inode);
2994
2995/*
2996 * inode_dio_begin - signal start of a direct I/O requests
2997 * @inode: inode the direct I/O happens on
2998 *
2999 * This is called once we've finished processing a direct I/O request,
3000 * and is used to wake up callers waiting for direct I/O to be quiesced.
3001 */
3002static inline void inode_dio_begin(struct inode *inode)
3003{
3004 atomic_inc(&inode->i_dio_count);
3005}
3006
3007/*
3008 * inode_dio_end - signal finish of a direct I/O requests
3009 * @inode: inode the direct I/O happens on
3010 *
3011 * This is called once we've finished processing a direct I/O request,
3012 * and is used to wake up callers waiting for direct I/O to be quiesced.
3013 */
3014static inline void inode_dio_end(struct inode *inode)
3015{
3016 if (atomic_dec_and_test(&inode->i_dio_count))
3017 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3018}
3019
3020extern void inode_set_flags(struct inode *inode, unsigned int flags,
3021 unsigned int mask);
3022
3023extern const struct file_operations generic_ro_fops;
3024
3025#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3026
3027extern int readlink_copy(char __user *, int, const char *);
3028extern int page_readlink(struct dentry *, char __user *, int);
3029extern const char *page_get_link(struct dentry *, struct inode *,
3030 struct delayed_call *);
3031extern void page_put_link(void *);
3032extern int __page_symlink(struct inode *inode, const char *symname, int len,
3033 int nofs);
3034extern int page_symlink(struct inode *inode, const char *symname, int len);
3035extern const struct inode_operations page_symlink_inode_operations;
3036extern void kfree_link(void *);
3037extern void generic_fillattr(struct inode *, struct kstat *);
3038extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3039extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3040void __inode_add_bytes(struct inode *inode, loff_t bytes);
3041void inode_add_bytes(struct inode *inode, loff_t bytes);
3042void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3043void inode_sub_bytes(struct inode *inode, loff_t bytes);
3044static inline loff_t __inode_get_bytes(struct inode *inode)
3045{
3046 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3047}
3048loff_t inode_get_bytes(struct inode *inode);
3049void inode_set_bytes(struct inode *inode, loff_t bytes);
3050const char *simple_get_link(struct dentry *, struct inode *,
3051 struct delayed_call *);
3052extern const struct inode_operations simple_symlink_inode_operations;
3053
3054extern int iterate_dir(struct file *, struct dir_context *);
3055
3056extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3057extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3058
3059static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3060{
3061 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3062 stat, STATX_BASIC_STATS);
3063}
3064static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3065{
3066 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3067 stat, STATX_BASIC_STATS);
3068}
3069static inline int vfs_fstatat(int dfd, const char __user *filename,
3070 struct kstat *stat, int flags)
3071{
3072 return vfs_statx(dfd, filename, flags, stat, STATX_BASIC_STATS);
3073}
3074static inline int vfs_fstat(int fd, struct kstat *stat)
3075{
3076 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3077}
3078
3079
3080extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3081extern int vfs_readlink(struct dentry *, char __user *, int);
3082
3083extern int __generic_block_fiemap(struct inode *inode,
3084 struct fiemap_extent_info *fieinfo,
3085 loff_t start, loff_t len,
3086 get_block_t *get_block);
3087extern int generic_block_fiemap(struct inode *inode,
3088 struct fiemap_extent_info *fieinfo, u64 start,
3089 u64 len, get_block_t *get_block);
3090
3091extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3092extern void put_filesystem(struct file_system_type *fs);
3093extern struct file_system_type *get_fs_type(const char *name);
3094extern struct super_block *get_super(struct block_device *);
3095extern struct super_block *get_super_thawed(struct block_device *);
3096extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3097extern struct super_block *get_active_super(struct block_device *bdev);
3098extern void drop_super(struct super_block *sb);
3099extern void drop_super_exclusive(struct super_block *sb);
3100extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3101extern void iterate_supers_type(struct file_system_type *,
3102 void (*)(struct super_block *, void *), void *);
3103
3104extern int dcache_dir_open(struct inode *, struct file *);
3105extern int dcache_dir_close(struct inode *, struct file *);
3106extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3107extern int dcache_readdir(struct file *, struct dir_context *);
3108extern int simple_setattr(struct dentry *, struct iattr *);
3109extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3110extern int simple_statfs(struct dentry *, struct kstatfs *);
3111extern int simple_open(struct inode *inode, struct file *file);
3112extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3113extern int simple_unlink(struct inode *, struct dentry *);
3114extern int simple_rmdir(struct inode *, struct dentry *);
3115extern int simple_rename(struct inode *, struct dentry *,
3116 struct inode *, struct dentry *, unsigned int);
3117extern int noop_fsync(struct file *, loff_t, loff_t, int);
3118extern int simple_empty(struct dentry *);
3119extern int simple_readpage(struct file *file, struct page *page);
3120extern int simple_write_begin(struct file *file, struct address_space *mapping,
3121 loff_t pos, unsigned len, unsigned flags,
3122 struct page **pagep, void **fsdata);
3123extern int simple_write_end(struct file *file, struct address_space *mapping,
3124 loff_t pos, unsigned len, unsigned copied,
3125 struct page *page, void *fsdata);
3126extern int always_delete_dentry(const struct dentry *);
3127extern struct inode *alloc_anon_inode(struct super_block *);
3128extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3129extern const struct dentry_operations simple_dentry_operations;
3130
3131extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3132extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3133extern const struct file_operations simple_dir_operations;
3134extern const struct inode_operations simple_dir_inode_operations;
3135extern void make_empty_dir_inode(struct inode *inode);
3136extern bool is_empty_dir_inode(struct inode *inode);
3137struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3138struct dentry *d_alloc_name(struct dentry *, const char *);
3139extern int simple_fill_super(struct super_block *, unsigned long,
3140 const struct tree_descr *);
3141extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3142extern void simple_release_fs(struct vfsmount **mount, int *count);
3143
3144extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3145 loff_t *ppos, const void *from, size_t available);
3146extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3147 const void __user *from, size_t count);
3148
3149extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3150extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3151
3152extern int generic_check_addressable(unsigned, u64);
3153
3154#ifdef CONFIG_MIGRATION
3155extern int buffer_migrate_page(struct address_space *,
3156 struct page *, struct page *,
3157 enum migrate_mode);
3158#else
3159#define buffer_migrate_page NULL
3160#endif
3161
3162extern int setattr_prepare(struct dentry *, struct iattr *);
3163extern int inode_newsize_ok(const struct inode *, loff_t offset);
3164extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3165
3166extern int file_update_time(struct file *file);
3167
3168static inline bool io_is_direct(struct file *filp)
3169{
3170 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3171}
3172
3173static inline bool vma_is_dax(struct vm_area_struct *vma)
3174{
3175 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3176}
3177
3178static inline int iocb_flags(struct file *file)
3179{
3180 int res = 0;
3181 if (file->f_flags & O_APPEND)
3182 res |= IOCB_APPEND;
3183 if (io_is_direct(file))
3184 res |= IOCB_DIRECT;
3185 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3186 res |= IOCB_DSYNC;
3187 if (file->f_flags & __O_SYNC)
3188 res |= IOCB_SYNC;
3189 return res;
3190}
3191
3192static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3193{
3194 if (unlikely(flags & ~RWF_SUPPORTED))
3195 return -EOPNOTSUPP;
3196
3197 if (flags & RWF_NOWAIT) {
3198 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3199 return -EOPNOTSUPP;
3200 ki->ki_flags |= IOCB_NOWAIT;
3201 }
3202 if (flags & RWF_HIPRI)
3203 ki->ki_flags |= IOCB_HIPRI;
3204 if (flags & RWF_DSYNC)
3205 ki->ki_flags |= IOCB_DSYNC;
3206 if (flags & RWF_SYNC)
3207 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3208 return 0;
3209}
3210
3211static inline ino_t parent_ino(struct dentry *dentry)
3212{
3213 ino_t res;
3214
3215 /*
3216 * Don't strictly need d_lock here? If the parent ino could change
3217 * then surely we'd have a deeper race in the caller?
3218 */
3219 spin_lock(&dentry->d_lock);
3220 res = dentry->d_parent->d_inode->i_ino;
3221 spin_unlock(&dentry->d_lock);
3222 return res;
3223}
3224
3225/* Transaction based IO helpers */
3226
3227/*
3228 * An argresp is stored in an allocated page and holds the
3229 * size of the argument or response, along with its content
3230 */
3231struct simple_transaction_argresp {
3232 ssize_t size;
3233 char data[0];
3234};
3235
3236#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3237
3238char *simple_transaction_get(struct file *file, const char __user *buf,
3239 size_t size);
3240ssize_t simple_transaction_read(struct file *file, char __user *buf,
3241 size_t size, loff_t *pos);
3242int simple_transaction_release(struct inode *inode, struct file *file);
3243
3244void simple_transaction_set(struct file *file, size_t n);
3245
3246/*
3247 * simple attribute files
3248 *
3249 * These attributes behave similar to those in sysfs:
3250 *
3251 * Writing to an attribute immediately sets a value, an open file can be
3252 * written to multiple times.
3253 *
3254 * Reading from an attribute creates a buffer from the value that might get
3255 * read with multiple read calls. When the attribute has been read
3256 * completely, no further read calls are possible until the file is opened
3257 * again.
3258 *
3259 * All attributes contain a text representation of a numeric value
3260 * that are accessed with the get() and set() functions.
3261 */
3262#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3263static int __fops ## _open(struct inode *inode, struct file *file) \
3264{ \
3265 __simple_attr_check_format(__fmt, 0ull); \
3266 return simple_attr_open(inode, file, __get, __set, __fmt); \
3267} \
3268static const struct file_operations __fops = { \
3269 .owner = THIS_MODULE, \
3270 .open = __fops ## _open, \
3271 .release = simple_attr_release, \
3272 .read = simple_attr_read, \
3273 .write = simple_attr_write, \
3274 .llseek = generic_file_llseek, \
3275}
3276
3277static inline __printf(1, 2)
3278void __simple_attr_check_format(const char *fmt, ...)
3279{
3280 /* don't do anything, just let the compiler check the arguments; */
3281}
3282
3283int simple_attr_open(struct inode *inode, struct file *file,
3284 int (*get)(void *, u64 *), int (*set)(void *, u64),
3285 const char *fmt);
3286int simple_attr_release(struct inode *inode, struct file *file);
3287ssize_t simple_attr_read(struct file *file, char __user *buf,
3288 size_t len, loff_t *ppos);
3289ssize_t simple_attr_write(struct file *file, const char __user *buf,
3290 size_t len, loff_t *ppos);
3291
3292struct ctl_table;
3293int proc_nr_files(struct ctl_table *table, int write,
3294 void __user *buffer, size_t *lenp, loff_t *ppos);
3295int proc_nr_dentry(struct ctl_table *table, int write,
3296 void __user *buffer, size_t *lenp, loff_t *ppos);
3297int proc_nr_inodes(struct ctl_table *table, int write,
3298 void __user *buffer, size_t *lenp, loff_t *ppos);
3299int __init get_filesystem_list(char *buf);
3300
3301#define __FMODE_EXEC ((__force int) FMODE_EXEC)
3302#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3303
3304#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3305#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3306 (flag & __FMODE_NONOTIFY)))
3307
3308static inline bool is_sxid(umode_t mode)
3309{
3310 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3311}
3312
3313static inline int check_sticky(struct inode *dir, struct inode *inode)
3314{
3315 if (!(dir->i_mode & S_ISVTX))
3316 return 0;
3317
3318 return __check_sticky(dir, inode);
3319}
3320
3321static inline void inode_has_no_xattr(struct inode *inode)
3322{
3323 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3324 inode->i_flags |= S_NOSEC;
3325}
3326
3327static inline bool is_root_inode(struct inode *inode)
3328{
3329 return inode == inode->i_sb->s_root->d_inode;
3330}
3331
3332static inline bool dir_emit(struct dir_context *ctx,
3333 const char *name, int namelen,
3334 u64 ino, unsigned type)
3335{
3336 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3337}
3338static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3339{
3340 return ctx->actor(ctx, ".", 1, ctx->pos,
3341 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3342}
3343static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3344{
3345 return ctx->actor(ctx, "..", 2, ctx->pos,
3346 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3347}
3348static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3349{
3350 if (ctx->pos == 0) {
3351 if (!dir_emit_dot(file, ctx))
3352 return false;
3353 ctx->pos = 1;
3354 }
3355 if (ctx->pos == 1) {
3356 if (!dir_emit_dotdot(file, ctx))
3357 return false;
3358 ctx->pos = 2;
3359 }
3360 return true;
3361}
3362static inline bool dir_relax(struct inode *inode)
3363{
3364 inode_unlock(inode);
3365 inode_lock(inode);
3366 return !IS_DEADDIR(inode);
3367}
3368
3369static inline bool dir_relax_shared(struct inode *inode)
3370{
3371 inode_unlock_shared(inode);
3372 inode_lock_shared(inode);
3373 return !IS_DEADDIR(inode);
3374}
3375
3376extern bool path_noexec(const struct path *path);
3377extern void inode_nohighmem(struct inode *inode);
3378
3379#endif /* _LINUX_FS_H */