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/xarray.h> 17 #include <linux/rbtree.h> 18 #include <linux/init.h> 19 #include <linux/pid.h> 20 #include <linux/bug.h> 21 #include <linux/mutex.h> 22 #include <linux/rwsem.h> 23 #include <linux/mm_types.h> 24 #include <linux/capability.h> 25 #include <linux/semaphore.h> 26 #include <linux/fcntl.h> 27 #include <linux/fiemap.h> 28 #include <linux/rculist_bl.h> 29 #include <linux/atomic.h> 30 #include <linux/shrinker.h> 31 #include <linux/migrate_mode.h> 32 #include <linux/uidgid.h> 33 #include <linux/lockdep.h> 34 #include <linux/percpu-rwsem.h> 35 #include <linux/workqueue.h> 36 #include <linux/delayed_call.h> 37 #include <linux/uuid.h> 38 #include <linux/errseq.h> 39 #include <linux/ioprio.h> 40 #include <linux/fs_types.h> 41 #include <linux/build_bug.h> 42 #include <linux/stddef.h> 43 44 #include <asm/byteorder.h> 45 #include <uapi/linux/fs.h> 46 47 struct backing_dev_info; 48 struct bdi_writeback; 49 struct bio; 50 struct export_operations; 51 struct hd_geometry; 52 struct iovec; 53 struct kiocb; 54 struct kobject; 55 struct pipe_inode_info; 56 struct poll_table_struct; 57 struct kstatfs; 58 struct vm_area_struct; 59 struct vfsmount; 60 struct cred; 61 struct swap_info_struct; 62 struct seq_file; 63 struct workqueue_struct; 64 struct iov_iter; 65 struct fscrypt_info; 66 struct fscrypt_operations; 67 struct fsverity_info; 68 struct fsverity_operations; 69 struct fs_context; 70 struct fs_parameter_description; 71 72 extern void __init inode_init(void); 73 extern void __init inode_init_early(void); 74 extern void __init files_init(void); 75 extern void __init files_maxfiles_init(void); 76 77 extern struct files_stat_struct files_stat; 78 extern unsigned long get_max_files(void); 79 extern unsigned int sysctl_nr_open; 80 extern struct inodes_stat_t inodes_stat; 81 extern int leases_enable, lease_break_time; 82 extern int sysctl_protected_symlinks; 83 extern int sysctl_protected_hardlinks; 84 extern int sysctl_protected_fifos; 85 extern int sysctl_protected_regular; 86 87 typedef __kernel_rwf_t rwf_t; 88 89 struct buffer_head; 90 typedef int (get_block_t)(struct inode *inode, sector_t iblock, 91 struct buffer_head *bh_result, int create); 92 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset, 93 ssize_t bytes, void *private); 94 95 #define MAY_EXEC 0x00000001 96 #define MAY_WRITE 0x00000002 97 #define MAY_READ 0x00000004 98 #define MAY_APPEND 0x00000008 99 #define MAY_ACCESS 0x00000010 100 #define MAY_OPEN 0x00000020 101 #define MAY_CHDIR 0x00000040 102 /* called from RCU mode, don't block */ 103 #define MAY_NOT_BLOCK 0x00000080 104 105 /* 106 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond 107 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open() 108 */ 109 110 /* file is open for reading */ 111 #define FMODE_READ ((__force fmode_t)0x1) 112 /* file is open for writing */ 113 #define FMODE_WRITE ((__force fmode_t)0x2) 114 /* file is seekable */ 115 #define FMODE_LSEEK ((__force fmode_t)0x4) 116 /* file can be accessed using pread */ 117 #define FMODE_PREAD ((__force fmode_t)0x8) 118 /* file can be accessed using pwrite */ 119 #define FMODE_PWRITE ((__force fmode_t)0x10) 120 /* File is opened for execution with sys_execve / sys_uselib */ 121 #define FMODE_EXEC ((__force fmode_t)0x20) 122 /* File is opened with O_NDELAY (only set for block devices) */ 123 #define FMODE_NDELAY ((__force fmode_t)0x40) 124 /* File is opened with O_EXCL (only set for block devices) */ 125 #define FMODE_EXCL ((__force fmode_t)0x80) 126 /* File is opened using open(.., 3, ..) and is writeable only for ioctls 127 (specialy hack for floppy.c) */ 128 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100) 129 /* 32bit hashes as llseek() offset (for directories) */ 130 #define FMODE_32BITHASH ((__force fmode_t)0x200) 131 /* 64bit hashes as llseek() offset (for directories) */ 132 #define FMODE_64BITHASH ((__force fmode_t)0x400) 133 134 /* 135 * Don't update ctime and mtime. 136 * 137 * Currently a special hack for the XFS open_by_handle ioctl, but we'll 138 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. 139 */ 140 #define FMODE_NOCMTIME ((__force fmode_t)0x800) 141 142 /* Expect random access pattern */ 143 #define FMODE_RANDOM ((__force fmode_t)0x1000) 144 145 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */ 146 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000) 147 148 /* File is opened with O_PATH; almost nothing can be done with it */ 149 #define FMODE_PATH ((__force fmode_t)0x4000) 150 151 /* File needs atomic accesses to f_pos */ 152 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) 153 /* Write access to underlying fs */ 154 #define FMODE_WRITER ((__force fmode_t)0x10000) 155 /* Has read method(s) */ 156 #define FMODE_CAN_READ ((__force fmode_t)0x20000) 157 /* Has write method(s) */ 158 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000) 159 160 #define FMODE_OPENED ((__force fmode_t)0x80000) 161 #define FMODE_CREATED ((__force fmode_t)0x100000) 162 163 /* File is stream-like */ 164 #define FMODE_STREAM ((__force fmode_t)0x200000) 165 166 /* File was opened by fanotify and shouldn't generate fanotify events */ 167 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) 168 169 /* File is capable of returning -EAGAIN if I/O will block */ 170 #define FMODE_NOWAIT ((__force fmode_t)0x8000000) 171 172 /* File represents mount that needs unmounting */ 173 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000) 174 175 /* File does not contribute to nr_files count */ 176 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000) 177 178 /* 179 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector 180 * that indicates that they should check the contents of the iovec are 181 * valid, but not check the memory that the iovec elements 182 * points too. 183 */ 184 #define CHECK_IOVEC_ONLY -1 185 186 /* 187 * Attribute flags. These should be or-ed together to figure out what 188 * has been changed! 189 */ 190 #define ATTR_MODE (1 << 0) 191 #define ATTR_UID (1 << 1) 192 #define ATTR_GID (1 << 2) 193 #define ATTR_SIZE (1 << 3) 194 #define ATTR_ATIME (1 << 4) 195 #define ATTR_MTIME (1 << 5) 196 #define ATTR_CTIME (1 << 6) 197 #define ATTR_ATIME_SET (1 << 7) 198 #define ATTR_MTIME_SET (1 << 8) 199 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ 200 #define ATTR_KILL_SUID (1 << 11) 201 #define ATTR_KILL_SGID (1 << 12) 202 #define ATTR_FILE (1 << 13) 203 #define ATTR_KILL_PRIV (1 << 14) 204 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ 205 #define ATTR_TIMES_SET (1 << 16) 206 #define ATTR_TOUCH (1 << 17) 207 208 /* 209 * Whiteout is represented by a char device. The following constants define the 210 * mode and device number to use. 211 */ 212 #define WHITEOUT_MODE 0 213 #define WHITEOUT_DEV 0 214 215 /* 216 * This is the Inode Attributes structure, used for notify_change(). It 217 * uses the above definitions as flags, to know which values have changed. 218 * Also, in this manner, a Filesystem can look at only the values it cares 219 * about. Basically, these are the attributes that the VFS layer can 220 * request to change from the FS layer. 221 * 222 * Derek Atkins <warlord@MIT.EDU> 94-10-20 223 */ 224 struct iattr { 225 unsigned int ia_valid; 226 umode_t ia_mode; 227 kuid_t ia_uid; 228 kgid_t ia_gid; 229 loff_t ia_size; 230 struct timespec64 ia_atime; 231 struct timespec64 ia_mtime; 232 struct timespec64 ia_ctime; 233 234 /* 235 * Not an attribute, but an auxiliary info for filesystems wanting to 236 * implement an ftruncate() like method. NOTE: filesystem should 237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). 238 */ 239 struct file *ia_file; 240 }; 241 242 /* 243 * Includes for diskquotas. 244 */ 245 #include <linux/quota.h> 246 247 /* 248 * Maximum number of layers of fs stack. Needs to be limited to 249 * prevent kernel stack overflow 250 */ 251 #define FILESYSTEM_MAX_STACK_DEPTH 2 252 253 /** 254 * enum positive_aop_returns - aop return codes with specific semantics 255 * 256 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has 257 * completed, that the page is still locked, and 258 * should be considered active. The VM uses this hint 259 * to return the page to the active list -- it won't 260 * be a candidate for writeback again in the near 261 * future. Other callers must be careful to unlock 262 * the page if they get this return. Returned by 263 * writepage(); 264 * 265 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has 266 * unlocked it and the page might have been truncated. 267 * The caller should back up to acquiring a new page and 268 * trying again. The aop will be taking reasonable 269 * precautions not to livelock. If the caller held a page 270 * reference, it should drop it before retrying. Returned 271 * by readpage(). 272 * 273 * address_space_operation functions return these large constants to indicate 274 * special semantics to the caller. These are much larger than the bytes in a 275 * page to allow for functions that return the number of bytes operated on in a 276 * given page. 277 */ 278 279 enum positive_aop_returns { 280 AOP_WRITEPAGE_ACTIVATE = 0x80000, 281 AOP_TRUNCATED_PAGE = 0x80001, 282 }; 283 284 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */ 285 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct 286 * helper code (eg buffer layer) 287 * to clear GFP_FS from alloc */ 288 289 /* 290 * oh the beauties of C type declarations. 291 */ 292 struct page; 293 struct address_space; 294 struct writeback_control; 295 296 /* 297 * Write life time hint values. 298 * Stored in struct inode as u8. 299 */ 300 enum rw_hint { 301 WRITE_LIFE_NOT_SET = 0, 302 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE, 303 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT, 304 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM, 305 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG, 306 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME, 307 }; 308 309 #define IOCB_EVENTFD (1 << 0) 310 #define IOCB_APPEND (1 << 1) 311 #define IOCB_DIRECT (1 << 2) 312 #define IOCB_HIPRI (1 << 3) 313 #define IOCB_DSYNC (1 << 4) 314 #define IOCB_SYNC (1 << 5) 315 #define IOCB_WRITE (1 << 6) 316 #define IOCB_NOWAIT (1 << 7) 317 318 struct kiocb { 319 struct file *ki_filp; 320 321 /* The 'ki_filp' pointer is shared in a union for aio */ 322 randomized_struct_fields_start 323 324 loff_t ki_pos; 325 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2); 326 void *private; 327 int ki_flags; 328 u16 ki_hint; 329 u16 ki_ioprio; /* See linux/ioprio.h */ 330 unsigned int ki_cookie; /* for ->iopoll */ 331 332 randomized_struct_fields_end 333 }; 334 335 static inline bool is_sync_kiocb(struct kiocb *kiocb) 336 { 337 return kiocb->ki_complete == NULL; 338 } 339 340 /* 341 * "descriptor" for what we're up to with a read. 342 * This allows us to use the same read code yet 343 * have multiple different users of the data that 344 * we read from a file. 345 * 346 * The simplest case just copies the data to user 347 * mode. 348 */ 349 typedef struct { 350 size_t written; 351 size_t count; 352 union { 353 char __user *buf; 354 void *data; 355 } arg; 356 int error; 357 } read_descriptor_t; 358 359 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, 360 unsigned long, unsigned long); 361 362 struct address_space_operations { 363 int (*writepage)(struct page *page, struct writeback_control *wbc); 364 int (*readpage)(struct file *, struct page *); 365 366 /* Write back some dirty pages from this mapping. */ 367 int (*writepages)(struct address_space *, struct writeback_control *); 368 369 /* Set a page dirty. Return true if this dirtied it */ 370 int (*set_page_dirty)(struct page *page); 371 372 /* 373 * Reads in the requested pages. Unlike ->readpage(), this is 374 * PURELY used for read-ahead!. 375 */ 376 int (*readpages)(struct file *filp, struct address_space *mapping, 377 struct list_head *pages, unsigned nr_pages); 378 379 int (*write_begin)(struct file *, struct address_space *mapping, 380 loff_t pos, unsigned len, unsigned flags, 381 struct page **pagep, void **fsdata); 382 int (*write_end)(struct file *, struct address_space *mapping, 383 loff_t pos, unsigned len, unsigned copied, 384 struct page *page, void *fsdata); 385 386 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ 387 sector_t (*bmap)(struct address_space *, sector_t); 388 void (*invalidatepage) (struct page *, unsigned int, unsigned int); 389 int (*releasepage) (struct page *, gfp_t); 390 void (*freepage)(struct page *); 391 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); 392 /* 393 * migrate the contents of a page to the specified target. If 394 * migrate_mode is MIGRATE_ASYNC, it must not block. 395 */ 396 int (*migratepage) (struct address_space *, 397 struct page *, struct page *, enum migrate_mode); 398 bool (*isolate_page)(struct page *, isolate_mode_t); 399 void (*putback_page)(struct page *); 400 int (*launder_page) (struct page *); 401 int (*is_partially_uptodate) (struct page *, unsigned long, 402 unsigned long); 403 void (*is_dirty_writeback) (struct page *, bool *, bool *); 404 int (*error_remove_page)(struct address_space *, struct page *); 405 406 /* swapfile support */ 407 int (*swap_activate)(struct swap_info_struct *sis, struct file *file, 408 sector_t *span); 409 void (*swap_deactivate)(struct file *file); 410 }; 411 412 extern const struct address_space_operations empty_aops; 413 414 /* 415 * pagecache_write_begin/pagecache_write_end must be used by general code 416 * to write into the pagecache. 417 */ 418 int pagecache_write_begin(struct file *, struct address_space *mapping, 419 loff_t pos, unsigned len, unsigned flags, 420 struct page **pagep, void **fsdata); 421 422 int pagecache_write_end(struct file *, struct address_space *mapping, 423 loff_t pos, unsigned len, unsigned copied, 424 struct page *page, void *fsdata); 425 426 /** 427 * struct address_space - Contents of a cacheable, mappable object. 428 * @host: Owner, either the inode or the block_device. 429 * @i_pages: Cached pages. 430 * @gfp_mask: Memory allocation flags to use for allocating pages. 431 * @i_mmap_writable: Number of VM_SHARED mappings. 432 * @nr_thps: Number of THPs in the pagecache (non-shmem only). 433 * @i_mmap: Tree of private and shared mappings. 434 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable. 435 * @nrpages: Number of page entries, protected by the i_pages lock. 436 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock. 437 * @writeback_index: Writeback starts here. 438 * @a_ops: Methods. 439 * @flags: Error bits and flags (AS_*). 440 * @wb_err: The most recent error which has occurred. 441 * @private_lock: For use by the owner of the address_space. 442 * @private_list: For use by the owner of the address_space. 443 * @private_data: For use by the owner of the address_space. 444 */ 445 struct address_space { 446 struct inode *host; 447 struct xarray i_pages; 448 gfp_t gfp_mask; 449 atomic_t i_mmap_writable; 450 #ifdef CONFIG_READ_ONLY_THP_FOR_FS 451 /* number of thp, only for non-shmem files */ 452 atomic_t nr_thps; 453 #endif 454 struct rb_root_cached i_mmap; 455 struct rw_semaphore i_mmap_rwsem; 456 unsigned long nrpages; 457 unsigned long nrexceptional; 458 pgoff_t writeback_index; 459 const struct address_space_operations *a_ops; 460 unsigned long flags; 461 errseq_t wb_err; 462 spinlock_t private_lock; 463 struct list_head private_list; 464 void *private_data; 465 } __attribute__((aligned(sizeof(long)))) __randomize_layout; 466 /* 467 * On most architectures that alignment is already the case; but 468 * must be enforced here for CRIS, to let the least significant bit 469 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. 470 */ 471 struct request_queue; 472 473 struct block_device { 474 dev_t bd_dev; /* not a kdev_t - it's a search key */ 475 int bd_openers; 476 struct inode * bd_inode; /* will die */ 477 struct super_block * bd_super; 478 struct mutex bd_mutex; /* open/close mutex */ 479 void * bd_claiming; 480 void * bd_holder; 481 int bd_holders; 482 bool bd_write_holder; 483 #ifdef CONFIG_SYSFS 484 struct list_head bd_holder_disks; 485 #endif 486 struct block_device * bd_contains; 487 unsigned bd_block_size; 488 u8 bd_partno; 489 struct hd_struct * bd_part; 490 /* number of times partitions within this device have been opened. */ 491 unsigned bd_part_count; 492 int bd_invalidated; 493 struct gendisk * bd_disk; 494 struct request_queue * bd_queue; 495 struct backing_dev_info *bd_bdi; 496 struct list_head bd_list; 497 /* 498 * Private data. You must have bd_claim'ed the block_device 499 * to use this. NOTE: bd_claim allows an owner to claim 500 * the same device multiple times, the owner must take special 501 * care to not mess up bd_private for that case. 502 */ 503 unsigned long bd_private; 504 505 /* The counter of freeze processes */ 506 int bd_fsfreeze_count; 507 /* Mutex for freeze */ 508 struct mutex bd_fsfreeze_mutex; 509 } __randomize_layout; 510 511 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */ 512 #define PAGECACHE_TAG_DIRTY XA_MARK_0 513 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1 514 #define PAGECACHE_TAG_TOWRITE XA_MARK_2 515 516 /* 517 * Returns true if any of the pages in the mapping are marked with the tag. 518 */ 519 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag) 520 { 521 return xa_marked(&mapping->i_pages, tag); 522 } 523 524 static inline void i_mmap_lock_write(struct address_space *mapping) 525 { 526 down_write(&mapping->i_mmap_rwsem); 527 } 528 529 static inline void i_mmap_unlock_write(struct address_space *mapping) 530 { 531 up_write(&mapping->i_mmap_rwsem); 532 } 533 534 static inline void i_mmap_lock_read(struct address_space *mapping) 535 { 536 down_read(&mapping->i_mmap_rwsem); 537 } 538 539 static inline void i_mmap_unlock_read(struct address_space *mapping) 540 { 541 up_read(&mapping->i_mmap_rwsem); 542 } 543 544 /* 545 * Might pages of this file be mapped into userspace? 546 */ 547 static inline int mapping_mapped(struct address_space *mapping) 548 { 549 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root); 550 } 551 552 /* 553 * Might pages of this file have been modified in userspace? 554 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff 555 * marks vma as VM_SHARED if it is shared, and the file was opened for 556 * writing i.e. vma may be mprotected writable even if now readonly. 557 * 558 * If i_mmap_writable is negative, no new writable mappings are allowed. You 559 * can only deny writable mappings, if none exists right now. 560 */ 561 static inline int mapping_writably_mapped(struct address_space *mapping) 562 { 563 return atomic_read(&mapping->i_mmap_writable) > 0; 564 } 565 566 static inline int mapping_map_writable(struct address_space *mapping) 567 { 568 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? 569 0 : -EPERM; 570 } 571 572 static inline void mapping_unmap_writable(struct address_space *mapping) 573 { 574 atomic_dec(&mapping->i_mmap_writable); 575 } 576 577 static inline int mapping_deny_writable(struct address_space *mapping) 578 { 579 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? 580 0 : -EBUSY; 581 } 582 583 static inline void mapping_allow_writable(struct address_space *mapping) 584 { 585 atomic_inc(&mapping->i_mmap_writable); 586 } 587 588 /* 589 * Use sequence counter to get consistent i_size on 32-bit processors. 590 */ 591 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 592 #include <linux/seqlock.h> 593 #define __NEED_I_SIZE_ORDERED 594 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) 595 #else 596 #define i_size_ordered_init(inode) do { } while (0) 597 #endif 598 599 struct posix_acl; 600 #define ACL_NOT_CACHED ((void *)(-1)) 601 #define ACL_DONT_CACHE ((void *)(-3)) 602 603 static inline struct posix_acl * 604 uncached_acl_sentinel(struct task_struct *task) 605 { 606 return (void *)task + 1; 607 } 608 609 static inline bool 610 is_uncached_acl(struct posix_acl *acl) 611 { 612 return (long)acl & 1; 613 } 614 615 #define IOP_FASTPERM 0x0001 616 #define IOP_LOOKUP 0x0002 617 #define IOP_NOFOLLOW 0x0004 618 #define IOP_XATTR 0x0008 619 #define IOP_DEFAULT_READLINK 0x0010 620 621 struct fsnotify_mark_connector; 622 623 /* 624 * Keep mostly read-only and often accessed (especially for 625 * the RCU path lookup and 'stat' data) fields at the beginning 626 * of the 'struct inode' 627 */ 628 struct inode { 629 umode_t i_mode; 630 unsigned short i_opflags; 631 kuid_t i_uid; 632 kgid_t i_gid; 633 unsigned int i_flags; 634 635 #ifdef CONFIG_FS_POSIX_ACL 636 struct posix_acl *i_acl; 637 struct posix_acl *i_default_acl; 638 #endif 639 640 const struct inode_operations *i_op; 641 struct super_block *i_sb; 642 struct address_space *i_mapping; 643 644 #ifdef CONFIG_SECURITY 645 void *i_security; 646 #endif 647 648 /* Stat data, not accessed from path walking */ 649 unsigned long i_ino; 650 /* 651 * Filesystems may only read i_nlink directly. They shall use the 652 * following functions for modification: 653 * 654 * (set|clear|inc|drop)_nlink 655 * inode_(inc|dec)_link_count 656 */ 657 union { 658 const unsigned int i_nlink; 659 unsigned int __i_nlink; 660 }; 661 dev_t i_rdev; 662 loff_t i_size; 663 struct timespec64 i_atime; 664 struct timespec64 i_mtime; 665 struct timespec64 i_ctime; 666 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ 667 unsigned short i_bytes; 668 u8 i_blkbits; 669 u8 i_write_hint; 670 blkcnt_t i_blocks; 671 672 #ifdef __NEED_I_SIZE_ORDERED 673 seqcount_t i_size_seqcount; 674 #endif 675 676 /* Misc */ 677 unsigned long i_state; 678 struct rw_semaphore i_rwsem; 679 680 unsigned long dirtied_when; /* jiffies of first dirtying */ 681 unsigned long dirtied_time_when; 682 683 struct hlist_node i_hash; 684 struct list_head i_io_list; /* backing dev IO list */ 685 #ifdef CONFIG_CGROUP_WRITEBACK 686 struct bdi_writeback *i_wb; /* the associated cgroup wb */ 687 688 /* foreign inode detection, see wbc_detach_inode() */ 689 int i_wb_frn_winner; 690 u16 i_wb_frn_avg_time; 691 u16 i_wb_frn_history; 692 #endif 693 struct list_head i_lru; /* inode LRU list */ 694 struct list_head i_sb_list; 695 struct list_head i_wb_list; /* backing dev writeback list */ 696 union { 697 struct hlist_head i_dentry; 698 struct rcu_head i_rcu; 699 }; 700 atomic64_t i_version; 701 atomic_t i_count; 702 atomic_t i_dio_count; 703 atomic_t i_writecount; 704 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) 705 atomic_t i_readcount; /* struct files open RO */ 706 #endif 707 union { 708 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ 709 void (*free_inode)(struct inode *); 710 }; 711 struct file_lock_context *i_flctx; 712 struct address_space i_data; 713 struct list_head i_devices; 714 union { 715 struct pipe_inode_info *i_pipe; 716 struct block_device *i_bdev; 717 struct cdev *i_cdev; 718 char *i_link; 719 unsigned i_dir_seq; 720 }; 721 722 __u32 i_generation; 723 724 #ifdef CONFIG_FSNOTIFY 725 __u32 i_fsnotify_mask; /* all events this inode cares about */ 726 struct fsnotify_mark_connector __rcu *i_fsnotify_marks; 727 #endif 728 729 #ifdef CONFIG_FS_ENCRYPTION 730 struct fscrypt_info *i_crypt_info; 731 #endif 732 733 #ifdef CONFIG_FS_VERITY 734 struct fsverity_info *i_verity_info; 735 #endif 736 737 void *i_private; /* fs or device private pointer */ 738 } __randomize_layout; 739 740 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode); 741 742 static inline unsigned int i_blocksize(const struct inode *node) 743 { 744 return (1 << node->i_blkbits); 745 } 746 747 static inline int inode_unhashed(struct inode *inode) 748 { 749 return hlist_unhashed(&inode->i_hash); 750 } 751 752 /* 753 * __mark_inode_dirty expects inodes to be hashed. Since we don't 754 * want special inodes in the fileset inode space, we make them 755 * appear hashed, but do not put on any lists. hlist_del() 756 * will work fine and require no locking. 757 */ 758 static inline void inode_fake_hash(struct inode *inode) 759 { 760 hlist_add_fake(&inode->i_hash); 761 } 762 763 /* 764 * inode->i_mutex nesting subclasses for the lock validator: 765 * 766 * 0: the object of the current VFS operation 767 * 1: parent 768 * 2: child/target 769 * 3: xattr 770 * 4: second non-directory 771 * 5: second parent (when locking independent directories in rename) 772 * 773 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two 774 * non-directories at once. 775 * 776 * The locking order between these classes is 777 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory 778 */ 779 enum inode_i_mutex_lock_class 780 { 781 I_MUTEX_NORMAL, 782 I_MUTEX_PARENT, 783 I_MUTEX_CHILD, 784 I_MUTEX_XATTR, 785 I_MUTEX_NONDIR2, 786 I_MUTEX_PARENT2, 787 }; 788 789 static inline void inode_lock(struct inode *inode) 790 { 791 down_write(&inode->i_rwsem); 792 } 793 794 static inline void inode_unlock(struct inode *inode) 795 { 796 up_write(&inode->i_rwsem); 797 } 798 799 static inline void inode_lock_shared(struct inode *inode) 800 { 801 down_read(&inode->i_rwsem); 802 } 803 804 static inline void inode_unlock_shared(struct inode *inode) 805 { 806 up_read(&inode->i_rwsem); 807 } 808 809 static inline int inode_trylock(struct inode *inode) 810 { 811 return down_write_trylock(&inode->i_rwsem); 812 } 813 814 static inline int inode_trylock_shared(struct inode *inode) 815 { 816 return down_read_trylock(&inode->i_rwsem); 817 } 818 819 static inline int inode_is_locked(struct inode *inode) 820 { 821 return rwsem_is_locked(&inode->i_rwsem); 822 } 823 824 static inline void inode_lock_nested(struct inode *inode, unsigned subclass) 825 { 826 down_write_nested(&inode->i_rwsem, subclass); 827 } 828 829 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass) 830 { 831 down_read_nested(&inode->i_rwsem, subclass); 832 } 833 834 void lock_two_nondirectories(struct inode *, struct inode*); 835 void unlock_two_nondirectories(struct inode *, struct inode*); 836 837 /* 838 * NOTE: in a 32bit arch with a preemptable kernel and 839 * an UP compile the i_size_read/write must be atomic 840 * with respect to the local cpu (unlike with preempt disabled), 841 * but they don't need to be atomic with respect to other cpus like in 842 * true SMP (so they need either to either locally disable irq around 843 * the read or for example on x86 they can be still implemented as a 844 * cmpxchg8b without the need of the lock prefix). For SMP compiles 845 * and 64bit archs it makes no difference if preempt is enabled or not. 846 */ 847 static inline loff_t i_size_read(const struct inode *inode) 848 { 849 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 850 loff_t i_size; 851 unsigned int seq; 852 853 do { 854 seq = read_seqcount_begin(&inode->i_size_seqcount); 855 i_size = inode->i_size; 856 } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); 857 return i_size; 858 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 859 loff_t i_size; 860 861 preempt_disable(); 862 i_size = inode->i_size; 863 preempt_enable(); 864 return i_size; 865 #else 866 return inode->i_size; 867 #endif 868 } 869 870 /* 871 * NOTE: unlike i_size_read(), i_size_write() does need locking around it 872 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount 873 * can be lost, resulting in subsequent i_size_read() calls spinning forever. 874 */ 875 static inline void i_size_write(struct inode *inode, loff_t i_size) 876 { 877 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 878 preempt_disable(); 879 write_seqcount_begin(&inode->i_size_seqcount); 880 inode->i_size = i_size; 881 write_seqcount_end(&inode->i_size_seqcount); 882 preempt_enable(); 883 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 884 preempt_disable(); 885 inode->i_size = i_size; 886 preempt_enable(); 887 #else 888 inode->i_size = i_size; 889 #endif 890 } 891 892 static inline unsigned iminor(const struct inode *inode) 893 { 894 return MINOR(inode->i_rdev); 895 } 896 897 static inline unsigned imajor(const struct inode *inode) 898 { 899 return MAJOR(inode->i_rdev); 900 } 901 902 extern struct block_device *I_BDEV(struct inode *inode); 903 904 struct fown_struct { 905 rwlock_t lock; /* protects pid, uid, euid fields */ 906 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ 907 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ 908 kuid_t uid, euid; /* uid/euid of process setting the owner */ 909 int signum; /* posix.1b rt signal to be delivered on IO */ 910 }; 911 912 /* 913 * Track a single file's readahead state 914 */ 915 struct file_ra_state { 916 pgoff_t start; /* where readahead started */ 917 unsigned int size; /* # of readahead pages */ 918 unsigned int async_size; /* do asynchronous readahead when 919 there are only # of pages ahead */ 920 921 unsigned int ra_pages; /* Maximum readahead window */ 922 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */ 923 loff_t prev_pos; /* Cache last read() position */ 924 }; 925 926 /* 927 * Check if @index falls in the readahead windows. 928 */ 929 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) 930 { 931 return (index >= ra->start && 932 index < ra->start + ra->size); 933 } 934 935 struct file { 936 union { 937 struct llist_node fu_llist; 938 struct rcu_head fu_rcuhead; 939 } f_u; 940 struct path f_path; 941 struct inode *f_inode; /* cached value */ 942 const struct file_operations *f_op; 943 944 /* 945 * Protects f_ep_links, f_flags. 946 * Must not be taken from IRQ context. 947 */ 948 spinlock_t f_lock; 949 enum rw_hint f_write_hint; 950 atomic_long_t f_count; 951 unsigned int f_flags; 952 fmode_t f_mode; 953 struct mutex f_pos_lock; 954 loff_t f_pos; 955 struct fown_struct f_owner; 956 const struct cred *f_cred; 957 struct file_ra_state f_ra; 958 959 u64 f_version; 960 #ifdef CONFIG_SECURITY 961 void *f_security; 962 #endif 963 /* needed for tty driver, and maybe others */ 964 void *private_data; 965 966 #ifdef CONFIG_EPOLL 967 /* Used by fs/eventpoll.c to link all the hooks to this file */ 968 struct list_head f_ep_links; 969 struct list_head f_tfile_llink; 970 #endif /* #ifdef CONFIG_EPOLL */ 971 struct address_space *f_mapping; 972 errseq_t f_wb_err; 973 } __randomize_layout 974 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ 975 976 struct file_handle { 977 __u32 handle_bytes; 978 int handle_type; 979 /* file identifier */ 980 unsigned char f_handle[0]; 981 }; 982 983 static inline struct file *get_file(struct file *f) 984 { 985 atomic_long_inc(&f->f_count); 986 return f; 987 } 988 #define get_file_rcu_many(x, cnt) \ 989 atomic_long_add_unless(&(x)->f_count, (cnt), 0) 990 #define get_file_rcu(x) get_file_rcu_many((x), 1) 991 #define file_count(x) atomic_long_read(&(x)->f_count) 992 993 #define MAX_NON_LFS ((1UL<<31) - 1) 994 995 /* Page cache limit. The filesystems should put that into their s_maxbytes 996 limits, otherwise bad things can happen in VM. */ 997 #if BITS_PER_LONG==32 998 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT) 999 #elif BITS_PER_LONG==64 1000 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX) 1001 #endif 1002 1003 #define FL_POSIX 1 1004 #define FL_FLOCK 2 1005 #define FL_DELEG 4 /* NFSv4 delegation */ 1006 #define FL_ACCESS 8 /* not trying to lock, just looking */ 1007 #define FL_EXISTS 16 /* when unlocking, test for existence */ 1008 #define FL_LEASE 32 /* lease held on this file */ 1009 #define FL_CLOSE 64 /* unlock on close */ 1010 #define FL_SLEEP 128 /* A blocking lock */ 1011 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */ 1012 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */ 1013 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */ 1014 #define FL_LAYOUT 2048 /* outstanding pNFS layout */ 1015 1016 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE) 1017 1018 /* 1019 * Special return value from posix_lock_file() and vfs_lock_file() for 1020 * asynchronous locking. 1021 */ 1022 #define FILE_LOCK_DEFERRED 1 1023 1024 /* legacy typedef, should eventually be removed */ 1025 typedef void *fl_owner_t; 1026 1027 struct file_lock; 1028 1029 struct file_lock_operations { 1030 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 1031 void (*fl_release_private)(struct file_lock *); 1032 }; 1033 1034 struct lock_manager_operations { 1035 fl_owner_t (*lm_get_owner)(fl_owner_t); 1036 void (*lm_put_owner)(fl_owner_t); 1037 void (*lm_notify)(struct file_lock *); /* unblock callback */ 1038 int (*lm_grant)(struct file_lock *, int); 1039 bool (*lm_break)(struct file_lock *); 1040 int (*lm_change)(struct file_lock *, int, struct list_head *); 1041 void (*lm_setup)(struct file_lock *, void **); 1042 }; 1043 1044 struct lock_manager { 1045 struct list_head list; 1046 /* 1047 * NFSv4 and up also want opens blocked during the grace period; 1048 * NLM doesn't care: 1049 */ 1050 bool block_opens; 1051 }; 1052 1053 struct net; 1054 void locks_start_grace(struct net *, struct lock_manager *); 1055 void locks_end_grace(struct lock_manager *); 1056 bool locks_in_grace(struct net *); 1057 bool opens_in_grace(struct net *); 1058 1059 /* that will die - we need it for nfs_lock_info */ 1060 #include <linux/nfs_fs_i.h> 1061 1062 /* 1063 * struct file_lock represents a generic "file lock". It's used to represent 1064 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to 1065 * note that the same struct is used to represent both a request for a lock and 1066 * the lock itself, but the same object is never used for both. 1067 * 1068 * FIXME: should we create a separate "struct lock_request" to help distinguish 1069 * these two uses? 1070 * 1071 * The varous i_flctx lists are ordered by: 1072 * 1073 * 1) lock owner 1074 * 2) lock range start 1075 * 3) lock range end 1076 * 1077 * Obviously, the last two criteria only matter for POSIX locks. 1078 */ 1079 struct file_lock { 1080 struct file_lock *fl_blocker; /* The lock, that is blocking us */ 1081 struct list_head fl_list; /* link into file_lock_context */ 1082 struct hlist_node fl_link; /* node in global lists */ 1083 struct list_head fl_blocked_requests; /* list of requests with 1084 * ->fl_blocker pointing here 1085 */ 1086 struct list_head fl_blocked_member; /* node in 1087 * ->fl_blocker->fl_blocked_requests 1088 */ 1089 fl_owner_t fl_owner; 1090 unsigned int fl_flags; 1091 unsigned char fl_type; 1092 unsigned int fl_pid; 1093 int fl_link_cpu; /* what cpu's list is this on? */ 1094 wait_queue_head_t fl_wait; 1095 struct file *fl_file; 1096 loff_t fl_start; 1097 loff_t fl_end; 1098 1099 struct fasync_struct * fl_fasync; /* for lease break notifications */ 1100 /* for lease breaks: */ 1101 unsigned long fl_break_time; 1102 unsigned long fl_downgrade_time; 1103 1104 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ 1105 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ 1106 union { 1107 struct nfs_lock_info nfs_fl; 1108 struct nfs4_lock_info nfs4_fl; 1109 struct { 1110 struct list_head link; /* link in AFS vnode's pending_locks list */ 1111 int state; /* state of grant or error if -ve */ 1112 unsigned int debug_id; 1113 } afs; 1114 } fl_u; 1115 } __randomize_layout; 1116 1117 struct file_lock_context { 1118 spinlock_t flc_lock; 1119 struct list_head flc_flock; 1120 struct list_head flc_posix; 1121 struct list_head flc_lease; 1122 }; 1123 1124 /* The following constant reflects the upper bound of the file/locking space */ 1125 #ifndef OFFSET_MAX 1126 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) 1127 #define OFFSET_MAX INT_LIMIT(loff_t) 1128 #define OFFT_OFFSET_MAX INT_LIMIT(off_t) 1129 #endif 1130 1131 extern void send_sigio(struct fown_struct *fown, int fd, int band); 1132 1133 #define locks_inode(f) file_inode(f) 1134 1135 #ifdef CONFIG_FILE_LOCKING 1136 extern int fcntl_getlk(struct file *, unsigned int, struct flock *); 1137 extern int fcntl_setlk(unsigned int, struct file *, unsigned int, 1138 struct flock *); 1139 1140 #if BITS_PER_LONG == 32 1141 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *); 1142 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, 1143 struct flock64 *); 1144 #endif 1145 1146 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); 1147 extern int fcntl_getlease(struct file *filp); 1148 1149 /* fs/locks.c */ 1150 void locks_free_lock_context(struct inode *inode); 1151 void locks_free_lock(struct file_lock *fl); 1152 extern void locks_init_lock(struct file_lock *); 1153 extern struct file_lock * locks_alloc_lock(void); 1154 extern void locks_copy_lock(struct file_lock *, struct file_lock *); 1155 extern void locks_copy_conflock(struct file_lock *, struct file_lock *); 1156 extern void locks_remove_posix(struct file *, fl_owner_t); 1157 extern void locks_remove_file(struct file *); 1158 extern void locks_release_private(struct file_lock *); 1159 extern void posix_test_lock(struct file *, struct file_lock *); 1160 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); 1161 extern int locks_delete_block(struct file_lock *); 1162 extern int vfs_test_lock(struct file *, struct file_lock *); 1163 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); 1164 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); 1165 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl); 1166 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type); 1167 extern void lease_get_mtime(struct inode *, struct timespec64 *time); 1168 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv); 1169 extern int vfs_setlease(struct file *, long, struct file_lock **, void **); 1170 extern int lease_modify(struct file_lock *, int, struct list_head *); 1171 1172 struct notifier_block; 1173 extern int lease_register_notifier(struct notifier_block *); 1174 extern void lease_unregister_notifier(struct notifier_block *); 1175 1176 struct files_struct; 1177 extern void show_fd_locks(struct seq_file *f, 1178 struct file *filp, struct files_struct *files); 1179 #else /* !CONFIG_FILE_LOCKING */ 1180 static inline int fcntl_getlk(struct file *file, unsigned int cmd, 1181 struct flock __user *user) 1182 { 1183 return -EINVAL; 1184 } 1185 1186 static inline int fcntl_setlk(unsigned int fd, struct file *file, 1187 unsigned int cmd, struct flock __user *user) 1188 { 1189 return -EACCES; 1190 } 1191 1192 #if BITS_PER_LONG == 32 1193 static inline int fcntl_getlk64(struct file *file, unsigned int cmd, 1194 struct flock64 __user *user) 1195 { 1196 return -EINVAL; 1197 } 1198 1199 static inline int fcntl_setlk64(unsigned int fd, struct file *file, 1200 unsigned int cmd, struct flock64 __user *user) 1201 { 1202 return -EACCES; 1203 } 1204 #endif 1205 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) 1206 { 1207 return -EINVAL; 1208 } 1209 1210 static inline int fcntl_getlease(struct file *filp) 1211 { 1212 return F_UNLCK; 1213 } 1214 1215 static inline void 1216 locks_free_lock_context(struct inode *inode) 1217 { 1218 } 1219 1220 static inline void locks_init_lock(struct file_lock *fl) 1221 { 1222 return; 1223 } 1224 1225 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl) 1226 { 1227 return; 1228 } 1229 1230 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) 1231 { 1232 return; 1233 } 1234 1235 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) 1236 { 1237 return; 1238 } 1239 1240 static inline void locks_remove_file(struct file *filp) 1241 { 1242 return; 1243 } 1244 1245 static inline void posix_test_lock(struct file *filp, struct file_lock *fl) 1246 { 1247 return; 1248 } 1249 1250 static inline int posix_lock_file(struct file *filp, struct file_lock *fl, 1251 struct file_lock *conflock) 1252 { 1253 return -ENOLCK; 1254 } 1255 1256 static inline int locks_delete_block(struct file_lock *waiter) 1257 { 1258 return -ENOENT; 1259 } 1260 1261 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) 1262 { 1263 return 0; 1264 } 1265 1266 static inline int vfs_lock_file(struct file *filp, unsigned int cmd, 1267 struct file_lock *fl, struct file_lock *conf) 1268 { 1269 return -ENOLCK; 1270 } 1271 1272 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) 1273 { 1274 return 0; 1275 } 1276 1277 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl) 1278 { 1279 return -ENOLCK; 1280 } 1281 1282 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type) 1283 { 1284 return 0; 1285 } 1286 1287 static inline void lease_get_mtime(struct inode *inode, 1288 struct timespec64 *time) 1289 { 1290 return; 1291 } 1292 1293 static inline int generic_setlease(struct file *filp, long arg, 1294 struct file_lock **flp, void **priv) 1295 { 1296 return -EINVAL; 1297 } 1298 1299 static inline int vfs_setlease(struct file *filp, long arg, 1300 struct file_lock **lease, void **priv) 1301 { 1302 return -EINVAL; 1303 } 1304 1305 static inline int lease_modify(struct file_lock *fl, int arg, 1306 struct list_head *dispose) 1307 { 1308 return -EINVAL; 1309 } 1310 1311 struct files_struct; 1312 static inline void show_fd_locks(struct seq_file *f, 1313 struct file *filp, struct files_struct *files) {} 1314 #endif /* !CONFIG_FILE_LOCKING */ 1315 1316 static inline struct inode *file_inode(const struct file *f) 1317 { 1318 return f->f_inode; 1319 } 1320 1321 static inline struct dentry *file_dentry(const struct file *file) 1322 { 1323 return d_real(file->f_path.dentry, file_inode(file)); 1324 } 1325 1326 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl) 1327 { 1328 return locks_lock_inode_wait(locks_inode(filp), fl); 1329 } 1330 1331 struct fasync_struct { 1332 rwlock_t fa_lock; 1333 int magic; 1334 int fa_fd; 1335 struct fasync_struct *fa_next; /* singly linked list */ 1336 struct file *fa_file; 1337 struct rcu_head fa_rcu; 1338 }; 1339 1340 #define FASYNC_MAGIC 0x4601 1341 1342 /* SMP safe fasync helpers: */ 1343 extern int fasync_helper(int, struct file *, int, struct fasync_struct **); 1344 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); 1345 extern int fasync_remove_entry(struct file *, struct fasync_struct **); 1346 extern struct fasync_struct *fasync_alloc(void); 1347 extern void fasync_free(struct fasync_struct *); 1348 1349 /* can be called from interrupts */ 1350 extern void kill_fasync(struct fasync_struct **, int, int); 1351 1352 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); 1353 extern int f_setown(struct file *filp, unsigned long arg, int force); 1354 extern void f_delown(struct file *filp); 1355 extern pid_t f_getown(struct file *filp); 1356 extern int send_sigurg(struct fown_struct *fown); 1357 1358 /* 1359 * sb->s_flags. Note that these mirror the equivalent MS_* flags where 1360 * represented in both. 1361 */ 1362 #define SB_RDONLY 1 /* Mount read-only */ 1363 #define SB_NOSUID 2 /* Ignore suid and sgid bits */ 1364 #define SB_NODEV 4 /* Disallow access to device special files */ 1365 #define SB_NOEXEC 8 /* Disallow program execution */ 1366 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */ 1367 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */ 1368 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */ 1369 #define SB_NOATIME 1024 /* Do not update access times. */ 1370 #define SB_NODIRATIME 2048 /* Do not update directory access times */ 1371 #define SB_SILENT 32768 1372 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */ 1373 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */ 1374 #define SB_I_VERSION (1<<23) /* Update inode I_version field */ 1375 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */ 1376 1377 /* These sb flags are internal to the kernel */ 1378 #define SB_SUBMOUNT (1<<26) 1379 #define SB_FORCE (1<<27) 1380 #define SB_NOSEC (1<<28) 1381 #define SB_BORN (1<<29) 1382 #define SB_ACTIVE (1<<30) 1383 #define SB_NOUSER (1<<31) 1384 1385 /* 1386 * Umount options 1387 */ 1388 1389 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ 1390 #define MNT_DETACH 0x00000002 /* Just detach from the tree */ 1391 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ 1392 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ 1393 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ 1394 1395 /* sb->s_iflags */ 1396 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */ 1397 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */ 1398 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */ 1399 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */ 1400 1401 /* sb->s_iflags to limit user namespace mounts */ 1402 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */ 1403 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020 1404 #define SB_I_UNTRUSTED_MOUNTER 0x00000040 1405 1406 /* Possible states of 'frozen' field */ 1407 enum { 1408 SB_UNFROZEN = 0, /* FS is unfrozen */ 1409 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ 1410 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ 1411 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop 1412 * internal threads if needed) */ 1413 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ 1414 }; 1415 1416 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) 1417 1418 struct sb_writers { 1419 int frozen; /* Is sb frozen? */ 1420 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */ 1421 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS]; 1422 }; 1423 1424 struct super_block { 1425 struct list_head s_list; /* Keep this first */ 1426 dev_t s_dev; /* search index; _not_ kdev_t */ 1427 unsigned char s_blocksize_bits; 1428 unsigned long s_blocksize; 1429 loff_t s_maxbytes; /* Max file size */ 1430 struct file_system_type *s_type; 1431 const struct super_operations *s_op; 1432 const struct dquot_operations *dq_op; 1433 const struct quotactl_ops *s_qcop; 1434 const struct export_operations *s_export_op; 1435 unsigned long s_flags; 1436 unsigned long s_iflags; /* internal SB_I_* flags */ 1437 unsigned long s_magic; 1438 struct dentry *s_root; 1439 struct rw_semaphore s_umount; 1440 int s_count; 1441 atomic_t s_active; 1442 #ifdef CONFIG_SECURITY 1443 void *s_security; 1444 #endif 1445 const struct xattr_handler **s_xattr; 1446 #ifdef CONFIG_FS_ENCRYPTION 1447 const struct fscrypt_operations *s_cop; 1448 struct key *s_master_keys; /* master crypto keys in use */ 1449 #endif 1450 #ifdef CONFIG_FS_VERITY 1451 const struct fsverity_operations *s_vop; 1452 #endif 1453 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */ 1454 struct list_head s_mounts; /* list of mounts; _not_ for fs use */ 1455 struct block_device *s_bdev; 1456 struct backing_dev_info *s_bdi; 1457 struct mtd_info *s_mtd; 1458 struct hlist_node s_instances; 1459 unsigned int s_quota_types; /* Bitmask of supported quota types */ 1460 struct quota_info s_dquot; /* Diskquota specific options */ 1461 1462 struct sb_writers s_writers; 1463 1464 /* 1465 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and 1466 * s_fsnotify_marks together for cache efficiency. They are frequently 1467 * accessed and rarely modified. 1468 */ 1469 void *s_fs_info; /* Filesystem private info */ 1470 1471 /* Granularity of c/m/atime in ns (cannot be worse than a second) */ 1472 u32 s_time_gran; 1473 /* Time limits for c/m/atime in seconds */ 1474 time64_t s_time_min; 1475 time64_t s_time_max; 1476 #ifdef CONFIG_FSNOTIFY 1477 __u32 s_fsnotify_mask; 1478 struct fsnotify_mark_connector __rcu *s_fsnotify_marks; 1479 #endif 1480 1481 char s_id[32]; /* Informational name */ 1482 uuid_t s_uuid; /* UUID */ 1483 1484 unsigned int s_max_links; 1485 fmode_t s_mode; 1486 1487 /* 1488 * The next field is for VFS *only*. No filesystems have any business 1489 * even looking at it. You had been warned. 1490 */ 1491 struct mutex s_vfs_rename_mutex; /* Kludge */ 1492 1493 /* 1494 * Filesystem subtype. If non-empty the filesystem type field 1495 * in /proc/mounts will be "type.subtype" 1496 */ 1497 const char *s_subtype; 1498 1499 const struct dentry_operations *s_d_op; /* default d_op for dentries */ 1500 1501 /* 1502 * Saved pool identifier for cleancache (-1 means none) 1503 */ 1504 int cleancache_poolid; 1505 1506 struct shrinker s_shrink; /* per-sb shrinker handle */ 1507 1508 /* Number of inodes with nlink == 0 but still referenced */ 1509 atomic_long_t s_remove_count; 1510 1511 /* Pending fsnotify inode refs */ 1512 atomic_long_t s_fsnotify_inode_refs; 1513 1514 /* Being remounted read-only */ 1515 int s_readonly_remount; 1516 1517 /* AIO completions deferred from interrupt context */ 1518 struct workqueue_struct *s_dio_done_wq; 1519 struct hlist_head s_pins; 1520 1521 /* 1522 * Owning user namespace and default context in which to 1523 * interpret filesystem uids, gids, quotas, device nodes, 1524 * xattrs and security labels. 1525 */ 1526 struct user_namespace *s_user_ns; 1527 1528 /* 1529 * The list_lru structure is essentially just a pointer to a table 1530 * of per-node lru lists, each of which has its own spinlock. 1531 * There is no need to put them into separate cachelines. 1532 */ 1533 struct list_lru s_dentry_lru; 1534 struct list_lru s_inode_lru; 1535 struct rcu_head rcu; 1536 struct work_struct destroy_work; 1537 1538 struct mutex s_sync_lock; /* sync serialisation lock */ 1539 1540 /* 1541 * Indicates how deep in a filesystem stack this SB is 1542 */ 1543 int s_stack_depth; 1544 1545 /* s_inode_list_lock protects s_inodes */ 1546 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp; 1547 struct list_head s_inodes; /* all inodes */ 1548 1549 spinlock_t s_inode_wblist_lock; 1550 struct list_head s_inodes_wb; /* writeback inodes */ 1551 } __randomize_layout; 1552 1553 /* Helper functions so that in most cases filesystems will 1554 * not need to deal directly with kuid_t and kgid_t and can 1555 * instead deal with the raw numeric values that are stored 1556 * in the filesystem. 1557 */ 1558 static inline uid_t i_uid_read(const struct inode *inode) 1559 { 1560 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid); 1561 } 1562 1563 static inline gid_t i_gid_read(const struct inode *inode) 1564 { 1565 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid); 1566 } 1567 1568 static inline void i_uid_write(struct inode *inode, uid_t uid) 1569 { 1570 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid); 1571 } 1572 1573 static inline void i_gid_write(struct inode *inode, gid_t gid) 1574 { 1575 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid); 1576 } 1577 1578 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran); 1579 extern struct timespec64 current_time(struct inode *inode); 1580 1581 /* 1582 * Snapshotting support. 1583 */ 1584 1585 void __sb_end_write(struct super_block *sb, int level); 1586 int __sb_start_write(struct super_block *sb, int level, bool wait); 1587 1588 #define __sb_writers_acquired(sb, lev) \ 1589 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) 1590 #define __sb_writers_release(sb, lev) \ 1591 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) 1592 1593 /** 1594 * sb_end_write - drop write access to a superblock 1595 * @sb: the super we wrote to 1596 * 1597 * Decrement number of writers to the filesystem. Wake up possible waiters 1598 * wanting to freeze the filesystem. 1599 */ 1600 static inline void sb_end_write(struct super_block *sb) 1601 { 1602 __sb_end_write(sb, SB_FREEZE_WRITE); 1603 } 1604 1605 /** 1606 * sb_end_pagefault - drop write access to a superblock from a page fault 1607 * @sb: the super we wrote to 1608 * 1609 * Decrement number of processes handling write page fault to the filesystem. 1610 * Wake up possible waiters wanting to freeze the filesystem. 1611 */ 1612 static inline void sb_end_pagefault(struct super_block *sb) 1613 { 1614 __sb_end_write(sb, SB_FREEZE_PAGEFAULT); 1615 } 1616 1617 /** 1618 * sb_end_intwrite - drop write access to a superblock for internal fs purposes 1619 * @sb: the super we wrote to 1620 * 1621 * Decrement fs-internal number of writers to the filesystem. Wake up possible 1622 * waiters wanting to freeze the filesystem. 1623 */ 1624 static inline void sb_end_intwrite(struct super_block *sb) 1625 { 1626 __sb_end_write(sb, SB_FREEZE_FS); 1627 } 1628 1629 /** 1630 * sb_start_write - get write access to a superblock 1631 * @sb: the super we write to 1632 * 1633 * When a process wants to write data or metadata to a file system (i.e. dirty 1634 * a page or an inode), it should embed the operation in a sb_start_write() - 1635 * sb_end_write() pair to get exclusion against file system freezing. This 1636 * function increments number of writers preventing freezing. If the file 1637 * system is already frozen, the function waits until the file system is 1638 * thawed. 1639 * 1640 * Since freeze protection behaves as a lock, users have to preserve 1641 * ordering of freeze protection and other filesystem locks. Generally, 1642 * freeze protection should be the outermost lock. In particular, we have: 1643 * 1644 * sb_start_write 1645 * -> i_mutex (write path, truncate, directory ops, ...) 1646 * -> s_umount (freeze_super, thaw_super) 1647 */ 1648 static inline void sb_start_write(struct super_block *sb) 1649 { 1650 __sb_start_write(sb, SB_FREEZE_WRITE, true); 1651 } 1652 1653 static inline int sb_start_write_trylock(struct super_block *sb) 1654 { 1655 return __sb_start_write(sb, SB_FREEZE_WRITE, false); 1656 } 1657 1658 /** 1659 * sb_start_pagefault - get write access to a superblock from a page fault 1660 * @sb: the super we write to 1661 * 1662 * When a process starts handling write page fault, it should embed the 1663 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get 1664 * exclusion against file system freezing. This is needed since the page fault 1665 * is going to dirty a page. This function increments number of running page 1666 * faults preventing freezing. If the file system is already frozen, the 1667 * function waits until the file system is thawed. 1668 * 1669 * Since page fault freeze protection behaves as a lock, users have to preserve 1670 * ordering of freeze protection and other filesystem locks. It is advised to 1671 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault 1672 * handling code implies lock dependency: 1673 * 1674 * mmap_sem 1675 * -> sb_start_pagefault 1676 */ 1677 static inline void sb_start_pagefault(struct super_block *sb) 1678 { 1679 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true); 1680 } 1681 1682 /* 1683 * sb_start_intwrite - get write access to a superblock for internal fs purposes 1684 * @sb: the super we write to 1685 * 1686 * This is the third level of protection against filesystem freezing. It is 1687 * free for use by a filesystem. The only requirement is that it must rank 1688 * below sb_start_pagefault. 1689 * 1690 * For example filesystem can call sb_start_intwrite() when starting a 1691 * transaction which somewhat eases handling of freezing for internal sources 1692 * of filesystem changes (internal fs threads, discarding preallocation on file 1693 * close, etc.). 1694 */ 1695 static inline void sb_start_intwrite(struct super_block *sb) 1696 { 1697 __sb_start_write(sb, SB_FREEZE_FS, true); 1698 } 1699 1700 static inline int sb_start_intwrite_trylock(struct super_block *sb) 1701 { 1702 return __sb_start_write(sb, SB_FREEZE_FS, false); 1703 } 1704 1705 1706 extern bool inode_owner_or_capable(const struct inode *inode); 1707 1708 /* 1709 * VFS helper functions.. 1710 */ 1711 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool); 1712 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t); 1713 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t); 1714 extern int vfs_symlink(struct inode *, struct dentry *, const char *); 1715 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **); 1716 extern int vfs_rmdir(struct inode *, struct dentry *); 1717 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **); 1718 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int); 1719 extern int vfs_whiteout(struct inode *, struct dentry *); 1720 1721 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode, 1722 int open_flag); 1723 1724 int vfs_mkobj(struct dentry *, umode_t, 1725 int (*f)(struct dentry *, umode_t, void *), 1726 void *); 1727 1728 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 1729 1730 #ifdef CONFIG_COMPAT 1731 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd, 1732 unsigned long arg); 1733 #else 1734 #define compat_ptr_ioctl NULL 1735 #endif 1736 1737 /* 1738 * VFS file helper functions. 1739 */ 1740 extern void inode_init_owner(struct inode *inode, const struct inode *dir, 1741 umode_t mode); 1742 extern bool may_open_dev(const struct path *path); 1743 /* 1744 * VFS FS_IOC_FIEMAP helper definitions. 1745 */ 1746 struct fiemap_extent_info { 1747 unsigned int fi_flags; /* Flags as passed from user */ 1748 unsigned int fi_extents_mapped; /* Number of mapped extents */ 1749 unsigned int fi_extents_max; /* Size of fiemap_extent array */ 1750 struct fiemap_extent __user *fi_extents_start; /* Start of 1751 fiemap_extent array */ 1752 }; 1753 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical, 1754 u64 phys, u64 len, u32 flags); 1755 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags); 1756 1757 /* 1758 * This is the "filldir" function type, used by readdir() to let 1759 * the kernel specify what kind of dirent layout it wants to have. 1760 * This allows the kernel to read directories into kernel space or 1761 * to have different dirent layouts depending on the binary type. 1762 */ 1763 struct dir_context; 1764 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, 1765 unsigned); 1766 1767 struct dir_context { 1768 filldir_t actor; 1769 loff_t pos; 1770 }; 1771 1772 struct block_device_operations; 1773 1774 /* These macros are for out of kernel modules to test that 1775 * the kernel supports the unlocked_ioctl and compat_ioctl 1776 * fields in struct file_operations. */ 1777 #define HAVE_COMPAT_IOCTL 1 1778 #define HAVE_UNLOCKED_IOCTL 1 1779 1780 /* 1781 * These flags let !MMU mmap() govern direct device mapping vs immediate 1782 * copying more easily for MAP_PRIVATE, especially for ROM filesystems. 1783 * 1784 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) 1785 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) 1786 * NOMMU_MAP_READ: Can be mapped for reading 1787 * NOMMU_MAP_WRITE: Can be mapped for writing 1788 * NOMMU_MAP_EXEC: Can be mapped for execution 1789 */ 1790 #define NOMMU_MAP_COPY 0x00000001 1791 #define NOMMU_MAP_DIRECT 0x00000008 1792 #define NOMMU_MAP_READ VM_MAYREAD 1793 #define NOMMU_MAP_WRITE VM_MAYWRITE 1794 #define NOMMU_MAP_EXEC VM_MAYEXEC 1795 1796 #define NOMMU_VMFLAGS \ 1797 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) 1798 1799 /* 1800 * These flags control the behavior of the remap_file_range function pointer. 1801 * If it is called with len == 0 that means "remap to end of source file". 1802 * See Documentation/filesystems/vfs.rst for more details about this call. 1803 * 1804 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate) 1805 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request 1806 */ 1807 #define REMAP_FILE_DEDUP (1 << 0) 1808 #define REMAP_FILE_CAN_SHORTEN (1 << 1) 1809 1810 /* 1811 * These flags signal that the caller is ok with altering various aspects of 1812 * the behavior of the remap operation. The changes must be made by the 1813 * implementation; the vfs remap helper functions can take advantage of them. 1814 * Flags in this category exist to preserve the quirky behavior of the hoisted 1815 * btrfs clone/dedupe ioctls. 1816 */ 1817 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN) 1818 1819 struct iov_iter; 1820 1821 struct file_operations { 1822 struct module *owner; 1823 loff_t (*llseek) (struct file *, loff_t, int); 1824 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 1825 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 1826 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); 1827 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); 1828 int (*iopoll)(struct kiocb *kiocb, bool spin); 1829 int (*iterate) (struct file *, struct dir_context *); 1830 int (*iterate_shared) (struct file *, struct dir_context *); 1831 __poll_t (*poll) (struct file *, struct poll_table_struct *); 1832 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 1833 long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 1834 int (*mmap) (struct file *, struct vm_area_struct *); 1835 unsigned long mmap_supported_flags; 1836 int (*open) (struct inode *, struct file *); 1837 int (*flush) (struct file *, fl_owner_t id); 1838 int (*release) (struct inode *, struct file *); 1839 int (*fsync) (struct file *, loff_t, loff_t, int datasync); 1840 int (*fasync) (int, struct file *, int); 1841 int (*lock) (struct file *, int, struct file_lock *); 1842 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 1843 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 1844 int (*check_flags)(int); 1845 int (*flock) (struct file *, int, struct file_lock *); 1846 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); 1847 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); 1848 int (*setlease)(struct file *, long, struct file_lock **, void **); 1849 long (*fallocate)(struct file *file, int mode, loff_t offset, 1850 loff_t len); 1851 void (*show_fdinfo)(struct seq_file *m, struct file *f); 1852 #ifndef CONFIG_MMU 1853 unsigned (*mmap_capabilities)(struct file *); 1854 #endif 1855 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, 1856 loff_t, size_t, unsigned int); 1857 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in, 1858 struct file *file_out, loff_t pos_out, 1859 loff_t len, unsigned int remap_flags); 1860 int (*fadvise)(struct file *, loff_t, loff_t, int); 1861 } __randomize_layout; 1862 1863 struct inode_operations { 1864 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); 1865 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *); 1866 int (*permission) (struct inode *, int); 1867 struct posix_acl * (*get_acl)(struct inode *, int); 1868 1869 int (*readlink) (struct dentry *, char __user *,int); 1870 1871 int (*create) (struct inode *,struct dentry *, umode_t, bool); 1872 int (*link) (struct dentry *,struct inode *,struct dentry *); 1873 int (*unlink) (struct inode *,struct dentry *); 1874 int (*symlink) (struct inode *,struct dentry *,const char *); 1875 int (*mkdir) (struct inode *,struct dentry *,umode_t); 1876 int (*rmdir) (struct inode *,struct dentry *); 1877 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t); 1878 int (*rename) (struct inode *, struct dentry *, 1879 struct inode *, struct dentry *, unsigned int); 1880 int (*setattr) (struct dentry *, struct iattr *); 1881 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int); 1882 ssize_t (*listxattr) (struct dentry *, char *, size_t); 1883 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, 1884 u64 len); 1885 int (*update_time)(struct inode *, struct timespec64 *, int); 1886 int (*atomic_open)(struct inode *, struct dentry *, 1887 struct file *, unsigned open_flag, 1888 umode_t create_mode); 1889 int (*tmpfile) (struct inode *, struct dentry *, umode_t); 1890 int (*set_acl)(struct inode *, struct posix_acl *, int); 1891 } ____cacheline_aligned; 1892 1893 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio, 1894 struct iov_iter *iter) 1895 { 1896 return file->f_op->read_iter(kio, iter); 1897 } 1898 1899 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio, 1900 struct iov_iter *iter) 1901 { 1902 return file->f_op->write_iter(kio, iter); 1903 } 1904 1905 static inline int call_mmap(struct file *file, struct vm_area_struct *vma) 1906 { 1907 return file->f_op->mmap(file, vma); 1908 } 1909 1910 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, 1911 unsigned long nr_segs, unsigned long fast_segs, 1912 struct iovec *fast_pointer, 1913 struct iovec **ret_pointer); 1914 1915 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *); 1916 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); 1917 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); 1918 extern ssize_t vfs_readv(struct file *, const struct iovec __user *, 1919 unsigned long, loff_t *, rwf_t); 1920 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *, 1921 loff_t, size_t, unsigned int); 1922 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in, 1923 struct file *file_out, loff_t pos_out, 1924 size_t len, unsigned int flags); 1925 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, 1926 struct file *file_out, loff_t pos_out, 1927 loff_t *count, 1928 unsigned int remap_flags); 1929 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in, 1930 struct file *file_out, loff_t pos_out, 1931 loff_t len, unsigned int remap_flags); 1932 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in, 1933 struct file *file_out, loff_t pos_out, 1934 loff_t len, unsigned int remap_flags); 1935 extern int vfs_dedupe_file_range(struct file *file, 1936 struct file_dedupe_range *same); 1937 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos, 1938 struct file *dst_file, loff_t dst_pos, 1939 loff_t len, unsigned int remap_flags); 1940 1941 1942 struct super_operations { 1943 struct inode *(*alloc_inode)(struct super_block *sb); 1944 void (*destroy_inode)(struct inode *); 1945 void (*free_inode)(struct inode *); 1946 1947 void (*dirty_inode) (struct inode *, int flags); 1948 int (*write_inode) (struct inode *, struct writeback_control *wbc); 1949 int (*drop_inode) (struct inode *); 1950 void (*evict_inode) (struct inode *); 1951 void (*put_super) (struct super_block *); 1952 int (*sync_fs)(struct super_block *sb, int wait); 1953 int (*freeze_super) (struct super_block *); 1954 int (*freeze_fs) (struct super_block *); 1955 int (*thaw_super) (struct super_block *); 1956 int (*unfreeze_fs) (struct super_block *); 1957 int (*statfs) (struct dentry *, struct kstatfs *); 1958 int (*remount_fs) (struct super_block *, int *, char *); 1959 void (*umount_begin) (struct super_block *); 1960 1961 int (*show_options)(struct seq_file *, struct dentry *); 1962 int (*show_devname)(struct seq_file *, struct dentry *); 1963 int (*show_path)(struct seq_file *, struct dentry *); 1964 int (*show_stats)(struct seq_file *, struct dentry *); 1965 #ifdef CONFIG_QUOTA 1966 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 1967 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 1968 struct dquot **(*get_dquots)(struct inode *); 1969 #endif 1970 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t); 1971 long (*nr_cached_objects)(struct super_block *, 1972 struct shrink_control *); 1973 long (*free_cached_objects)(struct super_block *, 1974 struct shrink_control *); 1975 }; 1976 1977 /* 1978 * Inode flags - they have no relation to superblock flags now 1979 */ 1980 #define S_SYNC 1 /* Writes are synced at once */ 1981 #define S_NOATIME 2 /* Do not update access times */ 1982 #define S_APPEND 4 /* Append-only file */ 1983 #define S_IMMUTABLE 8 /* Immutable file */ 1984 #define S_DEAD 16 /* removed, but still open directory */ 1985 #define S_NOQUOTA 32 /* Inode is not counted to quota */ 1986 #define S_DIRSYNC 64 /* Directory modifications are synchronous */ 1987 #define S_NOCMTIME 128 /* Do not update file c/mtime */ 1988 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ 1989 #define S_PRIVATE 512 /* Inode is fs-internal */ 1990 #define S_IMA 1024 /* Inode has an associated IMA struct */ 1991 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */ 1992 #define S_NOSEC 4096 /* no suid or xattr security attributes */ 1993 #ifdef CONFIG_FS_DAX 1994 #define S_DAX 8192 /* Direct Access, avoiding the page cache */ 1995 #else 1996 #define S_DAX 0 /* Make all the DAX code disappear */ 1997 #endif 1998 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */ 1999 #define S_CASEFOLD 32768 /* Casefolded file */ 2000 #define S_VERITY 65536 /* Verity file (using fs/verity/) */ 2001 2002 /* 2003 * Note that nosuid etc flags are inode-specific: setting some file-system 2004 * flags just means all the inodes inherit those flags by default. It might be 2005 * possible to override it selectively if you really wanted to with some 2006 * ioctl() that is not currently implemented. 2007 * 2008 * Exception: SB_RDONLY is always applied to the entire file system. 2009 * 2010 * Unfortunately, it is possible to change a filesystems flags with it mounted 2011 * with files in use. This means that all of the inodes will not have their 2012 * i_flags updated. Hence, i_flags no longer inherit the superblock mount 2013 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org 2014 */ 2015 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) 2016 2017 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; } 2018 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb) 2019 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \ 2020 ((inode)->i_flags & S_SYNC)) 2021 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \ 2022 ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) 2023 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK) 2024 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME) 2025 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION) 2026 2027 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) 2028 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) 2029 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) 2030 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL) 2031 2032 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) 2033 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) 2034 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) 2035 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) 2036 #define IS_IMA(inode) ((inode)->i_flags & S_IMA) 2037 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) 2038 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) 2039 #define IS_DAX(inode) ((inode)->i_flags & S_DAX) 2040 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED) 2041 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD) 2042 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY) 2043 2044 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ 2045 (inode)->i_rdev == WHITEOUT_DEV) 2046 2047 static inline bool HAS_UNMAPPED_ID(struct inode *inode) 2048 { 2049 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid); 2050 } 2051 2052 static inline enum rw_hint file_write_hint(struct file *file) 2053 { 2054 if (file->f_write_hint != WRITE_LIFE_NOT_SET) 2055 return file->f_write_hint; 2056 2057 return file_inode(file)->i_write_hint; 2058 } 2059 2060 static inline int iocb_flags(struct file *file); 2061 2062 static inline u16 ki_hint_validate(enum rw_hint hint) 2063 { 2064 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1; 2065 2066 if (hint <= max_hint) 2067 return hint; 2068 return 0; 2069 } 2070 2071 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) 2072 { 2073 *kiocb = (struct kiocb) { 2074 .ki_filp = filp, 2075 .ki_flags = iocb_flags(filp), 2076 .ki_hint = ki_hint_validate(file_write_hint(filp)), 2077 .ki_ioprio = get_current_ioprio(), 2078 }; 2079 } 2080 2081 /* 2082 * Inode state bits. Protected by inode->i_lock 2083 * 2084 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC, 2085 * I_DIRTY_DATASYNC and I_DIRTY_PAGES. 2086 * 2087 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, 2088 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at 2089 * various stages of removing an inode. 2090 * 2091 * Two bits are used for locking and completion notification, I_NEW and I_SYNC. 2092 * 2093 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on 2094 * fdatasync(). i_atime is the usual cause. 2095 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of 2096 * these changes separately from I_DIRTY_SYNC so that we 2097 * don't have to write inode on fdatasync() when only 2098 * mtime has changed in it. 2099 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. 2100 * I_NEW Serves as both a mutex and completion notification. 2101 * New inodes set I_NEW. If two processes both create 2102 * the same inode, one of them will release its inode and 2103 * wait for I_NEW to be released before returning. 2104 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can 2105 * also cause waiting on I_NEW, without I_NEW actually 2106 * being set. find_inode() uses this to prevent returning 2107 * nearly-dead inodes. 2108 * I_WILL_FREE Must be set when calling write_inode_now() if i_count 2109 * is zero. I_FREEING must be set when I_WILL_FREE is 2110 * cleared. 2111 * I_FREEING Set when inode is about to be freed but still has dirty 2112 * pages or buffers attached or the inode itself is still 2113 * dirty. 2114 * I_CLEAR Added by clear_inode(). In this state the inode is 2115 * clean and can be destroyed. Inode keeps I_FREEING. 2116 * 2117 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are 2118 * prohibited for many purposes. iget() must wait for 2119 * the inode to be completely released, then create it 2120 * anew. Other functions will just ignore such inodes, 2121 * if appropriate. I_NEW is used for waiting. 2122 * 2123 * I_SYNC Writeback of inode is running. The bit is set during 2124 * data writeback, and cleared with a wakeup on the bit 2125 * address once it is done. The bit is also used to pin 2126 * the inode in memory for flusher thread. 2127 * 2128 * I_REFERENCED Marks the inode as recently references on the LRU list. 2129 * 2130 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). 2131 * 2132 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to 2133 * synchronize competing switching instances and to tell 2134 * wb stat updates to grab the i_pages lock. See 2135 * inode_switch_wbs_work_fn() for details. 2136 * 2137 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper 2138 * and work dirs among overlayfs mounts. 2139 * 2140 * I_CREATING New object's inode in the middle of setting up. 2141 * 2142 * Q: What is the difference between I_WILL_FREE and I_FREEING? 2143 */ 2144 #define I_DIRTY_SYNC (1 << 0) 2145 #define I_DIRTY_DATASYNC (1 << 1) 2146 #define I_DIRTY_PAGES (1 << 2) 2147 #define __I_NEW 3 2148 #define I_NEW (1 << __I_NEW) 2149 #define I_WILL_FREE (1 << 4) 2150 #define I_FREEING (1 << 5) 2151 #define I_CLEAR (1 << 6) 2152 #define __I_SYNC 7 2153 #define I_SYNC (1 << __I_SYNC) 2154 #define I_REFERENCED (1 << 8) 2155 #define __I_DIO_WAKEUP 9 2156 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) 2157 #define I_LINKABLE (1 << 10) 2158 #define I_DIRTY_TIME (1 << 11) 2159 #define __I_DIRTY_TIME_EXPIRED 12 2160 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED) 2161 #define I_WB_SWITCH (1 << 13) 2162 #define I_OVL_INUSE (1 << 14) 2163 #define I_CREATING (1 << 15) 2164 2165 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC) 2166 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES) 2167 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) 2168 2169 extern void __mark_inode_dirty(struct inode *, int); 2170 static inline void mark_inode_dirty(struct inode *inode) 2171 { 2172 __mark_inode_dirty(inode, I_DIRTY); 2173 } 2174 2175 static inline void mark_inode_dirty_sync(struct inode *inode) 2176 { 2177 __mark_inode_dirty(inode, I_DIRTY_SYNC); 2178 } 2179 2180 extern void inc_nlink(struct inode *inode); 2181 extern void drop_nlink(struct inode *inode); 2182 extern void clear_nlink(struct inode *inode); 2183 extern void set_nlink(struct inode *inode, unsigned int nlink); 2184 2185 static inline void inode_inc_link_count(struct inode *inode) 2186 { 2187 inc_nlink(inode); 2188 mark_inode_dirty(inode); 2189 } 2190 2191 static inline void inode_dec_link_count(struct inode *inode) 2192 { 2193 drop_nlink(inode); 2194 mark_inode_dirty(inode); 2195 } 2196 2197 enum file_time_flags { 2198 S_ATIME = 1, 2199 S_MTIME = 2, 2200 S_CTIME = 4, 2201 S_VERSION = 8, 2202 }; 2203 2204 extern bool atime_needs_update(const struct path *, struct inode *); 2205 extern void touch_atime(const struct path *); 2206 static inline void file_accessed(struct file *file) 2207 { 2208 if (!(file->f_flags & O_NOATIME)) 2209 touch_atime(&file->f_path); 2210 } 2211 2212 extern int file_modified(struct file *file); 2213 2214 int sync_inode(struct inode *inode, struct writeback_control *wbc); 2215 int sync_inode_metadata(struct inode *inode, int wait); 2216 2217 struct file_system_type { 2218 const char *name; 2219 int fs_flags; 2220 #define FS_REQUIRES_DEV 1 2221 #define FS_BINARY_MOUNTDATA 2 2222 #define FS_HAS_SUBTYPE 4 2223 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ 2224 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */ 2225 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ 2226 int (*init_fs_context)(struct fs_context *); 2227 const struct fs_parameter_description *parameters; 2228 struct dentry *(*mount) (struct file_system_type *, int, 2229 const char *, void *); 2230 void (*kill_sb) (struct super_block *); 2231 struct module *owner; 2232 struct file_system_type * next; 2233 struct hlist_head fs_supers; 2234 2235 struct lock_class_key s_lock_key; 2236 struct lock_class_key s_umount_key; 2237 struct lock_class_key s_vfs_rename_key; 2238 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; 2239 2240 struct lock_class_key i_lock_key; 2241 struct lock_class_key i_mutex_key; 2242 struct lock_class_key i_mutex_dir_key; 2243 }; 2244 2245 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) 2246 2247 #ifdef CONFIG_BLOCK 2248 extern struct dentry *mount_bdev(struct file_system_type *fs_type, 2249 int flags, const char *dev_name, void *data, 2250 int (*fill_super)(struct super_block *, void *, int)); 2251 #else 2252 static inline struct dentry *mount_bdev(struct file_system_type *fs_type, 2253 int flags, const char *dev_name, void *data, 2254 int (*fill_super)(struct super_block *, void *, int)) 2255 { 2256 return ERR_PTR(-ENODEV); 2257 } 2258 #endif 2259 extern struct dentry *mount_single(struct file_system_type *fs_type, 2260 int flags, void *data, 2261 int (*fill_super)(struct super_block *, void *, int)); 2262 extern struct dentry *mount_nodev(struct file_system_type *fs_type, 2263 int flags, void *data, 2264 int (*fill_super)(struct super_block *, void *, int)); 2265 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); 2266 void generic_shutdown_super(struct super_block *sb); 2267 #ifdef CONFIG_BLOCK 2268 void kill_block_super(struct super_block *sb); 2269 #else 2270 static inline void kill_block_super(struct super_block *sb) 2271 { 2272 BUG(); 2273 } 2274 #endif 2275 void kill_anon_super(struct super_block *sb); 2276 void kill_litter_super(struct super_block *sb); 2277 void deactivate_super(struct super_block *sb); 2278 void deactivate_locked_super(struct super_block *sb); 2279 int set_anon_super(struct super_block *s, void *data); 2280 int set_anon_super_fc(struct super_block *s, struct fs_context *fc); 2281 int get_anon_bdev(dev_t *); 2282 void free_anon_bdev(dev_t); 2283 struct super_block *sget_fc(struct fs_context *fc, 2284 int (*test)(struct super_block *, struct fs_context *), 2285 int (*set)(struct super_block *, struct fs_context *)); 2286 struct super_block *sget(struct file_system_type *type, 2287 int (*test)(struct super_block *,void *), 2288 int (*set)(struct super_block *,void *), 2289 int flags, void *data); 2290 2291 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ 2292 #define fops_get(fops) \ 2293 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) 2294 #define fops_put(fops) \ 2295 do { if (fops) module_put((fops)->owner); } while(0) 2296 /* 2297 * This one is to be used *ONLY* from ->open() instances. 2298 * fops must be non-NULL, pinned down *and* module dependencies 2299 * should be sufficient to pin the caller down as well. 2300 */ 2301 #define replace_fops(f, fops) \ 2302 do { \ 2303 struct file *__file = (f); \ 2304 fops_put(__file->f_op); \ 2305 BUG_ON(!(__file->f_op = (fops))); \ 2306 } while(0) 2307 2308 extern int register_filesystem(struct file_system_type *); 2309 extern int unregister_filesystem(struct file_system_type *); 2310 extern struct vfsmount *kern_mount(struct file_system_type *); 2311 extern void kern_unmount(struct vfsmount *mnt); 2312 extern int may_umount_tree(struct vfsmount *); 2313 extern int may_umount(struct vfsmount *); 2314 extern long do_mount(const char *, const char __user *, 2315 const char *, unsigned long, void *); 2316 extern struct vfsmount *collect_mounts(const struct path *); 2317 extern void drop_collected_mounts(struct vfsmount *); 2318 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *, 2319 struct vfsmount *); 2320 extern int vfs_statfs(const struct path *, struct kstatfs *); 2321 extern int user_statfs(const char __user *, struct kstatfs *); 2322 extern int fd_statfs(int, struct kstatfs *); 2323 extern int freeze_super(struct super_block *super); 2324 extern int thaw_super(struct super_block *super); 2325 extern bool our_mnt(struct vfsmount *mnt); 2326 extern __printf(2, 3) 2327 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...); 2328 extern int super_setup_bdi(struct super_block *sb); 2329 2330 extern int current_umask(void); 2331 2332 extern void ihold(struct inode * inode); 2333 extern void iput(struct inode *); 2334 extern int generic_update_time(struct inode *, struct timespec64 *, int); 2335 2336 /* /sys/fs */ 2337 extern struct kobject *fs_kobj; 2338 2339 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK) 2340 2341 #ifdef CONFIG_MANDATORY_FILE_LOCKING 2342 extern int locks_mandatory_locked(struct file *); 2343 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char); 2344 2345 /* 2346 * Candidates for mandatory locking have the setgid bit set 2347 * but no group execute bit - an otherwise meaningless combination. 2348 */ 2349 2350 static inline int __mandatory_lock(struct inode *ino) 2351 { 2352 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; 2353 } 2354 2355 /* 2356 * ... and these candidates should be on SB_MANDLOCK mounted fs, 2357 * otherwise these will be advisory locks 2358 */ 2359 2360 static inline int mandatory_lock(struct inode *ino) 2361 { 2362 return IS_MANDLOCK(ino) && __mandatory_lock(ino); 2363 } 2364 2365 static inline int locks_verify_locked(struct file *file) 2366 { 2367 if (mandatory_lock(locks_inode(file))) 2368 return locks_mandatory_locked(file); 2369 return 0; 2370 } 2371 2372 static inline int locks_verify_truncate(struct inode *inode, 2373 struct file *f, 2374 loff_t size) 2375 { 2376 if (!inode->i_flctx || !mandatory_lock(inode)) 2377 return 0; 2378 2379 if (size < inode->i_size) { 2380 return locks_mandatory_area(inode, f, size, inode->i_size - 1, 2381 F_WRLCK); 2382 } else { 2383 return locks_mandatory_area(inode, f, inode->i_size, size - 1, 2384 F_WRLCK); 2385 } 2386 } 2387 2388 #else /* !CONFIG_MANDATORY_FILE_LOCKING */ 2389 2390 static inline int locks_mandatory_locked(struct file *file) 2391 { 2392 return 0; 2393 } 2394 2395 static inline int locks_mandatory_area(struct inode *inode, struct file *filp, 2396 loff_t start, loff_t end, unsigned char type) 2397 { 2398 return 0; 2399 } 2400 2401 static inline int __mandatory_lock(struct inode *inode) 2402 { 2403 return 0; 2404 } 2405 2406 static inline int mandatory_lock(struct inode *inode) 2407 { 2408 return 0; 2409 } 2410 2411 static inline int locks_verify_locked(struct file *file) 2412 { 2413 return 0; 2414 } 2415 2416 static inline int locks_verify_truncate(struct inode *inode, struct file *filp, 2417 size_t size) 2418 { 2419 return 0; 2420 } 2421 2422 #endif /* CONFIG_MANDATORY_FILE_LOCKING */ 2423 2424 2425 #ifdef CONFIG_FILE_LOCKING 2426 static inline int break_lease(struct inode *inode, unsigned int mode) 2427 { 2428 /* 2429 * Since this check is lockless, we must ensure that any refcounts 2430 * taken are done before checking i_flctx->flc_lease. Otherwise, we 2431 * could end up racing with tasks trying to set a new lease on this 2432 * file. 2433 */ 2434 smp_mb(); 2435 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2436 return __break_lease(inode, mode, FL_LEASE); 2437 return 0; 2438 } 2439 2440 static inline int break_deleg(struct inode *inode, unsigned int mode) 2441 { 2442 /* 2443 * Since this check is lockless, we must ensure that any refcounts 2444 * taken are done before checking i_flctx->flc_lease. Otherwise, we 2445 * could end up racing with tasks trying to set a new lease on this 2446 * file. 2447 */ 2448 smp_mb(); 2449 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2450 return __break_lease(inode, mode, FL_DELEG); 2451 return 0; 2452 } 2453 2454 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) 2455 { 2456 int ret; 2457 2458 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK); 2459 if (ret == -EWOULDBLOCK && delegated_inode) { 2460 *delegated_inode = inode; 2461 ihold(inode); 2462 } 2463 return ret; 2464 } 2465 2466 static inline int break_deleg_wait(struct inode **delegated_inode) 2467 { 2468 int ret; 2469 2470 ret = break_deleg(*delegated_inode, O_WRONLY); 2471 iput(*delegated_inode); 2472 *delegated_inode = NULL; 2473 return ret; 2474 } 2475 2476 static inline int break_layout(struct inode *inode, bool wait) 2477 { 2478 smp_mb(); 2479 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2480 return __break_lease(inode, 2481 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK, 2482 FL_LAYOUT); 2483 return 0; 2484 } 2485 2486 #else /* !CONFIG_FILE_LOCKING */ 2487 static inline int break_lease(struct inode *inode, unsigned int mode) 2488 { 2489 return 0; 2490 } 2491 2492 static inline int break_deleg(struct inode *inode, unsigned int mode) 2493 { 2494 return 0; 2495 } 2496 2497 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) 2498 { 2499 return 0; 2500 } 2501 2502 static inline int break_deleg_wait(struct inode **delegated_inode) 2503 { 2504 BUG(); 2505 return 0; 2506 } 2507 2508 static inline int break_layout(struct inode *inode, bool wait) 2509 { 2510 return 0; 2511 } 2512 2513 #endif /* CONFIG_FILE_LOCKING */ 2514 2515 /* fs/open.c */ 2516 struct audit_names; 2517 struct filename { 2518 const char *name; /* pointer to actual string */ 2519 const __user char *uptr; /* original userland pointer */ 2520 int refcnt; 2521 struct audit_names *aname; 2522 const char iname[]; 2523 }; 2524 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0); 2525 2526 extern long vfs_truncate(const struct path *, loff_t); 2527 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, 2528 struct file *filp); 2529 extern int vfs_fallocate(struct file *file, int mode, loff_t offset, 2530 loff_t len); 2531 extern long do_sys_open(int dfd, const char __user *filename, int flags, 2532 umode_t mode); 2533 extern struct file *file_open_name(struct filename *, int, umode_t); 2534 extern struct file *filp_open(const char *, int, umode_t); 2535 extern struct file *file_open_root(struct dentry *, struct vfsmount *, 2536 const char *, int, umode_t); 2537 extern struct file * dentry_open(const struct path *, int, const struct cred *); 2538 extern struct file * open_with_fake_path(const struct path *, int, 2539 struct inode*, const struct cred *); 2540 static inline struct file *file_clone_open(struct file *file) 2541 { 2542 return dentry_open(&file->f_path, file->f_flags, file->f_cred); 2543 } 2544 extern int filp_close(struct file *, fl_owner_t id); 2545 2546 extern struct filename *getname_flags(const char __user *, int, int *); 2547 extern struct filename *getname(const char __user *); 2548 extern struct filename *getname_kernel(const char *); 2549 extern void putname(struct filename *name); 2550 2551 extern int finish_open(struct file *file, struct dentry *dentry, 2552 int (*open)(struct inode *, struct file *)); 2553 extern int finish_no_open(struct file *file, struct dentry *dentry); 2554 2555 /* fs/ioctl.c */ 2556 2557 extern int ioctl_preallocate(struct file *filp, int mode, void __user *argp); 2558 2559 /* fs/dcache.c */ 2560 extern void __init vfs_caches_init_early(void); 2561 extern void __init vfs_caches_init(void); 2562 2563 extern struct kmem_cache *names_cachep; 2564 2565 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) 2566 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) 2567 2568 #ifdef CONFIG_BLOCK 2569 extern int register_blkdev(unsigned int, const char *); 2570 extern void unregister_blkdev(unsigned int, const char *); 2571 extern void bdev_unhash_inode(dev_t dev); 2572 extern struct block_device *bdget(dev_t); 2573 extern struct block_device *bdgrab(struct block_device *bdev); 2574 extern void bd_set_size(struct block_device *, loff_t size); 2575 extern void bd_forget(struct inode *inode); 2576 extern void bdput(struct block_device *); 2577 extern void invalidate_bdev(struct block_device *); 2578 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *); 2579 extern int sync_blockdev(struct block_device *bdev); 2580 extern void kill_bdev(struct block_device *); 2581 extern struct super_block *freeze_bdev(struct block_device *); 2582 extern void emergency_thaw_all(void); 2583 extern void emergency_thaw_bdev(struct super_block *sb); 2584 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb); 2585 extern int fsync_bdev(struct block_device *); 2586 2587 extern struct super_block *blockdev_superblock; 2588 2589 static inline bool sb_is_blkdev_sb(struct super_block *sb) 2590 { 2591 return sb == blockdev_superblock; 2592 } 2593 #else 2594 static inline void bd_forget(struct inode *inode) {} 2595 static inline int sync_blockdev(struct block_device *bdev) { return 0; } 2596 static inline void kill_bdev(struct block_device *bdev) {} 2597 static inline void invalidate_bdev(struct block_device *bdev) {} 2598 2599 static inline struct super_block *freeze_bdev(struct block_device *sb) 2600 { 2601 return NULL; 2602 } 2603 2604 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb) 2605 { 2606 return 0; 2607 } 2608 2609 static inline int emergency_thaw_bdev(struct super_block *sb) 2610 { 2611 return 0; 2612 } 2613 2614 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg) 2615 { 2616 } 2617 2618 static inline bool sb_is_blkdev_sb(struct super_block *sb) 2619 { 2620 return false; 2621 } 2622 #endif 2623 extern int sync_filesystem(struct super_block *); 2624 extern const struct file_operations def_blk_fops; 2625 extern const struct file_operations def_chr_fops; 2626 #ifdef CONFIG_BLOCK 2627 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); 2628 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); 2629 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); 2630 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder); 2631 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode, 2632 void *holder); 2633 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, 2634 void *holder); 2635 extern struct block_device *bd_start_claiming(struct block_device *bdev, 2636 void *holder); 2637 extern void bd_finish_claiming(struct block_device *bdev, 2638 struct block_device *whole, void *holder); 2639 extern void bd_abort_claiming(struct block_device *bdev, 2640 struct block_device *whole, void *holder); 2641 extern void blkdev_put(struct block_device *bdev, fmode_t mode); 2642 2643 #ifdef CONFIG_SYSFS 2644 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk); 2645 extern void bd_unlink_disk_holder(struct block_device *bdev, 2646 struct gendisk *disk); 2647 #else 2648 static inline int bd_link_disk_holder(struct block_device *bdev, 2649 struct gendisk *disk) 2650 { 2651 return 0; 2652 } 2653 static inline void bd_unlink_disk_holder(struct block_device *bdev, 2654 struct gendisk *disk) 2655 { 2656 } 2657 #endif 2658 #endif 2659 2660 /* fs/char_dev.c */ 2661 #define CHRDEV_MAJOR_MAX 512 2662 /* Marks the bottom of the first segment of free char majors */ 2663 #define CHRDEV_MAJOR_DYN_END 234 2664 /* Marks the top and bottom of the second segment of free char majors */ 2665 #define CHRDEV_MAJOR_DYN_EXT_START 511 2666 #define CHRDEV_MAJOR_DYN_EXT_END 384 2667 2668 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); 2669 extern int register_chrdev_region(dev_t, unsigned, const char *); 2670 extern int __register_chrdev(unsigned int major, unsigned int baseminor, 2671 unsigned int count, const char *name, 2672 const struct file_operations *fops); 2673 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, 2674 unsigned int count, const char *name); 2675 extern void unregister_chrdev_region(dev_t, unsigned); 2676 extern void chrdev_show(struct seq_file *,off_t); 2677 2678 static inline int register_chrdev(unsigned int major, const char *name, 2679 const struct file_operations *fops) 2680 { 2681 return __register_chrdev(major, 0, 256, name, fops); 2682 } 2683 2684 static inline void unregister_chrdev(unsigned int major, const char *name) 2685 { 2686 __unregister_chrdev(major, 0, 256, name); 2687 } 2688 2689 /* fs/block_dev.c */ 2690 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 2691 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ 2692 2693 #ifdef CONFIG_BLOCK 2694 #define BLKDEV_MAJOR_MAX 512 2695 extern const char *__bdevname(dev_t, char *buffer); 2696 extern const char *bdevname(struct block_device *bdev, char *buffer); 2697 extern struct block_device *lookup_bdev(const char *); 2698 extern void blkdev_show(struct seq_file *,off_t); 2699 2700 #else 2701 #define BLKDEV_MAJOR_MAX 0 2702 #endif 2703 2704 extern void init_special_inode(struct inode *, umode_t, dev_t); 2705 2706 /* Invalid inode operations -- fs/bad_inode.c */ 2707 extern void make_bad_inode(struct inode *); 2708 extern bool is_bad_inode(struct inode *); 2709 2710 #ifdef CONFIG_BLOCK 2711 extern int revalidate_disk(struct gendisk *); 2712 extern int check_disk_change(struct block_device *); 2713 extern int __invalidate_device(struct block_device *, bool); 2714 extern int invalidate_partition(struct gendisk *, int); 2715 #endif 2716 unsigned long invalidate_mapping_pages(struct address_space *mapping, 2717 pgoff_t start, pgoff_t end); 2718 2719 static inline void invalidate_remote_inode(struct inode *inode) 2720 { 2721 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 2722 S_ISLNK(inode->i_mode)) 2723 invalidate_mapping_pages(inode->i_mapping, 0, -1); 2724 } 2725 extern int invalidate_inode_pages2(struct address_space *mapping); 2726 extern int invalidate_inode_pages2_range(struct address_space *mapping, 2727 pgoff_t start, pgoff_t end); 2728 extern int write_inode_now(struct inode *, int); 2729 extern int filemap_fdatawrite(struct address_space *); 2730 extern int filemap_flush(struct address_space *); 2731 extern int filemap_fdatawait_keep_errors(struct address_space *mapping); 2732 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, 2733 loff_t lend); 2734 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping, 2735 loff_t start_byte, loff_t end_byte); 2736 2737 static inline int filemap_fdatawait(struct address_space *mapping) 2738 { 2739 return filemap_fdatawait_range(mapping, 0, LLONG_MAX); 2740 } 2741 2742 extern bool filemap_range_has_page(struct address_space *, loff_t lstart, 2743 loff_t lend); 2744 extern int filemap_write_and_wait(struct address_space *mapping); 2745 extern int filemap_write_and_wait_range(struct address_space *mapping, 2746 loff_t lstart, loff_t lend); 2747 extern int __filemap_fdatawrite_range(struct address_space *mapping, 2748 loff_t start, loff_t end, int sync_mode); 2749 extern int filemap_fdatawrite_range(struct address_space *mapping, 2750 loff_t start, loff_t end); 2751 extern int filemap_check_errors(struct address_space *mapping); 2752 extern void __filemap_set_wb_err(struct address_space *mapping, int err); 2753 2754 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart, 2755 loff_t lend); 2756 extern int __must_check file_check_and_advance_wb_err(struct file *file); 2757 extern int __must_check file_write_and_wait_range(struct file *file, 2758 loff_t start, loff_t end); 2759 2760 static inline int file_write_and_wait(struct file *file) 2761 { 2762 return file_write_and_wait_range(file, 0, LLONG_MAX); 2763 } 2764 2765 /** 2766 * filemap_set_wb_err - set a writeback error on an address_space 2767 * @mapping: mapping in which to set writeback error 2768 * @err: error to be set in mapping 2769 * 2770 * When writeback fails in some way, we must record that error so that 2771 * userspace can be informed when fsync and the like are called. We endeavor 2772 * to report errors on any file that was open at the time of the error. Some 2773 * internal callers also need to know when writeback errors have occurred. 2774 * 2775 * When a writeback error occurs, most filesystems will want to call 2776 * filemap_set_wb_err to record the error in the mapping so that it will be 2777 * automatically reported whenever fsync is called on the file. 2778 */ 2779 static inline void filemap_set_wb_err(struct address_space *mapping, int err) 2780 { 2781 /* Fastpath for common case of no error */ 2782 if (unlikely(err)) 2783 __filemap_set_wb_err(mapping, err); 2784 } 2785 2786 /** 2787 * filemap_check_wb_error - has an error occurred since the mark was sampled? 2788 * @mapping: mapping to check for writeback errors 2789 * @since: previously-sampled errseq_t 2790 * 2791 * Grab the errseq_t value from the mapping, and see if it has changed "since" 2792 * the given value was sampled. 2793 * 2794 * If it has then report the latest error set, otherwise return 0. 2795 */ 2796 static inline int filemap_check_wb_err(struct address_space *mapping, 2797 errseq_t since) 2798 { 2799 return errseq_check(&mapping->wb_err, since); 2800 } 2801 2802 /** 2803 * filemap_sample_wb_err - sample the current errseq_t to test for later errors 2804 * @mapping: mapping to be sampled 2805 * 2806 * Writeback errors are always reported relative to a particular sample point 2807 * in the past. This function provides those sample points. 2808 */ 2809 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping) 2810 { 2811 return errseq_sample(&mapping->wb_err); 2812 } 2813 2814 static inline int filemap_nr_thps(struct address_space *mapping) 2815 { 2816 #ifdef CONFIG_READ_ONLY_THP_FOR_FS 2817 return atomic_read(&mapping->nr_thps); 2818 #else 2819 return 0; 2820 #endif 2821 } 2822 2823 static inline void filemap_nr_thps_inc(struct address_space *mapping) 2824 { 2825 #ifdef CONFIG_READ_ONLY_THP_FOR_FS 2826 atomic_inc(&mapping->nr_thps); 2827 #else 2828 WARN_ON_ONCE(1); 2829 #endif 2830 } 2831 2832 static inline void filemap_nr_thps_dec(struct address_space *mapping) 2833 { 2834 #ifdef CONFIG_READ_ONLY_THP_FOR_FS 2835 atomic_dec(&mapping->nr_thps); 2836 #else 2837 WARN_ON_ONCE(1); 2838 #endif 2839 } 2840 2841 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, 2842 int datasync); 2843 extern int vfs_fsync(struct file *file, int datasync); 2844 2845 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes, 2846 unsigned int flags); 2847 2848 /* 2849 * Sync the bytes written if this was a synchronous write. Expect ki_pos 2850 * to already be updated for the write, and will return either the amount 2851 * of bytes passed in, or an error if syncing the file failed. 2852 */ 2853 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count) 2854 { 2855 if (iocb->ki_flags & IOCB_DSYNC) { 2856 int ret = vfs_fsync_range(iocb->ki_filp, 2857 iocb->ki_pos - count, iocb->ki_pos - 1, 2858 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1); 2859 if (ret) 2860 return ret; 2861 } 2862 2863 return count; 2864 } 2865 2866 extern void emergency_sync(void); 2867 extern void emergency_remount(void); 2868 #ifdef CONFIG_BLOCK 2869 extern sector_t bmap(struct inode *, sector_t); 2870 #endif 2871 extern int notify_change(struct dentry *, struct iattr *, struct inode **); 2872 extern int inode_permission(struct inode *, int); 2873 extern int generic_permission(struct inode *, int); 2874 extern int __check_sticky(struct inode *dir, struct inode *inode); 2875 2876 static inline bool execute_ok(struct inode *inode) 2877 { 2878 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); 2879 } 2880 2881 static inline void file_start_write(struct file *file) 2882 { 2883 if (!S_ISREG(file_inode(file)->i_mode)) 2884 return; 2885 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true); 2886 } 2887 2888 static inline bool file_start_write_trylock(struct file *file) 2889 { 2890 if (!S_ISREG(file_inode(file)->i_mode)) 2891 return true; 2892 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false); 2893 } 2894 2895 static inline void file_end_write(struct file *file) 2896 { 2897 if (!S_ISREG(file_inode(file)->i_mode)) 2898 return; 2899 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE); 2900 } 2901 2902 /* 2903 * get_write_access() gets write permission for a file. 2904 * put_write_access() releases this write permission. 2905 * This is used for regular files. 2906 * We cannot support write (and maybe mmap read-write shared) accesses and 2907 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode 2908 * can have the following values: 2909 * 0: no writers, no VM_DENYWRITE mappings 2910 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist 2911 * > 0: (i_writecount) users are writing to the file. 2912 * 2913 * Normally we operate on that counter with atomic_{inc,dec} and it's safe 2914 * except for the cases where we don't hold i_writecount yet. Then we need to 2915 * use {get,deny}_write_access() - these functions check the sign and refuse 2916 * to do the change if sign is wrong. 2917 */ 2918 static inline int get_write_access(struct inode *inode) 2919 { 2920 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; 2921 } 2922 static inline int deny_write_access(struct file *file) 2923 { 2924 struct inode *inode = file_inode(file); 2925 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; 2926 } 2927 static inline void put_write_access(struct inode * inode) 2928 { 2929 atomic_dec(&inode->i_writecount); 2930 } 2931 static inline void allow_write_access(struct file *file) 2932 { 2933 if (file) 2934 atomic_inc(&file_inode(file)->i_writecount); 2935 } 2936 static inline bool inode_is_open_for_write(const struct inode *inode) 2937 { 2938 return atomic_read(&inode->i_writecount) > 0; 2939 } 2940 2941 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) 2942 static inline void i_readcount_dec(struct inode *inode) 2943 { 2944 BUG_ON(!atomic_read(&inode->i_readcount)); 2945 atomic_dec(&inode->i_readcount); 2946 } 2947 static inline void i_readcount_inc(struct inode *inode) 2948 { 2949 atomic_inc(&inode->i_readcount); 2950 } 2951 #else 2952 static inline void i_readcount_dec(struct inode *inode) 2953 { 2954 return; 2955 } 2956 static inline void i_readcount_inc(struct inode *inode) 2957 { 2958 return; 2959 } 2960 #endif 2961 extern int do_pipe_flags(int *, int); 2962 2963 #define __kernel_read_file_id(id) \ 2964 id(UNKNOWN, unknown) \ 2965 id(FIRMWARE, firmware) \ 2966 id(FIRMWARE_PREALLOC_BUFFER, firmware) \ 2967 id(MODULE, kernel-module) \ 2968 id(KEXEC_IMAGE, kexec-image) \ 2969 id(KEXEC_INITRAMFS, kexec-initramfs) \ 2970 id(POLICY, security-policy) \ 2971 id(X509_CERTIFICATE, x509-certificate) \ 2972 id(MAX_ID, ) 2973 2974 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM, 2975 #define __fid_stringify(dummy, str) #str, 2976 2977 enum kernel_read_file_id { 2978 __kernel_read_file_id(__fid_enumify) 2979 }; 2980 2981 static const char * const kernel_read_file_str[] = { 2982 __kernel_read_file_id(__fid_stringify) 2983 }; 2984 2985 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id) 2986 { 2987 if ((unsigned)id >= READING_MAX_ID) 2988 return kernel_read_file_str[READING_UNKNOWN]; 2989 2990 return kernel_read_file_str[id]; 2991 } 2992 2993 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t, 2994 enum kernel_read_file_id); 2995 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t, 2996 enum kernel_read_file_id); 2997 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t, 2998 enum kernel_read_file_id); 2999 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *); 3000 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *); 3001 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *); 3002 extern struct file * open_exec(const char *); 3003 3004 /* fs/dcache.c -- generic fs support functions */ 3005 extern bool is_subdir(struct dentry *, struct dentry *); 3006 extern bool path_is_under(const struct path *, const struct path *); 3007 3008 extern char *file_path(struct file *, char *, int); 3009 3010 #include <linux/err.h> 3011 3012 /* needed for stackable file system support */ 3013 extern loff_t default_llseek(struct file *file, loff_t offset, int whence); 3014 3015 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); 3016 3017 extern int inode_init_always(struct super_block *, struct inode *); 3018 extern void inode_init_once(struct inode *); 3019 extern void address_space_init_once(struct address_space *mapping); 3020 extern struct inode * igrab(struct inode *); 3021 extern ino_t iunique(struct super_block *, ino_t); 3022 extern int inode_needs_sync(struct inode *inode); 3023 extern int generic_delete_inode(struct inode *inode); 3024 static inline int generic_drop_inode(struct inode *inode) 3025 { 3026 return !inode->i_nlink || inode_unhashed(inode); 3027 } 3028 3029 extern struct inode *ilookup5_nowait(struct super_block *sb, 3030 unsigned long hashval, int (*test)(struct inode *, void *), 3031 void *data); 3032 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, 3033 int (*test)(struct inode *, void *), void *data); 3034 extern struct inode *ilookup(struct super_block *sb, unsigned long ino); 3035 3036 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval, 3037 int (*test)(struct inode *, void *), 3038 int (*set)(struct inode *, void *), 3039 void *data); 3040 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); 3041 extern struct inode * iget_locked(struct super_block *, unsigned long); 3042 extern struct inode *find_inode_nowait(struct super_block *, 3043 unsigned long, 3044 int (*match)(struct inode *, 3045 unsigned long, void *), 3046 void *data); 3047 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); 3048 extern int insert_inode_locked(struct inode *); 3049 #ifdef CONFIG_DEBUG_LOCK_ALLOC 3050 extern void lockdep_annotate_inode_mutex_key(struct inode *inode); 3051 #else 3052 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; 3053 #endif 3054 extern void unlock_new_inode(struct inode *); 3055 extern void discard_new_inode(struct inode *); 3056 extern unsigned int get_next_ino(void); 3057 extern void evict_inodes(struct super_block *sb); 3058 3059 extern void __iget(struct inode * inode); 3060 extern void iget_failed(struct inode *); 3061 extern void clear_inode(struct inode *); 3062 extern void __destroy_inode(struct inode *); 3063 extern struct inode *new_inode_pseudo(struct super_block *sb); 3064 extern struct inode *new_inode(struct super_block *sb); 3065 extern void free_inode_nonrcu(struct inode *inode); 3066 extern int should_remove_suid(struct dentry *); 3067 extern int file_remove_privs(struct file *); 3068 3069 extern void __insert_inode_hash(struct inode *, unsigned long hashval); 3070 static inline void insert_inode_hash(struct inode *inode) 3071 { 3072 __insert_inode_hash(inode, inode->i_ino); 3073 } 3074 3075 extern void __remove_inode_hash(struct inode *); 3076 static inline void remove_inode_hash(struct inode *inode) 3077 { 3078 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash)) 3079 __remove_inode_hash(inode); 3080 } 3081 3082 extern void inode_sb_list_add(struct inode *inode); 3083 3084 #ifdef CONFIG_BLOCK 3085 extern int bdev_read_only(struct block_device *); 3086 #endif 3087 extern int set_blocksize(struct block_device *, int); 3088 extern int sb_set_blocksize(struct super_block *, int); 3089 extern int sb_min_blocksize(struct super_block *, int); 3090 3091 extern int generic_file_mmap(struct file *, struct vm_area_struct *); 3092 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); 3093 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); 3094 extern int generic_remap_checks(struct file *file_in, loff_t pos_in, 3095 struct file *file_out, loff_t pos_out, 3096 loff_t *count, unsigned int remap_flags); 3097 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out); 3098 extern int generic_copy_file_checks(struct file *file_in, loff_t pos_in, 3099 struct file *file_out, loff_t pos_out, 3100 size_t *count, unsigned int flags); 3101 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); 3102 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); 3103 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); 3104 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *); 3105 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t); 3106 3107 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos, 3108 rwf_t flags); 3109 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos, 3110 rwf_t flags); 3111 3112 /* fs/block_dev.c */ 3113 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to); 3114 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from); 3115 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end, 3116 int datasync); 3117 extern void block_sync_page(struct page *page); 3118 3119 /* fs/splice.c */ 3120 extern ssize_t generic_file_splice_read(struct file *, loff_t *, 3121 struct pipe_inode_info *, size_t, unsigned int); 3122 extern ssize_t iter_file_splice_write(struct pipe_inode_info *, 3123 struct file *, loff_t *, size_t, unsigned int); 3124 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, 3125 struct file *out, loff_t *, size_t len, unsigned int flags); 3126 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, 3127 loff_t *opos, size_t len, unsigned int flags); 3128 3129 3130 extern void 3131 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); 3132 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); 3133 extern loff_t no_llseek(struct file *file, loff_t offset, int whence); 3134 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); 3135 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); 3136 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, 3137 int whence, loff_t maxsize, loff_t eof); 3138 extern loff_t fixed_size_llseek(struct file *file, loff_t offset, 3139 int whence, loff_t size); 3140 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t); 3141 extern loff_t no_seek_end_llseek(struct file *, loff_t, int); 3142 extern int generic_file_open(struct inode * inode, struct file * filp); 3143 extern int nonseekable_open(struct inode * inode, struct file * filp); 3144 extern int stream_open(struct inode * inode, struct file * filp); 3145 3146 #ifdef CONFIG_BLOCK 3147 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode, 3148 loff_t file_offset); 3149 3150 enum { 3151 /* need locking between buffered and direct access */ 3152 DIO_LOCKING = 0x01, 3153 3154 /* filesystem does not support filling holes */ 3155 DIO_SKIP_HOLES = 0x02, 3156 }; 3157 3158 void dio_end_io(struct bio *bio); 3159 3160 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, 3161 struct block_device *bdev, struct iov_iter *iter, 3162 get_block_t get_block, 3163 dio_iodone_t end_io, dio_submit_t submit_io, 3164 int flags); 3165 3166 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, 3167 struct inode *inode, 3168 struct iov_iter *iter, 3169 get_block_t get_block) 3170 { 3171 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, 3172 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES); 3173 } 3174 #endif 3175 3176 void inode_dio_wait(struct inode *inode); 3177 3178 /* 3179 * inode_dio_begin - signal start of a direct I/O requests 3180 * @inode: inode the direct I/O happens on 3181 * 3182 * This is called once we've finished processing a direct I/O request, 3183 * and is used to wake up callers waiting for direct I/O to be quiesced. 3184 */ 3185 static inline void inode_dio_begin(struct inode *inode) 3186 { 3187 atomic_inc(&inode->i_dio_count); 3188 } 3189 3190 /* 3191 * inode_dio_end - signal finish of a direct I/O requests 3192 * @inode: inode the direct I/O happens on 3193 * 3194 * This is called once we've finished processing a direct I/O request, 3195 * and is used to wake up callers waiting for direct I/O to be quiesced. 3196 */ 3197 static inline void inode_dio_end(struct inode *inode) 3198 { 3199 if (atomic_dec_and_test(&inode->i_dio_count)) 3200 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); 3201 } 3202 3203 /* 3204 * Warn about a page cache invalidation failure diring a direct I/O write. 3205 */ 3206 void dio_warn_stale_pagecache(struct file *filp); 3207 3208 extern void inode_set_flags(struct inode *inode, unsigned int flags, 3209 unsigned int mask); 3210 3211 extern const struct file_operations generic_ro_fops; 3212 3213 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) 3214 3215 extern int readlink_copy(char __user *, int, const char *); 3216 extern int page_readlink(struct dentry *, char __user *, int); 3217 extern const char *page_get_link(struct dentry *, struct inode *, 3218 struct delayed_call *); 3219 extern void page_put_link(void *); 3220 extern int __page_symlink(struct inode *inode, const char *symname, int len, 3221 int nofs); 3222 extern int page_symlink(struct inode *inode, const char *symname, int len); 3223 extern const struct inode_operations page_symlink_inode_operations; 3224 extern void kfree_link(void *); 3225 extern void generic_fillattr(struct inode *, struct kstat *); 3226 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int); 3227 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int); 3228 void __inode_add_bytes(struct inode *inode, loff_t bytes); 3229 void inode_add_bytes(struct inode *inode, loff_t bytes); 3230 void __inode_sub_bytes(struct inode *inode, loff_t bytes); 3231 void inode_sub_bytes(struct inode *inode, loff_t bytes); 3232 static inline loff_t __inode_get_bytes(struct inode *inode) 3233 { 3234 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes; 3235 } 3236 loff_t inode_get_bytes(struct inode *inode); 3237 void inode_set_bytes(struct inode *inode, loff_t bytes); 3238 const char *simple_get_link(struct dentry *, struct inode *, 3239 struct delayed_call *); 3240 extern const struct inode_operations simple_symlink_inode_operations; 3241 3242 extern int iterate_dir(struct file *, struct dir_context *); 3243 3244 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32); 3245 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int); 3246 3247 static inline int vfs_stat(const char __user *filename, struct kstat *stat) 3248 { 3249 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT, 3250 stat, STATX_BASIC_STATS); 3251 } 3252 static inline int vfs_lstat(const char __user *name, struct kstat *stat) 3253 { 3254 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT, 3255 stat, STATX_BASIC_STATS); 3256 } 3257 static inline int vfs_fstatat(int dfd, const char __user *filename, 3258 struct kstat *stat, int flags) 3259 { 3260 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT, 3261 stat, STATX_BASIC_STATS); 3262 } 3263 static inline int vfs_fstat(int fd, struct kstat *stat) 3264 { 3265 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0); 3266 } 3267 3268 3269 extern const char *vfs_get_link(struct dentry *, struct delayed_call *); 3270 extern int vfs_readlink(struct dentry *, char __user *, int); 3271 3272 extern int __generic_block_fiemap(struct inode *inode, 3273 struct fiemap_extent_info *fieinfo, 3274 loff_t start, loff_t len, 3275 get_block_t *get_block); 3276 extern int generic_block_fiemap(struct inode *inode, 3277 struct fiemap_extent_info *fieinfo, u64 start, 3278 u64 len, get_block_t *get_block); 3279 3280 extern struct file_system_type *get_filesystem(struct file_system_type *fs); 3281 extern void put_filesystem(struct file_system_type *fs); 3282 extern struct file_system_type *get_fs_type(const char *name); 3283 extern struct super_block *get_super(struct block_device *); 3284 extern struct super_block *get_super_thawed(struct block_device *); 3285 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev); 3286 extern struct super_block *get_active_super(struct block_device *bdev); 3287 extern void drop_super(struct super_block *sb); 3288 extern void drop_super_exclusive(struct super_block *sb); 3289 extern void iterate_supers(void (*)(struct super_block *, void *), void *); 3290 extern void iterate_supers_type(struct file_system_type *, 3291 void (*)(struct super_block *, void *), void *); 3292 3293 extern int dcache_dir_open(struct inode *, struct file *); 3294 extern int dcache_dir_close(struct inode *, struct file *); 3295 extern loff_t dcache_dir_lseek(struct file *, loff_t, int); 3296 extern int dcache_readdir(struct file *, struct dir_context *); 3297 extern int simple_setattr(struct dentry *, struct iattr *); 3298 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int); 3299 extern int simple_statfs(struct dentry *, struct kstatfs *); 3300 extern int simple_open(struct inode *inode, struct file *file); 3301 extern int simple_link(struct dentry *, struct inode *, struct dentry *); 3302 extern int simple_unlink(struct inode *, struct dentry *); 3303 extern int simple_rmdir(struct inode *, struct dentry *); 3304 extern int simple_rename(struct inode *, struct dentry *, 3305 struct inode *, struct dentry *, unsigned int); 3306 extern int noop_fsync(struct file *, loff_t, loff_t, int); 3307 extern int noop_set_page_dirty(struct page *page); 3308 extern void noop_invalidatepage(struct page *page, unsigned int offset, 3309 unsigned int length); 3310 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter); 3311 extern int simple_empty(struct dentry *); 3312 extern int simple_readpage(struct file *file, struct page *page); 3313 extern int simple_write_begin(struct file *file, struct address_space *mapping, 3314 loff_t pos, unsigned len, unsigned flags, 3315 struct page **pagep, void **fsdata); 3316 extern int simple_write_end(struct file *file, struct address_space *mapping, 3317 loff_t pos, unsigned len, unsigned copied, 3318 struct page *page, void *fsdata); 3319 extern int always_delete_dentry(const struct dentry *); 3320 extern struct inode *alloc_anon_inode(struct super_block *); 3321 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **); 3322 extern const struct dentry_operations simple_dentry_operations; 3323 3324 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); 3325 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); 3326 extern const struct file_operations simple_dir_operations; 3327 extern const struct inode_operations simple_dir_inode_operations; 3328 extern void make_empty_dir_inode(struct inode *inode); 3329 extern bool is_empty_dir_inode(struct inode *inode); 3330 struct tree_descr { const char *name; const struct file_operations *ops; int mode; }; 3331 struct dentry *d_alloc_name(struct dentry *, const char *); 3332 extern int simple_fill_super(struct super_block *, unsigned long, 3333 const struct tree_descr *); 3334 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); 3335 extern void simple_release_fs(struct vfsmount **mount, int *count); 3336 3337 extern ssize_t simple_read_from_buffer(void __user *to, size_t count, 3338 loff_t *ppos, const void *from, size_t available); 3339 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, 3340 const void __user *from, size_t count); 3341 3342 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); 3343 extern int generic_file_fsync(struct file *, loff_t, loff_t, int); 3344 3345 extern int generic_check_addressable(unsigned, u64); 3346 3347 #ifdef CONFIG_MIGRATION 3348 extern int buffer_migrate_page(struct address_space *, 3349 struct page *, struct page *, 3350 enum migrate_mode); 3351 extern int buffer_migrate_page_norefs(struct address_space *, 3352 struct page *, struct page *, 3353 enum migrate_mode); 3354 #else 3355 #define buffer_migrate_page NULL 3356 #define buffer_migrate_page_norefs NULL 3357 #endif 3358 3359 extern int setattr_prepare(struct dentry *, struct iattr *); 3360 extern int inode_newsize_ok(const struct inode *, loff_t offset); 3361 extern void setattr_copy(struct inode *inode, const struct iattr *attr); 3362 3363 extern int file_update_time(struct file *file); 3364 3365 static inline bool io_is_direct(struct file *filp) 3366 { 3367 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host); 3368 } 3369 3370 static inline bool vma_is_dax(struct vm_area_struct *vma) 3371 { 3372 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host); 3373 } 3374 3375 static inline bool vma_is_fsdax(struct vm_area_struct *vma) 3376 { 3377 struct inode *inode; 3378 3379 if (!vma->vm_file) 3380 return false; 3381 if (!vma_is_dax(vma)) 3382 return false; 3383 inode = file_inode(vma->vm_file); 3384 if (S_ISCHR(inode->i_mode)) 3385 return false; /* device-dax */ 3386 return true; 3387 } 3388 3389 static inline int iocb_flags(struct file *file) 3390 { 3391 int res = 0; 3392 if (file->f_flags & O_APPEND) 3393 res |= IOCB_APPEND; 3394 if (io_is_direct(file)) 3395 res |= IOCB_DIRECT; 3396 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host)) 3397 res |= IOCB_DSYNC; 3398 if (file->f_flags & __O_SYNC) 3399 res |= IOCB_SYNC; 3400 return res; 3401 } 3402 3403 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags) 3404 { 3405 if (unlikely(flags & ~RWF_SUPPORTED)) 3406 return -EOPNOTSUPP; 3407 3408 if (flags & RWF_NOWAIT) { 3409 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT)) 3410 return -EOPNOTSUPP; 3411 ki->ki_flags |= IOCB_NOWAIT; 3412 } 3413 if (flags & RWF_HIPRI) 3414 ki->ki_flags |= IOCB_HIPRI; 3415 if (flags & RWF_DSYNC) 3416 ki->ki_flags |= IOCB_DSYNC; 3417 if (flags & RWF_SYNC) 3418 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC); 3419 if (flags & RWF_APPEND) 3420 ki->ki_flags |= IOCB_APPEND; 3421 return 0; 3422 } 3423 3424 static inline ino_t parent_ino(struct dentry *dentry) 3425 { 3426 ino_t res; 3427 3428 /* 3429 * Don't strictly need d_lock here? If the parent ino could change 3430 * then surely we'd have a deeper race in the caller? 3431 */ 3432 spin_lock(&dentry->d_lock); 3433 res = dentry->d_parent->d_inode->i_ino; 3434 spin_unlock(&dentry->d_lock); 3435 return res; 3436 } 3437 3438 /* Transaction based IO helpers */ 3439 3440 /* 3441 * An argresp is stored in an allocated page and holds the 3442 * size of the argument or response, along with its content 3443 */ 3444 struct simple_transaction_argresp { 3445 ssize_t size; 3446 char data[0]; 3447 }; 3448 3449 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) 3450 3451 char *simple_transaction_get(struct file *file, const char __user *buf, 3452 size_t size); 3453 ssize_t simple_transaction_read(struct file *file, char __user *buf, 3454 size_t size, loff_t *pos); 3455 int simple_transaction_release(struct inode *inode, struct file *file); 3456 3457 void simple_transaction_set(struct file *file, size_t n); 3458 3459 /* 3460 * simple attribute files 3461 * 3462 * These attributes behave similar to those in sysfs: 3463 * 3464 * Writing to an attribute immediately sets a value, an open file can be 3465 * written to multiple times. 3466 * 3467 * Reading from an attribute creates a buffer from the value that might get 3468 * read with multiple read calls. When the attribute has been read 3469 * completely, no further read calls are possible until the file is opened 3470 * again. 3471 * 3472 * All attributes contain a text representation of a numeric value 3473 * that are accessed with the get() and set() functions. 3474 */ 3475 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ 3476 static int __fops ## _open(struct inode *inode, struct file *file) \ 3477 { \ 3478 __simple_attr_check_format(__fmt, 0ull); \ 3479 return simple_attr_open(inode, file, __get, __set, __fmt); \ 3480 } \ 3481 static const struct file_operations __fops = { \ 3482 .owner = THIS_MODULE, \ 3483 .open = __fops ## _open, \ 3484 .release = simple_attr_release, \ 3485 .read = simple_attr_read, \ 3486 .write = simple_attr_write, \ 3487 .llseek = generic_file_llseek, \ 3488 } 3489 3490 static inline __printf(1, 2) 3491 void __simple_attr_check_format(const char *fmt, ...) 3492 { 3493 /* don't do anything, just let the compiler check the arguments; */ 3494 } 3495 3496 int simple_attr_open(struct inode *inode, struct file *file, 3497 int (*get)(void *, u64 *), int (*set)(void *, u64), 3498 const char *fmt); 3499 int simple_attr_release(struct inode *inode, struct file *file); 3500 ssize_t simple_attr_read(struct file *file, char __user *buf, 3501 size_t len, loff_t *ppos); 3502 ssize_t simple_attr_write(struct file *file, const char __user *buf, 3503 size_t len, loff_t *ppos); 3504 3505 struct ctl_table; 3506 int proc_nr_files(struct ctl_table *table, int write, 3507 void __user *buffer, size_t *lenp, loff_t *ppos); 3508 int proc_nr_dentry(struct ctl_table *table, int write, 3509 void __user *buffer, size_t *lenp, loff_t *ppos); 3510 int proc_nr_inodes(struct ctl_table *table, int write, 3511 void __user *buffer, size_t *lenp, loff_t *ppos); 3512 int __init get_filesystem_list(char *buf); 3513 3514 #define __FMODE_EXEC ((__force int) FMODE_EXEC) 3515 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) 3516 3517 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) 3518 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ 3519 (flag & __FMODE_NONOTIFY))) 3520 3521 static inline bool is_sxid(umode_t mode) 3522 { 3523 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP)); 3524 } 3525 3526 static inline int check_sticky(struct inode *dir, struct inode *inode) 3527 { 3528 if (!(dir->i_mode & S_ISVTX)) 3529 return 0; 3530 3531 return __check_sticky(dir, inode); 3532 } 3533 3534 static inline void inode_has_no_xattr(struct inode *inode) 3535 { 3536 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC)) 3537 inode->i_flags |= S_NOSEC; 3538 } 3539 3540 static inline bool is_root_inode(struct inode *inode) 3541 { 3542 return inode == inode->i_sb->s_root->d_inode; 3543 } 3544 3545 static inline bool dir_emit(struct dir_context *ctx, 3546 const char *name, int namelen, 3547 u64 ino, unsigned type) 3548 { 3549 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0; 3550 } 3551 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) 3552 { 3553 return ctx->actor(ctx, ".", 1, ctx->pos, 3554 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0; 3555 } 3556 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) 3557 { 3558 return ctx->actor(ctx, "..", 2, ctx->pos, 3559 parent_ino(file->f_path.dentry), DT_DIR) == 0; 3560 } 3561 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) 3562 { 3563 if (ctx->pos == 0) { 3564 if (!dir_emit_dot(file, ctx)) 3565 return false; 3566 ctx->pos = 1; 3567 } 3568 if (ctx->pos == 1) { 3569 if (!dir_emit_dotdot(file, ctx)) 3570 return false; 3571 ctx->pos = 2; 3572 } 3573 return true; 3574 } 3575 static inline bool dir_relax(struct inode *inode) 3576 { 3577 inode_unlock(inode); 3578 inode_lock(inode); 3579 return !IS_DEADDIR(inode); 3580 } 3581 3582 static inline bool dir_relax_shared(struct inode *inode) 3583 { 3584 inode_unlock_shared(inode); 3585 inode_lock_shared(inode); 3586 return !IS_DEADDIR(inode); 3587 } 3588 3589 extern bool path_noexec(const struct path *path); 3590 extern void inode_nohighmem(struct inode *inode); 3591 3592 /* mm/fadvise.c */ 3593 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len, 3594 int advice); 3595 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len, 3596 int advice); 3597 3598 #if defined(CONFIG_IO_URING) 3599 extern struct sock *io_uring_get_socket(struct file *file); 3600 #else 3601 static inline struct sock *io_uring_get_socket(struct file *file) 3602 { 3603 return NULL; 3604 } 3605 #endif 3606 3607 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags, 3608 unsigned int flags); 3609 3610 int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa, 3611 struct fsxattr *fa); 3612 3613 static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags) 3614 { 3615 memset(fa, 0, sizeof(*fa)); 3616 fa->fsx_xflags = xflags; 3617 } 3618 3619 /* 3620 * Flush file data before changing attributes. Caller must hold any locks 3621 * required to prevent further writes to this file until we're done setting 3622 * flags. 3623 */ 3624 static inline int inode_drain_writes(struct inode *inode) 3625 { 3626 inode_dio_wait(inode); 3627 return filemap_write_and_wait(inode->i_mapping); 3628 } 3629 3630 #endif /* _LINUX_FS_H */ 3631